diff --git a/.agents/notes/README.i18n.yaml b/.agents/notes/README.i18n.yaml
index 6c7c2c635b..7f5b16b9b8 100644
--- a/.agents/notes/README.i18n.yaml
+++ b/.agents/notes/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write .agents/notes/README.md
 README.md: 3cfbb5154713046846a3bfcb2ccea62c0e4cb6c0
-README.zh.md: ddecac79519219c4a76cf9ba19edea312eea9d0d
+README.zh.md: a3369a94c6761e27567b1408d98a81665443f2ef
diff --git a/.agents/notes/README.zh.md b/.agents/notes/README.zh.md
index ddecac7951..a3369a94c6 100644
--- a/.agents/notes/README.zh.md
+++ b/.agents/notes/README.zh.md
@@ -11,7 +11,7 @@
 - **生命周期**（顶层文件夹）是 Agent Note 的状态，Agent Note 随状态变化在文件夹之间移动：
   - **`proposed/`**：实施前评审的提案；尚未构建（或仅部分构建）。
   - **`implemented/`**：决策已交付。文件记录做了什么决定、否决了什么，并**与实际交付的内容保持同步**：当代码后续移动文件、重命名包（package）或更改键名/默认值时，Agent Note 在同一个变更中同步更新（仅限事实——路径、名称、结构——而非决策本身）。见 [implemented/AGENTS.md](implemented/AGENTS.md)。
-  - **`rejected/`**：提案经过讨论后被否决。仅当其决策依据仍能避免一种诱人且影响重大的错误时保留；否则删除完整的三个配对文件。
+  - **`rejected/`**：提案经过讨论后被否决。仅当其决策依据仍能避免一种诱人且影响重大的错误时保留；否则删除完整的英文、中文和伴随记录三文件组。
 - **类别**（嵌套文件夹）是决策的*种类*——见下方[分类](#classification)。
 
 文件名中的日期是该主题**首次提出**的时间（以 git 历史为准）。Agent Note 之间的交叉引用使用相对 Markdown 链接（`[topic](../../implemented/architecture/2026-…-….md)`），从不使用纯文字或编号，这样既可机械检查，也能在文件夹间移动时保持有效。
@@ -28,7 +28,7 @@
 |---|---|
 | `feature` | 面向用户或模型的新功能。 |
 | `bug-fix` | 修正缺陷或弥补事故复盘（postmortem）发现的缺口。 |
-| `simplification` | 在不增加功能的前提下移除代码、行为或对外表面积。 |
+| `simplification` | 在不增加功能的前提下移除代码、行为或对外范围。 |
 | `architecture` | 关于**交付源码**的结构性决策：包之间的关系、运行时词汇。 |
 | `process` | 代码**周边**的工具、策略或工作流——门禁、包管理器、vendor 化——不涉及运行时行为。 |
 | `testing` | 测试基础设施与策略。 |
@@ -39,13 +39,13 @@
 
 当一份 implemented Agent Note 记录的交付决策已经完整落地，且其决策依据不太可能再指导未来工作时，将其归档。如果其中的备选方案、归属边界、否定性保证、持久化语义或协议语义、安全规则，或者重新引入条件仍有价值，则继续作为活跃记录保留。绝不归档 proposed Agent Note：过时的提案应转为 rejected。仅当 rejected Agent Note 仍能避免一种可能发生的错误时保留；否则一并删除其英文、中文和伴随记录文件。请使用经过校准的 [`dsh-archive-agent-notes`](../skills/dsh-archive-agent-notes/SKILL.md) 工作流，不要根据字数、存续时间或目标配额来判断。
 
-归档路径编码为 `archived/{class}/yyyy-mm-dd-topic-title.md`；其中有意省略 `implemented`，因为只有 implemented Agent Note 可以进入归档。归档变更会移动完整的英文、中文和伴随记录三个文件，保留 `Status: implemented`，在两种语言的文件中紧接该状态行插入相同的 `Archived: YYYY-MM-DD` 行，重新记录伴随文件，并修复或删除入站链接。归档时只允许对内容做这些更改。
+归档路径编码为 `archived/{class}/yyyy-mm-dd-topic-title.md`；其中有意省略 `implemented`，因为只有 implemented Agent Note 可以进入归档。归档变更会移动完整的英文、中文和伴随记录三个文件，保留 `Status: implemented`，在两种语言的文件中紧接该状态行插入相同的 `Archived: YYYY-MM-DD` 行，重新记录伴随记录，并修复或删除入站链接。归档时只允许对内容做这些更改。
 
 封存后，每组归档文件都永久冻结。禁止编辑、翻译、重新格式化、更新、移动或删除，也不得将其视为当前行为的权威依据。文档门禁会跳过归档源文件，包括其中的出站链接；当活跃文档有意引用历史时，仍可链接到归档 Agent Note。[`verify-archived-agent-notes`](../../scripts/verify-archived-agent-notes.ts) 强制执行封闭的类别目录树、完整的三文件配对、归档元数据、伴随记录 hash，以及仅追加的冻结内容 manifest。[归档政策 Agent Note](implemented/process/2026-07-26-frozen-agent-note-archive.md) 记录了设计依据。
 
 ## 何时需要写一份
 
-每个非平凡变更都必须在同一 PR（Pull Request）中新增或更新至少一份 Agent Note。如果变更修改了行为、架构、跨文件或跨包契约、流程或工具、测试策略、磁盘、协议或配置格式，或者其他维护者可能合理重新审视的决策，就属于非平凡变更。对未来重大工作的提案从 `proposed/` 开始；已经做出的决策从 `implemented/` 开始。选择与决策匹配的类别文件夹（见[分类](#classification)）。
+每个非平凡变更都必须在同一 PR（Pull Request）中新增或更新至少一份 Agent Note。如果变更修改了行为、架构、跨文件或跨包契约、流程或工具、测试策略、磁盘存储格式、协议格式（wire format）或配置格式，或者其他维护者可能合理重新审视的决策，就属于非平凡变更。对未来重大工作的提案从 `proposed/` 开始；已经做出的决策从 `implemented/` 开始。选择与决策匹配的类别文件夹（见[分类](#classification)）。
 
 更新已经拥有该决策的 Agent Note 即可满足规则；不要创建重复记录。只有不涉及行为、契约、结构、流程或理由变化的纯机械性或局部编辑才可豁免。Agent Note 永远不会被编辑为一个*不同的决策*：用新 Agent Note 取代旧记录，并让两个记录保持互相链接，除非后续依据下方规则完全合并旧记录。编辑 `implemented/` Agent Note 以跟踪其现有决策的所在位置是必需的，而非禁止的；见 [implemented/AGENTS.md](implemented/AGENTS.md)。
 
@@ -112,7 +112,7 @@ Status: <status>
 
 ### 曾考虑的替代方案——必需
 
-每份 Agent Note 都必须包含 `## Alternatives considered` 章节：每个真实的替代方案及其落选原因，每个替代方案用一个加粗引导的段落，或对争议较大的替代方案用 `### Why not <X>?` 子节。记录决策时不记录它击败了什么，就是在邀请反复争论——正是这些 Agent Note 存在的意义所要防止的。
+每份 Agent Note 都必须包含 `## Alternatives considered` 章节：每个真实的替代方案及其落选原因，每个替代方案用一个加粗引导的段落，或对争议较大的替代方案用 `### Why not <X>?` 子节。记录决策时不记录它击败了什么，就是在邀请反复争论——这正是 Agent Note 旨在防止的问题。
 
 替代方案是记录下来的，不是凭空编造的。日期早于 2026-07-05 且替代方案无法从记录中重建的 Agent Note，在该章节位置放置以下精确注释，门禁仅对格式规范之前的文件接受此注释：
 
@@ -122,8 +122,8 @@ Status: <status>
 
 ### 在生命周期之间移动
 
-将文件在生命周期文件夹之间移动意味着在同一个变更中更新 `Status:` 行并满足目标文件夹的骨架要求——否则门禁会失败。具体而言，`proposed/` → `implemented/` 将 `## Proposal` 改写为现在时态的 `## Decision`，将 `## Acceptance criteria` 和 `## Risks` 折入 `## Consequences`（或折入一个现在时态的 `## Testing`/`## Verification` 章节，用于描述现在锁定该行为的内容），并用实际交付的内容替换计划——即 [implemented/AGENTS.md](implemented/AGENTS.md) 所要求的改写，使之机械化。`proposed/` → `rejected/` 仅在 `Status:` 行添加原因并冻结文件。
+将文件在生命周期文件夹之间移动意味着在同一个变更中更新 `Status:` 行并满足目标文件夹的骨架要求——否则门禁会失败。具体而言，`proposed/` → `implemented/` 将 `## Proposal` 改写为现在时态的 `## Decision`，将 `## Acceptance criteria` 和 `## Risks` 折入 `## Consequences`（或折入一个现在时态的 `## Testing`/`## Verification` 章节，用于描述现在锁定该行为的内容），并用实际交付的内容替换计划——也就是将 [implemented/AGENTS.md](implemented/AGENTS.md) 所要求的改写变成可机械检查的规则。`proposed/` → `rejected/` 仅在 `Status:` 行添加原因并冻结文件。
 
 ### 中文对侧文件
 
-`.zh.md` 对侧文件按 [i18n 契约](../../docs/i18n/README.md)逐章节镜像其英文兄弟文件的结构；机器检查的头部标记（`# Agent Note: ` 和 `Status:` 行）保持英文原样不翻译。格式门禁跳过 `.zh.md` 文件——配对门禁负责它们的一致性。
+`.zh.md` 对侧文件按 [i18n 契约](../../docs/i18n/README.md)逐章节镜像其英文对侧文件的结构；机器检查的头部标记（`# Agent Note: ` 和 `Status:` 行）保持英文原样不翻译。格式门禁跳过 `.zh.md` 文件——配对门禁负责它们的一致性。
diff --git a/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.i18n.yaml
index 055804b569..70618612a2 100644
--- a/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-19-gui-layering-and-rpc-protocol.md: 63db4786adcc007d09b7a58824a59f4d1e1e8be1
-2026-07-19-gui-layering-and-rpc-protocol.zh.md: b3037ceb8c172925581d2862ea675e53a7f8c54e
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.md
+2026-07-19-gui-layering-and-rpc-protocol.md: b9718da4725316c64686adef24827e2984d8723d
+2026-07-19-gui-layering-and-rpc-protocol.zh.md: 2add148054e8f97c65600cd719fb4f8e0283f52d
diff --git a/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.md b/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.md
index 63db4786ad..b9718da472 100644
--- a/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.md
+++ b/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.md
@@ -165,7 +165,7 @@ One example row (the table structure is the reading key):
 |---|---|---|---|
 | `session.list` | `{ cursor?: string }` (cursor is a reserved seat, unimplemented) | `{ items: SessionSummary[] }` | persisted sessions, updatedAt descending; v1 builds no index |
 
-The remaining methods (`session.create`/`session.history`/`session.prompt`/`session.cancel`/`host.describe`) are not re-copied here — signatures are the source of truth; see `api/sessions.ts`, `api/host.ts`, and `RpcMethodMap`.
+The remaining methods (`session.create`/`session.history`/`session.rename`/`session.prompt`/`session.cancel`/`host.describe`) are not re-copied here — signatures are the source of truth; see `api/sessions.ts`, `api/host.ts`, and `RpcMethodMap`.
 
 ### Frames (server→client, named unions)
 
@@ -187,7 +187,7 @@ The remaining frame types are not re-copied here; the full unions are `MuxFrame`
 - **Cold sessions resume implicitly**: when `history`/`prompt` hits an unattached session the impl auto-resumes, deduplicating concurrent triggers with an in-flight table; attachment status is not exposed to clients (`running` already covers it).
 - **Approvals/questions**: the requested frame mints a stable rpcId on acceptance; first answer wins, and the host's in-memory pending table (keyed by rpcId) is the only referee; after a mux reopen, still-pending requested frames replay after the subscribed frame (rpcId reused verbatim — refresh recovery). The audit events `approval/asked`/`decided` continue through the durable log — frames = the live control plane, events = the durable audit. **Status**: the contract and frame types are shipped; the host-side pending table/wire answerer is unimplemented (`respond` in `api-proxy.ts` is a stub, always `not-pending`); PendingCard v1 is display-only.
 - **No protocol version**: client and host release bound together; `host.describe` has no protocolVersion field; introduce one when an independently released client appears.
-- **Reserved-seam discipline**: the map holds only implemented methods; an unknown method fails loud at envelope parse (`bad-request`) — no not-implemented fallback code. The reservation list (implementing = copy the signature into the domain interface + add the map row + add the schema pair): `session.fork`, `prompt.mode` gaining `'inject'`, `task.list`, `host.listModels`, describe gaining `hostInstanceId`.
+- **Reserved-seam discipline**: the map holds only implemented methods; an unknown method fails loud at envelope parse (`bad-request`) — no not-implemented fallback code. The reservation list (implementing = copy the signature into the domain interface + add the map row + add the schema pair): `session.fork`, `prompt.mode` gaining `'inject'`, `task.list`, `host.listModels`, describe gaining `hostInstanceId`. (`session.rename` graduated from this list: it appends a user-source `session/title` event.)
 
 ## The client carrier: the AbstractApiClient class family (`fetch/client.ts`)
 
diff --git a/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.zh.md b/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.zh.md
index b3037ceb8c..2add148054 100644
--- a/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-19-gui-layering-and-rpc-protocol.zh.md
@@ -163,7 +163,7 @@ export type ResponseValue<K> =
 |---|---|---|---|
 | `session.list` | `{ cursor?: string }`（cursor 留座不实现） | `{ items: SessionSummary[] }` | 已持久化 session，updatedAt 倒序；v1 不建索引 |
 
-其余方法（`session.create`/`session.history`/`session.prompt`/`session.cancel`/`host.describe`）的参数与返回不在此复写——签名即事实源，见 `api/sessions.ts`、`api/host.ts` 与 `RpcMethodMap`。
+其余方法（`session.create`/`session.history`/`session.rename`/`session.prompt`/`session.cancel`/`host.describe`）的参数与返回不在此复写——签名即事实源，见 `api/sessions.ts`、`api/host.ts` 与 `RpcMethodMap`。
 
 ### 帧（server→client，具名 union）
 
@@ -185,7 +185,7 @@ export type ResponseValue<K> =
 - **冷 session 隐式 resume**：`history`/`prompt` 命中未 attach 的 session 时 impl 自动 resume，并发触发用在途表去重；attach 与否不对客暴露（`running` 已覆盖）。
 - **审批/问答**：requested 帧受理时 mint 稳定 rpcId；先到先赢，host 内存 pending 表（keyed by rpcId）是唯一裁判；mux 重开后在 subscribed 帧后重放仍 pending 的 requested 帧（rpcId 原样复用，刷新恢复）。审计事件 `approval/asked`/`decided` 照旧走 durable 日志——帧=live 控制面，事件=durable 审计。**现状**：契约与帧类型已 shipped，host 侧 pending 表/wire answerer 未实现（`api-proxy.ts` 的 `respond` 是 stub，恒回 `not-pending`）；PendingCard v1 只展示。
 - **不设协议版本**：client 与 host 绑定发布，`host.describe` 无 protocolVersion 字段；出现独立发布的 client 时再引入。
-- **预留接缝纪律**：map 只含已实现方法，未知 method 在信封 parse 即 fail loud（`bad-request`），不设 not-implemented 兜底码。预留清单（实现时把签名抄进域接口+map 加行+schema 加对即升格）：`session.fork`、`prompt.mode` 加 `'inject'`、`task.list`、`host.listModels`、describe 加 `hostInstanceId`。
+- **预留接缝纪律**：map 只含已实现方法，未知 method 在信封 parse 即 fail loud（`bad-request`），不设 not-implemented 兜底码。预留清单（实现时把签名抄进域接口+map 加行+schema 加对即升格）：`session.fork`、`prompt.mode` 加 `'inject'`、`task.list`、`host.listModels`、describe 加 `hostInstanceId`。（`session.rename` 已从本清单毕业：追加 user 来源的 `session/title` 事件。）
 
 ## 客户端载体：AbstractApiClient 类体系（`fetch/client.ts`）
 
diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml
index 121879629c..7abc19ba70 100644
--- a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md
-2026-07-25-web-input-machine-and-slash-pipeline.md: 8cf3be7b3b7579d0c37898a58fb0ab4990fd71bf
-2026-07-25-web-input-machine-and-slash-pipeline.zh.md: b1488893558d8b2bf9c104faca968435d46a9640
+2026-07-25-web-input-machine-and-slash-pipeline.md: f446f42c9e202afcb404c7a551a4f715228bb8e5
+2026-07-25-web-input-machine-and-slash-pipeline.zh.md: 8f5e449bb878811b70bc5bc29e4a09bbc1a33bfa
diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md
index 8cf3be7b3b..f446f42c9e 100644
--- a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md
+++ b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.md
@@ -62,8 +62,8 @@ Calls that stay un-evented (registry registration → explicit call → await):
 
 A trigger/menu/pick pipeline with zero knowledge of "commands":
 
-- The service holds only the source registry (`SlashSource{trigger: '/'|'@', name, candidates, onPick, matchSpace?, matchEnter?}`; (trigger,name) unique, registration order = group order = polling order) and `sessionOf(sctx)`. Implementing a match hook IS the declaration of participation in space/enter adjudication; the pipeline polls in registration order, the first non-undefined answer wins, and no claimant means the default sink. matchSpace is synchronous (space fires mid-keystroke; hot cache only); matchEnter is asynchronous (it may await the source's own warmup, and a warmup failure rejects).
-- The controller holds the single authoritative hit (span included; retained for Space after the menu closes), the per-session menu store, the candidate-fetch generation, keyboard arbitration (combobox mode: focus stays in the textarea, ↑↓/Enter/Escape are intercepted and all pass the IME composition guard, with the single exception Shift+Enter unconditionally going first), and pick orchestration (outcome → self-dispatched bail events); at each session scope's birth it runs `warm(projection)` once over the source roster — within that scope the projection holds only the stable sessionId, with no published/capability transitions; the scope disposer tears down the controller.
+- The service holds only the source registry (`SlashSource{trigger: '/'|'@', name, order?, candidates, onPick, matchSpace?, matchEnter?}`; (trigger,name) unique; the optional `order` sorts the roster — lower first, default 0, ties keep registration order — and that sorted roster is both group order and polling order) and `sessionOf(sctx)`. Implementing a match hook IS the declaration of participation in space/enter adjudication; the pipeline polls in roster order, the first non-undefined answer wins, and no claimant means the default sink. matchSpace is synchronous (space fires mid-keystroke; hot cache only); matchEnter is asynchronous (it may await the source's own warmup, and a warmup failure rejects).
+- The controller holds the single authoritative hit (span included; retained for Space after the menu closes), the per-session menu store, the candidate-fetch generation, keyboard arbitration (combobox mode: focus stays in the textarea, ↑↓/Enter/Escape are intercepted and all pass the IME composition guard, with the single exception Shift+Enter unconditionally going first), and pick orchestration (outcome → self-dispatched bail events); a `dismiss()` verb backs MenuView's injected `onDismiss` (a pointer down outside both the menu and the surrounding composer card closes the menu; MenuView also localizes group titles through the `slash.menu` locale namespace and clamps its height to the viewport space above the composer via ui-primitives' `useAnchoredMaxHeight`); at each session scope's birth it runs `warm(projection)` once over the source roster — within that scope the projection holds only the stable sessionId, with no published/capability transitions; the scope disposer tears down the controller.
 - Trigger-detection word boundaries (`user@host` and URL `/` never trigger) and the guard tiers (plain: `/` everywhere + `@` inline / claimed: `/` suppressed, `@` live / frozen: none) are the frozen pure core.
 
 ### hub / facade: the resident shell and the strict-session input body
diff --git a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md
index b148889355..8f5e449bb8 100644
--- a/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-25-web-input-machine-and-slash-pipeline.zh.md
@@ -62,8 +62,8 @@ occurrence 表与 chip 三投影：
 
 对"命令"零知识的触发/菜单/pick 管线：
 
-- service 只有 source 注册表（`SlashSource{trigger: '/'|'@', name, candidates, onPick, matchSpace?, matchEnter?}`；(trigger,name) 唯一、注册序 = 组序 = 轮询序）与 `sessionOf(sctx)`。实现 match 钩子即参与空格/回车裁决的声明；管线按注册序轮询，首个非 undefined 应答胜出，无人认领落 default sink。matchSpace 同步（空格在击键中触发，只许热缓存）；matchEnter 异步（可 await 源自身预热，预热失败即 reject）。
-- controller 持有唯一权威 hit（含 span；菜单关闭后为 Space 保留）、per-session menu store、候选 fetch generation、键盘仲裁（combobox 模式：焦点始终在 textarea，↑↓/Enter/Escape 拦截且全程过 IME composition 守卫，唯一例外 Shift+Enter 无条件先行）、pick 编排（outcome → 自派 bail 事件）；每个 session scope 出生时对 source roster 做一次 `warm(projection)`，projection 在该 scope 内只有稳定的 sessionId，无 published/能力跃迁；scope disposer 拆除 controller。
+- service 只有 source 注册表（`SlashSource{trigger: '/'|'@', name, order?, candidates, onPick, matchSpace?, matchEnter?}`；(trigger,name) 唯一；可选 `order` 对 roster 排序——越小越靠前、默认 0、同值保持注册序——排序后的 roster 同时是组序与轮询序）与 `sessionOf(sctx)`。实现 match 钩子即参与空格/回车裁决的声明；管线按 roster 序轮询，首个非 undefined 应答胜出，无人认领落 default sink。matchSpace 同步（空格在击键中触发，只许热缓存）；matchEnter 异步（可 await 源自身预热，预热失败即 reject）。
+- controller 持有唯一权威 hit（含 span；菜单关闭后为 Space 保留）、per-session menu store、候选 fetch generation、键盘仲裁（combobox 模式：焦点始终在 textarea，↑↓/Enter/Escape 拦截且全程过 IME composition 守卫，唯一例外 Shift+Enter 无条件先行）、pick 编排（outcome → 自派 bail 事件）；`dismiss()` 动词支撑 MenuView 注入的 `onDismiss`（指针落在菜单与所在 composer 卡片之外即关闭菜单；MenuView 还经 `slash.menu` locale 命名空间本地化组标题，并经 ui-primitives 的 `useAnchoredMaxHeight` 把高度收敛到 composer 上方的视口空间）；每个 session scope 出生时对 source roster 做一次 `warm(projection)`，projection 在该 scope 内只有稳定的 sessionId，无 published/能力跃迁；scope disposer 拆除 controller。
 - 触发检测词边界（`user@host`、URL `/` 永不触发）、守卫分档（plain：`/` 到处 + `@` 行内 / claimed：`/` 抑制、`@` 活 / frozen：全无）为冻结纯核。
 
 ### hub / facade：常驻外壳与严格 session 输入体
diff --git a/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.i18n.yaml
index 82047e9d87..5e423f9383 100644
--- a/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-26-subprocess-consumer-migration.md: b31f69f1251a7219e168ed7800ba16ff7d6b328c
-2026-07-26-subprocess-consumer-migration.zh.md: 47e7519a72f5482f255d15d87b483e9295f6cd2c
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.md
+2026-07-26-subprocess-consumer-migration.md: 477dffc2271db08b8986b34645067b247ad7ace7
+2026-07-26-subprocess-consumer-migration.zh.md: 5e1035872c6e2101e41d2801cccdfb5ef2c088fc
diff --git a/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.md b/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.md
index b31f69f125..477dffc227 100644
--- a/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.md
+++ b/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.md
@@ -6,7 +6,7 @@ English | [中文](2026-07-26-subprocess-consumer-migration.zh.md)
 
 ## Problem
 
-The [subprocess seam](2026-07-26-subprocess-seam.md) shipped shaped for exactly one consumer family: batch-collected stdout/stderr, batch stdin, a single escalating `kill()`. That was deliberate scope control, and its own note records "migrate the other spawn sites" as rejected-for-now. Review on the introducing PR reversed that deferral: the stacked follow-up should reshape the interface toward Node's API and move the remaining process-running places onto the service. The remaining spawners each carried a private copy of some slice of the same mechanics — lsp-local had its own detached-tree signalling (POSIX group + Windows taskkill + liveness polling), subagent-subprocess had the dispose ladder and its own scrub, mcp-client and pty-local and the SDK helper each had a third/fourth/fifth copy of the credential scrub — and none of it was swappable or centrally testable.
+The [subprocess seam](2026-07-26-subprocess-seam.md) shipped shaped for exactly one consumer family: batch-collected stdout/stderr, batch stdin, a single escalating `kill()`. That was deliberate scope control, and its own note records "migrate the other spawn sites" as rejected-for-now. Review on the introducing PR reversed that deferral: the stacked follow-up should reshape the interface toward Node's API and move the remaining process-running places onto the service. The remaining spawners each carried a private copy of some slice of the same mechanics — lsp-local had its own detached-tree signalling (POSIX group + Windows taskkill + liveness polling), subagent-subprocess had the dispose ladder and its own scrub, and mcp-client, pty-local, the SDK helper, and the TUI Git probe each carried another credential scrub — and none of it was swappable or centrally testable.
 
 ## Decision
 
@@ -15,7 +15,7 @@ The seam's vocabulary is now Node-shaped, and every spawner that can ride the se
 - **Per-stream stdio dispositions** on `SubprocessSpawnSpec`: `'pipe'` (the raw `Readable`/`Writable`, for consumer-owned protocol framing), `'inherit'` (diagnostics to the parent's stream), and collect mode `{ maxBytes, spill? }` — the original bounded tail-keep shape, with the spill file now optional so a diagnostic tail (a language server's stderr) buffers without touching disk. stdin is `'ignore'`, `'pipe'`, or `{ data }` (write-and-close batch).
 - **`SubprocessOutcome` carries exit facts only** (Node's close-event vocabulary); collected output stays readable through `handle.collected` after settlement (spill fds seal at the settle boundary), so batch and streaming callers share one access path and nothing is copied into the outcome.
 - **Tree-scoped termination behind one verb**: `terminate()` owns the SIGTERM→grace→SIGKILL escalation (serves the spec's abort signal too, and is a no-op once the tree is gone) — the handle exposes no single-signal `kill(signal?)`, so a consumer cannot skip the grace window; `waitForExit()` polls tree liveness (POSIX group probe; direct-child boundary on Windows); Windows tree termination (`taskkill /T`, injectable) moved in from lsp-local, so tree semantics are platform-correct for every consumer. (The stdin-EOF-first dispose ladder initially absorbed from `subagent-subprocess` later moved back out to its one consumer — see the [ladder-ownership Agent Note](2026-07-27-dispose-ladder-to-consumer.md).)
-- **One scrub definition**: `scrubbedParentEnv()`/`SENSITIVE_ENV_PATTERN` live on the seam. Spawners that cannot route the spawn itself through the service — pty-local (node-pty owns the fork) and mcp-client (the MCP SDK owns the transport spawn) — import the function, so environment policy is single-sourced even where process ownership is not; the SDK helper's `scrubEnvironment()` defaults through it as well.
+- **One scrub definition**: `scrubbedParentEnv()`/`SENSITIVE_ENV_PATTERN` live on the seam, and credential-shaped names contain `KEY`, `PASSWORD`, `SECRET`, or `TOKEN`, case-insensitively. Spawners that cannot route the spawn itself through the service — pty-local (node-pty owns the fork), mcp-client (the MCP SDK owns the transport spawn), and the TUI's synchronous Git branch probe — import the function, so environment policy is single-sourced even where process ownership is not. The SDK helper's `scrubEnvironment()` defaults through the function and applies the exported pattern when its caller supplies an explicit environment; the pattern remains public for this production consumer.
 
 Migrations landed with the reshape: **bash-local/bash-sandbox** (collect modes + batch stdin; the bash `kill()` maps to `terminate()` so `task_kill` keeps escalation semantics), **lsp-local** (piped protocol streams + a no-spill collected stderr tail; `LspConnection` takes the seam's spawn function; its private tree-op helpers deleted), **subagent-acp** (piped ndjson streams + inherited stderr; spawn failure surfaces through `done` rejection into the same startup race; disposal is the backend-owned `disposeAcpChild` ladder over the seam's verbs, with the plugin's configured graces). **`dsh-subagent-subprocess` is deleted** — the dispose ladder and scrub are the seam's; the unused isolated-config-dir helper died with it (no consumer existed).
 
@@ -23,16 +23,16 @@ Compositions mounting lsp-local or subagent-acp now load `dsh-subprocess-local`
 
 ## Alternatives considered
 
-**Keep the batch-only seam and let stream consumers stay bespoke.** The introducing note's position, rejected by review: it leaves three private copies of tree signalling and five of the scrub, and any future runner (containerized executor, remote process host) would have to pick which private copy to fork. The Node-shaped dispositions cover all three observed stream shapes without widening the outcome type or buffering piped streams.
+**Keep the batch-only seam and let stream consumers stay bespoke.** The introducing note's position, rejected by review: it leaves three private copies of tree signalling and six of the scrub, and any future runner (containerized executor, remote process host) would have to pick which private copy to fork. The Node-shaped dispositions cover all three observed stream shapes without widening the outcome type or buffering piped streams.
 
 **A single `stdio: 'pipe' | 'inherit' | 'collect'` mode for all three streams at once.** Rejected: real consumers mix modes per stream (lsp: pipe/pipe/collect; acp: pipe/pipe/inherit; bash: data/collect/collect). Per-stream dispositions are exactly Node's shape and avoid a second spawn call for the mixed cases.
 
 **Migrate pty-local and mcp-client spawns too.** Rejected on ownership grounds, not scope: node-pty's `fork()` allocates the terminal itself, and the MCP SDK's `StdioClientTransport` spawns internally — neither call site is ours to route. They adopt the shared scrub (the part that is policy), and their READMEs say why the spawn stays put.
 
-**Migrate the test-support launchers (acp-snapshot, loader-smoke) and the SDK package-manager runner.** Rejected: the support packages are deliberately dependency-light test infrastructure that must not depend on product seams, and the SDK wizard's `stdio: 'inherit'`-with-redirect semantics plus its out-of-composition lifecycle (no cordis context at all) make the service a poor fit; it shares the scrub instead.
+**Migrate the test-support launchers (acp-snapshot, loader-smoke), the SDK package-manager runner, and the TUI Git probe.** Rejected: the support packages are deliberately dependency-light test infrastructure that must not depend on product seams; the SDK wizard's `stdio: 'inherit'`-with-redirect semantics plus its out-of-composition lifecycle (no cordis context at all) make the service a poor fit; and the TUI probe is synchronous. The production callers share the scrub instead.
 
 ## Consequences
 
 Bought: one implementation of tree signalling, escalation, bounded collection, and the scrub, tested once in `dsh-subprocess-local`'s suites (including injected-platform Windows coverage that lsp-local's private copy never had); lsp-local and subagent-acp shed their process plumbing and their children now survive plugin reloads and die with composition teardown like bash's; a whole package (`dsh-subagent-subprocess`) is gone. The seam README's "one consumer family" limitation is retired.
 
-Cost: the seam is wider — three stdio modes and the terminate/waitForExit/dispose lifecycle surface instead of one mode and one verb — so a future backend implements more surface; the compositions for lsp-local/subagent-acp each carry the subprocess row now; and `SubprocessOutcome` no longer carries output, a breaking shape change inside the still-unreleased stack (the PR2 layer was updated in place rather than shimmed, per the pre-release stance). pty-local/mcp-client/SDK/test-support spawns remain outside the service by ownership, with the scrub as the shared floor.
+Cost: the seam is wider — three stdio modes and the terminate/waitForExit/dispose lifecycle surface instead of one mode and one verb — so a future backend implements more surface; the compositions for lsp-local/subagent-acp each carry the subprocess row now; and `SubprocessOutcome` no longer carries output, a breaking shape change inside the still-unreleased stack (the PR2 layer was updated in place rather than shimmed, per the pre-release stance). pty-local/mcp-client/SDK/TUI/test-support spawns remain outside the service by ownership or execution shape, with the scrub as the shared floor and its pattern intentionally exported for explicit-environment consumers.
diff --git a/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.zh.md b/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.zh.md
index 47e7519a72..5e1035872c 100644
--- a/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-26-subprocess-consumer-migration.zh.md
@@ -6,7 +6,7 @@ Status: implemented
 
 ## 问题
 
-[进程 seam](2026-07-26-subprocess-seam.md) 交付时恰好只为一个消费方家族塑形：批量收集的 stdout/stderr、批量 stdin、单一的升级式 `kill()`。那是有意的范围控制，其自身的 Agent Note 也把「迁移其余 spawn 调用点」记为暂缓否决项。引入该 seam 的 PR（Pull Request）上的评审推翻了这一暂缓决定：堆叠其上的后续变更应当把接口向 Node 的 API 方向重塑，并把其余运行进程之处迁到该服务上。其余各 spawn 调用点此前各自持有同一套机制中某个切片的私有副本——lsp-local 自带 detached 进程树信号发送（POSIX 进程组 + Windows taskkill + 存活轮询），subagent-subprocess 自带 dispose（资源释放）阶梯和自己的凭据清除，mcp-client、pty-local 与 SDK helper 则各自持有凭据清除的第三、第四、第五份副本——而这一切既不可替换，也无法集中测试。
+[进程 seam](2026-07-26-subprocess-seam.md) 交付时恰好只为一个消费方家族塑形：批量收集的 stdout/stderr、批量 stdin、单一的升级式 `kill()`。那是有意的范围控制，其自身的 Agent Note 也把「迁移其余 spawn 调用点」记为暂缓否决项。引入该 seam 的 PR（Pull Request）上的评审推翻了这一暂缓决定：堆叠其上的后续变更应当把接口向 Node 的 API 方向重塑，并把其余运行进程之处迁到该服务上。其余各 spawn 调用点此前各自持有同一套机制中某个切片的私有副本——lsp-local 自带 detached 进程树信号发送（POSIX 进程组 + Windows taskkill + 存活轮询），subagent-subprocess 自带 dispose（资源释放）阶梯和自己的凭据清除，mcp-client、pty-local、SDK helper 与 TUI Git 探测则各自持有另一份凭据清除——而这一切既不可替换，也无法集中测试。
 
 ## 决策
 
@@ -15,7 +15,7 @@ Status: implemented
 - **按流划分的 stdio 处置方式（disposition）**，位于 `SubprocessSpawnSpec` 上：`'pipe'`（原始的 `Readable`/`Writable`，供消费方自有的协议分帧使用）、`'inherit'`（诊断输出直通父进程的流），以及收集模式（collect）`{ maxBytes, spill? }`——即最初的有界尾部保留形状，只是 spill 文件改为可选，使诊断尾部（例如语言服务器的 stderr）无需落盘即可缓冲。stdin 则为 `'ignore'`、`'pipe'` 或 `{ data }`（写完即关闭的批量形式）。
 - **`SubprocessOutcome` 只承载退出事实**（Node close 事件的词汇）；收集到的输出在结算后仍可经 `handle.collected` 读取（spill 文件描述符在结算边界封存），因此批量与流式调用方共用一条访问路径，也没有任何内容被复制进这份结果。
 - **以进程树为范围的终止，集中在一个动词后面**：`terminate()` 拥有 SIGTERM→宽限期→SIGKILL 升级（也承接 spec 的 abort 信号，进程树消亡后为空操作）——句柄不暴露单信号的 `kill(signal?)`，因此消费方无法跳过宽限窗口；`waitForExit()` 轮询进程树存活状态（POSIX 进程组探测；Windows 上以直接子进程为界）。Windows 进程树终止（`taskkill /T`，可注入）自 lsp-local 迁入，因此每个消费方拿到的进程树语义在各平台上都正确。（最初从 `subagent-subprocess` 吸收的以 stdin EOF 打头的 dispose 阶梯，后来又移回其唯一消费方——见[阶梯归属 Agent Note](2026-07-27-dispose-ladder-to-consumer.md)。）
-- **凭据清除只有一份定义**：`scrubbedParentEnv()`/`SENSITIVE_ENV_PATTERN` 定义在 seam 上。无法把 spawn 本身路由到该服务的调用点——pty-local（node-pty 拥有 fork）与 mcp-client（MCP SDK 拥有传输层的 spawn）——改为导入该函数，因此即便进程所有权无法统一，环境策略仍是单一来源；SDK helper 的 `scrubEnvironment()` 默认同样委托给它。
+- **凭据清除只有一份定义**：`scrubbedParentEnv()`/`SENSITIVE_ENV_PATTERN` 定义在 seam 上，且凭据形状的名称不区分大小写地包含 `KEY`、`PASSWORD`、`SECRET` 或 `TOKEN`。无法把 spawn 本身路由到该服务的调用点——pty-local（node-pty 拥有 fork）、mcp-client（MCP SDK 拥有传输层的 spawn）与 TUI 的同步 Git 分支探测——改为导入该函数，因此即便进程所有权无法统一，环境策略仍是单一来源。SDK helper 的 `scrubEnvironment()` 默认委托给该函数，并在调用方显式传入环境时应用导出的正则；该正则因此生产消费方而保持公开。
 
 各项迁移随这次重塑一并落地：**bash-local/bash-sandbox**（收集模式 + 批量 stdin；bash 的 `kill()` 映射到 `terminate()`，因此 `task_kill` 保有升级语义），**lsp-local**（管道化的协议流 + 无 spill 的 stderr 收集尾部；`LspConnection` 改为接收 seam 的 spawn 函数；其私有的进程树操作辅助函数已删除），**subagent-acp**（管道化的 ndjson 流 + inherit 的 stderr；spawn 失败经 `done` 的 reject 汇入同一个启动竞态；dispose 是后端自有的 `disposeAcpChild` 阶梯，经由 seam 的动词运行，携带插件所配置的宽限期）。**`dsh-subagent-subprocess` 已删除**——dispose 阶梯与凭据清除归 seam 所有；无人使用的隔离配置目录辅助函数随之消亡（其消费方本就不存在）。
 
@@ -23,16 +23,16 @@ Status: implemented
 
 ## 曾考虑的替代方案
 
-**保持只支持批量的 seam，让流式消费方继续各自为政。**这正是引入该 seam 的 Agent Note 当初的立场，评审将其否决：这样会留下三份进程树信号发送的私有副本和五份凭据清除的私有副本，而未来任何运行器（容器化执行器、远程进程宿主）都得挑选去 fork 哪一份私有副本。Node 形状的处置方式覆盖已观察到的全部三种流形状，既不拓宽结果类型，也不缓冲管道化的流。
+**保持只支持批量的 seam，让流式消费方继续各自为政。**这正是引入该 seam 的 Agent Note 当初的立场，评审将其否决：这样会留下三份进程树信号发送的私有副本和六份凭据清除的私有副本，而未来任何运行器（容器化执行器、远程进程宿主）都得挑选去 fork 哪一份私有副本。Node 形状的处置方式覆盖已观察到的全部三种流形状，既不拓宽结果类型，也不缓冲管道化的流。
 
 **用单个 `stdio: 'pipe' | 'inherit' | 'collect'` 模式一次性统辖全部三条流。**否决：真实消费方按流混用模式（lsp：pipe/pipe/collect；acp：pipe/pipe/inherit；bash：data/collect/collect）。按流划分的处置方式恰好就是 Node 的形状，也免去了混用场景的第二个 spawn 调用。
 
 **把 pty-local 与 mcp-client 的 spawn 也一并迁移。**基于所有权而非范围否决：node-pty 的 `fork()` 自行分配终端，MCP SDK 的 `StdioClientTransport` 在内部完成 spawn——这两处调用点都不归我们路由。它们采纳共享的凭据清除（那正是属于策略的部分），并在各自的 README 中说明 spawn 为何留在原地。
 
-**迁移 test-support 启动器（acp-snapshot、loader-smoke）与 SDK package-manager 运行器。**否决：support 各包（package）是刻意保持轻依赖的测试基础设施，不得依赖产品 seam；而 SDK 向导那套附带重定向的 `stdio: 'inherit'` 语义，加上其完全脱离组合的生命周期（根本没有 cordis 上下文），使该服务并不合用；它改为共享凭据清除。
+**迁移 test-support 启动器（acp-snapshot、loader-smoke）、SDK package-manager 运行器与 TUI Git 探测。**否决：support 各包（package）是刻意保持轻依赖的测试基础设施，不得依赖产品 seam；SDK 向导那套附带重定向的 `stdio: 'inherit'` 语义，加上其完全脱离组合的生命周期（根本没有 cordis 上下文），使该服务并不合用；TUI 探测则是同步调用。这些生产调用点改为共享凭据清除。
 
 ## 后果
 
 换来的是：进程树信号发送、升级、有界收集与凭据清除各自只剩一份实现，且只在 `dsh-subprocess-local` 的测试套件中测试一次（其中包括 lsp-local 的私有副本从未有过的、以注入平台方式实现的 Windows 覆盖）；lsp-local 与 subagent-acp 卸下了自己的进程管道，其子进程如今像 bash 的一样，在插件重载后存活、随组合拆除而终止；一个完整的包（`dsh-subagent-subprocess`）就此消失。seam README 中「只有一个消费方家族」的限制说明也随之退役。
 
-代价是：这道 seam 变宽了（stdio 模式从一种变为三种、终止动词换成 terminate/waitForExit/dispose 这组生命周期表面），未来的后端因此要实现更宽的表面；lsp-local/subagent-acp 的各组合如今都多出 subprocess 这一行组合配置；`SubprocessOutcome` 也不再承载输出，这是仍未发布的堆叠变更内部的一次破坏性形状变更（依照预发布立场，PR2 那一层被就地更新，而非加 shim）。pty-local/mcp-client/SDK/test-support 的 spawn 因所有权归属留在该服务之外，以凭据清除作为共享底线。
+代价是：这道 seam 变宽了（stdio 模式从一种变为三种、终止动词换成 terminate/waitForExit/dispose 这组生命周期表面），未来的后端因此要实现更宽的表面；lsp-local/subagent-acp 的各组合如今都多出 subprocess 这一行组合配置；`SubprocessOutcome` 也不再承载输出，这是仍未发布的堆叠变更内部的一次破坏性形状变更（依照预发布立场，PR2 那一层被就地更新，而非加 shim）。pty-local/mcp-client/SDK/TUI/test-support 的 spawn 因所有权归属或执行形状留在该服务之外，以凭据清除作为共底线，且为显式环境的生产消费方有意保持该正则导出。
diff --git a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml
index a96c124e92..f3a3226c11 100644
--- a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md
-2026-07-28-dsh-native-typescript-source-launch.md: 019389f3e5e9229f4359bbd58c95dbb2f14eb24b
-2026-07-28-dsh-native-typescript-source-launch.zh.md: 2cfff25d228e67ac85a9bc9087fa09ddb64213a0
+2026-07-28-dsh-native-typescript-source-launch.md: 1ba1dd2663038ad7c49af71f8b428245f7fa3e2b
+2026-07-28-dsh-native-typescript-source-launch.zh.md: 02f84f34820469ad9e810ae17d79d3fe12b0cd4c
diff --git a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md
index 019389f3e5..1ba1dd2663 100644
--- a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md
+++ b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md
@@ -4,6 +4,8 @@ Status: implemented
 
 English | [中文](2026-07-28-dsh-native-typescript-source-launch.zh.md)
 
+> The Node-native launch vector is superseded by [dsh source launch through the tsx ESM hook](2026-07-29-dsh-source-launch-tsx-esm.md): Node 26.0.0 removed `--experimental-transform-types`, and the paths loader described here is deleted. The Cordis-config declaration gate (`verify-cordis-config`), the app-boot fail-loud plugin diagnostic, and the vendored `import type` marks remain current.
+
 ## Problem
 
 The `dsh` source entry point originally used `tsx` to run `apps/cli/src/bin.ts`, with the same third-party loader implicitly handling both TypeScript transformation and the root tsconfig's `paths` resolution. With Node handling TypeScript natively, it does not apply tsconfig path mappings; resolving through package exports would instead mix potentially stale or nonexistent `lib/` artifacts into the source launch.
diff --git a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md
index 2cfff25d22..02f84f3482 100644
--- a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md
@@ -4,6 +4,8 @@ Status: implemented
 
 [English](2026-07-28-dsh-native-typescript-source-launch.md) | 中文
 
+> Node 原生启动向量已被 [dsh 通过 tsx ESM hook 源码启动](2026-07-29-dsh-source-launch-tsx-esm.md) 取代：Node 26.0.0 移除了 `--experimental-transform-types`，本文描述的 paths loader 已删除。Cordis 配置声明门禁（`verify-cordis-config`）、app-boot 的 fail-loud 插件诊断以及 vendor 中的 `import type` 标注仍然有效。
+
 ## 问题
 
 `dsh` 源码入口原本使用 `tsx` 运行 `apps/cli/src/bin.ts`，TypeScript 转换和根 tsconfig 的 `paths` 解析都由同一个第三方 loader 隐式处理。改由 Node 原生处理 TypeScript 后，Node 不会应用 tsconfig 路径映射；如果改为通过包导出解析，源码启动会混入可能陈旧或不存在的 `lib/` 产物。
diff --git a/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.i18n.yaml
new file mode 100644
index 0000000000..ab618103c2
--- /dev/null
+++ b/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.md
+2026-07-29-dsh-source-launch-tsx-esm.md: 93fbb248b45efde37d5fbdb1ec4b812ab3332088
+2026-07-29-dsh-source-launch-tsx-esm.zh.md: 48f410bd846e5808cc95180279348a0ac5ba1c95
diff --git a/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.md b/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.md
new file mode 100644
index 0000000000..93fbb248b4
--- /dev/null
+++ b/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.md
@@ -0,0 +1,38 @@
+# Agent Note: dsh source launch through the tsx ESM hook
+
+Status: implemented
+
+English | [中文](2026-07-29-dsh-source-launch-tsx-esm.zh.md)
+
+> Supersedes [native TypeScript source launch](2026-07-28-dsh-native-typescript-source-launch.md): Node removed the capability that decision was built on.
+
+## Problem
+
+The [native source-launch decision](2026-07-28-dsh-native-typescript-source-launch.md) ran `apps/cli/src/bin.ts` under `node --experimental-transform-types` with a resolve-only paths loader, so Node owned TypeScript transformation. Node 26.0.0 removed `--experimental-transform-types` (the process rejects the flag with `bad option`), keeping only strip mode, and strip mode rejects syntax this source graph requires: vendored Cordis parameter properties (`constructor(private ctx: Context)`), the `@Inject` decorators in `vendor/hmr`, and runtime enums/namespaces throughout `vendor/` and `packages/workflow`. The repository's engines range (`^22.19.0 || >=24.0.0`) includes Node 26, so the native launch chain could not start at all there — and no CI job executed the real launch vector, so the incompatibility shipped silently.
+
+Startup latency also mattered: the off-thread `module.register()` hooks worker serialized every resolution across threads (~440ms of `makeSyncRequest` wait during TUI boot), and the full tsx default (`--import tsx`) pays ~0.4s in its CJS hook's resolution amplification.
+
+## Decision
+
+The `dsh` TUI, Web, and headless source launches run `node --import tsx/esm`: tsx's ESM-only hook owns both TypeScript transformation and tsconfig `paths` projection. `bin/dsh`, the root `dsh`/`demo:tui`/`demo:web` scripts, and the Code Mode TUI overlay use the same vector; `bin/dsh` references the hook and tsconfig by absolute checkout paths (bare `tsx/esm` does not resolve from an arbitrary cwd) and pins `TSX_TSCONFIG_PATH` to the root tsconfig. The CJS hook stays off because the CLI source graph is ESM-only; measured TUI time-to-banner is ~0.7s versus ~1.1s under the full tsx default and ~0.75s under the removed native chain.
+
+`scripts/tspath-loader.ts` and `apps/cli/src/tsconfig-paths-loader.ts` are deleted. With them went the loader's runtime rule of mapping a workspace import only for declared runtime dependencies — tsx applies the `paths` map unconditionally. Declaration completeness now rests on the static gates alone: `verify-cordis-config` for configured bare plugins, and workspace constraints for manifests. (That runtime rule found real bugs: `dsh-plan-mode` and `dsh-tool-tasks` imported `@deepseek-ai/dsh-llm` while declaring it only in devDependencies; fixed alongside this change.)
+
+The node-compat CI matrix (Node 22.19 and 26) gains `dsh-source-launch-smoke` (`apps/cli/tests/source-launch.compat.spec.ts`): a keyless piped-stdio launch of the exact production vector asserting the non-zero-exit TTY refusal. Any future Node change to module hooks or TypeScript handling turns this gate red instead of breaking developers' `pnpm dsh`.
+
+## Alternatives considered
+
+**Keep the native chain on Node ≤25 and branch by version.** Rejected: two transformation semantics (amaro versus esbuild) diverge on edge syntax, the launcher grows version probing, and the node-compat matrix must cover both paths — heavy maintenance for an experimental flag that already changed under us. amaro also rejects the `@Inject` decorators `vendor/hmr` uses, so the native path could not boot the shipped default TUI config anyway.
+
+**Make the source graph erasable-only so Node 26 strip mode accepts it.** Rejected: parameter properties and value namespaces pervade vendored Cordis/cosmokit/loader/schemastery; rewriting them is unbounded churn re-applied on every vendor sync.
+
+**A repo-owned in-thread loader (`module.registerHooks()` + esbuild or `@swc/core` transform).** Rejected for now: prototypes measured ~0.45s (esbuild path untested end-to-end; SWC breaks on `vendor/hmr`'s decorator + namespace merge in both decorator modes), but it means owning transform correctness and a resolve hook that tsx already provides. Revisit only if the ~0.3s gap becomes a real cost; the profiling evidence lives in the PR discussion.
+
+**Run built `lib/` for Node 26 and keep native for 24.** Rejected: loses the zero-build development loop on the newest Node line and mixes source and artifact planes.
+
+## Consequences
+
+- One launch vector across the whole engines range, including future Node lines that change native TypeScript support; the smoke gate enforces it per matrix line.
+- TypeScript transformation is delegated to tsx/esbuild again, reversing the prior note's goal of proving Node-native transformation; that goal is unreachable while vendored sources use non-erasable syntax and Node ships no transform mode.
+- The runtime declared-dependency enforcement in source launches is gone; undeclared workspace imports now surface only through static gates or built-mode resolution failures.
+- Startup improves ~0.4s over the full tsx default (`demo:headless` and ACP keep `--import tsx`; their graphs were not audited for CJS-hook dependence and their launch latency is not on the interactive path).
diff --git a/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.zh.md b/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.zh.md
new file mode 100644
index 0000000000..48f410bd84
--- /dev/null
+++ b/.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.zh.md
@@ -0,0 +1,38 @@
+# Agent Note: dsh 通过 tsx ESM hook 源码启动
+
+Status: implemented
+
+[English](2026-07-29-dsh-source-launch-tsx-esm.md) | 中文
+
+> 取代[原生 TypeScript 源码启动](2026-07-28-dsh-native-typescript-source-launch.md)：Node 移除了该决策所依赖的能力。
+
+## 问题
+
+[原生源码启动决策](2026-07-28-dsh-native-typescript-source-launch.md)让 `apps/cli/src/bin.ts` 在 `node --experimental-transform-types` 下运行，配合一个只做解析的 paths loader，由 Node 负责 TypeScript 转换。Node 26.0.0 移除了 `--experimental-transform-types`（进程以 `bad option` 拒绝该 flag），只保留 strip 模式，而 strip 模式无法接受这个源码图必需的语法：vendor Cordis 中的参数属性（`constructor(private ctx: Context)`）、`vendor/hmr` 中的 `@Inject` 装饰器，以及遍布 `vendor/` 与 `packages/workflow` 的运行时 enum/namespace。仓库的 engines 范围（`^22.19.0 || >=24.0.0`）包含 Node 26，因此原生启动链在其上完全无法启动——且没有任何 CI 任务执行过真实启动向量，这一不兼容悄然发布。
+
+启动延迟同样是问题：off-thread 的 `module.register()` hooks worker 把每次解析都跨线程序列化（TUI 启动期间约 440ms 的 `makeSyncRequest` 等待），而完整 tsx 默认形态（`--import tsx`）的 CJS hook 解析放大要多付约 0.4s。
+
+## 决策
+
+`dsh` 的 TUI、Web 与无头源码启动运行 `node --import tsx/esm`：由 tsx 的 ESM-only hook 同时负责 TypeScript 转换与 tsconfig `paths` 投影。`bin/dsh`、根目录的 `dsh`/`demo:tui`/`demo:web` 脚本以及 Code Mode TUI overlay 使用同一向量；`bin/dsh` 以 checkout 的绝对路径引用 hook 与 tsconfig（裸的 `tsx/esm` 无法从任意 cwd 解析），并将 `TSX_TSCONFIG_PATH` 固定到根 tsconfig。CJS hook 保持关闭，因为 CLI 源码图是纯 ESM；实测 TUI 到 banner 约 0.7s，对比完整 tsx 默认形态约 1.1s、已移除的原生链约 0.75s。
+
+`scripts/tspath-loader.ts` 与 `apps/cli/src/tsconfig-paths-loader.ts` 已删除。随之消失的还有该 loader "仅为已声明运行时依赖映射 workspace import" 的运行时规则——tsx 无条件应用 `paths` 映射。声明完整性现在仅由静态门禁保障：配置的裸插件走 `verify-cordis-config`，manifest 走 workspace constraints。（该运行时规则确实发现过真实缺陷：`dsh-plan-mode` 与 `dsh-tool-tasks` 导入 `@deepseek-ai/dsh-llm` 却只声明在 devDependencies；已随本变更修复。）
+
+node-compat CI 矩阵（Node 22.19 与 26）新增 `dsh-source-launch-smoke`（`apps/cli/tests/source-launch.compat.spec.ts`）：以精确的生产启动向量做 keyless 管道 stdio 启动，断言非零退出的 TTY 拒绝。未来 Node 对模块 hook 或 TypeScript 处理的任何改动都会让该门禁变红，而不是破坏开发者的 `pnpm dsh`。
+
+## 备选方案
+
+**在 Node ≤25 保留原生链并按版本分叉。** 拒绝：两套转换语义（amaro 与 esbuild）在边缘语法上会分歧，启动器要加版本探测，node-compat 矩阵要覆盖两条路径——为一个已经变动过的 experimental flag 付出沉重维护。而且 amaro 也不支持 `vendor/hmr` 使用的 `@Inject` 装饰器，原生路径本来就无法启动随附的默认 TUI 配置。
+
+**把源码图改成 erasable-only 以适配 Node 26 strip 模式。** 拒绝：参数属性与值 namespace 遍布 vendor 的 Cordis/cosmokit/loader/schemastery；改写是无界 churn，且每次 vendor sync 都要重做。
+
+**仓库自有的同线程 loader（`module.registerHooks()` + esbuild 或 `@swc/core` 转换）。** 暂拒：原型实测约 0.45s（esbuild 路径未端到端验证；SWC 在 `vendor/hmr` 的装饰器 + namespace 合并上两种装饰器模式都会崩），但意味着自行负责转换正确性和一个 tsx 已经提供的 resolve hook。仅当约 0.3s 的差距成为真实成本时再重启；profiling 证据在 PR 讨论中。
+
+**Node 26 运行构建产物 `lib/`，24 保留原生。** 拒绝：在最新 Node 版本线上失去零构建开发循环，且混淆源码面与产物面。
+
+## 结果
+
+- 整个 engines 范围（包括未来改变原生 TypeScript 支持的 Node 版本线）只有一个启动向量；冒烟门禁按矩阵行强制执行。
+- TypeScript 转换重新委托给 tsx/esbuild，逆转了前一篇 note "证明 Node 原生转换可用" 的目标；在 vendor 源码使用不可擦除语法且 Node 不再提供 transform 模式的情况下，该目标不可达。
+- 源码启动中的运行时依赖声明强制不复存在；未声明的 workspace import 现在只能通过静态门禁或构建模式的解析失败暴露。
+- 启动相比完整 tsx 默认形态快约 0.4s（`demo:headless` 与 ACP 保持 `--import tsx`：其依赖图未就 CJS hook 依赖性做审计，且其启动延迟不在交互路径上）。
diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.i18n.yaml
new file mode 100644
index 0000000000..5689b73bf8
--- /dev/null
+++ b/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.md
+2026-07-29-sticky-composer-conversation-scroll.md: 7ceae95dafffdb756ef49bb5612cd4e711eb59ca
+2026-07-29-sticky-composer-conversation-scroll.zh.md: d925d82f94635b5fe67b0be119c041d003def393
diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.md b/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.md
new file mode 100644
index 0000000000..7ceae95daf
--- /dev/null
+++ b/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.md
@@ -0,0 +1,29 @@
+# Agent Note: Fixed header, sticky composer inside the transcript scrollport
+
+Status: implemented
+
+English | [中文](2026-07-29-sticky-composer-conversation-scroll.zh.md)
+
+## Problem
+
+The active conversation column split scrolling: the chat (and trajectory) view owned `overflow-y: auto`, while the composer stack sat as a sibling below that scrollport. A wheel gesture over the stats line or input therefore hit a non-scrolling region and did nothing — the transcript only moved when the pointer was over the message list. Long drafts made it worse: the textarea is itself a scrollport, so wheel over the composer could be trapped there. The session header must occupy the top of the column as ordinary chrome (not `position: sticky` inside the scrollport), while the composer must stick to the bottom of the same scrollport as the transcript so wheel over the footer moves the flow.
+
+## Decision
+
+While a session exists, `ConversationRoot` always supplies a `wrapActiveBody` owner callback that wraps the view ring in a `data-conversation-scroll` body and places a `data-composer-seat` around the whole `'conversation.composer'` chain output (fallback + elected overlay siblings from `overlay: true`). Active CSS sticks that seat with `position: sticky; bottom: 0` so Question/Approval takeovers stay visible when the user is not pinned to the floor; hero CSS centers the fallback stack inside the scroll body. `ConversationSession` keeps a chrome-hidden header + body shell while blank so that tree seat does not change on the first send. The session header remains `flex: none` column chrome above the scrollport when visible. ChatView and Trajectory/Waterfall keep a local scroller only when mounted outside that host (unit tests); under the host they set `overflow: visible` and resolve bottom-follow / prepend anchoring through `closest('[data-conversation-scroll]')`.
+
+Session stats live on `'conversation.composer.dock'` (above `'conversation.input.dock'`). The InputBar textarea, when inside the host, chains `wheel` with `{ passive: false }`: while the capped textarea can still scroll in that direction it keeps the native gesture; only at its own edge does it `preventDefault` and apply `deltaY` to the host.
+
+## Alternatives considered
+
+**Sticky header and sticky composer inside one column scrollport.** Rejected for the header: it must occupy the top as fixed layout chrome, not participate in the scrollport's sticky layer.
+
+**Fixed flex-none composer below the scrollport with wheel forwarding.** Rejected: the product requires the composer to stick inside the transcript scrollport so the footer is part of that scroll hit-testing surface, not a sibling that only forwards deltas.
+
+**Portal the composer into ChatView's scroller.** Rejected: the composer is shared across view tabs; the wrap target is the Session body owned by the resident shell.
+
+**Keep StatsLine inside ChatView below the message column.** Rejected: outside the sticky composer it would scroll away while the input stayed pinned.
+
+## Consequences
+
+Wheel over the footer scrolls the transcript; the visible layout is a fixed header, scrolling transcript, and sticky bottom composer. Stats appear on every active view tab. Nested view scrollers under the host are suppressed so sticky Turn headers in Trajectory stick to the column host. Hero → active keeps the same textarea DOM node (assembled slash-flow snapshot) and the InputHub draft.
diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.zh.md b/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.zh.md
new file mode 100644
index 0000000000..d925d82f94
--- /dev/null
+++ b/.agents/notes/implemented/bug-fix/2026-07-29-sticky-composer-conversation-scroll.zh.md
@@ -0,0 +1,29 @@
+# Agent Note: 固定标题栏，sticky 编辑器位于 transcript 滚动容器内
+
+Status: implemented
+
+[English](2026-07-29-sticky-composer-conversation-scroll.md) | 中文
+
+## Problem
+
+活跃会话列把滚动拆成两段：聊天（以及 trajectory）视图自有 `overflow-y: auto`，编辑器栈则作为该滚动容器的兄弟节点坐在下方。指针落在统计行或输入区上时，滚轮打在不可滚动区域上因而毫无效果——只有指针在消息列表上时 transcript 才会移动。草稿变长时更糟：textarea 本身也是滚动容器，编辑器上的滚轮可能被截在那里。会话标题栏必须以普通 chrome 占据列顶（不能在滚动容器内 `position: sticky`），而编辑器必须与 transcript 贴在同一滚动容器底部，使页脚上的滚轮能带动内容流动。
+
+## Decision
+
+只要存在会话，`ConversationRoot` 就会始终提供 `wrapActiveBody` owner 回调，将视图环包进 `data-conversation-scroll` 主体，并用 `data-composer-seat` 包住整条 `'conversation.composer'` chain 输出（`overlay: true` 下的 fallback 与选举出的 overlay 兄弟节点）。活跃阶段 CSS 以 `position: sticky; bottom: 0` 钉住该 seat，使用户未贴底时 Question／Approval 接管仍可见；hero CSS 在滚动主体内居中 fallback 栈。`ConversationSession` 在 blank 时保留隐藏 chrome 的 header + body 壳，使首次发送时树座位不变。可见时会话标题栏仍是滚动容器之上的 `flex: none` 列 chrome。ChatView 与 Trajectory/Waterfall 仅在宿主之外挂载时（单元测试）保留本地 scroller；位于宿主下时设为 `overflow: visible`，并通过 `closest('[data-conversation-scroll]')` 解析贴底跟随与前置锚定。
+
+会话统计挂在 `'conversation.composer.dock'`（位于 `'conversation.input.dock'` 之上）。InputBar 的 textarea 在宿主内以 `{ passive: false }` 链式处理 `wheel`：在限高 textarea 仍能沿该方向滚动时保留原生手势；仅在自身边缘才 `preventDefault` 并将 `deltaY` 施加到宿主。
+
+## Alternatives considered
+
+**标题栏与编辑器都在同一列滚动容器内 sticky。** 标题栏否决：它必须作为固定布局 chrome 占据顶部，而不是参与滚动容器的 sticky 层。
+
+**滚动容器下方 flex-none 固定编辑器并转发滚轮。** 否决：产品要求编辑器 sticky 在 transcript 滚动容器内，使页脚成为该滚动命中面的一部分，而不是仅转发增量的兄弟节点。
+
+**把编辑器 portal 进 ChatView 的 scroller。** 否决：编辑器跨视图标签共享；包装目标是常驻壳拥有的 Session 主体。
+
+**把 StatsLine 留在 ChatView 消息列下方。** 否决：落在 sticky 编辑器之外会随内容滚走，而输入区仍钉在底部。
+
+## Consequences
+
+在页脚上滚轮会滚动 transcript；可见布局是固定标题栏、可滚动 transcript 与 sticky 底部编辑器。统计出现在每一个活跃视图标签上。宿主下的嵌套视图 scroller 被抑制，因而 Trajectory 的 sticky Turn 标题贴在列宿主上。hero → active 保持同一 textarea DOM 节点（assembled slash-flow 快照）以及 InputHub 草稿。
diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.i18n.yaml
new file mode 100644
index 0000000000..060de23ba3
--- /dev/null
+++ b/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.md
+2026-07-29-web-details-session-lifecycle.md: d9e0255768f165bed0631b9324e971b57ec7dcae
+2026-07-29-web-details-session-lifecycle.zh.md: 09452ba80ff240ddca76df239b40ea661566f8e2
diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.md b/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.md
new file mode 100644
index 0000000000..d9e0255768
--- /dev/null
+++ b/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.md
@@ -0,0 +1,29 @@
+# Agent Note: Web details follow the current Session lifecycle
+
+Status: implemented
+
+English | [中文](2026-07-29-web-details-session-lifecycle.zh.md)
+
+## Problem
+
+The details entry is Session-scoped, but its preferred grid width is root-scoped. Selecting a different Session replaced the details content without closing that root preference, so the new owner inherited stale viewing geometry. Hero and other unselected states render no Session-scoped details; they need a derived zero track without becoming false owners in the comparison.
+
+## Decision
+
+`AppFrame` reads the current Session id and its `blank` summary flag from the authoritative Session projection. It records the last non-blank selected id only when that Session can own details, so hero and other unselected states neither trigger closure nor replace the last Session owner; their rendered details track derives as zero without changing the stored preference. The first Session keeps the default details width; returning to the same Session restores its current width; selecting a different Session closes the root-scoped details preference through the layout store before paint. The per-Session chat selection remains owned by the session-scoped store described by the [slot system standard](../architecture/2026-07-22-slot-type-chain-implementation.md).
+
+The layout store is transient and starts details at its default width. It neither reads nor writes `localStorage`, so reload resets both panel widths and needs no Session-baseline exception. Manual close and reopen inside one unchanged Session retain their existing behavior. The lifecycle effect changes neither the [Workspace-owned New Session flow](../feature/2026-07-25-workspace-ui-product-flow.md), composer drafts, Session navigation, nor concession-chain resizing.
+
+## Alternatives considered
+
+**Close details in the New Session click handler.** Rejected because an unselected surface has no Session-scoped details and must not mutate geometry. Closure belongs to the later comparison between two defined Session owners.
+
+**Persist panel geometry per Session.** Rejected because the product contract needs stale context removed, not a new map of remembered widths. Per-Session geometry would also reopen details when users return, contrary to the chosen close-on-leave behavior.
+
+**Preserve persisted layout after the Session baseline is ready.** Rejected because it duplicates startup lifecycle in a presentation component solely to validate stale viewing state. Transient defaults make reload deterministic without a readiness flag.
+
+**Treat every current-projection change as a Session switch.** Rejected because startup materialization, hero, clearing selection, and invalidation are not transitions between two Session owners.
+
+## Consequences
+
+Details is open by default, including when the first Session materializes. Switching to a different Session forgets the dragged details width because close writes zero and reopen uses the contract default. Unselected states derive a zero rendered track while leaving the preferred geometry unchanged; returning to the same Session through one of those states restores its width. Reload forgets sidebar and details geometry. The layout behavior test covers initial defaults, first materialization, direct and hero-mediated Session switches, same-Session return, and the absence of layout storage; the keyless browser e2e drives the same owner transitions through the shipped composition while checking the full grid track and browser errors.
diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.zh.md b/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.zh.md
new file mode 100644
index 0000000000..09452ba80f
--- /dev/null
+++ b/.agents/notes/implemented/bug-fix/2026-07-29-web-details-session-lifecycle.zh.md
@@ -0,0 +1,29 @@
+# Agent Note: Web 详情栏遵循当前会话生命周期
+
+Status: implemented
+
+[English](2026-07-29-web-details-session-lifecycle.md) | 中文
+
+## 问题
+
+详情入口由会话作用域拥有，而其首选网格宽度由根作用域拥有。选择不同会话时，系统会替换详情内容，却不会关闭根作用域的该首选宽度，因此新 owner 会继承陈旧的查看几何信息。hero 和其他未选中状态不会渲染会话作用域的详情；其轨道需派生为零宽度，但不能因此在比较中成为伪 owner。
+
+## 决策
+
+`AppFrame` 从权威会话投影读取当前会话 id 及其摘要中的 `blank` 标志。它只在该会话能够拥有详情时记录最后一个选中的非 blank 会话 id，因此 hero 和其他未选中状态既不会触发关闭，也不会替换最后一个会话 owner；这些状态下，详情栏轨道的渲染宽度派生为零，但存储的首选宽度不变。首个会话保留详情栏的默认宽度；返回同一会话时恢复其当前宽度；选择不同会话时，系统会先通过布局 store 关闭根作用域存储的详情栏首选宽度，再进行绘制。逐会话的聊天选中项继续由 [slot 体系标准](../architecture/2026-07-22-slot-type-chain-implementation.md)所述的会话作用域 store 拥有。
+
+布局 store 是瞬时状态，详情栏以默认宽度启动。它既不读取也不写入 `localStorage`，因此重新加载会重置两个面板的宽度，无需会话基线例外。在同一个未变化的会话内手动关闭和重新打开详情栏，仍保持原有行为。该生命周期 effect 不改变 [Workspace 拥有的 New Session 动线](../feature/2026-07-25-workspace-ui-product-flow.md)、composer 草稿、会话导航或让步链缩放。
+
+## 考虑过的替代方案
+
+**在 New Session 点击处理器中关闭详情栏。** 之所以否决：未选中表面没有会话作用域的详情，不得修改几何信息。详情栏是否关闭，应由随后对两个已定义会话 owner 的比较决定。
+
+**按会话持久化面板几何信息。** 之所以否决：产品契约需要移除陈旧上下文，而不是新增一张保存各宽度的映射。按会话保存几何信息还会在用户返回时重新打开详情栏，与选定的离开即关闭行为相悖。
+
+**在会话基线就绪后保留持久化布局。** 之所以否决：这会仅为验证陈旧的查看状态，在呈现组件中重复实现启动生命周期。瞬时默认值无需就绪标志即可使重新加载具有确定性。
+
+**将当前投影的每次变化都视为会话切换。** 之所以否决：启动时的物化、hero、清除选中项和选中状态失效都不是两个会话 owner 之间的过渡。
+
+## 后果
+
+详情栏默认打开，首次会话物化时亦然。切换到不同会话会忘记拖动后的详情宽度，因为关闭操作会写入零值，重新打开时则使用契约默认值。未选中状态会将轨道的渲染宽度派生为零，同时保持首选几何信息不变；经由这些状态返回同一会话时，会恢复其宽度。重新加载会忘记侧边栏与详情栏的几何信息。布局行为测试覆盖初始默认值、首次物化、直接及经 hero 中转的会话切换、返回同一会话，以及不存在布局存储的情况；无密钥浏览器 e2e 则通过已交付的组合驱动相同的 owner 过渡，同时检查完整网格轨道和浏览器错误。
diff --git a/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.i18n.yaml b/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.i18n.yaml
index 64d1f57f51..774b9cc52e 100644
--- a/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-06-22-acp-subagent-backend.md: a45ce5e34873249969bbe4dabb87a89d10246b3d
-2026-06-22-acp-subagent-backend.zh.md: 3b6a11efc1ae0427eb5ce3b30029b77b00d6801a
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.md
+2026-06-22-acp-subagent-backend.md: 5f12aa1c08d4f4cfaa35f2f7f4b09ad341c3eae8
+2026-06-22-acp-subagent-backend.zh.md: 61359246b0c552f6126cd82ae843d489de809a38
diff --git a/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.md b/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.md
index a45ce5e348..5f12aa1c08 100644
--- a/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.md
+++ b/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.md
@@ -34,7 +34,7 @@ ACP `StopReason` → harness `SubagentStopReason`: `end_turn`→`completed`, `ma
 
 ### Security: scrubbed child environment
 
-The child is a separate process, so it inherits an environment. Credential-shaped ambient vars (`/KEY|SECRET|TOKEN/i`) are NOT forwarded by default — the parent harness's own secrets must not leak into a spawned process implicitly (the same policy the bash executor applies). The child's OWN credentials (it needs a model key) are supplied EXPLICITLY via `config.env`, layered AFTER the scrub, so an intended `DEEPSEEK_API_KEY` survives while an incidental `AWS_SECRET_ACCESS_KEY` does not. Child stderr is inherited to the parent's stderr (diagnostics surface naturally); a spawn-level `error` event (e.g. ENOENT for a bad command) is captured and raced against the ACP drive, so a bad command settles `error` instead of crashing the parent with an unhandled error.
+The child is a separate process, so it inherits an environment. Credential-shaped ambient vars (`/KEY|PASSWORD|SECRET|TOKEN/i`) are NOT forwarded by default — the parent harness's own secrets must not leak into a spawned process implicitly (the same policy the bash executor applies). The child's OWN credentials (it needs a model key) are supplied EXPLICITLY via `config.env`, layered AFTER the scrub, so an intended `DEEPSEEK_API_KEY` survives while an incidental `AWS_SECRET_ACCESS_KEY` does not. Child stderr is inherited to the parent's stderr (diagnostics surface naturally); a spawn-level `error` event (e.g. ENOENT for a bad command) is captured and raced against the ACP drive, so a bad command settles `error` instead of crashing the parent with an unhandled error.
 
 ## Testing
 
diff --git a/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.zh.md b/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.zh.md
index 3b6a11efc1..61359246b0 100644
--- a/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.zh.md
+++ b/.agents/notes/implemented/feature/2026-06-22-acp-subagent-backend.zh.md
@@ -34,7 +34,7 @@ ACP `StopReason` → harness `SubagentStopReason`：`end_turn`→`completed`、`
 
 ### 安全：清洗子进程环境
 
-子 agent 是独立进程，因此会继承环境变量。形如凭证的环境变量（`/KEY|SECRET|TOKEN/i`）默认不转发——父 harness 自身的密钥不得隐式泄露到派生进程中（与 bash 执行器采用的策略相同）。子 agent 自己的凭证（它需要模型密钥）通过 `config.env` 显式提供，在清洗之后叠加，因此有意传入的 `DEEPSEEK_API_KEY` 得以保留，而偶然存在的 `AWS_SECRET_ACCESS_KEY` 则不会。子进程的 stderr 继承到父进程的 stderr（诊断信息自然浮现）；spawn 级别的 `error` 事件（如命令不存在时的 ENOENT）被捕获并与 ACP 驱动竞速，因此错误命令解析为 `error` 而非以未处理错误崩溃父进程。
+子 agent 是独立进程，因此会继承环境变量。形如凭证的环境变量（`/KEY|PASSWORD|SECRET|TOKEN/i`）默认不转发——父 harness 自身的密钥不得隐式泄露到派生进程中（与 bash 执行器采用的策略相同）。子 agent 自己的凭证（它需要模型密钥）通过 `config.env` 显式提供，在清洗之后叠加，因此有意传入的 `DEEPSEEK_API_KEY` 得以保留，而偶然存在的 `AWS_SECRET_ACCESS_KEY` 则不会。子进程的 stderr 继承到父进程的 stderr（诊断信息自然浮现）；spawn 级别的 `error` 事件（如命令不存在时的 ENOENT）被捕获并与 ACP 驱动竞速，因此错误命令解析为 `error` 而非以未处理错误崩溃父进程。
 
 ## 测试
 
diff --git a/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.i18n.yaml b/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.i18n.yaml
index 940908457f..7e9f4d52e6 100644
--- a/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-07-mcp-client-plugin.md: 2a8ad62e3bad1f2ae47100294f5dba25ceb47d12
-2026-07-07-mcp-client-plugin.zh.md: 9860d5ec805b394eb0c5f59f54349264593a5599
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.md
+2026-07-07-mcp-client-plugin.md: 756a5c4dc9f1152ecb1955d93b8dc47fcf07c661
+2026-07-07-mcp-client-plugin.zh.md: d2ff04d68033402fbb2718f8624e44d35d860db3
diff --git a/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.md b/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.md
index 2a8ad62e3b..756a5c4dc9 100644
--- a/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.md
+++ b/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.md
@@ -149,7 +149,7 @@ A unified `execute` handler for all tools from one MCP server:
 
 ### Subprocess environment (stdio transport)
 
-Replicate the `buildChildEnv` + `SENSITIVE_ENV_PATTERN` scrub from `dsh-subagent-acp`: filter ambient env (strip credential-shaped vars matching `/KEY|SECRET|TOKEN/i`), then merge `config.env` on top. Explicit env overrides survive the scrub.
+Build the child environment from the subprocess seam's shared `scrubbedParentEnv()` base, which removes ambient names matching `/KEY|PASSWORD|SECRET|TOKEN/i` and ambient `DSH_*` names, then merge `config.env` on top. Explicit env overrides survive the scrub.
 
 ### Disconnection / crash
 
diff --git a/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.zh.md b/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.zh.md
index 9860d5ec80..d2ff04d680 100644
--- a/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-07-mcp-client-plugin.zh.md
@@ -149,7 +149,7 @@ MCP 仅保证工具名在[单个服务器内](https://modelcontextprotocol.io/sp
 
 ### 子进程环境（stdio 传输）
 
-复用 `dsh-subagent-acp` 的 `buildChildEnv` + `SENSITIVE_ENV_PATTERN` 清洗逻辑：过滤环境变量（剥离匹配 `/KEY|SECRET|TOKEN/i` 的凭证形变量），然后将 `config.env` 覆盖合并到顶层。显式配置的 env 不受清洗影响。
+以子进程服务边界共享的 `scrubbedParentEnv()` 为基础构建子进程环境；该基础环境会移除环境中匹配 `/KEY|PASSWORD|SECRET|TOKEN/i` 的名称以及 `DSH_*` 名称，然后在其上合并 `config.env`。显式配置的 env 覆盖在清洗后仍会保留。
 
 ### 断连 / 崩溃
 
diff --git a/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.i18n.yaml b/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.i18n.yaml
index 0c7363827e..28ca5e0ae4 100644
--- a/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-10-parallel-tool-call-execution.md: c67ae61939a3e7974f9bf729058a57f5576308a1
-2026-07-10-parallel-tool-call-execution.zh.md: a80317aa951cbf3a9cae0651348c99712a4193d5
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md
+2026-07-10-parallel-tool-call-execution.md: 19f5dc189821433052edfa72613980a2e94e2cae
+2026-07-10-parallel-tool-call-execution.zh.md: e90e357180ae3684500adf7cba41c5fc1dac5743
diff --git a/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md b/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md
index c67ae61939..19f5dc1898 100644
--- a/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md
+++ b/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md
@@ -10,7 +10,7 @@ An assistant message may contain several sibling `tool-call` blocks. Running the
 
 Concurrency is a host scheduling concern, not model-facing tool metadata. The loop needs to decide which calls may overlap without hardcoding tool names or exposing scheduler policy in the JSON schema.
 
-The session log remains authoritative: every started call has an audit event, every started call receives a result, and model history observes results in the original call order regardless of completion order.
+The session log remains authoritative: every started call has an audit event, ordinary completion and cancellation pair calls with results, and model history observes committed results in the original call order regardless of completion order.
 
 ## Decision
 
@@ -46,7 +46,7 @@ Only dispatch and the tool body overlap. `tools/pre-execute` and `tools/post-exe
 
 Each started call appends `tool/call` immediately before its pre-execute gate. Completed dispatches occupy model-order slots, and a commit cursor appends `tool/result` and collects `additionalContexts` only when the next slot is ready. Live surfaces may show several pending calls, but results and post-tool context remain model-ordered.
 
-An abort before a group starts records no calls from that group. An abort during a group stops replenishment, waits for already-started calls, commits their results in order, drains accepted batch context after those results, and then ends the step through the existing abort path. Calls that never start have no audit event.
+An abort before a group starts records no calls from that group. An abort during a group stops replenishment, waits for already-started calls, commits their results in order, drains accepted batch context after those results, and then ends the step through the existing abort path. Calls that never start have no audit event. An unexpected scheduler failure stops new dispatches, waits for every already-started dispatch to settle, and rethrows the first failure. Because that failure is terminal internal state rather than a tool outcome, the loop does not invent tool results for rejected or uncommitted calls.
 
 Code Mode remains outside this scheduler because the model emits one native `run_code` call. `run_code` and its internal dispatch queue remain serial; native sibling calls in `mode: 'both'` use the normal scheduler.
 
@@ -66,7 +66,7 @@ Filesystem read relies on a narrow recorder exception: its synchronous observati
 
 ## Verification
 
-Unit coverage pins fail-closed classification, typed argument validation, grouping, barriers, live reclassification after registry replacement, the rolling cap, distinct execution objects, middleware order, ordered results and context, and abort draining. First-party tests pin each parallel declaration.
+Unit coverage pins fail-closed classification, typed argument validation, grouping, barriers, live reclassification after registry replacement, the rolling cap, distinct execution objects, middleware order, ordered results and context, abort draining, and scheduler-failure quiescence. First-party tests pin each parallel declaration.
 
 Snapshot coverage pins the visible multi-call transcript: pending calls may overlap while completed results remain model-ordered. Code Mode coverage pins its serial boundary. No provider-backed e2e is required because scheduling is deterministic loop behavior.
 
@@ -84,6 +84,8 @@ Snapshot coverage pins the visible multi-call transcript: pending calls may over
 
 **Expose staged methods or a scheduling waterfall.** Public `prepare` / `dispatch` / `finalize` methods or a `tools/execution-mode` event add extension surface before another consumer needs it. The loop uses an internal scheduler view, while `executionMode(exec)` leaves an insertion point for a policy seam.
 
+**Convert scheduler failures into tool results.** AgentLoop cannot determine whether a rejected dispatch invoked the tool body; ToolRegistry owns body-invocation state and typed tool outcomes. Internal scheduler failures therefore remain terminal instead of being reclassified as `ABORTED` results.
+
 **Start calls while the model streams.** This may reduce latency further but changes assistant-message authority, replay, and call/result pairing. The scheduler starts only after the assistant message is complete.
 
 **Use fixed-size windows.** Waiting for every call in one window before starting the next leaves capacity idle behind a slow call. The rolling pool preserves the cap without that delay.
@@ -101,3 +103,5 @@ Ordered commits may hold a fast result behind a slow earlier sibling. This prese
 Concurrent external calls can compete for quota or process capacity. Providers own their capacity controls; the loop cap only limits calls from one agent step.
 
 Tool registration is a scheduling boundary. Registry mutations affect not-yet-started calls because the scheduler reclassifies after each barrier and before every pool replenishment. Already-started calls retain the scheduling decision under which they entered the pool.
+
+A terminal scheduler failure may leave recorded calls without results before the failed step closes. Waiting for live dispatches preserves quiescence without misreporting those internal failures as tool outcomes.
diff --git a/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.zh.md b/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.zh.md
index a80317aa95..e90e357180 100644
--- a/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.zh.md
@@ -10,7 +10,7 @@ Status: implemented
 
 并发属于宿主调度范畴，不是面向模型的工具元数据。循环需要在不硬编码工具名称、不向 JSON Schema 暴露调度策略的前提下，判断哪些调用可以重叠执行。
 
-会话日志仍是权威记录：每个已启动的调用都有审计事件，都会获得结果；无论完成顺序如何，模型历史都按原始调用顺序观察结果。
+会话日志仍是权威记录：每个已启动的调用都有审计事件，正常完成和取消都会使调用与结果配对；无论完成顺序如何，模型历史都按原始调用顺序观察已提交的结果。
 
 ## 决策
 
@@ -46,7 +46,7 @@ Status: implemented
 
 每个已启动的调用都会在进入 pre-execute 门禁之前立即追加 `tool/call`。已完成的派发占据模型顺序的槽位；提交游标只有在下一个槽位就绪时，才会追加 `tool/result` 并收集 `additionalContexts`。实时界面可以显示多个待处理调用，但结果和工具执行后的上下文仍按模型顺序排列。
 
-如果在一组启动前中止，系统不会记录该组的任何调用。如果在一组执行期间中止，系统会停止补充池，等待已启动的调用，按顺序提交其结果，在这些结果之后排空已接受的批次上下文，然后通过现有中止路径结束该步骤。从未启动的调用没有审计事件。
+如果在一组启动前中止，系统不会记录该组的任何调用。如果在一组执行期间中止，系统会停止补充池，等待已启动的调用，按顺序提交其结果，在这些结果之后排空已接受的批次上下文，然后通过现有中止路径结束该步骤。从未启动的调用没有审计事件。调度器发生意外故障时，会停止新的派发，等待每项已启动的派发结算，并重新抛出第一个故障。由于该故障是内部终态，而非工具结果，循环不会为被拒绝或未提交的调用虚构工具结果。
 
 Code Mode 仍不使用此调度器，因为模型只会发出一个原生 `run_code` 调用。`run_code` 及其内部派发队列仍按串行方式执行；`mode: 'both'` 中的原生并列调用使用常规调度器。
 
@@ -66,7 +66,7 @@ Code Mode 仍不使用此调度器，因为模型只会发出一个原生 `run_c
 
 ## 验证
 
-单元测试覆盖固定了安全退化的分类、类型化参数验证、分组、屏障、替换注册表后的实时重新分类、滚动上限、独立执行对象、中间件顺序、有序结果与上下文，以及中止排空。第一方测试固定每项并行声明。
+单元测试覆盖固定了安全退化的分类、类型化参数验证、分组、屏障、替换注册表后的实时重新分类、滚动上限、独立执行对象、中间件顺序、有序结果与上下文、中止排空，以及调度器故障后的完全停稳。第一方测试固定每项并行声明。
 
 快照覆盖固定了可见的多调用 transcript（文本记录）：待处理调用可以重叠执行，已完成结果仍按模型顺序排列。Code Mode 覆盖固定其串行边界。此调度属于确定性循环行为，因此无需依赖提供方的 e2e 测试。
 
@@ -84,6 +84,8 @@ Code Mode 仍不使用此调度器，因为模型只会发出一个原生 `run_c
 
 **公开分阶段方法或调度 waterfall。** 公开的 `prepare` / `dispatch` / `finalize` 方法或 `tools/execution-mode` 事件，会在出现另一个消费方之前扩大扩展接口。循环使用内部调度器视图，而 `executionMode(exec)` 为策略 seam 保留了插入点。
 
+**将调度器故障转换为工具结果。** AgentLoop 无法判断被拒绝的派发是否已调用工具主体；ToolRegistry 负责工具主体调用状态和类型化工具结果。因此，内部调度器故障保持为终态，而不会被重新分类为 `ABORTED` 结果。
+
 **在模型流式输出时启动调用。** 这可能进一步降低延迟，但会改变 assistant 消息的权威性、回放以及调用/结果配对。调度器只在 assistant 消息完成后才启动。
 
 **使用固定大小的窗口。** 如果在启动下一个窗口前等待当前窗口的每个调用，一个缓慢调用就会使容量闲置。滚动池在保持上限的同时避免了这项延迟。
@@ -101,3 +103,5 @@ Code Mode 仍不使用此调度器，因为模型只会发出一个原生 `run_c
 并发外部调用可能会争用配额或进程容量。提供方负责自身容量控制；循环上限只限制一个 agent 步骤中的调用数量。
 
 工具注册是调度边界。调度器会在每个屏障之后以及每次补充池之前重新分类，因此注册表变更会影响尚未启动的调用。已启动的调用保留它们进入池时所依据的调度决策。
+
+终态调度器故障可能会在故障步骤关闭前留下已记录但没有结果的调用。等待仍在运行的派发可确保完全停稳，而不会将这些内部故障误报为工具结果。
diff --git a/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.i18n.yaml b/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.i18n.yaml
index be4fc8f19f..2de2288506 100644
--- a/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md
-2026-07-16-persistent-pty-sessions.md: 691ccd4341837d63bb48d27c2a3fb007657fe7ba
-2026-07-16-persistent-pty-sessions.zh.md: 86d57f79ee74429542730ebcad60bc9f3c9f15ca
+2026-07-16-persistent-pty-sessions.md: d7d06dc8517780a37889e4f94dd0b99475dc5432
+2026-07-16-persistent-pty-sessions.zh.md: 7783fa4b4f3056ab3b3088a4e9618f45d030ccbf
diff --git a/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md b/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md
index 691ccd4341..d7d06dc851 100644
--- a/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md
+++ b/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md
@@ -40,7 +40,7 @@ Agent-scope disposal closes registrations first, then awaits quiescent teardown
 
 A registered `shell` backend constrains how a terminal starts; it does not constrain commands typed after startup. `dsh-pty-local` therefore applies two protections before spawning:
 
-- It builds a scrubbed child environment using the same credential-shaped-name policy as `bash-local`, removing ambient `*KEY*`, `*SECRET*`, `*TOKEN*`, and harness-managed variables unless an explicit trusted mapping supplies them.
+- It builds a scrubbed child environment using the same credential-shaped-name policy as `bash-local`, removing ambient `*KEY*`, `*PASSWORD*`, `*SECRET*`, `*TOKEN*`, and harness-managed variables unless an explicit trusted mapping supplies them.
 - It requires `ctx.sandbox` and the shared `ctx.sandboxPolicy`. At spawn, the backend resolves the owner's effective session mode over the deployment default and wraps the shell argv once; that mode and workspace root remain the process boundary for the PTY lifetime. A write that would change the effective `sandbox/mode` is rejected before commit while the owner has any open PTY or unpublished spawn, with an instruction to wait for creation to settle and close those sessions first; same-effective-mode writes remain valid. The pending reservation spans backend setup through publication, so there is no race in which a wider terminal appears after a downgrade. `danger-full-access` is the existing explicit unconfined choice rather than a PTY-specific bypass.
 
 Sandboxing confines local process effects but does not make arbitrary shell input safe: network calls and other external side effects remain governed by deployment policy. Tool descriptions state that PTY sessions are less auditable than one-shot tools and should be used only when persistence or interactive stdin is necessary.
diff --git a/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.zh.md b/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.zh.md
index 86d57f79ee..7783fa4b4f 100644
--- a/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.zh.md
@@ -40,7 +40,7 @@ agent scope dispose 时先关闭注册，再等待全部所属 PTY 静默退出
 
 注册的 `shell` 后端只约束终端如何启动，不约束启动后输入的命令。因此 `dsh-pty-local` 在 spawn 前应用两层保护：
 
-- 它使用与 `bash-local` 相同的凭证形态名称策略构建清洗后的子进程环境，移除环境中的 `*KEY*`、`*SECRET*`、`*TOKEN*` 和 harness 管理的变量，除非显式的可信映射提供这些值。
+- 它使用与 `bash-local` 相同的凭证形态名称策略构建清洗后的子进程环境，移除环境中的 `*KEY*`、`*PASSWORD*`、`*SECRET*`、`*TOKEN*` 和 harness 管理的变量，除非显式的可信映射提供这些值。
 - 它要求 `ctx.sandbox` 和共享的 `ctx.sandboxPolicy`。后端在 spawn 时，以部署默认值为底折叠 owner 的有效 session mode，并只包装一次 shell argv；该 mode 与 workspace root 在 PTY 的整个生命周期中充当进程边界。只要 owner 有任何已打开的 PTY 或尚未发布的 spawn，任何会改变生效 `sandbox/mode` 的写入都会在提交前被拒绝，并提示先等待创建操作结算，再关闭这些会话；不会改变生效模式的写入仍然有效。这项进行中的预留从后端 setup 持续到发布完成，因此不存在降级后又出现权限更宽的终端这一竞态。`danger-full-access` 是现有的显式无约束选择，不另设 PTY 私有 bypass。
 
 沙箱限制本地进程副作用，但不会让任意 shell 输入自动安全：网络调用和其他外部副作用仍由部署策略治理。工具描述会说明 PTY 会话比一次性工具更难审计，只应在确实需要持久状态或交互式 stdin 时使用。
diff --git a/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.i18n.yaml b/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.i18n.yaml
index b09dcbf17d..029cc4e19c 100644
--- a/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.md
-2026-07-21-log-backed-session-titles.md: 1bd58e35ec625fb0b04c0c119ce425ff30a64881
-2026-07-21-log-backed-session-titles.zh.md: 37ec95efbca334f71d19d2bc3e18c22d50d9b5fb
+2026-07-21-log-backed-session-titles.md: 81ac687c6f55dd0ca1eaeb9d84c811edcfe17b5c
+2026-07-21-log-backed-session-titles.zh.md: b0c7e9d76a1b9365fa16dcb223b390b5aec3e174
diff --git a/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.md b/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.md
index 1bd58e35ec..81ac687c6f 100644
--- a/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.md
+++ b/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.md
@@ -36,9 +36,13 @@ Model providers require explicit word, CJK-character, input-byte, output-token,
 
 Automatic provider failures are nonfatal warnings and retain the latest title. Explicit refresh failures reject to the caller. Output must be non-empty text with unique ordered seqs drawn from the fixed request; the service normalizes and byte-limits it before log acceptance.
 
+### Explicit rename
+
+`rename(session, title)` accepts a user title synchronously: it normalizes the text under the accepted-title byte limit, supersedes in-flight automatic work, and appends a `session/title` event with the third source kind, `user`. A user-sourced latest title pins the session: `onUserMessage` schedules no automatic revision while it stands, under either cadence. An explicit `refresh()` remains the deliberate unpin — it appends a provider or fallback event over the pinned one whenever a replacement title is derivable (an underivable fallback, e.g. under a tiny byte cap, leaves the pin standing). The Web host exposes this as the `session.rename` unary method (resuming cold sessions first) and returns the normalized title plus its event seq so the client settles its `title` projection cell before the push frame arrives.
+
 ### Forks and consumers
 
-A fork inherits seed title events unchanged, like the rest of its source log. The first-message provider does not automatically retitle a fork. The all-messages provider may append a child-owned revision after a later child prompt, using inherited and new eligible messages.
+A fork inherits seed title events unchanged, like the rest of its source log — a pinned (user-sourced) title stays pinned in the child until an explicit refresh. The first-message provider does not automatically retitle a fork. The all-messages provider may append a child-owned revision after a later child prompt, using inherited and new eligible messages.
 
 `ctx.sessionQuery.readTitle()` folds one live-preferred or persisted log without loading titles during `listSessions()`. The TUI uses the latest title as its header subtitle and sets the terminal window title to `<session title> — <configured product title>` after terminal-safe rendering. The Web host folds the same log state into a validated mux control frame after each attached-session subscription baseline and immediately after forwarding a live raw title event. The browser retains only newer title event seqs even when the control frame precedes list or session-instance creation; sidebar labels, search, breadcrumbs, and the browser title then react to the projected revision. `session.list` remains metadata-only, so a cold persisted session uses the cwd basename or id until opening or resuming it attaches the log. The browser title uses `<session title> — <existing HTML title>` only for a selected titled session and otherwise preserves the product title. Consumers reporting agent completion use the core `findLastMessageTurnEnd()` fold, so a later between-turn title record cannot replace the preceding message-triggered outcome.
 
@@ -59,4 +63,4 @@ A fork inherits seed title events unchanged, like the rest of its source log. Th
 - A fallback appears immediately. Each fresh Web session adds one first-message auxiliary call; other compositions choose whether better titles justify model cost and whether later prompts should retitle a session.
 - Auxiliary request records and late accepted titles consume event seqs without consuming turn numbers, so persistence exposes both attempted dispatches and accepted updates even though model history and KV-cache identity do not change.
 - One provider and monotonic per-session revisions make disposal, supersession, and stale-result rejection explicit, at the cost of leaving multi-strategy precedence to a composite provider.
-- Manual rename, deletion, generated-versus-user precedence, search, and list indexing remain outside the capability.
+- Deletion (unpinning without an explicit refresh), search, and list indexing remain outside the capability.
diff --git a/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.zh.md b/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.zh.md
index 37ec95efbc..b0c7e9d76a 100644
--- a/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.zh.md
@@ -36,9 +36,13 @@ Status: implemented
 
 自动提供方故障只会发出非致命警告，并保留最新标题。显式刷新失败则会向调用方返回拒绝。输出必须是非空文本，并包含来自固定请求、唯一且有序的 seq；服务会在日志接受前对其进行规范化并施加字节限制。
 
+### 显式重命名
+
+`rename(session, title)` 同步接受用户标题：按已接受标题的字节上限规范化文本、取代在途自动工作，并追加一条第三种来源 `user` 的 `session/title` 事件。最新标题来源为 user 即钉住该会话：只要它还在，`onUserMessage` 在任一节奏下都不再安排自动修订。显式 `refresh()` 仍是有意的解钉手段——只要能推导出替代标题，它就在被钉住的标题之上追加提供方或回退事件（推导不出回退标题时，例如字节上限过小，钉住状态保持不变）。Web host 将其暴露为 `session.rename` unary 方法（冷会话先恢复），并返回规范化后的标题及其事件 seq，使 client 在推送帧到达前就结算自己的 `title` 投影格。
+
 ### Fork 与消费方
 
-与源日志的其他部分相同，fork 会原样继承作为种子的标题事件。首消息提供方不会自动为 fork 重新生成标题。全部消息提供方可以在子会话出现后续提示词后追加一项归子会话所有的修订，并使用继承的合格消息和新增的合格消息。
+与源日志的其他部分相同，fork 会原样继承作为种子的标题事件——被钉住（user 来源）的标题在子会话中保持钉住，直到显式 refresh。首消息提供方不会自动为 fork 重新生成标题。全部消息提供方可以在子会话出现后续提示词后追加一项归子会话所有的修订，并使用继承的合格消息和新增的合格消息。
 
 `ctx.sessionQuery.readTitle()` 会折叠一份实时优先或已持久化的日志，而不会在 `listSessions()` 期间加载标题。TUI 使用最新标题作为其标题栏副标题，并在完成终端安全渲染后，将终端窗口标题设置为 `<session title> — <configured product title>`。Web host 会在每个已附加会话的订阅基线之后，以及转发实时原始标题事件后立即，将同一份日志状态折叠为经过校验的 mux 控制帧。即使控制帧先于列表或会话实例创建抵达，浏览器也只保留标题事件 seq 较新的版本；侧边栏标签、搜索、面包屑和浏览器标题会随投影后的修订更新。`session.list` 仍只包含元数据，因此尚未打开的持久化会话会继续以 cwd 基名或 id 作为回退，直至打开或恢复会话时附加其日志。浏览器仅在选中已有标题的会话时将标题设置为 `<session title> — <existing HTML title>`，否则保留产品标题。报告 agent 完成情况的消费方使用核心的 `findLastMessageTurnEnd()` 折叠逻辑，因此后续的轮次间标题记录无法取代此前由消息触发的结果。
 
@@ -59,4 +63,4 @@ Status: implemented
 - 回退标题会立即出现。每个新建的 Web 会话都会增加一次针对首消息的辅助调用；其他组合可以自行决定更优标题是否值得模型成本，以及后续提示词是否需要重新生成会话标题。
 - 辅助请求记录和延迟接受的标题会占用事件 seq，但不会占用轮次编号，因此持久化会同时呈现尝试发起的调用与已接受的更新，尽管模型历史和 KV 缓存标识保持不变。
 - 单个提供方和每会话单调递增的修订号让释放、取代和陈旧结果拒绝行为明确可见，但多策略优先级必须由复合提供方负责。
-- 手动重命名、删除、生成标题与用户标题的优先级、搜索和列表索引不在此功能范围内。
+- 删除（不经显式 refresh 的解钉）、搜索和列表索引不在此功能范围内。
diff --git a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml
index 48a75f3449..187c4dfe94 100644
--- a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md
-2026-07-22-docked-web-goal-bar.md: 52a7d223ce3522c5ba977b1126dcd63bd2f6366f
-2026-07-22-docked-web-goal-bar.zh.md: e4842a03ccb8a29b35c7af0c03c51b1324b6ab36
+2026-07-22-docked-web-goal-bar.md: 110aea299a260896b0098f10b337734e0c0aebcf
+2026-07-22-docked-web-goal-bar.zh.md: cc0a5eda6815e6c02e97fd02197659764d4f2d69
diff --git a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md
index 52a7d223ce..110aea299a 100644
--- a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md
+++ b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md
@@ -12,9 +12,9 @@ The web UI had no goal surface at all: the goal stack shipped with model tools,
 
 `GoalBar` (`packages/client/ui-goal/src/client/GoalBar.tsx`) is a new props-driven, self-contained component; `ConversationRoot` mounts it immediately before the composer `InputBar`. The strip's CSS mirrors the composer's horizontal geometry (32px side padding, 776px centered cap) plus the mock's 12px inset, and a -10px bottom margin eats InputBar's 8px top padding and tucks its square bottom edge 2px under the composer card's top edge. All strip states share one fixed 38px height so switching between them never resizes it. Loading (`goal === undefined`), absent (`goal === null`), and `phase === 'complete'` render nothing — a completed goal is history, not chrome.
 
-Visibility drives the label and actions: active shows "Ongoing Goal" with edit/clear; paused shows "Paused Goal" and adds a resume icon button; blocked shows "Blocked Goal" and carries `blockedReason.message` as the strip's `title` tooltip. Goal creation lives on the `/goal` command, not in the bar. The pencil swaps the strip for an inline edit form prefilled with the current objective: Enter or the check button saves through `GoalBarActions.onEdit(objective)`, Esc cancels, and an all-whitespace objective keeps save disabled. The form closes only when the edit succeeds; a failure preserves the draft and displays the error in the bar. Resume and clear failures are displayed there as well. Clear otherwise calls `onClear` directly with no confirmation — a clear keeps a durable tombstone, so nothing is unrecoverable. An effect keyed on the goal's id drops the edit form when the goal's identity changes, so a surviving draft can never be written over the goal that replaced it.
+Visibility drives the label and actions: active shows "Ongoing Goal" with pause/edit/clear; paused shows "Paused Goal" and swaps pause for a resume icon button; blocked shows "Blocked Goal" and carries `blockedReason.message` as the strip's `title` tooltip. Goal creation lives on the `/goal` command, not in the bar. The pencil swaps the strip for an inline edit form prefilled with the current objective: Enter or the check button saves through `GoalBarActions.onEdit(objective)`, Esc cancels, and an all-whitespace objective keeps save disabled. The form closes only when the edit succeeds; a failure preserves the draft and displays the error in the bar. Resume and clear failures are displayed there as well. Clear otherwise calls `onClear` directly with no confirmation — a clear keeps a durable tombstone, so nothing is unrecoverable. An effect keyed on the goal's id drops the edit form when the goal's identity changes, so a surviving draft can never be written over the goal that replaced it.
 
-`GoalBarActions` lives in the contract layer (`contract/slots.ts`, next to the `ConversationInjected.goalActions` slot it feeds) and carries exactly the rendered verbs: `onEdit`/`onResume`/`onClear`. Each callback asynchronously returns an explicit success/failure result so `GoalBar` owns its transitions and error display. `apply.ts` wires them to the runtime session methods; the runtime session resolves the current goal's compare-and-set ref internally, so the UI passes no ref.
+`GoalBarActions` lives in ui-goal's slot contract (`packages/client/ui-goal/src/client/slots.ts`) and carries exactly the rendered verbs: `onEdit`/`onPause`/`onResume`/`onClear`. Each callback asynchronously returns an explicit success/failure result so `GoalBar` owns its transitions and error display. `apply.ts` wires them to the runtime session methods; the runtime session resolves the current goal's compare-and-set ref internally, so the UI passes no ref.
 
 The runtime session gains the goal surface the strip (and future UI) needs: `fetchGoal` populates the snapshot on open, and a live `context/message` carrying `goal/change` meta triggers a coalesced refetch — concurrent triggers share the in-flight `goal.get`, while a trigger received during that read schedules one coalesced trailing read so independently ordered notifications and GET responses cannot leave stale state. Window replays never refetch, and matching the meta kind (rather than a goal key) also catches clear tombstones written by other clients. The six mutation verbs fold transport failures into `{ ok: false }` results like every sibling session method, and a get result older than a mutation response that landed mid-flight is dropped.
 
@@ -22,18 +22,18 @@ The strip's background is `--dsw-alias-interactive-bg-hover` rather than the moc
 
 ## Testing
 
-`packages/client/ui-goal/tests/goalbar.spec.tsx` pins the behavior through props alone: loading/absent/complete render nothing, the active strip renders label/objective and fires clear, the edit form prefills, rejects empty, saves on Enter, cancels on Esc, and resets when the goal's identity changes, the paused strip fires resume, and the blocked strip exposes the reason tooltip. Component failure-path cases prove that a failed edit preserves its draft and that edit/resume/clear errors remain visible in the bar. The skeleton specs mount `ConversationRoot` with and without `goalActions`; the undefined case is seeded with an active goal, so the missing gate — not the missing goal — is what hides the strip. Runtime session specs pin the folded-error results, the live-only in-flight-plus-trailing refetch, and the stale-read guard. A keyless real-browser smoke boots the assembled application through `boot → RPC → runtime → GoalBar` and records an inline snapshot of the rendered label, objective, and actions.
+`packages/client/ui-goal/tests/goalbar.spec.tsx` pins the behavior through props alone: loading/absent/complete render nothing, the active strip renders label/objective and fires clear, the edit form prefills, rejects empty, saves on Enter, cancels on Esc, and resets when the goal's identity changes, the active strip fires pause, the paused strip fires resume, and the blocked strip exposes the reason tooltip. Component failure-path cases prove that a failed edit preserves its draft and that edit/resume/clear errors remain visible in the bar. The skeleton specs mount `ConversationRoot` with and without `goalActions`; the undefined case is seeded with an active goal, so the missing gate — not the missing goal — is what hides the strip. Runtime session specs pin the folded-error results, the live-only in-flight-plus-trailing refetch, and the stale-read guard. A keyless real-browser smoke boots the assembled application through `boot → RPC → runtime → GoalBar` and records an inline snapshot of the rendered label, objective, and actions.
 
 ## Alternatives considered
 
 - **Put the strip in the session header** — rejected because the redesign's premise is that goal presence belongs to the composer's context; a header strip cannot dock into the composer card.
 - **Render a "Loading goal…" placeholder for `undefined`** — rejected: the strip would flash and collapse on every session open, chrome noise for a sub-second state.
 - **Include an inline create affordance when no goal is set** — rejected after implementation review: goal creation lives on the `/goal` command, matching the pattern where the model creates goals on request; the bar is a status indicator, not a creation surface.
-- **Carry the full verb set (`onPause`/`onComplete`) in `GoalBarActions`** — rejected as speculative generality: no consumer calls them, so the interface carries only the rendered verbs.
+- **Carry the full verb set (`onComplete` included) in `GoalBarActions`** — rejected as speculative generality: the interface carries only the rendered verbs (`onPause` joined it when the active strip gained its pause action).
 
 ## Consequences
 
-- Goal presence in the web UI is a composer-docked strip: sparkle, phase label, truncated objective, and edit/clear (plus resume when paused) — the browser client's first goal surface.
+- Goal presence in the web UI is a composer-docked strip: sparkle, phase label, truncated objective, and pause/edit/clear (resume replacing pause when paused) — the browser client's first goal surface.
 - The runtime session exposes the goal verbs over RPC with folded transport errors, and refreshes the snapshot's goal on open and on live goal-change meta (coalesced, guarded against stale reads).
-- Objective editing is reachable from the UI for the first time, through `goal.edit` with the runtime-owned ref; pause/complete remain available to other surfaces (`/goal`, model tools).
+- Objective editing is reachable from the UI for the first time, through `goal.edit` with the runtime-owned ref; complete remains available to other surfaces (`/goal`, model tools).
 - `goal === null` renders nothing; the composer carries no persistent create affordance — creation is the `/goal` command's job.
diff --git a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md
index e4842a03cc..cc0a5eda68 100644
--- a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md
@@ -12,9 +12,9 @@ Web UI 此前没有任何目标相关的界面：目标栈已随模型工具、T
 
 `GoalBar`（`packages/client/ui-goal/src/client/GoalBar.tsx`）是一个新的、由 props 驱动的自包含组件；`ConversationRoot` 将它挂载在输入框 `InputBar` 紧上方。横条的 CSS 对齐输入框的水平几何（两侧 32px 内边距、776px 居中上限），再加上设计稿的 12px 内缩，并用 -10px 的下外边距吃掉 InputBar 的 8px 上内边距，使它方形的底边收进输入框卡片顶边之下 2px。横条的所有状态共享固定的 38px 高度，状态切换不会引起尺寸变化。加载中（`goal === undefined`）、无目标（`goal === null`）和 `phase === 'complete'` 时不渲染任何内容：已完成的目标是历史记录，不是常驻界面元素。
 
-可见性决定标签和操作：active 状态显示 "Ongoing Goal" 并提供编辑／清除；paused 状态显示 "Paused Goal"，并增加一个恢复图标按钮；blocked 状态显示 "Blocked Goal"，并把 `blockedReason.message` 作为横条的 `title` 悬浮提示。创建目标的入口在 `/goal` 命令上，不在横条里。铅笔图标把横条切换为内联编辑表单，预填当前目标内容：Enter 或勾选按钮通过 `GoalBarActions.onEdit(objective)` 保存，Esc 取消，目标内容全为空白字符时保存按钮保持禁用。编辑成功后表单才会关闭；编辑失败时保留草稿，并在横条中显示错误。恢复和清除失败也显示在横条中。除此之外，清除直接调用 `onClear`，不做确认——清除会保留 durable 墓碑，没有不可恢复的损失。一个以目标 id 为键的 effect 会在目标身份变化时丢弃编辑表单，因此存留的草稿绝不可能覆盖掉替换它的新目标。
+可见性决定标签和操作：active 状态显示 "Ongoing Goal" 并提供暂停／编辑／清除；paused 状态显示 "Paused Goal"，把暂停换成一个恢复图标按钮；blocked 状态显示 "Blocked Goal"，并把 `blockedReason.message` 作为横条的 `title` 悬浮提示。创建目标的入口在 `/goal` 命令上，不在横条里。铅笔图标把横条切换为内联编辑表单，预填当前目标内容：Enter 或勾选按钮通过 `GoalBarActions.onEdit(objective)` 保存，Esc 取消，目标内容全为空白字符时保存按钮保持禁用。编辑成功后表单才会关闭；编辑失败时保留草稿，并在横条中显示错误。恢复和清除失败也显示在横条中。除此之外，清除直接调用 `onClear`，不做确认——清除会保留 durable 墓碑，没有不可恢复的损失。一个以目标 id 为键的 effect 会在目标身份变化时丢弃编辑表单，因此存留的草稿绝不可能覆盖掉替换它的新目标。
 
-`GoalBarActions` 位于 contract 层（`contract/slots.ts`，紧挨它所喂给的 `ConversationInjected.goalActions` 槽位），只携带实际渲染的动词：`onEdit`/`onResume`/`onClear`。每个回调都会异步返回显式成功／失败结果，因此 `GoalBar` 自行负责界面转换和错误显示。`apply.ts` 把它们接到运行时会话方法上；运行时会话在内部解析当前目标的 compare-and-set ref，因此 UI 不传 ref。
+`GoalBarActions` 位于 ui-goal 的槽位契约（`packages/client/ui-goal/src/client/slots.ts`），只携带实际渲染的动词：`onEdit`/`onPause`/`onResume`/`onClear`。每个回调都会异步返回显式成功／失败结果，因此 `GoalBar` 自行负责界面转换和错误显示。`apply.ts` 把它们接到运行时会话方法上；运行时会话在内部解析当前目标的 compare-and-set ref，因此 UI 不传 ref。
 
 运行时会话获得了横条（以及未来 UI）所需的目标表面：`fetchGoal` 在打开时填充快照；携带 `goal/change` 元数据的 live `context/message` 触发合并重新拉取——并发触发器共享正在执行的 `goal.get`，读取期间收到的触发器会安排一次合并后的尾随读取，避免彼此独立排序的通知和 GET 响应留下陈旧状态。窗口重放绝不触发重新拉取，且匹配元数据 kind（而不是 goal 键）还能捕获其他客户端写入的清除墓碑。六个变更动词与所有同类会话方法一样，把传输层失败折叠为 `{ ok: false }` 结果；比在拉取途中落地的变更响应更旧的 get 结果会被丢弃。
 
@@ -22,18 +22,18 @@ Web UI 此前没有任何目标相关的界面：目标栈已随模型工具、T
 
 ## 测试
 
-`packages/client/ui-goal/tests/goalbar.spec.tsx` 仅通过 props 固定这些行为：加载中／无目标／已完成时不渲染；active 横条渲染标签和目标内容并触发清除；编辑表单预填内容、拒绝空值、按 Enter 保存、按 Esc 取消，并在目标身份变化时重置；paused 横条触发恢复；blocked 横条暴露原因悬浮提示。组件失败路径用例证明编辑失败时保留草稿，并且编辑／恢复／清除错误持续显示在横条中。skeleton 规格测试分别挂载带与不带 `goalActions` 的 `ConversationRoot`；未定义的情形预置了一个 active 目标，因此隐藏横条的是缺失的挂载门，而不是缺失的目标。运行时会话规格测试固定了折叠错误结果、仅 live 的执行中读取加尾随读取，以及陈旧读取守卫。一个无密钥真实浏览器冒烟测试通过 `boot → RPC → runtime → GoalBar` 启动组装后的应用，并以内联快照记录渲染出的标签、目标内容和操作。
+`packages/client/ui-goal/tests/goalbar.spec.tsx` 仅通过 props 固定这些行为：加载中／无目标／已完成时不渲染；active 横条渲染标签和目标内容并触发清除；编辑表单预填内容、拒绝空值、按 Enter 保存、按 Esc 取消，并在目标身份变化时重置；active 横条触发暂停；paused 横条触发恢复；blocked 横条暴露原因悬浮提示。组件失败路径用例证明编辑失败时保留草稿，并且编辑／恢复／清除错误持续显示在横条中。skeleton 规格测试分别挂载带与不带 `goalActions` 的 `ConversationRoot`；未定义的情形预置了一个 active 目标，因此隐藏横条的是缺失的挂载门，而不是缺失的目标。运行时会话规格测试固定了折叠错误结果、仅 live 的执行中读取加尾随读取，以及陈旧读取守卫。一个无密钥真实浏览器冒烟测试通过 `boot → RPC → runtime → GoalBar` 启动组装后的应用，并以内联快照记录渲染出的标签、目标内容和操作。
 
 ## 考虑过的替代方案
 
 - **把横条放在会话头部**：不予采纳，因为重新设计的前提是目标的存在感属于输入框的上下文；放在头部的横条无法停靠进输入框卡片。
 - **为 `undefined` 渲染 "Loading goal…" 占位**：不予采纳，每次打开会话横条都会闪现再坍缩，对一个不到一秒的状态来说只是界面噪音。
 - **未设置目标时在横条内提供内联创建入口**：实现评审后不予采纳，创建目标的职责在 `/goal` 命令上，与模型按请求创建目标的模式一致；横条是状态指示器，不是创建入口。
-- **在 `GoalBarActions` 中携带完整动词集合（`onPause`/`onComplete`）**：作为投机性泛化不予采纳，没有消费方调用它们，接口只携带实际渲染的动词。
+- **在 `GoalBarActions` 中携带完整动词集合（含 `onComplete`）**：作为投机性泛化不予采纳，接口只携带实际渲染的动词（active 横条获得暂停操作后，`onPause` 随之加入）。
 
 ## 后果
 
-- Web UI 中目标的存在形式是停靠在输入框上方的横条：闪光图标、阶段标签、截断的目标内容，以及编辑／清除（暂停时另有恢复）——这是浏览器客户端的第一个目标界面。
+- Web UI 中目标的存在形式是停靠在输入框上方的横条：闪光图标、阶段标签、截断的目标内容，以及暂停／编辑／清除（暂停时恢复取代暂停）——这是浏览器客户端的第一个目标界面。
 - 运行时会话通过 RPC 暴露目标动词并折叠传输层错误，且在打开时和 live 目标变更元数据到达时刷新快照中的目标（合并拉取，带陈旧读取守卫）。
-- 目标内容首次可以从 UI 编辑，经由 `goal.edit`，ref 由运行时持有；暂停／完成对其他界面（`/goal`、模型工具）照常可用。
+- 目标内容首次可以从 UI 编辑，经由 `goal.edit`，ref 由运行时持有；完成对其他界面（`/goal`、模型工具）照常可用。
 - `goal === null` 时不渲染任何内容；输入框不提供常驻的创建入口，创建是 `/goal` 命令的职责。
diff --git a/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.i18n.yaml
new file mode 100644
index 0000000000..b5fa2609a4
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-tmux-location-context.md
+2026-07-27-tmux-location-context.md: bac5861f7f55c259de04d153115f164d90c415ad
+2026-07-27-tmux-location-context.zh.md: 03cd722381c45604f7aae8f3d0a9f9fbb8b12bb5
diff --git a/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.md b/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.md
new file mode 100644
index 0000000000..bac5861f7f
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.md
@@ -0,0 +1,61 @@
+# Agent Note: tmux-location context
+
+Status: implemented
+
+English | [中文](2026-07-27-tmux-location-context.zh.md)
+
+## Problem
+
+An agent running inside tmux has no way to tell the model where it is: which session, window, and pane the process occupies, and how the window is laid out. A user directing several panes wants the model to orient itself to its own location so instructions like "the pane below" or "this window" resolve. The location must reach the model as durable, reconstructable context, not a system-prompt value rewritten in place, and must cost nothing when the location has not changed.
+
+tmux exposes this without a daemon: `$TMUX_PANE` names the process's pane, and `tmux display-message -t "$TMUX_PANE" -p '<format>'` prints any pane/window/session field. The open question was how to observe it — pull on each preparation, or push from a tmux hook — and how to avoid a per-step token cost and hidden process-local state.
+
+## Decision
+
+`@deepseek-ai/dsh-tmux-context` is an opt-in function plugin in `packages/context/tmux-context/`, alongside the other bounded request-context enrichments that define neither a tool nor a service. Shipped examples do not mount it because tmux-location disclosure and its token cost are deployment policy.
+
+**Pull on the first step of each turn, not a tmux push.** The plugin prepends an `agent/step` listener and acts only when `step === 1`. A pull model needs no background process, no hook installation in the user's tmux, and no teardown; it re-reads current state each turn so a moved, renamed, or re-laid-out pane is picked up naturally. Gating on the first step makes the reading per-turn: a location is stable within a turn, and re-querying every step would add cost without new information. A pane moved mid-turn is reflected on the next turn, which is the accepted tradeoff for the simpler design.
+
+**Read through the `ctx.bash` seam, never raw `child_process`.** The listener runs the tmux/`ps` read commands through `ctx.bash`, so the deployment's sandbox and policy apply and the plugin owns no subprocess code. Absent `ctx.bash`, absent tmux env, a wrong field count, or an empty pane id each make the attempt a no-op, matching how `workspace-context` no-ops without an `fs` provider.
+
+**Detect a real pane by tty, not by `$TMUX_PANE` alone.** `$TMUX_PANE` is inherited: a terminal launched from a tmux shell (a VS Code integrated terminal, a desktop launcher) carries `$TMUX`/`$TMUX_PANE` from that ancestor even though the process does not live in that pane, which otherwise injects a stale, wrong location. The command resolves this process's controlling terminal with `ps -o tty= -p <pid>` (the agent's own pid, passed in-process) and compares it to the pane's `#{pane_tty}`; fields are emitted only on a match. A genuine pane owns this process's tty; an inherited environment names some other pane's tty and reads as "not in tmux". Checking `$TMUX` instead does not help — it is inherited identically. This is the definitive discriminator and needs no allowlist of terminal emulators.
+
+**Own location and layout only.** The queried fields are session name, window index/name, pane index/id, window/pane active flags, and `window_layout`. Pane and window pixel sizes are excluded (layout tree conveys structure; sizes are noisy and change on every terminal resize). Sibling-pane contents are never captured (`capture-pane`), keeping the reading small and avoiding scraping unrelated, possibly sensitive, output.
+
+**Inject only on change, with optional interval floor.** When due, the plugin calls `agent.inject()` for one `user/message` with source `{ kind: 'plugin', plugin: 'tmux-context' }`. Change suppression compares the rendered state block (everything after the turn preamble line) against the latest injection of this source, found by scanning raw durable session events — so the schedule survives compaction and process resume without a process-local cache. The optional `refreshIntervalMs` (manually validated as a non-negative safe integer at plugin load) additionally suppresses injections within that window of the latest one.
+
+### Text
+
+```text
+tmux location (turn <turn>):
+session <session>, window <index> "<name>", pane <index> <pane-id>
+window active=<0|1>, pane active=<0|1>, layout <window-layout>
+```
+
+The turn preamble is the volatile first line; the two-line state block below it is the unit compared for change suppression, so re-injection is driven by tmux state, not loop position.
+
+### Durability and request reconstruction
+
+Each reading is a normal surface node until compaction shadows it; the plugin contributes nothing to system-prompt assembly and `request/header` carries no tmux-context text. The reading records a preparation attempt, not a committed step: because the prepended listener runs first, its append may remain when a later `agent/step` listener cancels or fails the attempt, and the append-only log performs no rollback.
+
+The published `./invariant` companion registers no runtime check: a reading is a per-turn snapshot of external tmux state, so the session holds no cross-event relation to validate, and scheduling and format stay pinned by the package's pipeline tests.
+
+## Consequences
+
+An agent booted inside tmux now receives its own session/window/pane location and window layout as durable, source-attributed context, updated per turn when the location changes. Deployments opt in through cordis.yml; the default spine and shipped examples stay silent. Outside a real tmux pane — including a terminal that merely inherited `$TMUX`/`$TMUX_PANE` — or without a `ctx.bash` executor, the plugin is inert with no error, so composing it is safe everywhere. Because the reading is one durable `user/message`, it survives compaction as ordinary history, contributes nothing to system-prompt assembly or request headers, and costs at most one two-line message per changed turn. The pull model adds one `tmux display-message` subprocess (through the sandboxed bash seam) on the first step of each turn that is due. The optional interval floor is checked before the query and so suppresses both; an unchanged location is detected only by comparing the returned state, so it suppresses the injection while still paying for the query.
+
+## Testing
+
+Unit tests pin: first-step injection and source/surface metadata; the `$TMUX_PANE`-keyed command including its `#{pane_tty}`-vs-`ps -o tty=` guard; step-gating; change suppression across turns and re-injection on a moved pane; positive-interval suppression and threshold; every no-op path (no bash, nonzero exit, wrong field count, empty pane id, aborted signal, and a contained executor rejection from either `resolve()` or `run()` that warns instead of failing the turn); prepended ordering before ordinary `agent/step` listeners; resilience to a corrupt prior reading (non-text block, single-line text); and config rejection of negative and non-integer intervals. Per-file coverage is 100%.
+
+## Alternatives considered
+
+- **Push from a tmux hook / background watcher** — rejected: requires installing hooks in the user's tmux and a background process with teardown, to gain mid-step freshness that per-turn context does not need.
+- **Run every step** — rejected: location is stable within a turn; re-querying adds token cost without new information. Gating on `step === 1` yields per-turn readings.
+- **Raw `child_process`** — rejected: bypasses the sandbox/policy seam and hand-rolls subprocess code the `ctx.bash` executor already owns.
+- **Include pane/window pixel sizes** — rejected: sizes churn on every resize and add noise; the layout tree already conveys structure.
+- **Scrape sibling panes with `capture-pane`** — rejected: large, noisy, and privacy-sensitive; out of scope for "own location".
+- **Dynamic system-prompt section** — rejected: replacing a value erases the earlier readings behind prior reasoning and is not reconstructable; one durable attributed message records each location where it became visible.
+- **Trust `$TMUX_PANE` (or `$TMUX`) presence** — rejected: both are inherited by terminals launched from a tmux shell (VS Code integrated terminal), so a non-pane process injects a stale location. The pane `#{pane_tty}` vs. this process's controlling tty is the definitive check.
+- **Denylist known terminal emulators (e.g. `TERM_PROGRAM=vscode`)** — rejected: a partial, ever-growing list that still misses other launchers; the tty match is exact and launcher-agnostic.
+- **A runtime invariant validating each reading's turn, position, and format** — shipped initially, then removed: it re-derived the producer's own scheduling from the log and asserted a regex over text the same package had just rendered, so it restated `apply()` rather than checking an independent relation. Every failure it could report required an edit to this package, which its pipeline tests already catch. Reintroduce a companion check only for a relation the plugin does not itself compute — for example if readings gain cross-turn ordering or enclosure obligations that another package can violate.
diff --git a/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.zh.md b/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.zh.md
new file mode 100644
index 0000000000..03cd722381
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-tmux-location-context.zh.md
@@ -0,0 +1,61 @@
+# Agent Note：tmux 位置上下文
+
+Status: implemented
+
+[English](2026-07-27-tmux-location-context.md) | 中文
+
+## 问题
+
+运行在 tmux 内的 agent 无法告诉模型自己身在何处：进程占据哪个 session、window、pane，以及 window 如何布局。当用户操作多个 pane 时，希望模型能对自身位置有所定位，从而让"下方的 pane""这个 window"之类的指令得以解析。位置必须以持久、可重建的上下文形式送达模型，而非在原地被改写的系统提示值，并且当位置未变化时不产生任何成本。
+
+tmux 无需守护进程即可暴露这些信息：`$TMUX_PANE` 标识进程所在 pane，`tmux display-message -t "$TMUX_PANE" -p '<format>'` 可打印任意 pane/window/session 字段。待决问题在于如何观测——在每次准备时拉取，还是由 tmux hook 推送——以及如何避免逐步骤 token 成本与隐藏的进程内状态。
+
+## 决策
+
+`@deepseek-ai/dsh-tmux-context` 是位于 `packages/context/tmux-context/` 的可选启用型函数插件，与其他既不定义工具也不定义服务的有界请求上下文增强并列。随附示例不挂载它，因为 tmux 位置披露及其 token 成本属于部署策略。
+
+**在每轮的第一个 step 拉取，而非 tmux 推送。** 插件前置注册一个 `agent/step` 监听器，仅在 `step === 1` 时动作。拉取模型无需后台进程、无需在用户的 tmux 中安装 hook、也无需清理；它每轮重新读取当前状态，因此被移动、改名或重新布局的 pane 都会被自然感知。以第一个 step 为门槛使读数按轮次生成：位置在一轮内是稳定的，逐步骤重复查询只会增加成本而不带来新信息。轮次中途移动的 pane 会在下一轮反映，这是换取更简单设计所接受的取舍。
+
+**通过 `ctx.bash` seam 读取，绝不用裸 `child_process`。** 监听器通过 `ctx.bash` 运行 tmux／`ps` 只读命令，从而应用部署方的沙箱与策略，插件不拥有任何子进程代码。`ctx.bash` 缺失、tmux 环境缺失、字段数不符或 pane id 为空，都会使本次尝试成为空操作，与 `workspace-context` 在无 `fs` provider 时的空操作一致。
+
+**以 tty 判定真实 pane，而非仅凭 `$TMUX_PANE`。** `$TMUX_PANE` 会被继承：从 tmux shell 启动的终端（VS Code 集成终端、桌面启动器）会从该祖先进程带上 `$TMUX`／`$TMUX_PANE`，即使进程并不位于那个 pane 中，否则就会注入一个陈旧且错误的位置。命令用 `ps -o tty= -p <pid>`（在进程内传入 agent 自身的 pid）解析本进程的控制终端，并与 pane 的 `#{pane_tty}` 比较；只有匹配时才输出字段。真正的 pane 拥有本进程的 tty；继承而来的环境指向的是另一个 pane 的 tty，因而被读作"不在 tmux 中"。改为检查 `$TMUX` 也无济于事——它同样会被继承。这是决定性的判别依据，且无需维护终端模拟器名单。
+
+**仅自身位置与布局。** 查询字段为 session name、window index/name、pane index/id、window/pane 活动标志以及 `window_layout`。省略 pane 与 window 像素尺寸（布局树已传达结构；尺寸嘈杂且每次终端缩放都会变化）。从不采集相邻 pane 内容（`capture-pane`），使读数保持小巧，并避免抓取无关、可能敏感的输出。
+
+**仅在变化时注入，并可选间隔下限。** 需要时，插件调用 `agent.inject()` 注入一条来源为 `{ kind: 'plugin', plugin: 'tmux-context' }` 的 `user/message`。变化抑制将渲染出的状态块（轮次前缀行之后的全部内容）与该来源的最近一次注入比较，后者通过扫描原始持久会话事件获得——因此调度可跨压缩与进程恢复存续，无需进程内缓存。可选的 `refreshIntervalMs`（在插件加载时手动校验为非负安全整数）会额外抑制距最近一次注入不足该窗口的注入。
+
+### 文本
+
+```text
+tmux location (turn <turn>):
+session <session>, window <index> "<name>", pane <index> <pane-id>
+window active=<0|1>, pane active=<0|1>, layout <window-layout>
+```
+
+轮次前缀是易变的首行；其下的两行状态块才是变化抑制所比较的单元，因此重新注入由 tmux 状态驱动，而非循环位置。
+
+### 持久性与请求重建
+
+每条读数在被压缩遮蔽前都是普通表层节点；插件对系统提示装配毫无贡献，`request/header` 也不携带任何 tmux-context 文本。读数记录的是一次准备尝试，而非已提交的 step：由于前置监听器最先运行，当后续 `agent/step` 监听器取消或失败时其追加可能仍会保留，只追加的日志不做回滚。
+
+发布的 `./invariant` 伴生插件不注册任何运行时检查：读数是外部 tmux 状态的按轮快照，会话中不存在需要校验的跨事件关系，调度与格式由本包的管线测试固定。
+
+## 后果
+
+启动于 tmux 内的 agent 现在会以持久、带来源标记的上下文收到自身的 session/window/pane 位置及 window 布局，并在位置变化时按轮次更新。部署方通过 cordis.yml 选择启用；默认 spine 与随附示例保持沉默。在真实 tmux pane 之外——包括仅继承了 `$TMUX`／`$TMUX_PANE` 的终端——或没有 `ctx.bash` 执行器时，插件保持惰性且不报错，因此在任何地方组合它都安全。由于读数是一条持久的 `user/message`，它作为普通历史经受压缩，对系统提示装配与请求头毫无贡献，且每个发生变化的轮次至多花费一条两行消息。拉取模型在每个到期轮次的第一个 step 增加一次 `tmux display-message` 子进程（经沙箱化的 bash seam）。可选的间隔下限在查询之前检查，因此同时抑制查询与注入；而位置未变化只能通过比较查询返回的状态得知，因此它只抑制注入，查询开销仍会付出。
+
+## 测试
+
+单元测试固定了：首个 step 的注入及来源/表层元数据；以 `$TMUX_PANE` 为键的命令（含其 `#{pane_tty}` 与 `ps -o tty=` 的比对守卫）；step 门槛；跨轮次的变化抑制与 pane 移动时的重新注入；正间隔抑制与阈值；每条空操作路径（无 bash、非零退出、字段数不符、pane id 为空、信号已取消，以及 `resolve()` 或 `run()` 抛出的执行器拒绝被兜住并记录警告而非使该轮失败）；前置排序先于普通 `agent/step` 监听器；对损坏的历史读数（非文本块、单行文本）的容错；以及配置对负值与非整数间隔的拒绝。逐文件覆盖率为 100%。
+
+## 考虑过的替代方案
+
+- **由 tmux hook / 后台监视器推送**——否决：需要在用户的 tmux 中安装 hook，并引入带清理的后台进程，只为换取按轮次上下文并不需要的步内新鲜度。
+- **每个 step 都运行**——否决：位置在一轮内稳定；重复查询只增加 token 成本而无新信息。以 `step === 1` 为门槛得到按轮次读数。
+- **裸 `child_process`**——否决：绕过沙箱/策略 seam，并手写 `ctx.bash` 执行器已拥有的子进程代码。
+- **包含 pane/window 像素尺寸**——否决：尺寸每次缩放都变动、徒增噪声；布局树已传达结构。
+- **用 `capture-pane` 抓取相邻 pane**——否决：庞大、嘈杂且涉及隐私；超出"自身位置"范围。
+- **动态系统提示区块**——否决：替换某个值会抹去支撑先前推理的历史读数且不可重建；单条持久且带来源的消息在每个位置变得可见时予以记录。
+- **信任 `$TMUX_PANE`（或 `$TMUX`）存在即可**——否决：两者都会被从 tmux shell 启动的终端（VS Code 集成终端）继承，于是非 pane 进程会注入陈旧位置。pane 的 `#{pane_tty}` 与本进程控制终端的比对才是决定性检查。
+- **对已知终端模拟器设黑名单（如 `TERM_PROGRAM=vscode`）**——否决：名单不完整且会不断增长，仍会漏掉其他启动器；tty 比对精确且与启动器无关。
+- **用运行时 invariant 校验每条读数的轮次、位置与格式**——最初随包发布，随后移除：它从日志中重新推导生产者自身的调度，并对同一个包刚刚渲染出的文本断言正则，因此只是重述 `apply()`，而非检查一条独立关系。它能报出的每种失败都必须先修改本包，而这些本包的管线测试已经覆盖。仅当出现插件自身并不计算的关系时才重新引入伴生检查——例如读数将来具备可被其他包破坏的跨轮次顺序或包裹义务。
diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml
new file mode 100644
index 0000000000..6e9974fdce
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md
+2026-07-27-trajectory-inspection-ledger.md: 8c2a7c42b7898776de5d42459b09c0fb1737ec0b
+2026-07-27-trajectory-inspection-ledger.zh.md: 6c5733046dc2cfd3fd2bd005f4cf5c2d2bd110af
diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md
new file mode 100644
index 0000000000..8c2a7c42b7
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md
@@ -0,0 +1,43 @@
+# Agent Note: Trajectory inspection ledger
+
+Status: implemented
+
+English | [中文](2026-07-27-trajectory-inspection-ledger.zh.md)
+
+## Problem
+
+Trajectory has to make prose, machine payloads, token usage, timing, and nested tool activity readable in the same viewport. The earlier stacked Turn and Step cards preserved hierarchy but spent too much vertical space on repeated chrome, while a completely flat table would erase the causal structure that makes a trajectory useful. Role colors also risked borrowing success and warning semantics, which made visual decoration indistinguishable from runtime state.
+
+## Decision
+
+**Render a compact, turn-aware event ledger with a local record inspector, using the existing DeepSeek design system.**
+
+- The ledger keeps session events in sequence within rewind-delimited branches. Turn boundaries use a slightly heavier rule, the raw Turn id, and a continuous left rail; Request boundaries appear as small points integrated into that structure and use one chronological numbering space across ordinary and compaction requests.
+- Event kind and content form the two stable columns. Role tags align toward the content, nested subtools receive a small indentation, and CSS truncation preserves the available preview width. Token usage and duration stay in the inspector.
+- Product prose uses the existing sans stack. Turn ids, token counts, durations, tool calls, raw payloads, and other machine data use the existing code stack.
+- Existing theme tokens own both light and dark rendering. Neutral borders and surfaces form the structure; distinct low-emphasis role hues support scanning without carrying success or failure meaning, while business blue identifies selection, links, and focus.
+- The client runtime exposes a read-only history source independent from Session and SessionManager. Each activated source owns its raw entries, paging, live gap repair, and reconnect rebuild; the ordinary conversation snapshot remains the folded Chat projection. Trajectory subscribes to that source, exhausts its paging only while mounted, and lazily derives its event order, context lineage, schema index, and Requests instead of imposing those structures on every conversation consumer.
+- Ordinary generation and compaction calls form one chronological Request projection, distinguished by purpose rather than separate collections. Effective prompt state and its change ride the Request that introduced them; compaction and prompt changes are not independent inspection entities. Complete history makes global Request numbering and cumulative usage session-wide rather than tail-window-relative.
+- Call schemas come from the active recorded Request header. Keyless snapshot fixtures deliberately replace that catalog with the non-array `{{tools}}` token, which the durable inspection boundary treats as unavailable instead of attempting to project or fabricate schemas.
+- Selecting a record or Request opens an inspector inside Trajectory. Tabs and Summary sections follow the selected entity: Markdown messages expose rendered, source, provenance, and hierarchy views; tools add JSON payload/result and schema views; Requests add options, usage, timing, and result navigation. Images render as media rather than serialized data.
+- Turn folding removes all rows after its first record and replaces them with a compact step/tool-call count; Assistant folding applies the same interaction to its tool-call descendants. Global controls fold or expand both levels.
+- The separate Waterfall tab is removed. A fixed Overview above the ledger projects every record with known `startedAt` onto three semantic timing lanes using its own duration. Dragging left or right commits an inclusive interval filter: any record whose active interval overlaps either boundary remains visible, records without known timing leave the focused ledger, and clearing the selection restores the full branch. The Overview keeps the full time domain while focused so the selection can be resized or cleared without losing orientation.
+- This local inspector remains independent from the conversation-wide Chat details column. At narrow widths it overlays the ledger and remains dismissible by keyboard or pointer.
+
+## Alternatives considered
+
+**Copy Vite DevTools fonts, colors, glass surfaces, or component shapes.** Rejected: those choices express a different product identity. The implementation only adopts the transferable method: neutral structure, semantic accents, machine-data typography, dense scanning, and shadows reserved for floating layers.
+
+**Keep one card per Turn and Step.** Rejected: repeated card chrome reduced the number of visible records and made cross-step comparison slower.
+
+**Flatten every record without Turn or Request boundaries.** Rejected: a trajectory is not merely a log stream; those boundaries preserve the causal structure without consuming dedicated rows.
+
+**Reuse the global Chat details column.** Rejected: it would couple local inspection to conversation navigation and make a row click unexpectedly change another view's state.
+
+**Keep timing in a separate Waterfall tab.** Rejected: the placeholder summarized node counts rather than record timing and forced users to switch away from the rows they wanted to focus. A full-domain Overview keeps timing and filtered records in one visual context.
+
+**Change global theme tokens to match the reference.** Rejected: the existing theme already provides paired light and dark semantic layers, and a local redesign does not justify changing unrelated surfaces.
+
+## Consequences
+
+Trajectory shows more useful records per viewport while retaining Turn and Request orientation. Context rewrites and compactions remain inline with their surrounding history, while a rewind begins a successor branch that inherits only the retained prefix. The main ledger omits token usage and duration so content receives the available width; the local inspector exposes those facts together with full payloads, provenance, schemas, and request timing. The Overview uses recorded start/duration facts without fabricating live elapsed time, and its inclusive focus behavior matches the interaction users already know from Chrome DevTools Network. Focused component tests pin projection, folding, record and interval selection, entity-specific tabs, and running/error semantics; the assembled Web snapshot pins the ledger, Overview, and inspector through the real client composition.
diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md
new file mode 100644
index 0000000000..6c5733046d
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md
@@ -0,0 +1,43 @@
+# Agent Note：轨迹检查记录表
+
+Status: implemented
+
+[English](2026-07-27-trajectory-inspection-ledger.md) | 中文
+
+## 问题
+
+轨迹视图需要在同一视口内清晰呈现正文、机器载荷、token 用量、计时数据和嵌套工具活动。此前堆叠式的轮次与步骤卡片虽然保留了层级，却在重复界面框架上耗费了太多垂直空间；完全扁平化的表格又会抹去因果结构，而这种结构正是轨迹视图的价值所在。角色配色还可能借用成功与警告语义，使视觉装饰与运行时状态无法区分。
+
+## 决策
+
+**使用现有 DeepSeek 设计系统，渲染保留轮次结构的紧凑事件记录表，并提供局部记录检查器。**
+
+- 记录表在以 `rewind` 划分的分支内按会话事件顺序展示。轮次边界由稍粗的分割线、原始轮次 id 和连续的左侧竖线表示；请求边界以融入该结构的小圆点表示，普通请求与压缩（compaction）请求在整个时间序列中共用一套编号。
+- 事件类型与内容构成两个稳定列。角色标签朝内容侧对齐，嵌套子工具略微缩进，内容预览使用 CSS 截断以适应可用宽度。token 用量和耗时留在检查器中。
+- 产品正文使用现有无衬线字体栈。轮次 id、token 数、耗时、工具调用、原始载荷和其他机器数据使用现有代码字体栈。
+- 现有主题 token 同时负责亮色和暗色渲染。中性边框与表面构成整体结构；区分度较低的角色色帮助扫读而不表达成功或失败语义，业务蓝色则标识选择状态、链接和焦点。
+- 客户端 runtime 提供独立于 Session 和 SessionManager 的只读历史数据源。每个已激活的数据源自行拥有原始条目、分页、实时缺口修复和重连重建；普通会话快照仍然只是 Chat 所需的折叠投影。Trajectory 订阅该数据源，仅在挂载期间补齐全部历史，并按需派生事件顺序、上下文谱系、schema 索引和请求，避免让所有会话消费者承担这些结构。
+- 普通生成调用与压缩调用形成一条按时间排序的请求投影，以 purpose 区分而不是放入不同集合。生效的提示词状态及其变化附着在引入它们的请求上；压缩和提示词变化都不是独立检查实体。完整历史使全局请求编号和累计用量以整个会话为范围，而不是相对于末尾窗口。
+- 调用 schema 来自当前生效且已记录的请求头。无密钥快照 fixture（测试前置数据）有意将该目录替换为非数组 token `{{tools}}`，持久化检查边界会将其视为不可用，而不是尝试投影或虚构 schema。
+- 选择记录或请求后，轨迹视图内部会打开检查器，其标签页和概览区域随实体类型变化：Markdown 消息提供渲染、源码、来源和层级视图；工具提供 JSON 载荷／结果和 schema 视图；请求提供选项、用量、计时和结果跳转。图片以媒体形式渲染，而不是显示为序列化数据。
+- 折叠轮次时保留其第一条记录，将后续行替换为紧凑的步骤和工具调用数量；折叠助手时对其工具调用后代应用相同交互。全局控件可以分别折叠或展开这两个层级。
+- 移除独立的 waterfall（瀑布式事件）标签页。固定在记录表上方的 Overview 区域将所有 `startedAt` 已知的记录按各自耗时投影到三条语义计时轨道。向左或向右拖动会提交包含边界的区间筛选：任何活动区间与所选区间相交的记录都会保留，计时未知的记录会从聚焦后的记录表中移除，清除选择则恢复完整分支。聚焦后，Overview 区域仍保留完整时间范围，以便在不失去方位的情况下调整或清除选择。
+- 此局部检查器与会话级 Chat 详情栏相互独立。在窄屏下，检查器会覆盖记录表，并且仍可通过键盘或指针关闭。
+
+## 曾考虑的替代方案
+
+**照搬 Vite DevTools 的字体、颜色、玻璃表面或组件形状。** 不予采纳：这些选择表达的是另一种产品身份。实现仅吸收可迁移的方法，即中性结构、语义强调色、机器数据排版、紧凑扫读，以及只为浮层保留阴影。
+
+**每个轮次和步骤保留一张卡片。** 不予采纳：重复的卡片框架减少了可见记录数量，并降低了跨步骤比较的速度。
+
+**不使用轮次或请求边界，将所有记录完全扁平化。** 不予采纳：轨迹并非普通日志流；这些边界无需占用独立行，也能保留因果结构。
+
+**复用全局 Chat 详情栏。** 不予采纳：这会让局部检查与会话导航耦合，还会使行点击意外改变另一个视图的状态。
+
+**将计时保留在独立的 waterfall 标签页中。** 不予采纳：占位实现汇总的是节点数而非记录计时，并迫使用户离开想要聚焦的记录。保留完整时间范围的 Overview 区域让计时和筛选后的记录处于同一视觉上下文中。
+
+**修改全局主题 token 以匹配参考设计。** 不予采纳：现有主题已经提供配对的亮色与暗色语义层，局部重新设计不足以成为修改无关表面的理由。
+
+## 后果
+
+轨迹视图在保留轮次与请求定位的同时，每个视口可以显示更多有效记录。上下文 `rewrite` 与压缩保持在周边历史中的原始位置，`rewind` 则建立仅继承保留前缀的后继分支。主记录表省略 token 用量和耗时，让内容获得可用宽度；局部检查器展示这些数据以及完整载荷、来源、schema 和请求计时。Overview 区域使用记录的开始时间与耗时数据，而不虚构实时流逝时间，其包含边界的聚焦行为与用户熟悉的 Chrome DevTools Network 交互一致。针对性组件测试锁定投影、折叠、记录与区间选择、实体特定标签页和运行／错误语义；组装后的 Web 快照则通过真实客户端组合锁定记录表、Overview 区域与检查器。
diff --git a/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.i18n.yaml b/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.i18n.yaml
new file mode 100644
index 0000000000..d9ed8345dc
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-28-web-terminal-card.md
+2026-07-28-web-terminal-card.md: 14896b1d88e5cfd2e4c58830c7a1bca1e54ed823
+2026-07-28-web-terminal-card.zh.md: 16c9004f8f80b720b25b76ba5c04f308b0fccbaf
diff --git a/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md b/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md
new file mode 100644
index 0000000000..14896b1d88
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md
@@ -0,0 +1,70 @@
+# Agent Note: Web terminal card — the bash render intent reaches the browser
+
+Status: implemented
+
+English | [中文](2026-07-28-web-terminal-card.zh.md)
+
+## Problem
+
+The bash tool declares `card: 'terminal'` for both its call and its result ([render-intent union](../architecture/2026-07-02-tool-render-intent-union.md)): the call view carries the command, an optional model-authored description, and the working directory; the result view carries the output, exit code, and terminating signal. That view already reaches the browser — host, connection, and runtime deliver it onto `ConversationSnapshot` as `callView`/`resultView` — and the TUI already renders it as a `$`-prompt card with an exit line and a head/tail height cap.
+
+The Web client ignored it. `packages/client/ui-conversation/src/client/contract/tool-call-model.ts` derived every row from raw tool args, and `skeleton/DetailsPanel.tsx` flattened every tool's content blocks into one `<pre>` with `white-space: pre-wrap; word-break: break-word`. Two defects followed from soft-wrapping and from having no height bound: multi-column output (`ls`, a table, box drawing) folded into a paragraph and lost the column alignment that is the whole point of that output, and a long single-column listing stretched the details panel to the length of the listing.
+
+## Decision
+
+`TerminalBlock` is a `ui-primitives` component that renders a shell command as a terminal surface, and both Web render sites for a bash call consume the terminal render intent through it: the chat tool row's expanded body and the details panel's Output section. `ui-conversation/src/client/contract/terminal-card-model.ts` is the single place that turns the snapshot's `callView`/`resultView` pair into the component's props, so the two sites cannot disagree about a command, its cwd, or its exit status. It returns null — the generic path — whenever neither side declares `card: 'terminal'`, including a `card` value this client version does not know, and whenever a settled call's result view is generic, which is how the bash tool's execution errors and background starts keep their existing rendering. Two duties the render-intent contract assigns to the UI bridge land here rather than in the tool: a settled result's `title` REPLACES the pending one, and the working directory resolves against the session workspace — an absolute view cwd is used as-is, a relative one joins under the workspace, and an omitted one IS the workspace, which is the common case for a bash call with no `workdir`. A pure presenter cannot see the session cwd, which is why the resolution belongs at this seam; each render site supplies the cwd off the session list row. Only a PRESENT call view can mean "omitted, so use the workspace": when the paging window drops the call head there is no cwd anywhere — the result view carries none — and the original call may have used an explicit workdir, so the prompt draws a bare `$` rather than naming a directory it cannot know. The resolved path also normalizes its `.`/`..` segments, because the bash executor resolves the workdir before running: a `..` against `/w/app` runs in `/w`, so the prompt label has to read `w` rather than `..`. A UNC path's `server` and `share` are part of its root rather than poppable segments, since Windows cannot climb above a share. The call view's `description` rides the same derivation, since the contract renders it above the card and it must outrank the row's args-derived summary. All three render sites draw it: both chat-row shapes and the details panel. An expanded row draws it itself, because the collapsed summary is hidden while a row is open — without that the description would only ever be visible collapsed, which is the opposite of what "above the card" means.
+
+The component's contract:
+
+- **Prompt lines, one per command line.** Each line of the command gets its own row: label, then that line verbatim. A `command` carrying two shell commands on two lines therefore reads as the two commands it is, instead of collapsing into one ellipsized row. The label is the cwd's last path segment, or `~` when the cwd equals the `home` prop — a browser has no `$HOME`, so the caller supplies the absolute home directory and the collapse simply does not apply without it. A view with no cwd renders a plain `$`. A trailing newline is a terminator, not an empty final command. Only the FIRST row carries the label: the view knows one working directory — where the call started — and a later line may run somewhere else entirely, since a `cd` in the command is enough to move it. Repeating the label down the rows would state a directory per line that nothing here knows, which is the same reason the run-state dot appears once. Later rows keep a bare `$` so they still read as prompts.
+- **One run-state dot for the call, on the first row.** `StateDot` in three of its four states: the chase while running, red for the exit status that also renders the pill, green for a clean settle — the same indicator a tool row's leading icon uses, so a row and its own card cannot disagree about one command. The dot exists because the first question a reader has about a shell command is whether it is still running, and without it that had to be inferred from the absence of output — which a settled command producing no output also looks like. It sits out of flow in a gutter the card reserves as its OWN left padding, so it neither indents its command nor depends on the command's text metrics to line up. The reservation is padding rather than margin because every render site rewrites `margin` wholesale to set its own indent, which silently cancelled a margin-based gutter and let a container clip the dot. Exactly one dot, whatever the line count: the exit status the view carries is the whole call's, and bash reports no per-command status, so a dot per line would assert of a line that succeeded inside a failing call that the line itself failed. The single visually hidden text label carries the same scope, since `StateDot` is `aria-hidden` and one label per row would read to assistive technology as several distinct outcomes.
+- **No soft wrapping.** Output lines are `white-space: pre` inside a horizontally scrolling box. Column alignment survives; a long line scrolls instead of folding.
+- **Height cap with an expand control.** Output longer than `DEFAULT_TERMINAL_MAX_LINES` (16) lines shows `ceil(max/2)` head lines plus the remaining tail lines, with a button in between that reports the hidden count and expands. The count is of parsed lines after the trailing output terminator is dropped, so an N-line output ending in a newline is N lines. The split arithmetic is the same as the TUI transcript's collapsed tool card (`packages/ui/tui/src/components/transcript.ts`), so one command's head and tail slices agree between the two front ends.
+- **ANSI color.** `anser` splits the SGR runs; `ui-primitives/src/ansi.ts` resolves each run into an inline style rendered as React spans. A foreground-only run maps the basic 16 colors onto `--dsw-*` theme tokens so authored color stays legible under both themes; a run that paints its own background keeps anser's literal rgb for both so its intended contrast survives, as do 256-palette, truecolor, and the two basic colors this design system has no token for. Sequences that carry no color (OSC strings, non-CSI escapes, inert C0 controls) are stripped before parsing so they never reach the DOM as literal characters. Cursor movements resolve before that strip, into a per-line column buffer rather than by string surgery, because carriage return and backspace only MOVE the cursor — neither erases anything, so what a reader sees is whatever each column last had written to it. `100%` then a carriage return and `OK` shows `OK0%`, since the redraw is shorter than the frame beneath it; a trailing `abc` plus a backspace still shows `abc`, since nothing overwrote the `c`; `abc` plus two backspaces and `XY` shows `aXY`. Each of these was checked against a real terminal, because the earlier truncate-and-delete approximations looked right and were not. SGR state is stamped per column as a terminal stores it per cell, so a partial overwrite keeps each surviving character's own color: red `bad`, three backspaces, then `ok` shows `okd` with the `d` still red. A CSI sequence occupies no column and changes only the state later writes are stamped with, which is also why a carriage return does not reset color, and why SGR state threads from one line to the next rather than closing at each newline. Erase-in-line is part of the same replay, because `\r\x1b[K` is the single idiom every spinner and progress bar writes — modelling the `\r` alone left the previous frame's tail standing, which is text the terminal never showed. Only `m` accumulates into a cell's style; a cursor or erase sequence must not, or the state string grows per redraw and emits boundaries anser has to discard. SGR is held per cell as a NORMALIZED record (foreground, background, attribute set), not as the sequence history: accumulating raw sequences made every state boundary re-emit the whole chain, so output that switches color without a full reset emitted O(n^2) characters — 3200 such cells produced 25 MB and a `RangeError` well under bash's own output cap. The record also lets the attribute closers every chalk-based tool writes (`39`, `49`, `22`, `24`, …) actually close their attribute, and each boundary emits one canonical sequence for the state it opens. A run also has to CLOSE: the replay converges to the state the scan ended in, not the last written cell's, because a reset after the final write changes no cell yet ends the run — without that a line finishing in `\x1b[0m` leaked its color onto every later line. The cursor advances by terminal columns, so a tab reaches the next 8-column stop, a wide character takes two (its spacer blanking rather than closing the gap once the lead cell is overwritten), and a combining mark takes none. Width follows emoji PRESENTATION rather than the U+2600-U+27BF block: `\u2713`, the check every progress line writes, is one column, so treating the block as wide misaligned exactly the output this card exists for. Writing over either half of a wide pair blanks the other, since a terminal cannot leave one cell of a two-cell glyph standing: `a\tb` then a redraw of `XY` shows `XY      b`, since a two-character redraw cannot reach column 8.
+- **Exit status and copy.** A non-zero exit code or a signal renders a status pill, matching the exit-status distinction the bash tool's own renderer draws; a clean exit renders none, and settled empty output renders a dimmed placeholder — judged on the parsed lines the card renders, not on the raw text, since output that is only escapes or control bytes survives a `trim()` yet parses to nothing visible and would otherwise draw blank rows plus a copy control for invisible bytes. The copy control copies the raw output text, not the rendered tree, so the prompt line and the pill stay out of the clipboard.
+
+Geometry, radius, and fonts mirror `CodeBlock`, so a terminal card and a fenced code block match visually; `white-space: pre` plus horizontal scroll is the deliberate divergence. The clipboard write both components need moved out of `CodeBlock` into a package-internal `src/clipboard.ts`, unexported so it stays an implementation detail of the two blocks.
+
+### Inline output in the chat row reverses a stated convention
+
+`chat/ToolRow.tsx` and `contract/tool-call-model.ts` asserted "no inline output ever — full results live in the details panel". Showing the terminal block in the row reverses that, on the owner's explicit decision.
+
+The reason the reversal holds: for a shell command the output *is* the result the user is reading, so routing it exclusively to a panel makes the common case a two-step interaction. A bounded, height-capped, non-wrapping terminal block in the row is what makes a bash-heavy transcript readable in one pass. The old rule's actual concern was a row whose height was unbounded by the length of the output, and the height cap plus expand control is what keeps that from returning.
+
+The remaining bound: the row caps at `CHAT_TERMINAL_MAX_LINES` (8), half the primitive's default, which the panel keeps — the message flow is a summary surface read across many calls, the panel is the single-call reading surface. Only the terminal intent renders inline; a generic tool's content is still panel-only.
+
+One premise of that split has since weakened: [tool rows stopped being details-panel click targets](2026-07-28-tool-call-file-open-in-os.md) and nothing replaced the gesture, so the panel is currently unreachable in the assembled application. The in-row cap is therefore the only surface a reader actually has for a long output, which the expand control covers. Restoring a panel entry point is that change's follow-up, not this one's — but until it lands, "the panel stays the place for the full output" is not true, and the row's expand control carries that load alone.
+
+## Alternatives considered
+
+**Render the terminal block only in the details panel.** This keeps the stated no-inline-output convention and needs no reversal to record. Rejected by the owner's explicit decision: a shell command's output is what the user came to read, and putting it one click away costs more than the convention buys. Recorded here as the owner's call, not as a conclusion derived from the codebase.
+
+**Reuse `CodeBlock` with a `console` language instead of a new primitive.** Rejected: `CodeBlock` soft-wraps, which is the defect being fixed, and it has no exit status, no cwd prompt line, no height cap, and no ANSI handling. Adding four terminal-specific concerns to the shared code-fence component would impose them on every markdown fence. The two components share their geometry and font tokens instead, which is the only part where one implementation is correct for both.
+
+**Hand-roll the SGR parser.** Rejected: an SGR parser is exactly the surface [prefer maintained dependencies over hand-rolling](../process/2026-07-26-dependencies-over-hand-rolling.md) says not to own — its edge cases (256-palette and truecolor forms, `reverse`, multi-parameter runs, unterminated sequences) each fail on output nobody produces in a test, so a hand-rolled version stays subtly wrong for a long time. Stated honestly against that policy's bar: `anser` does **not** delete existing owned code. It is a capability addition, which that note distinguishes from a net-deletion simplification; the health and boundary-fit halves of the bar are what it clears. What stays hand-rolled is the part `anser` does not cover: the theme-token color mapping, the non-CSI sanitizing, the carriage-return redraw, and the per-line span folding the height cap slices.
+
+## Consequences
+
+`anser` is a new runtime dependency of `packages/client/ui-primitives`, so every consumer of that package pays for it once. A bash row in the Web chat carries output, which is a deliberate density increase over a summary-only row; the cap is what keeps it bounded, and a tighter cap is a props change, not a redesign.
+
+`TerminalBlock` reads only the terminal view's fields, so it stays a pure function of what the render intent carries — no session lookups, replay-safe like the presenters that produce the view. A UI without the terminal capability still gets the bridge's fenced fallback; nothing about the tool's result shape changed.
+
+A `run_code` sub-dispatch does not reach a terminal card on the shipped wire: `session.ts` folds `tool/code-dispatch(-start)` with `callView: null`/`resultView: null`, and the host's `viewFor` presents only top-level `tool/call`/`tool/result`, so a nested bash call keeps the generic flattened form. Both arms are pinned — the resolution path with views injected, and the no-view shape the wire actually delivers — so the gap is recorded rather than implied. Carrying presenter views through the code-dispatch wire is that seam's own change.
+
+Inline rendering is licensed for the terminal intent alone. A future intent that wants it needs its own bound and its own decision, argued against the reason recorded here rather than against the panel-only convention on its own.
+
+## Testing
+
+`packages/client/ui-primitives/tests/ansi.spec.ts` pins the parse layer: token mapping for the basic colors, literal rgb for the values with no token, the background-run pair, every decoration and the `textDecoration` collision between two of them, the sanitizing of OSC strings and non-CSI escapes and inert controls, the cursor replay (redraws leaving a longer frame's tail standing, a trailing backspace erasing nothing, erase-in-line in all three parameter forms, tab stops, wide characters, SGR threading across lines, and a cursor/erase sequence never entering a cell style), and CRLF preservation. Each replay case was checked against a real terminal first. `packages/client/ui-primitives/tests/terminal-block.spec.tsx` pins the component: cwd shortening, the running/empty/settled arms, signal outranking exit code, the trailing-newline terminator rule, the head/tail cap with its `aria-expanded` toggle, the run-state dot across all three reachable states plus its position ahead of the prompt label, the one-row-per-command-line prompt and its single dot on the first row, and the copy control asserting raw output on both the accepted and refused clipboard paths, plus `writeClipboard` directly.
+
+`packages/client/ui-conversation/tests/terminal-card.spec.tsx` pins the wiring at every render site: `terminalCardModel`'s derivation and each of its null arms, the result title replacing the pending one, the cwd resolving against the session workspace across all four of its cases, the panel resetting the card's expand state when the selection changes, the chat row's expand-gated body against the panel's full-height one, `BashRow`'s resident card and its agreement with its own summary row's state dot, and the panel's Output section including the run_code sub-dispatch and the out-of-window head. That file is written against no gate pressure — `packages/client/ui-conversation/src/*` sits on the coverage `exclude` list in `vitest.config.ts`, so a coverage run over this package measures none of these files.
+
+`apps/web/tests/terminal-card.snapshot.ts` pins the assembled application over the built client bundles: the same render intent at both conversation render sites and in both chat-row shapes, because a bash call reaches a resident card only through the keyed `BashRow` registration and every other terminal-declaring tool name lands on the render-site fallback row, whose body is expand-gated. Fixture turn 65 was named `bash` and turn 60 left as `fx-bash` so one fixture covers both shapes, and turn 60's command was made two lines so the built-bundle snapshot pins the per-line prompt and its single dot (`dotsPerPromptRow: [1, 0]`). That terminal turn is ordered BEFORE the todo turn on purpose: the standing plan retires at the next `turn/start`, so appending it after would have emptied the dock's plan strip and taken the todo surfaces' own coverage with it; that turn also carries what turn 60's two prompt rows cannot — SGR runs resolved to `--dsw-*` tokens, output past the chat cap, a nested cwd, and a non-zero exit authored beside the sample. The sample's body deliberately carries NO `[exit code: N]` line: the real bash presenter consumes that marker out of the body precisely because the card shows the exit as its own pill, so leaving it in would pin a frame showing the exit twice — one the product path cannot produce.
+
+`apps/web/tests/navigation-panes.e2e.ts` adds the real-browser scenario over its existing `echo NAVIGATION_OK` bash call, asserting what jsdom cannot compute: squeezing the output pane below its content width leaves the line at one row and gives the pane horizontal overflow, the run-state dot resolves to the green success token rather than to a literal color (a `--dsw-*` var has no computed value at all without the real theme stylesheet) and sits inside the card box yet left of the prompt label, which is the invariant the gutter padding owns, and the copy control reaches the page's own async Clipboard API rather than the `execCommand` fallback. Its `terminal-card.expected.md` golden records the resolved workspace in the prompt row, which is what a bash call with no `workdir` must show instead of a bare `$`.
+
+## Related
+
+- [Tagged render-intent union for tool-call presentation](../architecture/2026-07-02-tool-render-intent-union.md) — the `card`-tagged vocabulary this consumes; the Web client is now a full consumer of the `terminal` arm rather than of args alone.
+- [Web client syntax highlighting](../process/2026-07-26-web-syntax-highlighting-shiki.md) — owns `CodeBlock` and its shiki arm, and records why tool output deliberately stays unhighlighted; ANSI color here is authored color, not guessed grammar.
+- [Web client architecture](../architecture/2026-07-19-gui-web-client-architecture.md) — the slot and snapshot layering the two render sites sit in.
diff --git a/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.zh.md b/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.zh.md
new file mode 100644
index 0000000000..16c9004f8f
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-28-web-terminal-card.zh.md
@@ -0,0 +1,70 @@
+# Agent Note: Web terminal card — the bash render intent reaches the browser
+
+Status: implemented
+
+[English](2026-07-28-web-terminal-card.md) | 中文
+
+## Problem
+
+bash 工具的调用与结果都声明 `card: 'terminal'`（[渲染意图联合类型](../architecture/2026-07-02-tool-render-intent-union.md)）：调用视图携带命令、一段可选的模型撰写描述以及工作目录，结果视图携带输出、退出码与终止信号。该视图早已抵达浏览器——host、connection 与 runtime 把它投递到 `ConversationSnapshot` 的 `callView`/`resultView` 上——TUI 也早已把它渲染为带 `$` 提示符的卡片，附退出行与首尾高度上限。
+
+Web client 却对它视而不见。`packages/client/ui-conversation/src/client/contract/tool-call-model.ts` 仅从原始工具参数推导每一行，`skeleton/DetailsPanel.tsx` 则把所有工具的内容块压平进一个 `<pre>`，样式为 `white-space: pre-wrap; word-break: break-word`。软换行加上没有高度约束，带来两个缺陷：多列输出（`ls`、表格、制表符绘图）被折成一段文字，丢掉了这类输出赖以存在的列对齐；而单列的长列表会把详情面板拉长到与列表等长。
+
+## Decision
+
+`TerminalBlock` 是 `ui-primitives` 中把 shell 命令渲染为终端表面的组件，bash 调用在 Web 侧的两个渲染点都经由它消费 terminal 渲染意图：聊天工具行展开后的正文，以及详情面板的 Output 区。`ui-conversation/src/client/contract/terminal-card-model.ts` 是把快照上的 `callView`/`resultView` 这一对转换为该组件 props 的唯一位置，因此两个渲染点不可能在命令、cwd 或退出状态上产生分歧。当两侧都不声明 `card: 'terminal'` 时它返回 null，即走 generic 路径——包括本 client 版本不认识的 `card` 取值；当一个已落定调用的结果视图是 generic 时同样返回 null，这正是 bash 工具的执行错误与后台启动得以保持既有渲染的方式。渲染意图契约交给 UI 桥接层的两项职责也落在这里，而不在工具侧：已落定结果的 `title` **替换**待定标题；工作目录针对会话 workspace 解析——视图给出的绝对路径原样使用，相对路径在 workspace 之下拼接，省略则**就是** workspace，而这正是不带 `workdir` 的 bash 调用的常见情形。纯 presenter 看不到会话 cwd，因此该解析属于这道接缝；两个渲染点各自从会话列表行取出 cwd 传入。只有**存在**的调用视图才能表示「省略了 cwd，因此取 workspace」：当分页窗口丢掉调用头时，任何地方都不再有 cwd——结果视图并不携带它——而原调用完全可能使用过一个显式 workdir，因此提示行绘制一个裸 `$`，而不是命名一个它无法知晓的目录。解析后的路径还会归一化其 `.`／`..` 段，因为 bash 执行器在运行前就已解析 workdir：相对 `/w/app` 的 `..` 实际运行在 `/w`，因此提示标签必须读作 `w` 而不是 `..`。UNC 路径的 `server` 与 `share` 属于其根，而非可弹出的路径段，因为 Windows 无法越过一个共享向上。调用视图的 `description` 走同一处推导，因为契约把它渲染在卡片上方，且它必须优先于该行由参数推导出的摘要。三个渲染点都会绘制它：两种聊天行形态与详情面板。展开后的行自行绘制它，因为一行处于展开态时其折叠摘要是隐藏的——否则该描述将只在折叠时可见，这与「位于卡片上方」的含义正好相反。
+
+该组件的契约：
+
+- **提示符行，每条命令行一行。** 命令的每一行各占一行：标签，其后原样跟随该行。因此一个在两行上承载两条 shell 命令的 `command` 就读作它本身的两条命令，而不是被压成一行并省略号截断。标签取 cwd 的最后一段路径，当 cwd 等于 `home` prop 时取 `~`——浏览器没有 `$HOME`，因此由调用方提供绝对家目录，不提供时该折叠不生效。视图不带 cwd 时渲染一个纯 `$`。末尾换行是终止符，不是一条空的末命令。只有**第一行**携带该标签：视图只知道一个工作目录——调用开始处的那个——而后面的行完全可能在别处运行，命令里一个 `cd` 就足以改变它。把标签在各行重复，等于陈述一个此处无人知晓的逐行目录，这与运行状态点只出现一次是同一个理由。其余行保留一个裸 `$`，因此它们仍读作提示符。
+- **整次调用一枚运行状态点，位于第一行。** 它是 `StateDot` 四种状态中的三种：运行期间为追逐动画，与渲染状态徽章相同的退出状态为红色，干净落定为绿色——与工具行行首图标使用同一个指示器，因此一行与其自身的卡片不可能对同一条命令产生分歧。该状态点存在的理由是：读者对一条 shell 命令的第一个问题就是它是否仍在运行；没有它时，这一点只能从「没有输出」推断，而一条落定后无输出的命令看起来也一样。它以脱离文档流的方式落在卡片以**自身左内边距**预留的落区里，因此既不会缩进其命令，也不依赖命令自身的文本度量来与之对齐。该预留用 padding 而非 margin，是因为每个渲染点都会整条重写 `margin` 来设定自己的缩进——那会静默取消基于 margin 的落区，并让容器把状态点裁掉。无论有多少行，都只有一枚：视图携带的退出状态属于整次调用，而 bash 不报告逐条命令的状态，因此每行一枚状态点就等于在断言——一条在失败调用中其实成功了的命令行自身失败了。那一处视觉隐藏的文本标签具有相同的作用域，因为 `StateDot` 是 `aria-hidden`，而每行一个标签会被辅助技术读成好几个各自独立的结果。
+- **不软换行。** 输出行使用 `white-space: pre`，置于横向滚动的容器内。列对齐得以保留；长行滚动，而非折行。
+- **高度上限与展开控件。** 输出超过 `DEFAULT_TERMINAL_MAX_LINES`（16）行时，显示 `ceil(max/2)` 行首部加余下的尾部行数，中间是一个按钮，报告被隐藏的行数并可展开。计数针对的是剥除输出末尾终止符之后解析出的行，因此以换行结尾的 N 行输出就是 N 行。切分算法与 TUI transcript 折叠态工具卡片（`packages/ui/tui/src/components/transcript.ts`）完全一致，因此同一条命令的首尾切片在两个前端之间吻合。
+- **ANSI 颜色。** `anser` 切分 SGR 分段；`ui-primitives/src/ansi.ts` 把每段解析为内联样式，渲染成 React span。只设前景色的分段把基本 16 色映射到 `--dsw-*` 主题 token，使作者指定的颜色在两种主题下都可读；自行绘制背景的分段则前后景都保留 anser 给出的字面 rgb，以保住它意图中的对比度，256 色板、truecolor 以及本设计系统没有对应 token 的两种基本色同样如此。不承载颜色的转义序列（OSC 串、非 CSI 转义、无显示意义的 C0 控制符）在解析前被剥除，因此绝不会以字面字符抵达 DOM。光标移动在该剥除之前先行结算，且落在逐行的列缓冲里而不是靠字符串手术，因为回车与退格**只移动**光标——两者都不擦除任何东西，所以读者看到的就是每一列最后被写入的内容。`100%` 后接回车再接 `OK` 显示为 `OK0%`，因为这次重绘比它下面的帧更短；末尾 `abc` 加一个退格仍显示 `abc`，因为没有任何东西覆盖过那个 `c`；`abc` 加两个退格再接 `XY` 显示 `aXY`。这些用例都对照真实终端核实过，因为先前「截断加删除」的近似看起来是对的，实际并不对。SGR 状态按列打戳，与终端按单元格存储颜色的方式一致，因此部分覆盖会保留每个存活字符自身的颜色：红色 `bad`、三个退格、再写 `ok`，显示为 `okd` 且那个 `d` 仍是红的。CSI 序列不占列，只改变后续写入被打上的状态——这也正是回车不会重置颜色的原因，以及 SGR 状态会从一行延续到下一行、而不是在每个换行处关闭的原因。行内擦除属于同一次重放，因为 `\r\x1b[K` 是每个 spinner 与进度条都会写的同一个惯用法——只建模 `\r` 会让上一帧的尾巴留在原处，那是终端从未显示过的文本。只有 `m` 会累加进单元格样式；光标或擦除序列不能累加，否则状态串会随每次重绘线性增长，并发出 anser 只能丢弃的边界。SGR 按单元格以**归一化记录**保存（前景、背景、属性集合），而不是序列历史：累积原始序列会让每个状态边界重新发射整条链，因此不做完整 reset 的换色输出会发射 O(n^2) 个字符——3200 个这样的单元格产生 25 MB 并最终 `RangeError`，远低于 bash 自身的输出上限。该记录也让所有 chalk 系工具写出的属性闭合码（`39`、`49`、`22`、`24` 等）真正闭合其属性，且每个边界只为它开启的状态发射一条规范序列。一个分段也必须**收束**：重放收敛到扫描结束时的状态，而不是最后一个被写入单元格的状态——因为最后一次写入之后的 reset 不改变任何单元格，却结束了该分段；没有这一步，以 `\x1b[0m` 结尾的行会把颜色泄漏到其后所有行。光标按终端列推进，因此制表符前进到下一个 8 列制表位、宽字符占两列（其续列在首列被覆盖后变为空白而非合拢），组合标记不占列。宽度依据 emoji **presentation** 而非 U+2600–U+27BF 整个区块：`\u2713`——每条进度行都会写的对勾——只占一列，把该区块整体当作双宽恰好会错位这张卡片赖以存在的那类输出。写入宽字符对的任一半都会把另一半清成空白，因为终端无法让一个双格字形只留下一格：`a\tb` 之后用 `XY` 重绘显示为 `XY      b`，因为两个字符的重绘到不了第 8 列。
+- **退出状态与复制。** 非零退出码或信号渲染一枚状态徽章，与 bash 工具自身渲染器所作的退出状态区分一致；干净退出不渲染徽章，落定后的空输出渲染一处变暗的占位文字——该判定读的是卡片实际渲染的解析行，而非原始文本，因为只含转义或控制字节的输出能通过 `trim()` 却解析不出任何可见内容，否则就会画出一片空行外加一个把不可见字节写进剪贴板的复制控件。复制控件复制的是原始输出文本而非渲染后的树，因此提示符行与徽章不会进入剪贴板。
+
+几何尺寸、圆角与字体沿用 `CodeBlock`，因此终端卡片与围栏代码块在视觉上一致；`white-space: pre` 加横向滚动是有意的分歧。两个组件都需要的剪贴板写入从 `CodeBlock` 中提取到包内部的 `src/clipboard.ts`，不对外导出，因此它仍是这两个块的实现细节。
+
+### 聊天行内嵌输出推翻了一条既有约定
+
+`chat/ToolRow.tsx` 与 `contract/tool-call-model.ts` 都断言过「绝不内嵌输出——完整结果在详情面板」。在行内显示终端块推翻了这一点，依据是 owner 的明确决定。
+
+这次推翻成立的理由：对 shell 命令而言，输出**就是**用户要读的结果，把它专门收进面板会让最常见的情形变成两步交互。行内一个有界、限高、不换行的终端块，正是让 bash 密集的 transcript 一遍读完的条件。旧规则真正担心的是行高不受输出长度约束，而高度上限加展开控件正是防止其复现的机制。
+
+余下的约束：行内上限为 `CHAT_TERMINAL_MAX_LINES`（8），是组件默认值的一半，而面板沿用默认值——消息流是跨多次调用阅读的摘要表面，面板才是单次调用的阅读表面。只有 terminal 意图内嵌渲染；generic 工具的内容依旧只在面板中。
+
+这一划分的一个前提此后被削弱了：[工具行已不再是详情面板的点击目标](2026-07-28-tool-call-file-open-in-os.md)，且没有任何手势接替它，因此该面板在组装后的应用中当前不可达。于是行内上限成为读者实际拥有的唯一长输出表面，由展开控件承担。恢复面板入口属于那次改动的后续，而非本次改动——但在它落地之前，「面板仍是查看完整输出的地方」并不成立，行内的展开控件独自承担了这一职责。
+
+## Alternatives considered
+
+**只在详情面板渲染终端块。** 这样保留既有的「不内嵌输出」约定，也不需要记录任何推翻。已被 owner 的明确决定否决：shell 命令的输出正是用户来读的东西，把它挪到一次点击之外，代价高于该约定带来的收益。此处记录的是 owner 的裁决，而非从代码库推导出的结论。
+
+**复用 `CodeBlock` 并传入 `console` 语言，而不新建组件。** 已否决：`CodeBlock` 会软换行，而软换行正是本次要修的缺陷，且它没有退出状态、没有 cwd 提示符行、没有高度上限、也不处理 ANSI。把四项终端专属关注点加进共享的代码围栏组件，等于把它们强加给每一个 markdown 围栏。两个组件改为共享几何与字体 token，那是唯一一处「一套实现对两者都正确」的部分。
+
+**手写 SGR 解析器。** 已否决：SGR 解析器恰是[优先采用维护良好的依赖而非手写](../process/2026-07-26-dependencies-over-hand-rolling.md)所指明不该自持的那类实现——它的边界情形（256 色板与 truecolor 形式、`reverse`、多参数分段、未终止的序列）各自只在没人会写进测试的输出上失效，因此手写版本会在很长时间内一直微妙地出错。对照那条策略的门槛如实陈述：`anser` **并未**删除任何既有自持代码。它是一次能力增补，而那条 Agent Note 把这与净删除式的简化区分开来；它清过的是健康度与边界契合这两半门槛。`anser` 未覆盖而仍由我们手写的部分是：主题 token 的颜色映射、非 CSI 序列的剥除、回车重绘，以及供高度上限切片的逐行 span 折叠。
+
+## Consequences
+
+`anser` 成为 `packages/client/ui-primitives` 的一项新运行时依赖，因此该包的每个消费方都为它支付一次。Web 聊天中的 bash 行现在承载输出，相比只有摘要的行，这是有意提高的信息密度；上限是维持其有界的机制，而调紧上限是改一个 prop，不是重新设计。
+
+`TerminalBlock` 只读取 terminal 视图携带的字段，因此它始终是渲染意图内容的纯函数——不查会话状态，与产出该视图的 presenter 一样可安全回放。不具备终端能力的 UI 仍从桥接层拿到围栏式回退；工具的结果形态未作任何改动。
+
+在当前已交付的 wire 上，`run_code` 子派发不会得到终端卡片：`session.ts` 把 `tool/code-dispatch(-start)` 折叠为 `callView: null`／`resultView: null`，而 host 的 `viewFor` 只呈现顶层的 `tool/call`／`tool/result`，因此嵌套的 bash 调用保持通用的压平形式。两条分支都已钉住——注入视图后的解析路径，以及 wire 实际投递的无视图形态——因此这个缺口是被记录下来的，而非暗含的。把 presenter 视图贯穿 code-dispatch wire 属于那道接缝自身的改动。
+
+内嵌渲染的许可仅授予 terminal 意图。将来想要内嵌的意图需要有自己的边界与自己的决定，且需针对此处记录的理由来论证，而不是仅针对「只在面板」这条约定本身。
+
+## Testing
+
+`packages/client/ui-primitives/tests/ansi.spec.ts` 固定解析层：基本色的 token 映射、无对应 token 取值的字面 rgb、带背景分段的前后景配对、每一项装饰以及其中两项之间的 `textDecoration` 冲突、OSC 串与非 CSI 转义及无显示意义控制符的剥除、光标重放（较短重绘让上一帧尾巴留存、末尾退格不擦除任何东西、行内擦除的全部三种参数形式、制表位、宽字符、SGR 跨行延续，以及光标／擦除序列绝不进入单元格样式），以及 CRLF 的保留。每一条重放用例都先对照真实终端核实过。`packages/client/ui-primitives/tests/terminal-block.spec.tsx` 固定组件：cwd 缩短、运行中／空／已落定三条分支、信号优先于退出码、末尾终止符规则、首尾高度上限及其 `aria-expanded` 开关、运行状态点全部三种可达状态及其位于提示符标签之前的位置、每条命令行一行的提示区及其位于第一行的单枚状态点，以及复制控件在剪贴板接受与拒绝两条路径上都断言原始输出，另有对 `writeClipboard` 的直接固定。
+
+`packages/client/ui-conversation/tests/terminal-card.spec.tsx` 固定每个渲染点上的接线：`terminalCardModel` 的推导及其每一处 null 分支、结果标题替换待定标题、cwd 针对会话 workspace 解析的全部四种情形、切换选中调用时面板重置卡片展开态、对话行受展开控制的输出体与面板的全高输出体的对比、`BashRow` 的常驻卡片及其与自身摘要行状态点的一致性，以及面板 Output 区段（含 run_code 子派发与超出窗口的调用头）。该文件在没有门禁压力的情况下写成——`packages/client/ui-conversation/src/*` 位于 `vitest.config.ts` 的覆盖率 `exclude` 列表中，因此覆盖率运行不会统计其中任何文件。
+
+`apps/web/tests/terminal-card.snapshot.ts` 在构建后的客户端产物上固定组装完整的应用：同一渲染意图在两个对话渲染点、以及两种对话行形态下的表现——因为 bash 调用只有经由带键的 `BashRow` 注册才得到常驻卡片，而其他任何声明 terminal 的工具名都落到渲染点兜底行上，其输出体受展开控制。fixture 第 65 轮改名为 `bash`、第 60 轮保留 `fx-bash`，于是一份 fixture 覆盖两种形态，并把第 60 轮的命令改为两行，使构建产物快照钉住逐行提示区及其单枚状态点（`dotsPerPromptRow: [1, 0]`）。该终端轮有意排在 todo 轮**之前**：站立计划会在下一次 `turn/start` 时退役，若追加在其后就会让 dock 的计划条变空，并连带毁掉 todo 表面自身的覆盖；该轮还承载第 60 轮两个提示行无法覆盖的部分——解析到 `--dsw-*` token 的 SGR 分段、超出对话上限的输出、嵌套 cwd，以及在样本旁另行标注的非零退出码。样本正文有意**不含** `[exit code: N]` 行：真实的 bash presenter 正是因为卡片以徽章单独呈现退出状态，才把该标记从正文中消费掉；若保留它，钉住的将是一帧把退出状态显示两次的画面——而产品路径产不出这一帧。
+
+`apps/web/tests/navigation-panes.e2e.ts` 在其既有的 `echo NAVIGATION_OK` bash 调用上新增真实浏览器场景，断言 jsdom 无法计算的部分：把输出面板挤压到窄于内容宽度后，行仍保持单行且面板产生横向溢出；运行状态点解析为绿色的 success token，而不是字面颜色（没有真实主题样式表时，`--dsw-*` 变量根本不产生计算值），且它位于卡片盒之内、提示标签之左——这正是那道 gutter 内边距所拥有的不变量；复制控件走的是页面自身的异步 Clipboard API，而非 `execCommand` 兜底路径。其 `terminal-card.expected.md` 基准记录了提示行中已解析的 workspace——这正是不带 `workdir` 的 bash 调用应当显示的内容，而非一个裸 `$`。
+
+## Related
+
+- [Tagged render-intent union for tool-call presentation](../architecture/2026-07-02-tool-render-intent-union.md)——本次消费的 `card` 标签词汇；Web client 现在是 `terminal` 分支的完整消费方，而不再只消费参数。
+- [Web client syntax highlighting](../process/2026-07-26-web-syntax-highlighting-shiki.md)——它拥有 `CodeBlock` 及其 shiki 分支，并记录了工具输出为何有意不做语法高亮；这里的 ANSI 颜色是作者指定的颜色，不是猜出来的语法。
+- [Web client architecture](../architecture/2026-07-19-gui-web-client-architecture.md)——两个渲染点所处的 slot 与快照分层。
diff --git a/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.i18n.yaml b/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.i18n.yaml
new file mode 100644
index 0000000000..e0fc203ce6
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.md
+2026-07-29-web-message-icon-actions-and-clock.md: e0072458e4c0d3e37998b5564ad14ce17aa41515
+2026-07-29-web-message-icon-actions-and-clock.zh.md: 1cc25a9656e7a100d78dd3b6b3675ca490f455f9
diff --git a/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.md b/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.md
new file mode 100644
index 0000000000..e0072458e4
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.md
@@ -0,0 +1,27 @@
+# Agent Note: Web message IconActions and clocks
+
+Status: implemented
+
+English | [中文](2026-07-29-web-message-icon-actions-and-clock.zh.md)
+
+## Problem
+
+The web chat user bubble already had copy / branch / edit IconActions but no clock. Finalized assistant narration had no under-body action chrome at all, even though the Harness design shows a copy / branch / clock row after the answer settles. Streaming replies must not flash that chrome mid-token. Memoized rows also keep stable props across midnight, so a one-shot `Date.now()` would leave yesterday's messages stuck on `HH:mm`.
+
+## Decision
+
+**User bubbles prepend a date-aware local clock to the existing IconActions row; finalized assistant nodes append a copy / branch / clock row with `margin-top: 16px`; both seats re-format at the next local midnight.**
+
+Both seats format `node.time` through `formatMessageClock`: same calendar day → `HH:mm`, earlier this year → `M月D日 HH:mm`, other years → `YYYY年M月D日 HH:mm`. `useCalendarDay` is a component-local day tick (timeout to the next local midnight) so memoized rows re-render when the calendar day changes without a new framework hook. `MessageItem` places the label before copy (figma `388:20051`). `AssistantMarkdown` places it after branch (figma `43:32997`) and only when `streaming` is false with a known event time; the streaming tail omits the row. Copy writes joined text blocks. Branch stays a chrome stub. Hover-capable pointers keep both footers opacity-hidden until hover/focus-within. Clipboard write and the clock helpers live in `message-chrome.ts`. The assembled surface is pinned by `apps/web/tests/message-actions.e2e.ts` (cold-seeded history + aria golden); aria normalization collapses every clock shape to `{{clock}}`.
+
+## Alternatives considered
+
+**Show assistant IconActions during streaming.** Rejected: the request is to reveal the row only after output completes; mid-stream chrome would flicker and invite copying a partial answer.
+
+**Wire branch to a real session fork.** Rejected for this change: same rationale as the archived [user IconActions note](../../archived/feature/2026-07-27-user-message-icon-actions.md) — the mutation path is unspecified; the button reserves the design seat.
+
+**Publish the calendar day through a chat store or inject hook.** Rejected: the day tick is presentation-only local state with no cross-entry consumers; a component-local timeout matches the client rule that behavioral hooks may own state that does not subscribe to an external source.
+
+## Consequences
+
+Settled assistant answers expose copy and the event clock immediately; branch stays a stub. User and assistant clocks share the same day/year widening rules and refresh after midnight without a message mutation. Per-message paging remains a deferred footer seat in the package README. Package tests pin the three clock shapes and the midnight widen; the web e2e scenario pins the assembled IconActions chrome.
diff --git a/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.zh.md b/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.zh.md
new file mode 100644
index 0000000000..1cc25a9656
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-29-web-message-icon-actions-and-clock.zh.md
@@ -0,0 +1,27 @@
+# Agent Note: Web 消息 IconActions 与时钟
+
+Status: implemented
+
+[English](2026-07-29-web-message-icon-actions-and-clock.md) | 中文
+
+## 问题
+
+Web 聊天的用户气泡已有复制／分支／编辑 IconActions，但没有时钟。已定稿的 assistant 叙述下方完全没有操作栏，尽管 Harness 设计稿在回答结束后展示复制／分支／时钟。流式回复不得在 token 中途闪出该栏。经 memo 的行在跨午夜时仍保持稳定 props，因此一次性的 `Date.now()` 会让昨日消息一直卡在 `HH:mm`。
+
+## 决策
+
+**用户气泡在既有 IconActions 行前追加感知日期的本地时钟；已定稿的 assistant 节点在正文下追加带 `margin-top: 16px` 的复制／分支／时钟；两边都在下一个本地午夜重新格式化。**
+
+两边都通过 `formatMessageClock` 格式化 `node.time`：同一日历日 → `HH:mm`，同年更早 → `M月D日 HH:mm`，跨年 → `YYYY年M月D日 HH:mm`。`useCalendarDay` 是组件本地的日刻度（定时到下一个本地午夜），因此 memo 行在日历日变化时会重渲染，且不新增框架 hook。`MessageItem` 把标签放在复制之前（figma `388:20051`）。`AssistantMarkdown` 把它放在分支之后（figma `43:32997`），且仅在 `streaming` 为 false 且已知事件时间时渲染；流式尾部省略该行。复制写入拼接后的 text 块。分支仍是 chrome stub。具备 hover 能力的指针在 hover／focus-within 前保持两条 footer 透明。剪贴板写入与时钟辅助函数放在 `message-chrome.ts`。组装面由 `apps/web/tests/message-actions.e2e.ts`（冷 seed 历史 + aria golden）钉住；aria 归一化把每种时钟形态折叠为 `{{clock}}`。
+
+## 曾考虑的方案
+
+**在流式过程中展示 assistant IconActions。** 否决：需求是输出完成后才展示该行；中途 chrome 会闪烁，并诱使复制半截回答。
+
+**把分支接到真实的会话 fork。** 本次否决：与已归档的[用户 IconActions 笔记](../../archived/feature/2026-07-27-user-message-icon-actions.md)同一理由——变更路径尚未规定；按钮只预留设计座位。
+
+**通过 chat store 或 inject hook 发布日历日。** 否决：日刻度只是展示层本地状态，没有跨入口消费者；组件本地 timeout 符合「行为 hook 可拥有不订阅外部源的状态」这一客户端规则。
+
+## 后果
+
+已定稿的 assistant 回答立刻暴露复制与事件时钟；分支仍为 stub。用户与 assistant 时钟共用同一套跨天／跨年加宽规则，并在午夜后无需消息变更即可刷新。逐消息分页仍是包 README 中的暂缓 footer 座位。包级测试钉住三种时钟形态与午夜加宽；Web e2e 场景钉住组装后的 IconActions chrome。
diff --git a/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.i18n.yaml
index e90e257c46..789f1ec6ea 100644
--- a/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md
-2026-07-26-briefed-minimal-translation-updates.md: 63f25d5c36caecba435e9192534e0b494e1e0105
-2026-07-26-briefed-minimal-translation-updates.zh.md: 17cf2f895b187fb794e691c2b14d6e0bff78363d
+2026-07-26-briefed-minimal-translation-updates.md: a47251376771165d0eb229aaa0fb7f63589d7d77
+2026-07-26-briefed-minimal-translation-updates.zh.md: 5a6787b9bdb3286b71842abec075c5a0ebfc7991
diff --git a/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md b/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md
index 63f25d5c36..a472513767 100644
--- a/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md
+++ b/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md
@@ -14,7 +14,7 @@ Pair updates run on a generated briefing instead of the guidance corpus; only ne
 
 - **`pnpm run gen-translation-brief [--apply] [pair...]`** ([scripts/gen-translation-brief.ts](../../../../scripts/gen-translation-brief.ts), assembly in [scripts/translation-brief.ts](../../../../scripts/translation-brief.ts)) prints, per out-of-sync pair, the authored side's diff from its recorded last-confirmed blob to the working tree plus the change mapped at the narrowest safely aligned granularity, deterministically widening on mapping failure: a change confined to the pair's byte-identical code fences is computed outright (`--apply` splices it into the counterpart and validates the result against the pairing gate's structural signature before writing); otherwise changed Markdown units (headings, paragraphs, table rows, list items, code fences, block quotes, HTML blocks, thematic breaks, link definitions — matched by container-scoped kind sequences) each carry their last-confirmed source, current source, and current counterpart text with line numbers; units that do not align fall back to depth-matched heading sections; and when sections do not align either, or both sides drifted, the briefing says so and withholds the mapping instead of guessing. Terminology rows are matched against the changed spans only (word-boundary English matching with plural inflections), and for Chinese targets the briefing tracks each relevant term's document-wide first occurrence — when an edit moves it, the vacated and receiving spans join the briefing with an explanatory note, since the 首次出现 annotation must move with it. The unit mapping, code splice, and first-occurrence mechanics adopt the planner design from the [incremental prompt-pipeline work](https://github.com/deepseek-harness/deepseek-harness/pull/684), whose provider-backed bake-off independently validated the same scope ladder for the automated pipeline. The briefing is the translator's whole working set; the full sources of truth remain the escalation path for decisions the briefing cannot answer.
 - **The update path in [dsh-translate-docs](../../../skills/dsh-translate-docs/SKILL.md)** consumes the briefing: mechanical (code-fence-only) changes are applied with `--apply`, no subagent; prose diffs go to a subagent whose prompt is the briefing, not the corpus; verification is clause-by-clause on the changed spans, not the whole document.
-- **The pairing gate takes pair arguments.** `verify-translation-pairing [pair...]` checks just the named pairs (any of a pair's three files, or the bare stem, names it); the corpus-wide sweep remains the no-argument form that `doc-sync` and CI run. `--write` now requires naming the confirmed pairs — bare `--write` refuses, and re-recording everything is an explicit `--write --all` — because the old bare form silently blessed every drifted pair in the tree, including ones the caller never looked at, and a prose-only drift would then stay green forever. Each record's comment names its own scoped command.
+- **The pairing gate takes pair arguments.** `verify-translation-pairing [pair...]` checks just the named pairs (any of a pair's three files, or the bare stem, names it); the corpus-wide sweep remains the no-argument form that `doc-sync` and CI run. `--write` now requires naming the confirmed pairs — bare `--write` refuses, and re-recording everything is an explicit `--write --all` — because the old bare form silently blessed every drifted pair in the tree, including ones the caller never looked at, and a prose-only drift would then stay green forever. Each record's comment names its own scoped command. Before recording, `--write` stores each side's exact bytes with `git hash-object -w --stdin` and pins the blob under a content-addressed local `refs/dsh/translation-pairing/snapshots/` ref; an uncommitted last-confirmed snapshot is therefore available to the briefing generator's later `git cat-file`, not merely named by a hash that Git cannot resolve or left vulnerable to garbage collection.
 
 ## Benchmark
 
@@ -40,6 +40,7 @@ A second head-to-head replay on the same ten examples compared this note's shipp
 
 - A small prose edit's counterpart update now costs a briefing generation plus one small focused task — no corpus reads, no archaeology, no corpus-wide scans inside the loop — and the same PR obligation holds; the cheap path and the correct path point the same way.
 - The briefing generator is a second consumer of the consistency records: recorded blob hashes now also drive diff recovery and section mapping, strengthening the incentive to keep records honest.
+- Each distinct confirmed snapshot retains a content-addressed local ref and object. An abandoned re-record may therefore leave an extra durable pin, but it changes no branch or commit history; this local retention is the tradeoff that prevents garbage collection from invalidating an accepted pairing record.
 - `--write` without arguments no longer works; muscle-memory callers must name pairs or pass `--all`. That is the point — the bulk bless is now a visible, deliberate act.
 - Scoped checks mean an update loop can be green while an unrelated pair elsewhere is red; the corpus-wide check in `doc-sync`/CI still owns the tree-level invariant.
 - Span mapping trusts an alignment only when the kind sequences match across the last-confirmed source, current source, and current counterpart; a mapping failure widens deterministically (units → sections → whole document) rather than guessing, so a restructured document gets an explicit "locate the regions yourself" briefing, never a wrong map.
@@ -47,4 +48,4 @@ A second head-to-head replay on the same ten examples compared this note's shipp
 
 ## Testing
 
-[scripts/translation-brief.spec.ts](../../../../scripts/translation-brief.spec.ts) pins unit and section span extraction (container-scoped kinds, depth-only section alignment so translated heading text still maps, preamble), alignment and changed-index detection, the mechanical code splice and each of its refusal conditions, terminology row matching in both directions with word-boundary and plural-inflection discipline, first-occurrence movement tracking, fence escalation, and the rendered briefing's contract (unit bundles with three-way context, mechanical/sections/document scopes, per-direction digests, scoped finish commands). [scripts/translation-pairing.spec.ts](../../../../scripts/translation-pairing.spec.ts) pins argument normalization (any pair file or bare stem to the anchor) and the CLI matrix: scoped check, bare `--write` refusal, `--write <pair>`, `--write --all`, `--list` exclusivity, unknown flags.
+[scripts/translation-brief.spec.ts](../../../../scripts/translation-brief.spec.ts) pins unit and section span extraction (container-scoped kinds, depth-only section alignment so translated heading text still maps, preamble), alignment and changed-index detection, the mechanical code splice and each of its refusal conditions, terminology row matching in both directions with word-boundary and plural-inflection discipline, first-occurrence movement tracking, fence escalation, and the rendered briefing's contract (unit bundles with three-way context, mechanical/sections/document scopes, per-direction digests, scoped finish commands). [scripts/translation-pairing.spec.ts](../../../../scripts/translation-pairing.spec.ts) pins exact-byte persistence and recovery for uncommitted snapshots, failure before an unavailable object can enter a sidecar, argument normalization (any pair file or bare stem to the anchor), and the CLI matrix: scoped check, bare `--write` refusal, `--write <pair>`, `--write --all`, `--list` exclusivity, unknown flags.
diff --git a/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.zh.md b/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.zh.md
index 17cf2f895b..5a6787b9bd 100644
--- a/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.zh.md
@@ -14,7 +14,7 @@ Status: implemented
 
 - **`pnpm run gen-translation-brief [--apply] [pair...]`**（[scripts/gen-translation-brief.ts](../../../../scripts/gen-translation-brief.ts)，组装逻辑在 [scripts/translation-brief.ts](../../../../scripts/translation-brief.ts)）针对每个失去同步的配对，打印被改一侧从其记录在案的上次确认 blob 到当前工作区的 diff，并附上以能安全对齐的最窄粒度映射的这次改动，映射失败时粒度确定性地逐级放宽：仅落在配对中逐字节一致的围栏代码块内的改动会直接算出（`--apply` 会把它拼接进对侧文件，并在写入前用配对门禁的结构签名校验所得结果）；否则，每个有改动的 Markdown 单元（标题、段落、表格行、列表项、围栏代码块、块引用、HTML 块、分隔线、链接定义；匹配依据是以容器为作用域的种类序列）都带上各自的上次确认源文、当前源文与当前对侧文本及行号；无法对齐的单元回退到按深度匹配的标题章节；当章节也无法对齐或两侧同时漂移时，简报会明说这一点并省略映射，而不是靠猜。术语表行只与改动块匹配（英文术语按词边界匹配，含复数变形）；当目标侧是中文时，简报还会跟踪每个相关术语在整篇文档中的首次出现：一旦某次编辑使其移位，腾出的与接收的两处区间就会附一条解释性说明加入简报，因为「首次出现」括注必须随之移动。单元映射、代码拼接与首次出现机制采纳了[增量提示词流水线工作](https://github.com/deepseek-harness/deepseek-harness/pull/684)中的规划器设计；该项工作中接入提供方的对比评测，已为自动流水线独立验证了同一套范围阶梯。简报就是译者的全部工作集；简报回答不了的决策，仍以完整的真源文档作为升级求证路径。
 - **[dsh-translate-docs](../../../skills/dsh-translate-docs/SKILL.md) 中的更新路径**消费这份简报：机械类改动（只涉及围栏代码块）用 `--apply` 应用，不动用 subagent；行文类 diff 交给 subagent，其提示词就是简报本身，而非指导语料；核验只对改动块逐句进行，不覆盖整篇文档。
-- **配对门禁接受配对参数。**`verify-translation-pairing [pair...]` 只检查被点名的配对（配对三个文件中的任意一个，或其裸词干，都能指代该配对）；全语料扫描仍是 `doc-sync`（文档同步门禁）与 CI 运行的无参数形式。`--write` 现在要求点名已确认的配对：裸 `--write` 会拒绝执行，重新记录全部配对必须显式写 `--write --all`；原因是旧的裸形式会默默为树中每一个漂移的配对背书，包括调用者从未看过的那些，纯行文层面的漂移于是可以永远保持绿灯。每份记录的注释都写明针对该配对自身的按对命令。
+- **配对门禁接受配对参数。**`verify-translation-pairing [pair...]` 只检查被点名的配对（配对三个文件中的任意一个，或其裸词干，都能指代该配对）；全语料扫描仍是 `doc-sync`（文档同步门禁）与 CI 运行的无参数形式。`--write` 现在要求点名已确认的配对：裸 `--write` 会拒绝执行，重新记录全部配对必须显式写 `--write --all`；原因是旧的裸形式会默默为树中每一个漂移的配对背书，包括调用者从未看过的那些，纯行文层面的漂移于是可以永远保持绿灯。每份记录的注释都写明针对该配对自身的按对命令。写下记录之前，`--write` 用 `git hash-object -w --stdin` 存入每一侧的精确字节，并在内容寻址的本地 `refs/dsh/translation-pairing/snapshots/` ref 下固定该 blob；未提交的上次确认快照因此能被简报生成器之后的 `git cat-file` 取回，而不只是留下一个 Git 无法解析的 hash 名称或暴露于垃圾回收。
 
 ## 基准测试
 
@@ -40,6 +40,7 @@ Status: implemented
 
 - 一次小的行文修改，其对侧更新如今只需生成一份简报，外加一个小而聚焦的任务（不读指导语料、不翻查历史、循环内不做全语料扫描），同一 PR 内完成更新的义务保持不变；低成本的路径与正确的路径指向同一个方向。
 - 简报生成器成为一致性记录的第二个消费方：记录的 blob hash 如今还驱动 diff 还原与章节映射，这进一步强化了如实维护记录的动机。
+- 每个不同的已确认快照都会保留一个内容寻址的本地 ref 和对象。因此，中途放弃的重新记录可能留下额外的持久固定项，但它不会改变任何分支或提交历史；这种本地保留正是防止垃圾回收让已接受配对记录失效所付出的代价。
 - 不带参数的 `--write` 不再可用；靠肌肉记忆的调用者必须点名配对或传 `--all`。这正是目的所在：批量背书如今是一个可见的、有意为之的动作。
 - 按对检查意味着一个更新循环可以在别处某个无关配对处于红灯时自己保持绿灯；`doc-sync`/CI 中的全语料检查仍然承载树级不变式。
 - 区间映射只在上次确认源文、当前源文与当前对侧文本三方的种类序列一致时才信任一处对齐；映射失败时粒度确定性地逐级放宽（单元 → 章节 → 整篇文档）而不是靠猜，因此被重构过的文档拿到的是一份明确写着「请自行定位相关区域」的简报，绝不会是一张错误的地图。
@@ -47,4 +48,4 @@ Status: implemented
 
 ## 测试
 
-[scripts/translation-brief.spec.ts](../../../../scripts/translation-brief.spec.ts) 固定单元与章节的区间提取（以容器为作用域的种类、只按深度对齐章节从而让已翻译的标题文字仍能映射、首个标题前的序言）、对齐与改动索引检测、机械代码拼接及其每一个拒绝条件、带词边界与复数变形约束的双向术语行匹配、首次出现移位跟踪、围栏升级，以及渲染后简报的契约（带三方上下文的单元条目、机械／章节／整篇文档三种范围、分方向的规则摘要、按对的收尾命令）。[scripts/translation-pairing.spec.ts](../../../../scripts/translation-pairing.spec.ts) 固定参数归一化（配对的任一文件或裸词干都归一到锚点）与 CLI（命令行界面）用例矩阵：按对检查、裸 `--write` 拒绝执行、`--write <pair>`、`--write --all`、`--list` 的互斥性、未知标志。
+[scripts/translation-brief.spec.ts](../../../../scripts/translation-brief.spec.ts) 固定单元与章节的区间提取（以容器为作用域的种类、只按深度对齐章节从而让已翻译的标题文字仍能映射、首个标题前的序言）、对齐与改动索引检测、机械代码拼接及其每一个拒绝条件、带词边界与复数变形约束的双向术语行匹配、首次出现移位跟踪、围栏升级，以及渲染后简报的契约（带三方上下文的单元条目、机械／章节／整篇文档三种范围、分方向的规则摘要、按对的收尾命令）。[scripts/translation-pairing.spec.ts](../../../../scripts/translation-pairing.spec.ts) 固定未提交快照的精确字节持久化与取回、在不可用对象进入伴随记录前失败、参数归一化（配对的任一文件或裸词干都归一到锚点），以及 CLI（命令行界面）用例矩阵：按对检查、裸 `--write` 拒绝执行、`--write <pair>`、`--write --all`、`--list` 的互斥性、未知标志。
diff --git a/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.i18n.yaml
index 9fd37bcedb..72cbbc368f 100644
--- a/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-26-web-syntax-highlighting-shiki.md: b329e35f1d0ce7b3de454758403a09f67056b5af
-2026-07-26-web-syntax-highlighting-shiki.zh.md: 8e9d1f0d0c38ce64bcb5da1262538da762f70b12
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.md
+2026-07-26-web-syntax-highlighting-shiki.md: 48a1e4c43f19693f90906f210f0ed85db3f31687
+2026-07-26-web-syntax-highlighting-shiki.zh.md: 780b66a309c841c542f873f376226b8da454e050
diff --git a/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.md b/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.md
index b329e35f1d..48a1e4c43f 100644
--- a/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.md
+++ b/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.md
@@ -17,7 +17,7 @@ The client rendered every code surface — markdown fences in assistant prose, t
 - **Dependency**: `shiki/core` + `@shikijs/langs`, composed via `createHighlighterCoreSync` with `createJavaScriptRegexEngine({ forgiving: true })` — no oniguruma WASM, no async init, bundle-friendly. Grammar allowlist: `typescript` (embeds JS), `shellscript`, `json` — the languages the harness actually renders; everything else falls back to a geometry-identical plain block, never an error. Prior art: the VitePress site already renders all documentation code through shiki, and TextMate grammars materially beat regex highlighters on TypeScript — the payload that matters here.
 - **Singleton**: `ui-primitives/src/markdown/highlight.ts` creates one `HighlighterCore` per document and exposes `highlightToHtml(code, lang)` (undefined = render plain). Engine + grammar construction is a ~120-175ms long task, so the module pre-warms the singleton in a deferred task at plugin boot (the lazy path stays as the correctness fallback), keeping the cost off the render path where a stream's finalize swap would jank. The alias table is a `Map`, not an object: fence info strings are assistant-authored, so a label like `constructor` must miss instead of resolving an inherited property and crashing shiki. The shared `CodeBlock` component owns both arms; its shiki arm injects the generated span tree via `dangerouslySetInnerHTML` — sanctioned because shiki emits a static span tree computed from the code text (no user HTML passes through, no scripts/handlers), shiki's own documented consumption path.
 - **Theming**: shiki's `createCssVariablesTheme` routes every token color through `--shiki-*` custom properties; the VALUES live in a new `ui-theme/styles/shiki.css` token sheet (light on `:root`, dark on `body[data-ds-dark-theme]` — the same cascade as every other sheet), imported by the shell's `base.css` chain. Component CSS stays tokens-only; no literal color ever enters JS or component sheets. Background/foreground alias the existing markdown code-block tokens so highlighted and plain blocks agree.
-- **Surfaces**: markdown fences (`MarkdownText`'s `pre` component routes single-string fences through `CodeBlock`), the `run_code` expanded program body (ToolRow's code variant, `lang="typescript"`), and the details panel's Input args (`lang="json"`). Output stays plain deliberately — tool output is arbitrary text, and guessing a grammar would mis-highlight more than it helps.
+- **Surfaces**: markdown fences (`MarkdownText`'s `pre` component routes single-string fences through `CodeBlock`), the `run_code` expanded program body (ToolRow's code variant, `lang="typescript"`), and the details panel's Input args (`lang="json"`). Tool output is never syntax-highlighted — it is arbitrary text, and guessing a grammar would mis-highlight more than it helps; a bash card's output carries only the color its own ANSI sequences declare, through [the terminal card](../feature/2026-07-28-web-terminal-card.md).
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.zh.md b/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.zh.md
index 8e9d1f0d0c..780b66a309 100644
--- a/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-web-syntax-highlighting-shiki.zh.md
@@ -17,7 +17,7 @@ client 过去把每一处代码表面——assistant 正文里的 markdown 围
 - **依赖**：`shiki/core` + `@shikijs/langs`，经 `createHighlighterCoreSync` 搭配 `createJavaScriptRegexEngine({ forgiving: true })` 组装——不带 oniguruma WASM、没有异步初始化、对 bundle 友好。语法（grammar）白名单：`typescript`（内嵌 JS）、`shellscript`、`json`——即 harness 实际会渲染的那几种语言；其余一律回退到几何完全一致的纯文本块，绝不报错。先例：VitePress 站点已经通过 shiki 渲染全部文档代码；而在 TypeScript（正是此处要紧的载荷）上，TextMate 语法实质性优于正则高亮器。
 - **单例**：`ui-primitives/src/markdown/highlight.ts` 为每个 document 创建一个 `HighlighterCore`，并公开 `highlightToHtml(code, lang)`（undefined 即渲染为纯文本）。引擎加语法的构建是一次约 120-175ms 的长任务，因此模块在插件启动时用延迟任务预热单例（惰性路径保留为正确性兜底），把这笔开销挪出渲染路径——否则流式 finalize 交换的那一刻会卡顿。别名表用 `Map` 而非对象：fence 信息串由 assistant 撰写，诸如 `constructor` 这样的标签必须落空，而不是解析到继承属性并让 shiki 崩溃。共享的 `CodeBlock` 组件同时拥有两条分支；其 shiki 分支经 `dangerouslySetInnerHTML` 注入生成的 span 树——此用法获准，因为 shiki 输出的是从代码文本计算出的静态 span 树（不流经任何用户 HTML，没有脚本或事件处理器），这正是 shiki 自身文档载明的消费路径。
 - **主题化**：shiki 的 `createCssVariablesTheme` 让每一种 token 颜色都经由 `--shiki-*` 自定义属性路由；取值本身住在新增的 `ui-theme/styles/shiki.css` token 表里（亮色在 `:root`、暗色在 `body[data-ds-dark-theme]`——层叠方式与其余每张样式表相同），由壳的 `base.css` 导入链引入。组件 CSS 保持只用 token；任何字面颜色都不进入 JS 或组件样式表。背景/前景以别名指向既有的 markdown 代码块 token，使高亮块与纯文本块彼此一致。
-- **表面**：markdown 围栏代码块（`MarkdownText` 的 `pre` 组件把单字符串围栏路由到 `CodeBlock`）、`run_code` 展开后的程序正文（ToolRow 的 code 变体，`lang="typescript"`），以及 details 面板的 Input 参数（`lang="json"`）。输出有意保持纯文本——工具输出是任意文本，硬猜一种语法，带来的误高亮会多于帮助。
+- **表面**：markdown 围栏代码块（`MarkdownText` 的 `pre` 组件把单字符串围栏路由到 `CodeBlock`）、`run_code` 展开后的程序正文（ToolRow 的 code 变体，`lang="typescript"`），以及 details 面板的 Input 参数（`lang="json"`）。工具输出从不做语法高亮——它是任意文本，硬猜一种语法，带来的误高亮会多于帮助；bash 卡片的输出只承载其自身 ANSI 序列声明的颜色，经由[终端卡片](../feature/2026-07-28-web-terminal-card.md)渲染。
 
 ## 曾考虑的替代方案
 
diff --git a/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.i18n.yaml b/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.i18n.yaml
index 34dcf42c4f..4eb85b7d70 100644
--- a/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md
-2026-07-27-worktree-local-lefthook.md: d18f6c1bf8fe240759ad48f67ca6b231000eaf2c
-2026-07-27-worktree-local-lefthook.zh.md: 42a1625a3b2ec7b00942dc46b0c9c64058ecd2fc
+2026-07-27-worktree-local-lefthook.md: 75dfd47087356c34005ec4673e174e451a72c660
+2026-07-27-worktree-local-lefthook.zh.md: bc4902769561c3d33d2101de55e28e70d114f39b
diff --git a/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md b/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md
index d18f6c1bf8..75dfd47087 100644
--- a/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md
+++ b/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md
@@ -16,9 +16,9 @@ Hook installation is worktree-scoped. With `CI=true` or `GITHUB_ACTIONS=true`, t
 
 Before upgrading format 0, the installer refuses direct common-config `extensions.*`; it also refuses direct `core.worktree` or `core.bare=true` and non-empty dormant worktree configs that enabling the extension would activate. The migration removes direct `core.bare=false` because false is Git's default. The common repository config and every existing `config.worktree` must be regular files. These checks disable include expansion because Git's repository-format parser also ignores included targets. A repository-scoped lock serializes migration and hook writes; its process ID, random token, file identity, and exact contents must still match at release. Dead or invalid locks require manual recovery rather than automatic breaking.
 
-Each hook directory carries a JSON ownership marker containing the absolute path last published to worktree config. After a checkout moves, that marker permits replacement of only the exact stale owned value. Before Lefthook runs, the marker and every existing generated hook must be unaliased regular files. The installer resolves the effective scope, origin, and value of `core.hooksPath`, including active `config.worktree` includes; it refuses command-scoped paths, unowned worktree-scoped paths, and unowned reserved directories. An inherited system, global, or common-repository path requires `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`, which opts only the current worktree into Lefthook. Inactive `includeIf` targets are not recursively inspected because they do not affect the current configuration. Command-scoped Git configuration is removed from the Lefthook subprocess environment after validation.
+Each hook directory carries a JSON ownership marker containing the absolute path last published to worktree config. After a checkout moves, that marker permits replacement of only the exact stale owned value. Git seeds a new linked worktree's `config.worktree` from the main worktree; when that seed contains the marker-backed reserved hook path of a registered worktree, the installer replaces only the new worktree's config with its own path. Before Lefthook runs, the marker and every existing generated hook must be unaliased regular files. The installer resolves the effective scope, origin, and value of `core.hooksPath`, including active `config.worktree` includes; it refuses command-scoped paths, unowned worktree-scoped paths, and unowned reserved directories. An inherited system, global, or common-repository path requires `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`, which opts only the current worktree into Lefthook. Inactive `includeIf` targets are not recursively inspected because they do not affect the current configuration. Command-scoped Git configuration is removed from the Lefthook subprocess environment after validation.
 
-If Lefthook fails after changing `core.hooksPath`, the installer restores the previous worktree value; a rollback failure is reported alongside the installation failure. Existing files in `$GIT_COMMON_DIR/hooks` are never removed or rewritten. Focused installer tests pin isolation, migration refusal, ownership and relocation, concurrent installation, custom paths, and rollback.
+If Lefthook fails after changing `core.hooksPath`, the installer restores the previous worktree value; a rollback failure is reported alongside the installation failure. Existing files in `$GIT_COMMON_DIR/hooks` are never removed or rewritten. Focused installer tests pin isolation, copied new-worktree configuration, migration refusal, ownership and relocation, concurrent installation, custom paths, and rollback.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.zh.md b/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.zh.md
index 42a1625a3b..bc49027695 100644
--- a/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.zh.md
@@ -16,9 +16,9 @@ Lefthook 生成的钩子会优先使用安装时从对应 worktree 记录的绝
 
 升级格式 0 之前，安装程序会拒绝共用配置中直接设置的 `extensions.*`；它还会拒绝直接设置的 `core.worktree` 或 `core.bare=true`，以及启用扩展后将被激活的非空且尚未生效的 worktree 配置。迁移会移除直接设置的 `core.bare=false`，因为 false 是 Git 的默认值。共用仓库配置和每个已有的 `config.worktree` 都必须是常规文件。这些检查会禁用 include 展开，因为 Git 的仓库格式解析器也会忽略 include 目标。仓库级锁会串行化迁移和钩子写入；释放时，锁的进程 ID、随机令牌、文件身份和完整内容必须仍然匹配。所属进程已结束或内容无效的锁必须手动恢复，不会被自动破坏。
 
-每个钩子目录都有一个 JSON 所有权标记，其中包含上次写入 worktree 配置的绝对路径。检出目录移动后，该标记只允许替换确切的陈旧自有值。Lefthook 运行前，所有权标记和每个已有的生成钩子都必须是不带别名的常规文件。安装程序会解析 `core.hooksPath` 的生效作用域、来源和值，包括通过当前生效的 `config.worktree` include 加载的值；它会拒绝命令作用域路径、非自有的 worktree 作用域路径以及非自有的保留目录。继承自系统、全局或共用仓库配置的路径必须设置 `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`，从而只让当前 worktree 显式启用 Lefthook。未生效的 `includeIf` 目标不会被递归检查，因为它们不影响当前配置。完成验证后，Lefthook 子进程的环境会移除命令作用域的 Git 配置。
+每个钩子目录都有一个 JSON 所有权标记，其中包含上次写入 worktree 配置的绝对路径。检出目录移动后，该标记只允许替换确切的陈旧自有值。Git 会以主 worktree 的配置为新链接 worktree 初始化 `config.worktree`；当该初始配置包含某个已注册 worktree 中由所有权标记佐证的保留钩子路径时，安装程序只会在新 worktree 的配置中将其替换为新 worktree 自有的路径。Lefthook 运行前，所有权标记和每个已有的生成钩子都必须是不带别名的常规文件。安装程序会解析 `core.hooksPath` 的生效作用域、来源和值，包括通过当前生效的 `config.worktree` include 加载的值；它会拒绝命令作用域路径、非自有的 worktree 作用域路径以及非自有的保留目录。继承自系统、全局或共用仓库配置的路径必须设置 `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`，从而只让当前 worktree 显式启用 Lefthook。未生效的 `includeIf` 目标不会被递归检查，因为它们不影响当前配置。完成验证后，Lefthook 子进程的环境会移除命令作用域的 Git 配置。
 
-若 Lefthook 在更改 `core.hooksPath` 后失败，安装程序会恢复先前的 worktree 值；若回滚失败，会与安装失败一并报告。`$GIT_COMMON_DIR/hooks` 中的现有文件绝不会被移除或改写。聚焦的安装程序测试固定了隔离、迁移拒绝、所有权和检出目录移动、并发安装、自定义路径及回滚行为。
+若 Lefthook 在更改 `core.hooksPath` 后失败，安装程序会恢复先前的 worktree 值；若回滚失败，会与安装失败一并报告。`$GIT_COMMON_DIR/hooks` 中的现有文件绝不会被移除或改写。聚焦的安装程序测试固定了隔离、复制的新 worktree 配置、迁移拒绝、所有权和检出目录移动、并发安装、自定义路径及回滚行为。
 
 ## 考虑过的替代方案
 
diff --git a/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.i18n.yaml b/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.i18n.yaml
new file mode 100644
index 0000000000..7505370b2b
--- /dev/null
+++ b/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.md
+2026-07-28-local-json-tree-renderer.md: 5e8e6d6968319ada53e65a48e9e03b297dc0141c
+2026-07-28-local-json-tree-renderer.zh.md: 221880391da05551d05ad9e3dadb909e2e74fa68
diff --git a/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.md b/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.md
new file mode 100644
index 0000000000..5e8e6d6968
--- /dev/null
+++ b/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.md
@@ -0,0 +1,32 @@
+# Agent Note: Local JSON tree renderer
+
+Status: implemented
+
+English | [中文](2026-07-28-local-json-tree-renderer.zh.md)
+
+## Problem
+
+The read-only JSON inspector used by the [trajectory ledger](../feature/2026-07-27-trajectory-inspection-ledger.md) needs compact object and array previews, explicit array paths for copy actions, fixed-open and collapsible root modes, and keyboard navigation. `react-json-view-lite` exposes neither custom node rendering nor row identity, so satisfying those requirements through that dependency requires a package-manager patch against compiled distribution files and DOM traversal that reconstructs data paths from visible labels. The patch behaves as an untyped fork while its source maps and upstream source remain unchanged.
+
+## Decision
+
+`JsonTree` owns its recursive presentation in `dsh-client-ui-primitives`.
+
+- Each rendered row receives its value and property path directly. Object keys and array indexes extend that path during recursion, so copy actions never recover application data from rendered DOM text.
+- Expandable rows render the compact preview locally and mount child rows only while expanded. `expandTopLevel` selects between a fixed-open bracket frame and a collapsible root node without changing the public component contract.
+- The tree keeps one tabbable expander among visible nodes. Pointer activation claims that tab stop; Up and Down move it cyclically, while Left and Right collapse or expand the focused node.
+- `react-json-view-lite` is not a package dependency and has no pnpm patch. Focused component tests pin previews, expansion, keyboard focus, and array copy paths.
+
+## Alternatives considered
+
+**Keep the distribution patch.** Rejected because the application-specific renderer and array identity contract would remain hidden in generated third-party files, and every dependency update would require reviewing a fork without matching source maps.
+
+**Use the upstream renderer without previews.** Rejected because `{…}` and `[…]` discard the compact payload context that the trajectory inspector uses for scanning.
+
+**Inject previews and row metadata after render.** Rejected because effects or mutation observers would depend on the same private DOM structure while splitting one row between React ownership and imperative mutation.
+
+**Adopt a larger JSON viewer.** Rejected because editing, search, and theme systems are outside the current read-only contract; the added dependency surface would not remove the inspector-specific copy and layout code.
+
+## Consequences
+
+The JSON inspector has one source-level owner, explicit data flow, accurate array paths, and no patched dependency. The package now owns recursive rendering, expansion state, ARIA tree structure, and roving focus behavior, so changes to those semantics require focused component coverage. The implementation remains intentionally read-only and limited to the preview, navigation, and copy behavior used by current consumers.
diff --git a/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.zh.md b/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.zh.md
new file mode 100644
index 0000000000..221880391d
--- /dev/null
+++ b/.agents/notes/implemented/simplification/2026-07-28-local-json-tree-renderer.zh.md
@@ -0,0 +1,32 @@
+# Agent Note：本地 JSON 树渲染器
+
+Status: implemented
+
+[English](2026-07-28-local-json-tree-renderer.md) | 中文
+
+## 问题
+
+[轨迹检查记录表](../feature/2026-07-27-trajectory-inspection-ledger.md)使用的只读 JSON 检查器需要提供紧凑的对象和数组预览、供复制操作使用的明确数组路径、固定展开与可折叠两种根节点模式，以及键盘导航。`react-json-view-lite` 既不提供自定义节点渲染，也不提供行标识；要通过该依赖满足这些要求，就必须使用包管理器为编译后的发布文件打补丁，并遍历 DOM，从可见标签中还原数据路径。该补丁实际上相当于一个不受类型约束的 fork，但其源码映射与上游源码均未同步修改。
+
+## 决策
+
+`dsh-client-ui-primitives` 中的 `JsonTree` 自行负责递归呈现。
+
+- 每个渲染行都直接接收自身的值和属性路径。递归时，对象键和数组索引会附加到路径末尾，因此复制操作无需再从 DOM 渲染文本中反向还原应用数据。
+- 可展开行在本地渲染紧凑预览，仅在展开时挂载子行。`expandTopLevel` 可选择固定展开的括号框架或可折叠根节点，而不改变组件的公开契约。
+- 树中所有可见节点仅保留一个可通过 Tab 键聚焦的展开控件。使用指针激活控件后，该控件会成为 Tab 键焦点位置；上、下方向键循环移动焦点，左、右方向键折叠或展开当前焦点所在节点。
+- `react-json-view-lite` 不在 `dsh-client-ui-primitives` 的依赖项中，也没有 pnpm 补丁。针对性组件测试锁定预览、展开、键盘焦点和数组复制路径。
+
+## 曾考虑的替代方案
+
+**保留发布文件补丁。** 不予采纳：应用专用的渲染逻辑和数组标识契约仍会隐藏在生成的第三方文件中；每次更新依赖，都必须评审一个没有配套源码映射的 fork。
+
+**直接使用不提供预览的上游渲染器。** 不予采纳：`{…}` 和 `[…]` 会丢失轨迹检查器扫读时所需的紧凑载荷上下文。
+
+**渲染后注入预览和行元数据。** 不予采纳：Effect 或 MutationObserver 仍会依赖同一套私有 DOM 结构，同时使 React 和命令式变更机制分别负责同一行的不同部分。
+
+**采用更大型的 JSON 查看器。** 不予采纳：编辑、搜索与主题系统不在当前只读契约范围内；扩大依赖范围也无法去除检查器专用的复制和布局代码。
+
+## 后果
+
+JSON 检查器在源码层由单一实现负责，具有显式数据流和准确的数组路径，且没有经过补丁修改的依赖。`dsh-client-ui-primitives` 负责递归渲染、展开状态、ARIA 树结构和焦点循环移动行为，因此修改这些语义时必须提供针对性组件测试。实现有意保持只读，仅包含当前消费方使用的预览、导航与复制行为。
diff --git a/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.i18n.yaml b/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.i18n.yaml
index 43ca03dc7a..0283559c9c 100644
--- a/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.i18n.yaml
+++ b/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md
-2026-07-24-web-gui-browser-e2e-lane.md: d9e0a9660ecd6aeb75e835e68f92c0a268423872
-2026-07-24-web-gui-browser-e2e-lane.zh.md: e8c7d1c4596f20d88bd08423549fb6a9f7b0654b
+2026-07-24-web-gui-browser-e2e-lane.md: ce59dcce270d548c91e3719eee8e9c83aea0c154
+2026-07-24-web-gui-browser-e2e-lane.zh.md: bad3dd15ed7b98cc17340666a6c1094d0de057b1
diff --git a/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md b/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md
index d9e0a9660e..ce59dcce27 100644
--- a/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md
+++ b/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.md
@@ -42,7 +42,7 @@ The typecheck plane split is structural: the host scaffold, its support module,
 
 ### Coverage contract
 
-The lane covers three behavior families. Live-turn scenarios pin ordinary tool execution, cancellation, non-retryable failure, transient retry, resident questions, and mid-turn steering; synchronization uses durable events, `whenIdle()`, or an explicit replay marker rather than delays. Cold-history scenarios seed through the real persistence API and cover history rendering, sidebar search, trajectory and waterfall views, and tool details without model calls. Browser-lifecycle scenarios cover first-send workspace materialization, reload recovery, layout persistence, theme and locale preferences, and workspace create/rename/view operations. Each family asserts the browser surface and the authoritative host state; a stray model call or under-consumed fixture fails teardown.
+The lane covers three behavior families. Live-turn scenarios pin ordinary tool execution, cancellation, non-retryable failure, transient retry, resident questions, and mid-turn steering; synchronization uses durable events, `whenIdle()`, or an explicit replay marker rather than delays. Cold-history scenarios seed through the real persistence API and cover history rendering, sidebar search, trajectory and waterfall views, and tool details without model calls. Browser-lifecycle scenarios cover first-send workspace materialization, reload recovery, layout reset, theme and locale preferences, and workspace create/rename/view operations. Each family asserts the browser surface and the authoritative host state; a stray model call or under-consumed fixture fails teardown.
 
 ### CI stance
 
diff --git a/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.zh.md b/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.zh.md
index e8c7d1c459..bad3dd15ed 100644
--- a/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.zh.md
+++ b/.agents/notes/implemented/testing/2026-07-24-web-gui-browser-e2e-lane.zh.md
@@ -42,7 +42,7 @@ Web GUI 以一条真实组装链交付——chromium 页面 → client 插件 bu
 
 ### 覆盖契约
 
-该车道覆盖三类行为。实时轮次场景钉住普通工具执行、取消、不可重试失败、瞬态重试、常驻提问与轮次中途 steering；同步依赖持久事件、`whenIdle()` 或显式回放标记，而不使用延时。冷历史场景通过真实持久化 API 播种，在不调用模型的情况下覆盖历史渲染、侧栏搜索、Trajectory 与 Waterfall 视图及工具详情。浏览器生命周期场景覆盖首次发送时物化工作区、重新加载恢复、布局持久化、主题与语言偏好，以及工作区的创建、重命名和视图操作。每类场景都断言浏览器表面和权威的 host 状态；离群的模型调用或未耗尽的 fixture 会使拆卸失败。
+该车道覆盖三类行为。实时轮次场景钉住普通工具执行、取消、不可重试失败、瞬态重试、常驻提问与轮次中途 steering；同步依赖持久事件、`whenIdle()` 或显式回放标记，而不使用延时。冷历史场景通过真实持久化 API 播种，在不调用模型的情况下覆盖历史渲染、侧栏搜索、Trajectory 与 Waterfall 视图及工具详情。浏览器生命周期场景覆盖首次发送时物化工作区、重新加载恢复、布局重置、主题与语言偏好，以及工作区的创建、重命名和视图操作。每类场景都断言浏览器表面和权威的 host 状态；离群的模型调用或未耗尽的 fixture 会使拆卸失败。
 
 ### CI 立场
 
diff --git a/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.i18n.yaml b/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.i18n.yaml
index b14e969a84..62fc537c2f 100644
--- a/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.i18n.yaml
+++ b/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.md
-2026-07-04-prune-dead-core-spine-surface.md: a6c608617415f3af07de5c95fd20b0bde40bdef3
-2026-07-04-prune-dead-core-spine-surface.zh.md: 83603d8a8432b99d8f42222442b38005e196ac4e
+2026-07-04-prune-dead-core-spine-surface.md: 473f655ab0944b43f9b1193413801eb7a80286d8
+2026-07-04-prune-dead-core-spine-surface.zh.md: 00c6f7acca403c937ff765b6d73d3d0fa0c3153a
diff --git a/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.md b/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.md
index a6c6086174..473f655ab0 100644
--- a/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.md
+++ b/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.md
@@ -19,7 +19,7 @@ The production corpus is `packages/*/*/src`, example sources/config, and runtime
 | `code-runtime-worker` protocol/bootstrap re-exports | Outside-package production/e2e consumers use `WorkerCodeRuntime` and config, not `BootstrapPort`, `PatchableStream`, or worker message/boot types. | Keep the runtime class/config contract and make its wire/bootstrap vocabulary source-private. |
 | ACP `agentOptions` root export | The helper has only same-file and ACP-test consumers; the sole outside-package production consumer mounts the plugin namespace. | Keep `name`, `inject`, `Config`, `AcpConfig`, and `apply`; make `agentOptions` source-private and test it through bridge behavior. |
 | `providerWording` and `completedTurnPrefix` root exports | Each has one same-package production caller; only the balanced-prefix helper has a same-package white-box test. | Make them source-private and test provider behavior. |
-| `depthOf`, `SubagentDepthError`, `SENSITIVE_ENV_PATTERN`, `waitForExit`, and `exitsWithin` root exports | Production subagent backends consume the in-process runner and subprocess construction/disposal helpers, not these enforcement/test internals. | Keep depth/environment/exit behavior but make the helpers and error/regex source-private; test through spawn and disposal. |
+| `depthOf`, `SubagentDepthError`, `waitForExit`, and `exitsWithin` root exports | Production subagent backends consume the in-process runner and subprocess construction/disposal helpers, not these enforcement/test internals. `SENSITIVE_ENV_PATTERN` is excluded because the SDK helper applies it to caller-supplied environments. | Keep depth and exit behavior but make the remaining helpers and error source-private; test through spawn and disposal. Keep the shared credential pattern public. |
 | `PersistenceCoordinator.inits`, backend `inits` accessors, `seedCoversPrefix`, and `assertSerializable` | The accessors exist for white-box tests; `seedCoversPrefix` has no outside production importer; `assertSerializable` has no production caller and duplicates the coordinator append boundary's lossless snapshot. | Observe initialization through `session/flush`, make `seedCoversPrefix` source-private, and delete `assertSerializable`. Keep both backends, `SessionHeader`, and SQLite's version contract. |
 | `LlmError.status` and replay status | Adapters/replay populate it, but production branches on stable error code/message and never reads raw status. | Remove the unread field and replay plumbing while preserving error classification. |
 | `BlockAssembler.push()` return value | Both production callers ignore the returned completed block. | Return `void`; keep the deliberately public `blocks()`/`message()` contract. |
diff --git a/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.zh.md b/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.zh.md
index 83603d8a84..00c6f7acca 100644
--- a/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.zh.md
+++ b/.agents/notes/proposed/simplification/2026-07-04-prune-dead-core-spine-surface.zh.md
@@ -19,7 +19,7 @@ Status: proposed
 | `code-runtime-worker` 的 protocol/bootstrap 再导出 | 包外的生产/e2e 消费方使用 `WorkerCodeRuntime` 和配置，而非 `BootstrapPort`、`PatchableStream` 或 worker 消息/启动类型。 | 保留运行时类/配置契约，将其协议格式/bootstrap 词汇改为源码私有。 |
 | ACP 的 `agentOptions` 根导出 | 该辅助函数只有同文件和 ACP 测试消费方；唯一的包外生产消费方挂载的是插件命名空间。 | 保留 `name`、`inject`、`Config`、`AcpConfig` 和 `apply`；将 `agentOptions` 改为源码私有，通过桥接层行为测试。 |
 | `providerWording` 与 `completedTurnPrefix` 根导出 | 各有一个同包生产调用者；只有 balanced-prefix 辅助函数有一个同包白盒测试。 | 改为源码私有，测试提供方行为。 |
-| `depthOf`、`SubagentDepthError`、`SENSITIVE_ENV_PATTERN`、`waitForExit` 与 `exitsWithin` 根导出 | 生产 subagent 后端消费的是进程内 runner 和子进程构造/dispose（资源释放）辅助函数，而非这些强制/测试内部实现。 | 保留深度/环境/退出行为，但将辅助函数和 error/regex 改为源码私有；通过 spawn 和 dispose 测试。 |
+| `depthOf`、`SubagentDepthError`、`waitForExit` 与 `exitsWithin` 根导出 | 生产 subagent 后端消费的是进程内 runner 和子进程构造/dispose（资源释放）辅助函数，而非这些强制/测试内部实现。`SENSITIVE_ENV_PATTERN` 不在其中，因为 SDK helper 会将它应用于调用方传入的环境。 | 保留深度与退出行为，但将剩余辅助函数和 error 改为源码私有；通过 spawn 和 dispose 测试。保持共享凭据正则公开。 |
 | `PersistenceCoordinator.inits`、后端 `inits` 访问器、`seedCoversPrefix` 与 `assertSerializable` | 访问器为白盒测试而存在；`seedCoversPrefix` 没有包外生产导入者；`assertSerializable` 没有生产调用者，且与 coordinator append 边界的无损快照重复。 | 通过 `session/flush` 观察初始化，将 `seedCoversPrefix` 改为源码私有，删除 `assertSerializable`。保留两个后端、`SessionHeader` 和 SQLite 的版本契约。 |
 | `LlmError.status` 与回放 status | 适配器/回放填充它，但生产分支基于稳定的错误码/消息判断，从不读取原始 status。 | 移除未读字段和回放管道，保留错误分类。 |
 | `BlockAssembler.push()` 返回值 | 两个生产调用者都忽略返回的已完成块。 | 返回 `void`；保留有意公开的 `blocks()`/`message()` 契约。 |
diff --git a/AGENTS.md b/AGENTS.md
index 4ddfd1693c..b0f5e87e02 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -89,7 +89,7 @@ Real-API tests and demos read `DEEPSEEK_API_KEY`, optional `DEEPSEEK_BASE_URL`,
 ## Conventions
 
 - Every npm package is `@deepseek-ai/dsh-<name>`; vendored packages keep upstream names and are `private: true`. `cordis` is a peerDependency (+ dev) of every harness package.
-- ESM everywhere (`"type": "module"`). Cross-package imports use package names; in-package relative imports include `.ts`. Config subprocesses run built `lib/` under plain Node; source regressions use their declared launcher ([testing policy](docs/testing.md#test-subprocess-launch-modes)). CLI source-launch code and every module it reaches must support Node `--experimental-transform-types`: use `import type` for erased bindings and native ESM exports, with no TSX/JSX or tsx/esbuild-only transforms. TUI/Web `cordis.yml` bare plugins must appear in their resolver manifest's `dependencies`; `verify-cordis-config` enforces the [source-launch contract](.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md).
+- ESM everywhere (`"type": "module"`). Cross-package imports use package names; in-package relative imports include `.ts`. Config subprocesses run built `lib/` under plain Node; source regressions use their declared launcher ([testing policy](docs/testing.md#test-subprocess-launch-modes)). The `dsh` CLI source launch runs through tsx's ESM-only hook (`node --import tsx/esm`); modules it reaches must stay ESM (no CJS-only shapes) — Node's native TypeScript modes are unavailable across the engines range ([source-launch contract](.agents/notes/implemented/architecture/2026-07-29-dsh-source-launch-tsx-esm.md)). TUI/Web `cordis.yml` bare plugins must appear in their resolver manifest's `dependencies`; `verify-cordis-config` enforces it.
 - **Registrations are effects**: every contribution goes through `ctx.effect()` / `ctx.on()`; a registry's `register()` returns the disposer.
 - **Runtime invariants assert owned relationships.** Check authoritative event streams or mutable data, not service or method presence, plugin metadata or effects, or fixed pure examples. If a package has no plausible relationship, an explained empty companion is correct ([package contract](packages/AGENTS.md)).
 - **Typed events use declaration merging** and merge-extensible maps. Event JSDoc needs `@mode` and payload `@param`; scoped keys absent from payloads need `@dshScopeScan unsupported`. Public service methods document parameters and non-void returns.
diff --git a/apps/cli/README.i18n.yaml b/apps/cli/README.i18n.yaml
index 106c2f2c2a..bb5f370009 100644
--- a/apps/cli/README.i18n.yaml
+++ b/apps/cli/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write apps/cli/README.md
-README.md: 6b3a31a30a941e67341a518a7e32b1bf99eee8b0
-README.zh.md: 133c721612012b8fd1327344ae4e2951d05047cb
+README.md: 93c36d18abd06bbd7a80c918f520b92489180395
+README.zh.md: 85f4624a592eaf2ae44dc31fb4e18fb5657e62fd
diff --git a/apps/cli/README.md b/apps/cli/README.md
index 6b3a31a30a..93c36d18ab 100644
--- a/apps/cli/README.md
+++ b/apps/cli/README.md
@@ -28,6 +28,6 @@ Symlink the source-running launcher onto your PATH; it resolves the checkout thr
 ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh
 ```
 
-Source launches run `apps/cli/src/bin.ts` through Node's `--experimental-transform-types`; `scripts/tspath-loader.ts` only projects tsconfig `paths` into module resolution and does not transform code. Every module reachable from the CLI source entry follows Node's transform-types contract: erased bindings use `import type`, exports use native ESM, and the graph contains no TSX/JSX or transforms that only tsx/esbuild provides. The loader reads `TSX_TSCONFIG_PATH` when set (relative paths resolve from the invoking cwd), otherwise the repository's root tsconfig, using the root TypeScript development tool rather than an application dependency. It maps a workspace import only for a package self-reference or a declared runtime dependency. The TUI configs resolve bare plugins through `examples/package.json`, while the Web/headless `cordis.yml` resolves them through this package's `dependencies`; `verify-cordis-config` requires every configured bare plugin to be declared, while allowing unrelated dependencies.
+Source launches run `apps/cli/src/bin.ts` through tsx's ESM-only hook (`node --import tsx/esm`), which transforms TypeScript and projects the root tsconfig `paths` map into module resolution. Node's native TypeScript modes are not used: Node 26 removed `--experimental-transform-types`, and strip-only mode rejects syntax the source graph relies on (vendored parameter properties, decorators, runtime enums/namespaces). The CJS hook stays off because the source graph is ESM-only and the CJS resolver adds ~0.4s of startup. `bin/dsh` pins `TSX_TSCONFIG_PATH` to the checkout's root tsconfig so resolution is cwd-independent, and the `dsh-source-launch-smoke` node-compat gate runs this exact launch vector on every supported Node line. tsx applies the `paths` map without checking dependency declarations, so declaration completeness rests on the static gates: the TUI configs resolve bare plugins through `examples/package.json`, the Web/headless `cordis.yml` through this package's `dependencies`, and `verify-cordis-config` requires every configured bare plugin to be declared, while allowing unrelated dependencies.
 
 `pnpm run dsh` runs the same entry from the repo root and forwards arguments directly, for example `pnpm run dsh -p "task"`. The built form (`lib/bin.js`, via `pnpm run build`) boots the same config under plain Node.
diff --git a/apps/cli/README.zh.md b/apps/cli/README.zh.md
index 133c721612..85f4624a59 100644
--- a/apps/cli/README.zh.md
+++ b/apps/cli/README.zh.md
@@ -28,6 +28,6 @@ Web 和无头界面启动同一个共享组合（`cordis.yml`）：两者都将
 ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh
 ```
 
-源码启动会通过 Node 的 `--experimental-transform-types` 运行 `apps/cli/src/bin.ts`；`scripts/tspath-loader.ts` 只会将 tsconfig 的 `paths` 映射投射到模块解析中，而不会转换代码。从 CLI 源码入口可达的每个模块都遵守 Node transform-types 契约：会被擦除的绑定使用 `import type`，export 使用原生 ESM，整个依赖图不含 TSX/JSX，也不依赖仅由 tsx/esbuild 提供的转换。设置 `TSX_TSCONFIG_PATH` 时，loader 会读取该路径（相对路径从调用方的 cwd 解析），否则读取仓库根 tsconfig；它使用根目录的 TypeScript 开发工具，而不是应用依赖。仅当 workspace import 是包自身引用或已声明的运行时依赖时，loader 才会映射该 import。TUI 配置通过 `examples/package.json` 解析裸插件，而 Web／无头 `cordis.yml` 则通过本包的 `dependencies` 解析；`verify-cordis-config` 要求每个已配置的裸插件均已声明，同时允许存在无关依赖。
+源码启动会通过 tsx 的 ESM-only hook（`node --import tsx/esm`）运行 `apps/cli/src/bin.ts`，由它转换 TypeScript 并将根 tsconfig 的 `paths` 映射投射到模块解析中。不使用 Node 原生 TypeScript 模式：Node 26 移除了 `--experimental-transform-types`，而 strip-only 模式无法接受源码图依赖的语法（vendor 中的参数属性、装饰器、运行时 enum/namespace）。CJS hook 保持关闭，因为源码图是纯 ESM，而 CJS 解析器会增加约 0.4s 启动耗时。`bin/dsh` 将 `TSX_TSCONFIG_PATH` 固定到 checkout 的根 tsconfig，使解析与 cwd 无关；node-compat 门禁 `dsh-source-launch-smoke` 会在每条受支持的 Node 版本线上运行这一精确启动向量。tsx 应用 `paths` 映射时不检查依赖声明，声明完整性由静态门禁保障：TUI 配置通过 `examples/package.json` 解析裸插件，Web／无头 `cordis.yml` 通过本包的 `dependencies` 解析；`verify-cordis-config` 要求每个已配置的裸插件均已声明，同时允许存在无关依赖。
 
 `pnpm run dsh` 从仓库根目录运行同一入口并直接转发参数，例如 `pnpm run dsh -p "task"`。构建形式（`lib/bin.js`，通过 `pnpm run build`）会在普通 Node 下启动同一配置。
diff --git a/apps/cli/src/tsconfig-paths-loader.ts b/apps/cli/src/tsconfig-paths-loader.ts
deleted file mode 100644
index b7337998bc..0000000000
--- a/apps/cli/src/tsconfig-paths-loader.ts
+++ /dev/null
@@ -1,216 +0,0 @@
-/**
- * Node module resolve hook for the `dsh` source launcher. It projects the root
- * tsconfig `paths` map into Node resolution while leaving all TypeScript syntax
- * handling to Node's native transform-types runtime.
- * @module @deepseek-ai/dsh/tsconfig-paths-loader
- */
-
-import { readFile, stat } from 'node:fs/promises'
-import { dirname, extname, join, resolve } from 'node:path'
-import { fileURLToPath, pathToFileURL } from 'node:url'
-import type { ResolveHookContext, ResolveFnOutput } from 'node:module'
-import ts from 'typescript'
-
-interface LoaderData {
-  tsconfigPath: string
-}
-
-interface PackageManifest {
-  name?: string
-  dependencies?: Record<string, string>
-  optionalDependencies?: Record<string, string>
-  peerDependencies?: Record<string, string>
-}
-
-interface PathRule {
-  pattern: string
-  prefix: string
-  suffix: string
-  targets: readonly string[]
-}
-
-interface PathsCompilerOptions {
-  readonly baseUrl?: string
-  readonly paths?: ts.MapLike<string[]>
-  readonly pathsBasePath?: string
-}
-
-// Node's native TypeScript transform cannot parse JSX, so `.tsx` is excluded.
-const SOURCE_EXTENSIONS = ['.ts', '.mts', '.cts'] as const
-
-/**
- * Resolve package imports through one parsed tsconfig paths table.
- *
- * Manifest reads are process-scoped and memoized by path. Only matched source
- * aliases enter the cache, bounding it to directories participating in source
- * resolution.
- */
-export class TsconfigPathsResolver {
-  private readonly rules: readonly PathRule[]
-  private readonly configDirectory: string
-  private readonly manifests = new Map<string, Promise<PackageManifest | undefined>>()
-
-  private constructor(configDirectory: string, paths: ts.MapLike<string[]>) {
-    this.configDirectory = configDirectory
-    this.rules = Object.entries(paths)
-      .map(([pattern, targets]) => {
-        const wildcard = pattern.indexOf('*')
-        return {
-          pattern,
-          prefix: wildcard === -1 ? pattern : pattern.slice(0, wildcard),
-          suffix: wildcard === -1 ? '' : pattern.slice(wildcard + 1),
-          targets,
-        }
-      })
-      .sort((left, right) => {
-        const leftExact = left.pattern.includes('*') ? 0 : 1
-        const rightExact = right.pattern.includes('*') ? 0 : 1
-        return rightExact - leftExact || right.prefix.length - left.prefix.length || right.suffix.length - left.suffix.length
-      })
-  }
-
-  /**
-   * Parse a tsconfig including its `extends` chain.
-   * @param tsconfigPath Absolute tsconfig path supplying `compilerOptions.paths`.
-   * @returns A resolver backed by that path table.
-   */
-  static create(tsconfigPath: string): TsconfigPathsResolver {
-    let unrecoverable: ts.Diagnostic | undefined
-    const parsed = ts.getParsedCommandLineOfConfigFile(tsconfigPath, {}, {
-      ...ts.sys,
-      onUnRecoverableConfigFileDiagnostic(diagnostic) { unrecoverable = diagnostic },
-    })
-    if (parsed === undefined) {
-      const detail = unrecoverable === undefined
-        ? 'unknown configuration error'
-        : ts.flattenDiagnosticMessageText(unrecoverable.messageText, '\n')
-      throw new Error(`dsh source loader could not parse ${tsconfigPath}: ${detail}`)
-    }
-    const options = parsed.options as PathsCompilerOptions
-    const paths = options.paths
-    if (paths === undefined) throw new Error(`dsh source loader requires compilerOptions.paths in ${tsconfigPath}`)
-    const configDirectory = options.baseUrl ?? options.pathsBasePath ?? dirname(tsconfigPath)
-    return new TsconfigPathsResolver(configDirectory, paths)
-  }
-
-  /**
-   * Resolve one bare package specifier to a source file when the importing
-   * package (or config-directory owner) declares that package at runtime.
-   * @param specifier Module specifier passed to Node.
-   * @param parentURL Importing file or Loader config-directory URL.
-   * @returns Source file URL, or `undefined` when normal Node resolution owns the request.
-   */
-  async resolve(specifier: string, parentURL: string | undefined): Promise<string | undefined> {
-    const packageName = packageNameFromSpecifier(specifier)
-    if (packageName === undefined || parentURL === undefined || !parentURL.startsWith('file:')) return undefined
-    const matched = this.match(specifier)
-    if (matched === undefined) return undefined
-    const configParent = parentURL.endsWith('/')
-    const parentPath = fileURLToPath(parentURL)
-    const startDirectory = configParent ? parentPath : dirname(parentPath)
-    if (!await this.isDeclaredRuntimeDependency(startDirectory, packageName, configParent)) return undefined
-
-    for (const target of matched.targets) {
-      const substituted = target.replace('*', matched.wildcard)
-      const candidate = await existingSourcePath(resolve(this.configDirectory, substituted))
-      if (candidate !== undefined) return pathToFileURL(candidate).href
-    }
-    return undefined
-  }
-
-  private match(specifier: string): { targets: readonly string[]; wildcard: string } | undefined {
-    for (const rule of this.rules) {
-      if (!rule.pattern.includes('*')) {
-        if (specifier === rule.pattern) return { targets: rule.targets, wildcard: '' }
-        continue
-      }
-      if (!specifier.startsWith(rule.prefix) || !specifier.endsWith(rule.suffix)) continue
-      const wildcard = specifier.slice(rule.prefix.length, specifier.length - rule.suffix.length)
-      return { targets: rule.targets, wildcard }
-    }
-    return undefined
-  }
-
-  private async isDeclaredRuntimeDependency(
-    startDirectory: string,
-    packageName: string,
-    searchAncestors: boolean,
-  ): Promise<boolean> {
-    for (let directory = startDirectory; ; directory = dirname(directory)) {
-      const manifest = await this.readManifest(join(directory, 'package.json'))
-      if (manifest !== undefined) {
-        if (declaresRuntimeDependency(manifest, packageName)) return true
-        if (!searchAncestors) return false
-      }
-      const parent = dirname(directory)
-      if (parent === directory) return false
-    }
-  }
-
-  private readManifest(path: string): Promise<PackageManifest | undefined> {
-    let pending = this.manifests.get(path)
-    if (pending !== undefined) return pending
-    pending = readFile(path, 'utf8').then(
-      content => JSON.parse(content) as PackageManifest,
-      (error: unknown) => {
-        if (error instanceof Error && (error as NodeJS.ErrnoException).code === 'ENOENT') return undefined
-        throw error
-      },
-    )
-    this.manifests.set(path, pending)
-    return pending
-  }
-}
-
-let resolver: TsconfigPathsResolver | undefined
-
-/** Initialize the hook worker from the source-launch preloader. */
-export function initialize(data: LoaderData): void {
-  resolver = TsconfigPathsResolver.create(data.tsconfigPath)
-}
-
-/** Resolve declared workspace packages to source and delegate every other request to Node. */
-export async function resolveHook(
-  specifier: string,
-  context: ResolveHookContext,
-  nextResolve: (specifier: string, context: ResolveHookContext) => Promise<ResolveFnOutput>,
-): Promise<ResolveFnOutput> {
-  const url = await resolver?.resolve(specifier, context.parentURL)
-  return url === undefined ? nextResolve(specifier, context) : { url, shortCircuit: true }
-}
-
-// Node customization hooks discover this exact export name.
-export { resolveHook as resolve }
-
-function packageNameFromSpecifier(specifier: string): string | undefined {
-  if (specifier.startsWith('.') || specifier.startsWith('/') || /^[a-z][a-z+.-]*:/i.test(specifier)) {
-    return undefined
-  }
-  const segments = specifier.split('/')
-  return specifier.startsWith('@')
-    ? segments.length >= 2 ? `${segments[0]}/${segments[1]}` : undefined
-    : segments[0] || undefined
-}
-
-function declaresRuntimeDependency(manifest: PackageManifest, packageName: string): boolean {
-  return manifest.name === packageName
-    || packageName in (manifest.dependencies ?? {})
-    || packageName in (manifest.optionalDependencies ?? {})
-    || packageName in (manifest.peerDependencies ?? {})
-}
-
-async function existingSourcePath(base: string): Promise<string | undefined> {
-  const extension = extname(base)
-  if (extension === '.tsx') return undefined
-  const candidates = extension === ''
-    ? [base, ...SOURCE_EXTENSIONS.map(extension => `${base}${extension}`), ...SOURCE_EXTENSIONS.map(extension => join(base, `index${extension}`))]
-    : [base]
-  for (const candidate of candidates) {
-    try {
-      if ((await stat(candidate)).isFile()) return candidate
-    } catch (error) {
-      if ((error as NodeJS.ErrnoException).code !== 'ENOENT') throw error
-    }
-  }
-  return undefined
-}
diff --git a/apps/cli/tests/source-launch.compat.spec.ts b/apps/cli/tests/source-launch.compat.spec.ts
new file mode 100644
index 0000000000..bc6959261a
--- /dev/null
+++ b/apps/cli/tests/source-launch.compat.spec.ts
@@ -0,0 +1,36 @@
+import { fileURLToPath } from 'node:url'
+import { execa } from 'execa'
+import { describe, expect, it } from 'vitest'
+
+/**
+ * Keyless smoke for the SOURCE `dsh` launcher: run `apps/cli/src/bin.ts`
+ * with the exact production launch vector (`node --import tsx/esm`, the same
+ * shape as `bin/dsh` and the root `dsh`/`demo:tui`/`demo:web` scripts) and
+ * assert the piped-stdio TTY refusal. The Node compatibility matrix runs this
+ * WHOLE file, so a Node release changing module hooks or TypeScript handling
+ * breaks this gate instead of every developer's `pnpm dsh`; the built-bin
+ * suite covers the published `lib/` entry, not this source chain.
+ */
+
+const repoRoot = fileURLToPath(new URL('../../../', import.meta.url))
+const dshSourceBin = 'apps/cli/src/bin.ts'
+
+describe('dsh SOURCE launcher (node --import tsx/esm)', () => {
+  it('boots the source entry and refuses pipes LOUD (non-zero exit + stderr)', async () => {
+    const result = await execa(process.execPath, ['--import', 'tsx/esm', dshSourceBin], {
+      cwd: repoRoot,
+      input: '',
+      timeout: 25_000,
+      killSignal: 'SIGKILL',
+      reject: false,
+    })
+    if (result.timedOut) {
+      throw new Error(`dsh source launch did not exit within 25s. stdout:\n${result.stdout}\nstderr:\n${result.stderr}`)
+    }
+    expect(result.exitCode).not.toBe(0)
+    expect(result.stderr).toContain('requires stdin and stdout to be interactive TTYs')
+    expect(result.stderr).toContain('dsh -p')
+    // The refusal happens before any plugin mounts: stdout stays silent.
+    expect(result.stdout).toBe('')
+  }, 30_000)
+})
diff --git a/apps/cli/tests/tsconfig-paths-loader.spec.ts b/apps/cli/tests/tsconfig-paths-loader.spec.ts
deleted file mode 100644
index 834ee8c4d0..0000000000
--- a/apps/cli/tests/tsconfig-paths-loader.spec.ts
+++ /dev/null
@@ -1,180 +0,0 @@
-import { mkdirSync, mkdtempSync, rmSync, writeFileSync } from 'node:fs'
-import type { ResolveFnOutput, ResolveHookContext } from 'node:module'
-import { tmpdir } from 'node:os'
-import { dirname, join } from 'node:path'
-import { pathToFileURL } from 'node:url'
-import { afterEach, describe, expect, it, vi } from 'vitest'
-import { initialize, resolveHook, TsconfigPathsResolver } from '../src/tsconfig-paths-loader.ts'
-
-class ResolverFixture {
-  readonly root = mkdtempSync(join(tmpdir(), 'dsh-tsconfig-paths-'))
-
-  path(relativePath: string): string {
-    return join(this.root, relativePath)
-  }
-
-  write(relativePath: string, content = 'export {}\n'): string {
-    const path = this.path(relativePath)
-    mkdirSync(dirname(path), { recursive: true })
-    writeFileSync(path, content)
-    return path
-  }
-
-  writeJson(relativePath: string, value: unknown): string {
-    return this.write(relativePath, `${JSON.stringify(value)}\n`)
-  }
-
-  createResolver(paths: Record<string, string[]>): TsconfigPathsResolver {
-    const tsconfigPath = this.writeJson('tsconfig.json', { compilerOptions: { paths } })
-    return TsconfigPathsResolver.create(tsconfigPath)
-  }
-
-  parentURL(relativePath = 'consumer/src/nested/index.ts'): string {
-    return pathToFileURL(this.path(relativePath)).href
-  }
-
-  dispose(): void {
-    rmSync(this.root, { recursive: true, force: true })
-  }
-}
-
-const fixtures: ResolverFixture[] = []
-
-function fixture(): ResolverFixture {
-  const value = new ResolverFixture()
-  fixtures.push(value)
-  return value
-}
-
-afterEach(() => {
-  for (const value of fixtures.splice(0)) value.dispose()
-})
-
-describe('TsconfigPathsResolver', () => {
-  it('orders exact, longer-prefix, and longer-suffix path rules', async () => {
-    const files = fixture()
-    files.writeJson('consumer/package.json', {
-      dependencies: {
-        '@scope/feature-name': '*',
-        '@scope/feature-other': '*',
-        '@scope/plain-suffix': '*',
-      },
-    })
-    files.write('targets/exact.ts')
-    files.write('targets/prefix/other.ts')
-    files.write('targets/generic/feature-other.ts')
-    files.write('targets/suffix/plain.ts')
-    files.write('targets/generic/plain-suffix.ts')
-    const resolver = files.createResolver({
-      '@scope/*': ['./targets/generic/*'],
-      '@scope/*-suffix': ['./targets/suffix/*'],
-      '@scope/feature-*': ['./targets/prefix/*'],
-      '@scope/feature-name': ['./targets/exact.ts'],
-    })
-
-    await expect(resolver.resolve('@scope/feature-name', files.parentURL()))
-      .resolves.toBe(pathToFileURL(files.path('targets/exact.ts')).href)
-    await expect(resolver.resolve('@scope/feature-other', files.parentURL()))
-      .resolves.toBe(pathToFileURL(files.path('targets/prefix/other.ts')).href)
-    await expect(resolver.resolve('@scope/plain-suffix', files.parentURL()))
-      .resolves.toBe(pathToFileURL(files.path('targets/suffix/plain.ts')).href)
-  })
-
-  it('resolves only self-references and runtime dependencies from the nearest ancestor manifest', async () => {
-    const files = fixture()
-    files.writeJson('consumer/package.json', {
-      name: 'self-package',
-      dependencies: { dependency: '*' },
-      optionalDependencies: { optional: '*' },
-      peerDependencies: { peer: '*' },
-    })
-    for (const name of ['self-package', 'dependency', 'optional', 'peer', 'undeclared']) {
-      files.write(`targets/${name}.ts`)
-    }
-    const resolver = files.createResolver(Object.fromEntries(
-      ['self-package', 'dependency', 'optional', 'peer', 'undeclared']
-        .map(name => [name, [`./targets/${name}`]]),
-    ))
-
-    for (const name of ['self-package', 'dependency', 'optional', 'peer']) {
-      await expect(resolver.resolve(name, files.parentURL()))
-        .resolves.toBe(pathToFileURL(files.path(`targets/${name}.ts`)).href)
-    }
-    await expect(resolver.resolve('undeclared', files.parentURL())).resolves.toBeUndefined()
-  })
-
-  it('probes native TypeScript extensions and index files but excludes TSX and missing targets', async () => {
-    const files = fixture()
-    const names = ['plain-ts', 'module-mts', 'common-cts', 'directory', 'tsx-implicit', 'tsx-explicit', 'missing']
-    files.writeJson('consumer/package.json', {
-      dependencies: Object.fromEntries(names.map(name => [name, '*'])),
-    })
-    files.write('targets/plain.ts')
-    files.write('targets/module.mts')
-    files.write('targets/common.cts')
-    files.write('targets/directory/index.ts')
-    files.write('targets/component.tsx')
-    const resolver = files.createResolver({
-      'plain-ts': ['./targets/plain'],
-      'module-mts': ['./targets/module'],
-      'common-cts': ['./targets/common'],
-      'directory': ['./targets/directory'],
-      'tsx-implicit': ['./targets/component'],
-      'tsx-explicit': ['./targets/component.tsx'],
-      'missing': ['./targets/missing'],
-    })
-
-    for (const [name, target] of [
-      ['plain-ts', 'targets/plain.ts'],
-      ['module-mts', 'targets/module.mts'],
-      ['common-cts', 'targets/common.cts'],
-      ['directory', 'targets/directory/index.ts'],
-    ] as const) {
-      await expect(resolver.resolve(name, files.parentURL()))
-        .resolves.toBe(pathToFileURL(files.path(target)).href)
-    }
-    await expect(resolver.resolve('tsx-implicit', files.parentURL())).resolves.toBeUndefined()
-    await expect(resolver.resolve('tsx-explicit', files.parentURL())).resolves.toBeUndefined()
-    await expect(resolver.resolve('missing', files.parentURL())).resolves.toBeUndefined()
-  })
-
-  it('anchors inherited paths at the config that declared them', async () => {
-    const files = fixture()
-    files.writeJson('consumer/package.json', { dependencies: { custom: '*' } })
-    files.write('targets/custom.ts')
-    files.writeJson('base.json', { compilerOptions: { paths: { custom: ['./targets/custom'] } } })
-    const customTsconfig = files.writeJson('configs/custom.json', { extends: '../base.json' })
-    const resolver = TsconfigPathsResolver.create(customTsconfig)
-
-    await expect(resolver.resolve('custom', files.parentURL()))
-      .resolves.toBe(pathToFileURL(files.path('targets/custom.ts')).href)
-  })
-
-  it('short-circuits matched aliases and delegates unsupported schemes or unmatched requests', async () => {
-    const files = fixture()
-    files.writeJson('consumer/package.json', { dependencies: { matched: '*' } })
-    const target = files.write('targets/matched.ts')
-    const tsconfigPath = files.writeJson('tsconfig.json', {
-      compilerOptions: { paths: { matched: ['./targets/matched'] } },
-    })
-    initialize({ tsconfigPath })
-    const context: ResolveHookContext = {
-      conditions: [],
-      importAttributes: {},
-      parentURL: files.parentURL(),
-    }
-    const nextResolve = vi.fn(async (
-      specifier: string,
-      _context: ResolveHookContext,
-    ): Promise<ResolveFnOutput> => ({ url: `next:${specifier}` }))
-
-    await expect(resolveHook('matched', context, nextResolve))
-      .resolves.toEqual({ url: pathToFileURL(target).href, shortCircuit: true })
-    expect(nextResolve).not.toHaveBeenCalled()
-
-    for (const specifier of ['unmatched', 'node:fs', 'data:text/javascript,export default 1', 'https://example.test/mod.ts']) {
-      await expect(resolveHook(specifier, context, nextResolve)).resolves.toEqual({ url: `next:${specifier}` })
-      expect(nextResolve).toHaveBeenLastCalledWith(specifier, context)
-    }
-  })
-})
diff --git a/apps/web/tests/code-mode-fixture.snapshot.ts b/apps/web/tests/code-mode-fixture.snapshot.ts
index ce5a98b2eb..b7ab11f2c0 100644
--- a/apps/web/tests/code-mode-fixture.snapshot.ts
+++ b/apps/web/tests/code-mode-fixture.snapshot.ts
@@ -5,8 +5,9 @@
 // the code-variant parent row titled by the model-authored description, its
 // three always-visible nested sub-rows (bash through the sample registration,
 // read through GenericToolCard, the failing read wearing the error state),
-// the expanded program body, inert bash / file-link sub-row gestures, and
-// the trajectory/waterfall tabs' sub-call cells and timing lanes.
+// the expanded program body, inert bash / file-link sub-row gestures,
+// details-panel resolution of a sub-callId, and the Trajectory tab's sub-call
+// cells and timing overview.
 import { readFileSync } from 'node:fs'
 import { join } from 'node:path'
 import { act, cleanup, fireEvent, screen, waitFor, within } from '@testing-library/react'
@@ -197,7 +198,7 @@ it('expands the code row into the program body; sub-row clicks do not open detai
   `)
 })
 
-it('trajectory and waterfall surface the run_code sub-calls with real timing', async () => {
+it('trajectory surfaces run_code sub-calls in the ledger and timing overview', async () => {
   boot()
   await openFixtureSession()
 
@@ -208,51 +209,30 @@ it('trajectory and waterfall surface the run_code sub-calls with real timing', a
   }, { timeout: 10_000 })
   const subCells = [...document.querySelectorAll('[data-kind="subtool"]')]
   expect({
-    // Three Sub cells nested under the run_code Tool cell, in dispatch order,
-    // each with a real +N.Ns own-duration off the start/settle pair (the
-    // fixture spaces every event 800ms apart — never the em dash).
+    // Three Subtool cells nested under the run_code Tool cell in dispatch
+    // order, each paired with its result preview.
     subCells: subCells.map(cell => visibleText(cell)),
   }).toMatchInlineSnapshot(`
     {
       "subCells": [
-        "#51Subbash · {"command":"ls notes","description":"List notes"}+0.8s",
-        "#52Subread · {"path":"notes/demo.txt"}+0.8s",
-        "#53Subread · {"path":"notes/missing.txt"}+0.8s",
+        "SUBTOOLbash{"command":"ls notes","description":"List notes"}→demo.txt new-demo.txt",
+        "SUBTOOLread{"path":"notes/demo.txt"}→hello fixture",
+        "SUBTOOLread{"path":"notes/missing.txt"}→error",
       ],
     }
   `)
 
-  // Waterfall: each sub-call draws a measured lane scaled into the parent
-  // turn's dispatch window.
-  fireEvent.click(screen.getByRole('tab', { name: 'Waterfall' }))
-  await waitFor(() => {
-    expect(document.querySelector('[data-subspan]')).not.toBeNull()
-  }, { timeout: 10_000 })
-  const lanes = [...document.querySelectorAll('[data-subspan]')]
+  const timelineSubCalls = [...document.querySelectorAll('[data-timeline-span="subtool"]')]
   expect({
-    lanes: lanes.map(lane => ({
-      label: visibleText(lane.querySelector('[class*="subTag"]') ?? lane),
-      title: lane.querySelector('[data-timing]')?.getAttribute('title'),
-      timing: lane.querySelector('[data-timing]')?.getAttribute('data-timing'),
-    })),
+    count: timelineSubCalls.length,
+    measured: timelineSubCalls.map(span => span.getAttribute('title')?.endsWith(' · 800 ms')),
   }).toMatchInlineSnapshot(`
     {
-      "lanes": [
-        {
-          "label": "bash",
-          "timing": "measured",
-          "title": "bash · 0.80s",
-        },
-        {
-          "label": "read",
-          "timing": "measured",
-          "title": "read · 0.80s",
-        },
-        {
-          "label": "read",
-          "timing": "measured",
-          "title": "read · 0.80s",
-        },
+      "count": 3,
+      "measured": [
+        true,
+        true,
+        true,
       ],
     }
   `)
diff --git a/apps/web/tests/code-mode-round.e2e.ts b/apps/web/tests/code-mode-round.e2e.ts
index a7f493e448..9826d35f7c 100644
--- a/apps/web/tests/code-mode-round.e2e.ts
+++ b/apps/web/tests/code-mode-round.e2e.ts
@@ -118,14 +118,14 @@ describe('web e2e: Code Mode round renders nested sub-calls', () => {
     expect(await nest.locator('[data-state="error"]').count()).toBeGreaterThanOrEqual(1)
   }, 60_000)
 
-  it.skipIf(MODE === 'record')('a bash sub-row click leaves the details panel collapsed', async () => {
+  it.skipIf(MODE === 'record')('a bash sub-row click leaves the default details panel open', async () => {
     onTestFailed(() => saveFailureShot(page, 'web-e2e-code-mode-details'))
     const nest = page.locator('[data-subcalls]').first()
-    const frame = page.locator('[data-details-collapsed], [class*="frame"]').first()
-    expect(await frame.getAttribute('data-details-collapsed')).not.toBeNull()
+    const frame = page.locator('[style*="grid-template-columns"]').first()
+    expect(await frame.getAttribute('data-details-collapsed')).toBeNull()
     await nest.locator('[data-sample="bash-global"]').first().click()
-    // Tool rows no longer open details; the column stays width 0.
-    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).not.toBeNull()
+    // Tool rows do not drive layout geometry; the Session's default panel stays open.
+    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).toBeNull()
   })
 
   it.skipIf(MODE === 'record')('matches the conversation aria golden with stable anchors', async () => {
diff --git a/apps/web/tests/details-session-lifecycle.e2e.ts b/apps/web/tests/details-session-lifecycle.e2e.ts
new file mode 100644
index 0000000000..c4d6483245
--- /dev/null
+++ b/apps/web/tests/details-session-lifecycle.e2e.ts
@@ -0,0 +1,95 @@
+// Keyless browser regression for the details column's Session ownership.
+// The shipped composition retains geometry through unselected states and closes it only when a different Session takes ownership.
+import { readFile } from 'node:fs/promises'
+import { fileURLToPath } from 'node:url'
+import type { Browser, Page } from 'playwright'
+import { chromium } from 'playwright'
+import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest'
+import {
+  fixtureUserPrompts, launchWebScaffold, seedSession, watchConsole, webSnapshotMode, type WebScaffold,
+} from './scaffold.ts'
+import { connectFreshWorkspace, saveFailureShot } from './support.ts'
+
+const FIXTURE = fileURLToPath(new URL('./snapshots/lifecycle-chrome/session.jsonl', import.meta.url))
+const SEED_FIXTURE = fileURLToPath(new URL('./snapshots/seeded-history/seed.jsonl', import.meta.url))
+const PROMPT = 'Reply with the single word LIGHTHOUSE and stop.'
+const MODE = webSnapshotMode()
+
+/** Last AppFrame grid track in CSS pixels. */
+async function detailsTrack(page: Page): Promise<number> {
+  return await appFrame(page).evaluate((element) => {
+    const tracks = getComputedStyle(element).gridTemplateColumns.split(' ')
+    return Number.parseFloat(tracks.at(-1) ?? 'NaN')
+  })
+}
+
+/** AppFrame is the only product element with an inline grid track template. */
+function appFrame(page: Page) {
+  return page.locator('[style*="grid-template-columns"]').first()
+}
+
+describe.skipIf(MODE === 'record')('web e2e: details panel follows the current Session lifecycle', () => {
+  let scaffold: WebScaffold
+  let browser: Browser
+  let page: Page
+  let tripwire: ReturnType<typeof watchConsole>
+
+  beforeAll(async () => {
+    const fixture = await readFile(FIXTURE, 'utf8')
+    expect(fixtureUserPrompts(fixture)).toEqual([PROMPT])
+    scaffold = await launchWebScaffold({ replayFixture: FIXTURE, paceMs: 5 })
+    await seedSession(scaffold, await readFile(SEED_FIXTURE, 'utf8'), 'details-session-lifecycle-seed')
+    browser = await chromium.launch()
+    page = await browser.newPage({ viewport: { width: 1680, height: 1000 } })
+    tripwire = watchConsole(page)
+    await page.goto(scaffold.baseUrl, { waitUntil: 'load' })
+    await appFrame(page).waitFor({ timeout: 30_000 })
+    await connectFreshWorkspace(page)
+  }, 120_000)
+
+  afterAll(async () => {
+    await browser?.close()
+    await scaffold?.close()
+  })
+
+  it('retains geometry through hero and closes it for a different Session', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-details-session-lifecycle'))
+    const settled = scaffold.whenTurnSettled()
+    const input = page.locator('textarea').first()
+    await input.fill(PROMPT)
+    await input.press('Enter')
+    await settled
+    await page.getByText('LIGHTHOUSE', { exact: true }).waitFor({ timeout: 15_000 })
+
+    await expect.poll(() => detailsTrack(page), { timeout: 5_000 }).toBe(360)
+    expect(await page.getByText('详情', { exact: true }).count()).toBe(1)
+
+    await page.getByRole('button', { name: 'New session', exact: true }).last().click()
+    await page.getByText("Let's start building", { exact: false }).waitFor({ timeout: 15_000 })
+    await expect.poll(() => detailsTrack(page), { timeout: 5_000 }).toBe(0)
+    expect(await page.getByText('详情', { exact: true }).isVisible()).toBe(false)
+
+    const original = page.locator('[role=treeitem]').filter({ hasText: 'Reply with the single word' }).first()
+    await original.click()
+    await page.getByText('LIGHTHOUSE', { exact: true }).waitFor({ timeout: 15_000 })
+    await expect.poll(() => detailsTrack(page), { timeout: 5_000 }).toBe(360)
+    expect(await page.getByText('详情', { exact: true }).count()).toBe(1)
+
+    const ungrouped = page.getByText('Ungrouped', { exact: true })
+    const ungroupedRow = ungrouped.locator('..').locator('..')
+    const ungroupedSection = ungroupedRow.locator('..')
+    await expect.poll(async () => {
+      if (await ungroupedRow.getAttribute('aria-expanded') !== 'true') {
+        await ungrouped.click()
+        await page.waitForTimeout(50)
+      }
+      return await ungroupedRow.getAttribute('aria-expanded')
+    }, { timeout: 5_000 }).toBe('true')
+    const seeded = ungroupedSection.locator('[role="treeitem"]').nth(1)
+    await seeded.click()
+    await page.getByText('DONE', { exact: true }).waitFor({ timeout: 15_000 })
+    await expect.poll(() => detailsTrack(page), { timeout: 5_000 }).toBe(0)
+    expect(tripwire.pageErrors).toEqual([])
+    expect(tripwire.warnings).toEqual([])
+  }, 90_000)
+})
diff --git a/apps/web/tests/lifecycle-chrome.e2e.ts b/apps/web/tests/lifecycle-chrome.e2e.ts
index 4b54242495..a07db275fd 100644
--- a/apps/web/tests/lifecycle-chrome.e2e.ts
+++ b/apps/web/tests/lifecycle-chrome.e2e.ts
@@ -101,23 +101,15 @@ describe('web e2e: lifecycle & chrome (workspace flow / reload / dark mode)', ()
 
   it.skipIf(MODE === 'record')('recovers the whole surface across a reload from the log alone', async () => {
     onTestFailed(() => saveFailureShot(page, 'web-e2e-lifecycle-reload'))
-    // Fold a layout preference into the same reload: collapse the sidebar
-    // (persisted under dsh.layout.panels) before reloading.
-    await page.getByRole('button', { name: 'Collapse sidebar' }).click()
-    await expect.poll(() => page.getByRole('button', { name: 'Open sidebar' }).count(), { timeout: 10_000 }).toBe(1)
     const warningStart = tripwire.warnings.length
     await page.reload({ waitUntil: 'load' })
     await page.waitForSelector('[class*="frame"]', { timeout: 30_000 })
     acknowledgeReloadConnectionLoss(tripwire, warningStart)
-    // Layout persisted: the sidebar comes back collapsed.
-    await expect.poll(() => page.getByRole('button', { name: 'Open sidebar' }).count(), { timeout: 10_000 }).toBe(1)
     // Selection persisted (dsh.sessions.current) and history replayed: the
     // recorded turn re-renders from session.history with zero model calls —
     // the replay cursor was fully consumed before the reload, so any stray
     // request would fail the scenario loudly at close().
     await expect.poll(() => page.getByText('LIGHTHOUSE', { exact: true }).count(), { timeout: 15_000 }).toBeGreaterThanOrEqual(1)
-    // Expand back and confirm the tree still lists the materialized session.
-    await page.getByRole('button', { name: 'Open sidebar' }).click()
     await expect.poll(() => page.locator('[role="treeitem"][aria-selected="true"]').count(), { timeout: 10_000 }).toBe(1)
     // Golden of the recovered conversation region: rebuilt from the log, it
     // must render the same settled transcript the live turn produced.
diff --git a/apps/web/tests/message-actions.e2e.ts b/apps/web/tests/message-actions.e2e.ts
new file mode 100644
index 0000000000..a453d53cbc
--- /dev/null
+++ b/apps/web/tests/message-actions.e2e.ts
@@ -0,0 +1,93 @@
+// Web e2e scenario: message IconActions + clocks. Cold-seeds the seeded-history
+// fixture (zero model calls) and pins the settled conversation aria after the
+// user/assistant footers are focus-revealed — the surface package jsdom tests
+// cannot substitute for (docs/testing.md snapshot rule).
+import { mkdir, readFile, writeFile } from 'node:fs/promises'
+import { join } from 'node:path'
+import { fileURLToPath } from 'node:url'
+import type { Browser, Page } from 'playwright'
+import { chromium } from 'playwright'
+import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest'
+import {
+  assertFixtureInventory, captureStableAria, compareOrRefreshGolden, fixtureUserPrompts,
+  launchWebScaffold, seedSession, watchConsole, webSnapshotMode, type WebScaffold,
+} from './scaffold.ts'
+import { saveFailureShot } from './support.ts'
+
+const SNAPSHOT_DIR = fileURLToPath(new URL('./snapshots/message-actions', import.meta.url))
+// Borrowed read-only: this scenario needs any settled user+assistant pair, not
+// a new recording (workspace-management / sidebar-scrollbar pattern).
+const SEED = fileURLToPath(new URL('./snapshots/seeded-history/seed.jsonl', import.meta.url))
+const UI_EXPECTED = join(SNAPSHOT_DIR, 'ui.expected.md')
+const MODE = webSnapshotMode()
+const SEED_ID = 'message-actions-web-e2e'
+
+const PROMPT = 'Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop.'
+
+describe('web e2e: message IconActions and clocks on settled history', () => {
+  let scaffold: WebScaffold
+  let browser: Browser
+  let page: Page
+  let tripwire: ReturnType<typeof watchConsole>
+
+  beforeAll(async () => {
+    scaffold = await launchWebScaffold({})
+    const sessionCwd = join(scaffold.workspaceCwd, 'workspace')
+    await mkdir(sessionCwd, { recursive: true })
+    await writeFile(join(sessionCwd, 'a.txt'), 'alpha\n')
+    await writeFile(join(sessionCwd, 'b.txt'), 'beta\n')
+    const raw = await readFile(SEED, 'utf8')
+    expect(fixtureUserPrompts(raw), 'borrowed seed must carry the drive prompt').toEqual([PROMPT])
+    await seedSession(scaffold, raw, SEED_ID)
+    browser = await chromium.launch()
+    page = await browser.newPage({ viewport: { width: 1680, height: 1000 } })
+    tripwire = watchConsole(page)
+    await page.goto(scaffold.baseUrl, { waitUntil: 'load' })
+    await page.waitForSelector('[class*="frame"]', { timeout: 30_000 })
+  }, 120_000)
+
+  afterAll(async () => {
+    await browser?.close()
+    await scaffold?.close()
+  })
+
+  it.skipIf(MODE === 'record')('lists the seeded session and reveals user/assistant IconActions', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-message-actions'))
+    const groupRow = page.locator('[role="treeitem"]').first()
+    await groupRow.waitFor({ timeout: 15_000 })
+    await groupRow.click()
+    const sessionRow = page.locator('[role="treeitem"]').nth(1)
+    await sessionRow.waitFor({ timeout: 10_000 })
+    await sessionRow.click()
+    await expect.poll(() => page.getByText('DONE', { exact: true }).count(), { timeout: 15_000 }).toBe(1)
+
+    // Focus-reveal the footers (hover:hover keeps them opacity-hidden until
+    // hover/focus-within). User has three actions; each finalized assistant
+    // text node has copy + branch.
+    const copyButtons = page.getByRole('button', { name: '复制' })
+    await expect.poll(() => copyButtons.count(), { timeout: 10_000 }).toBeGreaterThanOrEqual(2)
+    await copyButtons.first().focus()
+    await expect.poll(() => page.getByRole('button', { name: '在新对话中分支' }).count(), { timeout: 5_000 })
+      .toBeGreaterThanOrEqual(2)
+    await expect.poll(() => page.getByRole('button', { name: '编辑' }).count(), { timeout: 5_000 }).toBe(1)
+  }, 60_000)
+
+  it.skipIf(MODE === 'record')('matches the conversation aria golden with IconActions and clocks', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-message-actions-aria'))
+    await page.getByRole('button', {
+      name: '选择模型，当前 deepseek-v4-flash',
+    }).waitFor({ timeout: 10_000 })
+    // Keep a footer focused so opacity-hidden actions stay in the a11y tree
+    // as an active/focused control during the capture.
+    await page.getByRole('button', { name: '复制' }).first().focus()
+    const snapshot = (await captureStableAria(page, '[class*="centerCol"]', scaffold.workspaceCwd))
+      .split(SEED_ID).join('{{seededId}}')
+    await compareOrRefreshGolden(UI_EXPECTED, snapshot, MODE)
+  })
+
+  it.skipIf(MODE === 'record')('issued zero model calls and kept a closed inventory', async () => {
+    expect(tripwire.pageErrors).toEqual([])
+    expect(tripwire.warnings).toEqual([])
+    await assertFixtureInventory(SNAPSHOT_DIR, ['ui.expected.md'])
+  })
+})
diff --git a/apps/web/tests/navigation-panes.e2e.ts b/apps/web/tests/navigation-panes.e2e.ts
index a27a61f80f..0f42d911aa 100644
--- a/apps/web/tests/navigation-panes.e2e.ts
+++ b/apps/web/tests/navigation-panes.e2e.ts
@@ -1,11 +1,11 @@
-// Web e2e scenarios: navigation & panes — the view tabs (Trajectory /
-// Waterfall) and sidebar search, all over ONE rich two-turn seeded fixture
-// rendered purely from the log (the seeded-history pattern: zero model calls
-// in replay, so every surface here is the client fold + host history RPC,
-// not replay binding). The seed is recorded live under the standard
-// discipline: turn 1 produces a bash call plus two parallel reads in one
-// assistant message (tool-call density for the trajectory/waterfall lanes),
-// turn 2 a markdown-rich reply (a second turn so the waterfall has two lanes).
+// Web e2e scenarios: navigation & panes — the Trajectory view and timing
+// overview, its local details inspector, and sidebar search, all over ONE rich
+// two-turn seeded fixture rendered purely from the log (the seeded-history
+// pattern: zero model calls in replay, so every surface here is the client
+// fold + host history RPC, not replay binding). The seed is recorded live
+// under the standard discipline: turn 1 produces a bash call plus two
+// parallel reads in one assistant message (tool-call density for the
+// trajectory ledger/timing lanes), turn 2 a markdown-rich reply.
 import { mkdir, readFile, writeFile } from 'node:fs/promises'
 import { fileURLToPath } from 'node:url'
 import { join } from 'node:path'
@@ -23,7 +23,7 @@ import { saveFailureShot } from './support.ts'
 const SNAPSHOT_DIR = fileURLToPath(new URL('./snapshots/navigation-panes', import.meta.url))
 const SEED = join(SNAPSHOT_DIR, 'seed.jsonl')
 const TRAJECTORY_EXPECTED = join(SNAPSHOT_DIR, 'trajectory.expected.md')
-const WATERFALL_EXPECTED = join(SNAPSHOT_DIR, 'waterfall.expected.md')
+const TERMINAL_EXPECTED = join(SNAPSHOT_DIR, 'terminal-card.expected.md')
 const MODE = webSnapshotMode()
 const SEED_ID = 'navigation-panes-web-e2e'
 
@@ -38,6 +38,7 @@ describe('web e2e: navigation & panes over a rich seeded session', () => {
   let browser: Browser
   let page: Page
   let tripwire: ReturnType<typeof watchConsole>
+  let slotErrors: string[]
 
   beforeAll(async () => {
     scaffold = await launchWebScaffold({})
@@ -57,6 +58,12 @@ describe('web e2e: navigation & panes over a rich seeded session', () => {
     browser = await chromium.launch()
     page = await browser.newPage({ viewport: { width: 1680, height: 1000 } })
     tripwire = watchConsole(page)
+    slotErrors = []
+    page.on('console', (message) => {
+      if (message.type() === 'error' && /slot entry crashed/i.test(message.text())) {
+        slotErrors.push(message.text())
+      }
+    })
     await page.goto(scaffold.baseUrl, { waitUntil: 'load' })
     await page.waitForSelector('[class*="frame"]', { timeout: 30_000 })
   }, 120_000)
@@ -122,58 +129,157 @@ describe('web e2e: navigation & panes over a rich seeded session', () => {
     await expect.poll(() => page.locator('[role="treeitem"]').count(), { timeout: 10_000 }).toBeGreaterThanOrEqual(1)
   }, 60_000)
 
-  it.skipIf(MODE === 'record')('renders the trajectory tab with turn sections and step cells', async () => {
+  it.skipIf(MODE === 'record')('renders the trajectory ledger and opens its local record inspector', async () => {
     onTestFailed(() => saveFailureShot(page, 'web-e2e-navigation-trajectory'))
     await page.getByRole('tab', { name: 'Trajectory' }).click()
-    // Two sticky turn sections; turn 1's step group summarizes its tool mix
-    // (bash + the two parallel reads collapse to 'bash read×2').
-    await expect.poll(() => page.getByText('Turn 1', { exact: true }).count(), { timeout: 15_000 }).toBe(1)
-    await expect.poll(() => page.getByText('Turn 2', { exact: true }).count(), { timeout: 10_000 }).toBe(1)
-    await expect.poll(() => page.getByText('bash read×2', { exact: false }).count(), { timeout: 10_000 }).toBe(1)
+    await page.waitForTimeout(100)
+    expect({
+      pageErrors: tripwire.pageErrors,
+      slotErrors,
+      warnings: tripwire.warnings,
+    }).toEqual({
+      pageErrors: [],
+      slotErrors: [],
+      warnings: [],
+    })
+    // Turn rules partition the ledger without restoring a separate header row.
+    await expect.poll(() => page.locator('tr[data-turn-start="true"]').count(), { timeout: 15_000 }).toBe(2)
+    await expect.poll(() => page.getByRole('columnheader').count(), { timeout: 10_000 }).toBe(0)
+    await page.locator('tr[data-kind="tool"]').first().click()
+    await expect.poll(() => page.getByRole('complementary', { name: 'Event details' }).count(), { timeout: 10_000 }).toBe(1)
+    await page.getByRole('tab', { name: 'Result' }).click()
+    await expect.poll(() => page.getByText('NAVIGATION_OK', { exact: false }).count(), { timeout: 10_000 }).toBeGreaterThanOrEqual(1)
     const snapshot = (await captureStableAria(page, '[class*="viewArea"]', scaffold.workspaceCwd))
       .split(SEED_ID).join('{{seededId}}')
     await compareOrRefreshGolden(TRAJECTORY_EXPECTED, snapshot, MODE)
+    await page.getByRole('complementary', { name: 'Event details' })
+      .getByRole('button', { name: 'Close details' }).click()
   }, 60_000)
 
-  it.skipIf(MODE === 'record')('renders the waterfall tab with span stats and one lane per span', async () => {
-    onTestFailed(() => saveFailureShot(page, 'web-e2e-navigation-waterfall'))
-    await page.getByRole('tab', { name: 'Waterfall' }).click()
-    // The stats header rides the waterfall body. The span fold counts THREE
-    // spans for this two-turn log: only assistant/steering nodes carry a turn
-    // number, so the first user message lands in a turn-0 prologue span (a
-    // P-I placeholder shape — pinned as-is; real spans are deferred to
-    // P-III per the view's deviation ledger). Calls: bash + two reads.
-    await expect.poll(() => page.getByText(/3 turns · \d+ steps · 3 tool calls/).count(), { timeout: 15_000 }).toBe(1)
-    // One lane per span, tagged by turn number, prologue included.
-    for (const tag of ['turn 0', 'turn 1', 'turn 2']) {
-      await expect.poll(() => page.getByText(tag, { exact: true }).count(), { timeout: 10_000 }).toBe(1)
-    }
-    const snapshot = (await captureStableAria(page, '[class*="viewArea"]', scaffold.workspaceCwd))
-      .split(SEED_ID).join('{{seededId}}')
-    await compareOrRefreshGolden(WATERFALL_EXPECTED, snapshot, MODE)
+  it.skipIf(MODE === 'record')('focuses the ledger by dragging an overview interval', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-navigation-timeline'))
+    const plot = page.getByLabel('Timeline overview; drag horizontally to focus events')
+    const before = await page.locator('tr[data-kind]').count()
+    const box = await plot.boundingBox()
+    if (box === null) throw new Error('trajectory timeline plot has no layout box')
+    await page.mouse.move(box.x + box.width * 0.55, box.y + box.height / 2)
+    await page.mouse.down()
+    await page.mouse.move(box.x + box.width * 0.9, box.y + box.height / 2)
+    await page.mouse.up()
+    await expect.poll(() => page.locator('tr[data-timeline-focus="outside"]').count(), { timeout: 10_000 })
+      .toBeGreaterThan(0)
+    await expect.poll(() => page.locator('tr[data-kind]').count(), { timeout: 10_000 }).toBe(before)
+    await plot.click({ button: 'right' })
+    await expect.poll(() => page.locator('tr[data-timeline-focus]').count(), { timeout: 10_000 }).toBe(0)
   }, 60_000)
 
-  it.skipIf(MODE === 'record')('bash and file-path rows leave the details column collapsed', async () => {
+  it.skipIf(MODE === 'record')('bash and file-path rows leave the default details column open', async () => {
     onTestFailed(() => saveFailureShot(page, 'web-e2e-navigation-details'))
     await page.getByRole('tab', { name: 'Chat' }).click()
     const bashRow = page.locator('[data-sample="bash-global"]').first()
     await bashRow.waitFor({ timeout: 15_000 })
-    const frame = page.locator('[data-details-collapsed], [class*="frame"]').first()
-    expect(await frame.getAttribute('data-details-collapsed')).not.toBeNull()
+    const frame = page.locator('[style*="grid-template-columns"]').first()
+    expect(await frame.getAttribute('data-details-collapsed')).toBeNull()
     await bashRow.click()
-    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).not.toBeNull()
+    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).toBeNull()
+    // The card's own controls are outside the summary row and must not open
+    // details either — the terminal card is read in place.
+    await page.locator('[data-sample="bash-global"] ~ [data-terminal] [class*="_copyButton_"]').first().click()
+    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).toBeNull()
     // Read summaries are host-open file links; they also must not open details.
     const fileLink = page.locator('[data-variant="read"] button').first()
     await fileLink.waitFor({ timeout: 10_000 })
     await fileLink.click()
-    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).not.toBeNull()
+    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).toBeNull()
+  }, 60_000)
+
+  it.skipIf(MODE === 'record')('renders the bash row as a terminal card in the real browser', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-navigation-terminal'))
+    await page.getByRole('tab', { name: 'Chat' }).click()
+    // The card is resident in the keyed bash row (no expand gesture): the
+    // recorded command's own output sits in the message flow, derived from the
+    // logged call/result presentations alone.
+    const card = page.locator('[data-sample="bash-global"] ~ [data-terminal], [data-sample="bash-global"] [data-terminal]').first()
+    await card.waitFor({ timeout: 15_000 })
+    // Real layout, not jsdom's stub (which computes no geometry at all):
+    // squeeze the output pane below its content width and the line must keep
+    // its single row and overflow sideways instead of folding. Soft-wrapping
+    // here is what shredded the column alignment this card exists to hold.
+    const layout = await card.locator('[class*="_output_"]').first().evaluate((node) => {
+      const pane = node as HTMLElement
+      const row = pane.querySelector<HTMLElement>('[class*="_line_"]')
+      if (row === null) throw new Error('output pane has no line')
+      const before = row.offsetHeight
+      const restore = pane.style.width
+      pane.style.width = '8px'
+      const squeezed = { wrapped: row.offsetHeight > before, scrollsSideways: pane.scrollWidth > pane.clientWidth }
+      pane.style.width = restore
+      return { whiteSpace: getComputedStyle(row).whiteSpace, overflowX: getComputedStyle(pane).overflowX, ...squeezed }
+    })
+    expect(layout).toEqual({ whiteSpace: 'pre', overflowX: 'auto', wrapped: false, scrollsSideways: true })
+    // The run-state dot's color is the whole point of it and is the one thing
+    // jsdom cannot report: --dsw-* tokens resolve only against the real theme
+    // stylesheet. This command settled cleanly, so the dot must be the green
+    // success token — a red one here would read as a failed command.
+    const dot = await card.locator('[class*="_runState_"][data-state]').first().evaluate((node) => {
+      // The token lives on body, so the probe must sit in the same cascade.
+      const probe = document.createElement('span')
+      probe.style.color = 'var(--dsw-alias-state-success-primary)'
+      document.body.appendChild(probe)
+      const success = getComputedStyle(probe).color
+      probe.remove()
+      return {
+        state: node.getAttribute('data-state'),
+        color: getComputedStyle(node as HTMLElement).color,
+        success,
+        // One label per card (the state is the call's), so it hangs off the
+        // prompt column rather than the row the dot sits in.
+        label: node.closest('[class*="_prompt_"]')?.querySelector('[class*="_runStateLabel_"]')?.textContent ?? null,
+        // The dot precedes the prompt label in document order, which is what
+        // puts it to the left of the `$`.
+        beforePrompt: node.compareDocumentPosition(node.parentElement!.querySelector('[class*="_cwd_"]')!)
+          === Node.DOCUMENT_POSITION_FOLLOWING,
+        // The dot lives in the card's OWN left padding, so it sits inside the
+        // card box yet left of the prompt text. Owning the reservation as padding
+        // rather than margin is what keeps a consumer's own margin from
+        // cancelling it and letting a container clip the dot — geometry jsdom
+        // cannot compute.
+        insideCard: (node as HTMLElement).getBoundingClientRect().left
+          >= (node.closest('[data-terminal]')?.getBoundingClientRect().left ?? Infinity),
+        leftOfPrompt: (node as HTMLElement).getBoundingClientRect().right
+          <= (node.closest('[class*="_promptLine_"]')
+            ?.querySelector('[class*="_cwd_"]')
+            ?.getBoundingClientRect().left ?? -Infinity),
+      }
+    })
+    expect(dot.state).toBe('done')
+    expect(dot.label).toBe('已完成')
+    expect(dot.beforePrompt).toBe(true)
+    expect(dot.insideCard).toBe(true)
+    expect(dot.leftOfPrompt).toBe(true)
+    // Resolved through the theme token, not a literal hex in the component.
+    expect(dot.success).toMatch(/^rgb/)
+    expect(dot.color).toBe(dot.success)
+    // Golden of the card at rest — captured before the copy click, whose
+    // confirmation label self-reverts on a timer and would not hold still.
+    const snapshot = (await captureStableAria(page, '[data-terminal]', scaffold.workspaceCwd))
+      .split(SEED_ID).join('{{seededId}}')
+    await compareOrRefreshGolden(TERMINAL_EXPECTED, snapshot, MODE)
+    // Copy writes the raw output through the browser's own clipboard, which in
+    // a real page is the async Clipboard API rather than the jsdom fallback.
+    await page.context().grantPermissions(['clipboard-read', 'clipboard-write'])
+    await card.locator('[class*="_copyButton_"]').first().click()
+    await expect.poll(() => card.locator('[class*="_copyButton_"]').first().textContent(), { timeout: 5_000 })
+      .toBe('复制成功')
+    expect(await page.evaluate(() => navigator.clipboard.readText())).toContain('NAVIGATION_OK')
   }, 60_000)
 
   it.skipIf(MODE === 'record')('issued zero model calls and stayed clean', async () => {
     expect(tripwire.pageErrors).toEqual([])
+    expect(slotErrors).toEqual([])
     expect(tripwire.warnings).toEqual([])
     await assertFixtureInventory(SNAPSHOT_DIR, [
-      'seed.jsonl', 'trajectory.expected.md', 'waterfall.expected.md',
+      'seed.jsonl', 'trajectory.expected.md', 'terminal-card.expected.md',
     ])
   })
 })
diff --git a/apps/web/tests/question-composer.e2e.ts b/apps/web/tests/question-composer.e2e.ts
index 6ecdb683b3..8689a74fc2 100644
--- a/apps/web/tests/question-composer.e2e.ts
+++ b/apps/web/tests/question-composer.e2e.ts
@@ -104,7 +104,7 @@ describe('web e2e: resident question composer round trip', () => {
             const inner = child.getBoundingClientRect()
             return Math.max(box.top - inner.top, inner.bottom - box.bottom)
           })))
-          const list = rows[0]?.parentElement ?? null
+          const list = card.querySelector<HTMLElement>('[data-question-scroll]')
           return {
             rows: rows.length,
             spill: Math.max(...spill),
diff --git a/apps/web/tests/scaffold.ts b/apps/web/tests/scaffold.ts
index f722ead61b..753cdc2953 100644
--- a/apps/web/tests/scaffold.ts
+++ b/apps/web/tests/scaffold.ts
@@ -389,6 +389,11 @@ function normalizeAria(snapshot: string, workspaceCwd: string): string {
     .split(base).join('{{workspace}}')
     .replace(/[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}/gi, '{{uuid}}')
     .replace(/\b\d+(?:\.\d+)?(?:ms|s|秒)\b/g, '{{duration}}')
+    // Message IconActions clocks widen by calendar day/year; collapse every
+    // shape so goldens stay stable across midnight and year boundaries.
+    .replace(/\d{4}年\d{1,2}月\d{1,2}日 \d{2}:\d{2}/g, '{{clock}}')
+    .replace(/\d{1,2}月\d{1,2}日 \d{2}:\d{2}/g, '{{clock}}')
+    .replace(/(?<!\d)\d{2}:\d{2}(?!\d)/g, '{{clock}}')
 }
 
 /**
diff --git a/apps/web/tests/seeded-history.e2e.ts b/apps/web/tests/seeded-history.e2e.ts
index d30b3d7c39..80d84d7b2f 100644
--- a/apps/web/tests/seeded-history.e2e.ts
+++ b/apps/web/tests/seeded-history.e2e.ts
@@ -139,10 +139,10 @@ describe('web e2e: seeded history renders through cold resume', () => {
     // capture; still zero model calls.
     const fileLink = page.locator('[data-variant="read"] button').first()
     await fileLink.waitFor({ timeout: 10_000 })
-    const frame = page.locator('[data-details-collapsed], [class*="frame"]').first()
-    expect(await frame.getAttribute('data-details-collapsed')).not.toBeNull()
+    const frame = page.locator('[style*="grid-template-columns"]').first()
+    expect(await frame.getAttribute('data-details-collapsed')).toBeNull()
     await fileLink.click()
-    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).not.toBeNull()
+    await expect.poll(() => frame.getAttribute('data-details-collapsed'), { timeout: 5_000 }).toBeNull()
     // Path label survives from the recorded args (a.txt).
     await expect.poll(() => page.getByText('a.txt', { exact: false }).count(), { timeout: 5_000 }).toBeGreaterThan(0)
   })
diff --git a/apps/web/tests/session-actions.snapshot.ts b/apps/web/tests/session-actions.snapshot.ts
new file mode 100644
index 0000000000..684afe5532
--- /dev/null
+++ b/apps/web/tests/session-actions.snapshot.ts
@@ -0,0 +1,130 @@
+// @vitest-environment jsdom
+// Session row actions in the assembled fixture app: Rename opens the
+// browser-owned dialog and settles the title from the unary response.
+import { readFileSync } from 'node:fs'
+import { join } from 'node:path'
+import { act, cleanup, fireEvent, screen, waitFor, within } from '@testing-library/react'
+import { afterEach, beforeEach, expect, it, vi } from 'vitest'
+import type { WebBootEntry } from '@deepseek-ai/dsh-client-modules/client'
+import { AppWebEntry } from '@deepseek-ai/dsh-client-web'
+
+const PLUGINS: readonly (WebBootEntry & { dir: string })[] = [
+  { id: '@deepseek-ai/dsh-client-connection', dir: 'connection', url: '/plugins/connection.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-runtime', dir: 'runtime', url: '/plugins/runtime.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-connection'], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-theme', dir: 'ui-theme', url: '/plugins/ui-theme.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-locale', dir: 'locale', url: '/plugins/locale.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-layout', dir: 'ui-layout', url: '/plugins/ui-layout.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-runtime'] },
+  { id: '@deepseek-ai/dsh-client-ui-sidebar', dir: 'ui-sidebar', url: '/plugins/ui-sidebar.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  { id: '@deepseek-ai/dsh-client-ui-conversation', dir: 'ui-conversation', url: '/plugins/ui-conversation.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  { id: '@deepseek-ai/dsh-client-ui-slash', dir: 'ui-slash', url: '/plugins/ui-slash.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-runtime', '@deepseek-ai/dsh-client-ui-conversation'] },
+  { id: '@deepseek-ai/dsh-client-ui-command', dir: 'ui-command', url: '/plugins/ui-command.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-slash', '@deepseek-ai/dsh-client-ui-conversation'] },
+  { id: '@deepseek-ai/dsh-client-ui-workspace', dir: 'ui-workspace', url: '/plugins/ui-workspace.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-runtime', '@deepseek-ai/dsh-client-ui-conversation', '@deepseek-ai/dsh-client-ui-sidebar'] },
+]
+
+const bundles = new Map(PLUGINS.map(plugin => [
+  plugin.url,
+  readFileSync(join(process.cwd(), 'packages/client', plugin.dir, 'lib/client.js'), 'utf8'),
+]))
+
+interface FixtureWindow extends Window {
+  __DSH_BOOT__?: { rev: string; entries: WebBootEntry[] }
+  __ModuleLoader__?: unknown
+}
+
+class ResizeObserverStub {
+  observe(): void {}
+  disconnect(): void {}
+  unobserve(): void {}
+}
+
+const win = window as FixtureWindow
+let unmount: (() => void) | undefined
+
+beforeEach(() => {
+  localStorage.clear()
+  history.replaceState(null, '', '/?fixture')
+  document.title = 'DeepSeek Harness'
+  const root = document.createElement('div')
+  root.id = 'root'
+  document.body.appendChild(root)
+  vi.stubGlobal('ResizeObserver', ResizeObserverStub)
+  vi.stubGlobal('requestAnimationFrame', (callback: FrameRequestCallback) =>
+    setTimeout(() => { callback(0) }, 0) as unknown as number)
+  vi.stubGlobal('cancelAnimationFrame', (id: number) => { clearTimeout(id) })
+  win.__DSH_BOOT__ = { rev: 'fx', entries: PLUGINS.map(({ dir: _dir, ...plugin }) => plugin) }
+})
+
+afterEach(() => {
+  act(() => { unmount?.() })
+  unmount = undefined
+  cleanup()
+  delete win.__DSH_BOOT__
+  delete win.__ModuleLoader__
+  delete (globalThis as Record<string, unknown>).__fxTiming
+  document.body.innerHTML = ''
+  document.head.querySelectorAll('style[data-plugin]').forEach((style) => { style.remove() })
+  document.title = ''
+  history.replaceState(null, '', '/')
+  vi.unstubAllGlobals()
+})
+
+async function bootApp(): Promise<void> {
+  const root = document.querySelector<HTMLElement>('#root')
+  if (root === null) throw new Error('snapshot root missing')
+  act(() => {
+    const entry = new AppWebEntry(root, {
+      fetchBundle: (url) => {
+        const code = bundles.get(url)
+        return code === undefined ? Promise.reject(new Error(`missing built bundle ${url}`)) : Promise.resolve(code)
+      },
+      executeBundle: (code) => { (0, eval)(code) },
+    })
+    void entry.run()
+    unmount = () => { entry.dispose() }
+  })
+  await screen.findByRole('tree', { name: 'Sessions' }, { timeout: 10_000 })
+}
+
+/** The session row element carrying the given visible label. */
+function rowOf(label: string): HTMLElement {
+  const tree = screen.getByRole('tree', { name: 'Sessions' })
+  const row = within(tree).getByText(label).closest<HTMLElement>('[role="treeitem"]')
+  if (row === null) throw new Error(`session row "${label}" missing`)
+  return row
+}
+
+/** Open the row's ... menu and click one action. The anchor button is
+ *  CSS-hover-revealed (real stylesheets are injected in this assembled run,
+ *  so role queries filter it as hidden); target it directly. */
+function pickRowAction(label: string, action: string): void {
+  const anchor = rowOf(label).querySelector<HTMLElement>(`button[aria-label="Session actions for ${label}"]`)
+  if (anchor === null) throw new Error(`row menu anchor for "${label}" missing`)
+  fireEvent.click(anchor)
+  fireEvent.click(screen.getByRole('menuitem', { name: action, hidden: true }))
+}
+
+it('renames a session through the row-menu dialog; the row settles from the unary response', async () => {
+  await bootApp()
+  const sourceLabel = 'Fixture 历史会话'
+  await screen.findByText(sourceLabel)
+
+  pickRowAction(sourceLabel, 'Rename')
+  const input = await screen.findByLabelText('Session name')
+  expect((input as HTMLInputElement).value).toBe(sourceLabel)
+  fireEvent.change(input, { target: { value: '  分叉  实验记录  ' } })
+  fireEvent.click(screen.getByRole('button', { name: 'Rename' }))
+
+  // Host-side normalization collapses whitespace; the dialog closes on
+  // acceptance and the row re-labels without any push-frame wait.
+  const renamed = '分叉 实验记录'
+  await waitFor(() => { expect(screen.queryByLabelText('Session name')).toBeNull() })
+  await screen.findByText(renamed)
+  const tree = screen.getByRole('tree', { name: 'Sessions' })
+  expect(within(tree).queryByText(sourceLabel)).toBeNull()
+
+  const rows = [...tree.querySelectorAll('[role="treeitem"]')].map(row => ({
+    label: row.textContent?.replace(/\s+/g, ' ').trim() ?? '',
+  }))
+  await expect(`${JSON.stringify(rows, null, 2)}\n`)
+    .toMatchFileSnapshot('./snapshots/session-actions/rename-rows.json')
+})
diff --git a/apps/web/tests/session-title.snapshot.ts b/apps/web/tests/session-title.snapshot.ts
index cc08da6cc0..4667b85079 100644
--- a/apps/web/tests/session-title.snapshot.ts
+++ b/apps/web/tests/session-title.snapshot.ts
@@ -143,7 +143,7 @@ it('projects titles and routes the next turn through the selected model in the b
   // allowed for the next turn; stop the fixture's resident run before sending
   // the route-report prompt.
   fireEvent.click(screen.getByRole('button', { name: 'Stop generating' }))
-  const composer = await screen.findByPlaceholderText('Message the agent')
+  const composer = await screen.findByPlaceholderText('给智能体发消息')
   fireEvent.change(composer, { target: { value: 'report model' } })
   fireEvent.keyDown(composer, { key: 'Enter' })
   await screen.findByText('当前模型：openai/gpt-5 · 推理等级：max', {}, { timeout: 10_000 })
diff --git a/apps/web/tests/slash-flow.snapshot.ts b/apps/web/tests/slash-flow.snapshot.ts
index ac47ca4cd0..c650d8e6de 100644
--- a/apps/web/tests/slash-flow.snapshot.ts
+++ b/apps/web/tests/slash-flow.snapshot.ts
@@ -57,12 +57,16 @@ class ResizeObserverStub {
   unobserve(): void {}
 }
 
+// jsdom has no scrollIntoView; the slash menu follows its highlighted option.
+const scrollIntoView = vi.fn()
 const win = window as FixtureWindow
 let unmount: (() => void) | undefined
 
 beforeEach(() => {
   localStorage.clear()
   document.title = 'DeepSeek Harness'
+  Element.prototype.scrollIntoView = scrollIntoView
+  scrollIntoView.mockClear()
   vi.stubGlobal('ResizeObserver', ResizeObserverStub)
   vi.stubGlobal('requestAnimationFrame', (callback: FrameRequestCallback) =>
     setTimeout(() => { callback(0) }, 0) as unknown as number)
@@ -80,6 +84,7 @@ afterEach(() => {
   document.head.querySelectorAll('style[data-plugin]').forEach((style) => { style.remove() })
   document.title = ''
   history.replaceState(null, '', '/')
+  Reflect.deleteProperty(Element.prototype, 'scrollIntoView')
   vi.unstubAllGlobals()
 })
 
diff --git a/apps/web/tests/smoke-real.e2e.ts b/apps/web/tests/smoke-real.e2e.ts
index 7b7721ce2c..aceb002d05 100644
--- a/apps/web/tests/smoke-real.e2e.ts
+++ b/apps/web/tests/smoke-real.e2e.ts
@@ -357,16 +357,20 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY || notReady.length > 0)('web smoke
     requireDist()
     sessionsDir = mkdtempSync(join(tmpdir(), 'dsh-web-w5-'))
     const port = await probeFreePort()
-    // tsx boot mirrors demo:web — lib/ may be unbuilt in this worktree. cwd is a
-    // temp dir (persistenceRoot is cwd-relative), so tsx needs the repo's loader
-    // and tsconfig paths pointed at explicitly.
+    // tsx boot mirrors demo:web — lib/ may be unbuilt in this worktree. Isolate
+    // the global Harness home inside the temp world; tsx also needs the repo's
+    // loader and tsconfig paths pointed at explicitly.
     const tsxLoader = pathToFileURL(createRequire(join(REPO_ROOT, 'package.json')).resolve('tsx')).href
     child = spawn(
       process.execPath,
       ['--import', tsxLoader, join(REPO_ROOT, 'apps/cli/src/bin.ts'), 'web', '--port', String(port)],
       {
         cwd: sessionsDir,
-        env: { ...process.env, TSX_TSCONFIG_PATH: join(REPO_ROOT, 'tsconfig.json') },
+        env: {
+          ...process.env,
+          DSH_HOME: join(sessionsDir, '.dsh'),
+          TSX_TSCONFIG_PATH: join(REPO_ROOT, 'tsconfig.json'),
+        },
         stdio: ['ignore', 'pipe', 'pipe'],
       },
     )
@@ -440,17 +444,17 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY || notReady.length > 0)('web smoke
     await screen(page, '04-round-complete')
   }, 150_000)
 
-  it('4 view tabs: Chat / Trajectory / Waterfall all switch', async () => {
+  it('view tabs: Chat and Trajectory switch', async () => {
     onTestFailed(() => saveFailureShot(page, 'w5-tabs'))
     await page.locator('button', { hasText: /Trajectory/i }).first().click()
     await screen(page, '05-trajectory-tab')
-    await page.locator('button', { hasText: /Waterfall/i }).first().click()
-    await screen(page, '06-waterfall-tab')
+    await page.getByLabel('Trajectory timeline').waitFor()
+    await expect.poll(() => page.getByRole('tab', { name: 'Waterfall' }).count()).toBe(0)
     await page.locator('button', { hasText: /^Chat$/i }).first().click()
     await screen(page, '07-back-to-chat')
   })
 
-  it('5 bash differential rendering: tool row click leaves the details column collapsed', async () => {
+  it('5 bash differential rendering: tool row click leaves the default details column open', async () => {
     onTestFailed(() => saveFailureShot(page, 'w5-tool-details'))
     const input = page.locator('textarea').first()
     await input.fill('请用 bash 工具运行命令 echo w5marker 然后告诉我结果')
@@ -462,14 +466,14 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY || notReady.length > 0)('web smoke
     const toolRow = page.locator('[data-sample="bash-global"]')
     await toolRow.waitFor({ timeout: 120_000 })
     await screen(page, '08-bash-round')
-    expect(await detailsTrack(page)).toBe(0)
+    expect(await detailsTrack(page)).toBe(360)
     await toolRow.click()
-    // Tool rows no longer drive layout.openDetails; the column stays closed.
-    expect(await detailsTrack(page)).toBe(0)
-    await screen(page, '09-details-closed')
+    // Tool rows no longer drive layout.openDetails; the default column stays open.
+    expect(await detailsTrack(page)).toBe(360)
+    await screen(page, '09-details-open')
   }, 150_000)
 
-  it('6 sidebar drag widens the column and persists across reload', async () => {
+  it('6 sidebar drag widens the column and resets across reload', async () => {
     onTestFailed(() => saveFailureShot(page, 'w5-drag'))
     const before = await firstTrack(page)
     const handle = page.locator('[class*="handle"]').first()
@@ -484,7 +488,7 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY || notReady.length > 0)('web smoke
     await screen(page, '10-sidebar-dragged')
     await page.reload({ waitUntil: 'load' })
     await page.waitForSelector('[class*="frame"]', { timeout: 30_000 })
-    expect(await firstTrack(page)).toBe(after)
+    expect(await firstTrack(page)).toBe(before)
   })
 
   it('7 dark mode: the body attribute cascades the token sheets', async () => {
diff --git a/apps/web/tests/snapshots/code-mode-round/session.jsonl b/apps/web/tests/snapshots/code-mode-round/session.jsonl
index 6e9e481129..b49160e3f9 100644
--- a/apps/web/tests/snapshots/code-mode-round/session.jsonl
+++ b/apps/web/tests/snapshots/code-mode-round/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1785013630399,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1785013630411,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1785013630411,"data":{"content":[{"type":"text","text":"Using ONE run_code program: run bash `echo CODE_ROUND_OK`, then read the file missing.txt catching its error in the program. Return an object with both outcomes. Then reply DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1785013630411,"data":{"content":[{"type":"text","text":"Using ONE run_code program: run bash `echo CODE_ROUND_OK`, then read the file missing.txt catching its error in the program. Return an object with both outcomes. Then reply DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"1c4a6a0f-a7bd-4642-919d-6ed7395df98b"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1785013630418,"data":{"title":"Using ONE run_code program: run","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1785013630479,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1785013630480,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -12,13 +12,13 @@
 {"type":"assistant/chunk","seq":202,"time":1785013633103,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_6VNoF1gDSerTBKoCfYSH3765","name":"run_code","arguments":"{\"description\": \"Run bash echo and catch missing file read\", \"code\": \"\\nconst bashResult = await tools.bash({\\n  command: \\\"echo CODE_ROUND_OK\\\",\\n  description: \\\"Echo CODE_ROUND_OK\\\"\\n});\\n\\nlet readError = null;\\ntry {\\n  await tools.read({ file_path: \\\"missing.txt\\\" });\\n} catch (e) {\\n  readError = {\\n    toolName: e.toolName,\\n    message: e.message\\n  };\\n}\\n\\nreturn { bash: bashResult.stdout.text.trim(), readError };\\n\"}"}}}}
 {"type":"assistant/chunk","seq":203,"time":1785013633104,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":8082,"outputTokens":235,"cacheReadTokens":384,"reasoningTokens":75}}}}
 {"type":"assistant/chunk","seq":204,"time":1785013633104,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":205,"time":1785013633108,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to write a single `run_code` program that:\n1. Runs bash to echo \"CODE_ROUND_OK\"\n2. Tries to read a file \"missing.txt\" and catches the error\n3. Returns an object with both outcomes\n4. They also want me to reply \"DONE\" and stop after\n\nLet me write this program."},{"type":"tool-call","id":"call_00_6VNoF1gDSerTBKoCfYSH3765","name":"run_code","arguments":"{\"description\": \"Run bash echo and catch missing file read\", \"code\": \"\\nconst bashResult = await tools.bash({\\n  command: \\\"echo CODE_ROUND_OK\\\",\\n  description: \\\"Echo CODE_ROUND_OK\\\"\\n});\\n\\nlet readError = null;\\ntry {\\n  await tools.read({ file_path: \\\"missing.txt\\\" });\\n} catch (e) {\\n  readError = {\\n    toolName: e.toolName,\\n    message: e.message\\n  };\\n}\\n\\nreturn { bash: bashResult.stdout.text.trim(), readError };\\n\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":8082,"outputTokens":235,"cacheReadTokens":384,"reasoningTokens":75}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204],"surfaceOp":"append"}
+{"type":"assistant/message","seq":205,"time":1785013633108,"data":{"turn":1,"step":1,"usage":{"inputTokens":8082,"outputTokens":235,"cacheReadTokens":384,"reasoningTokens":75},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to write a single `run_code` program that:\n1. Runs bash to echo \"CODE_ROUND_OK\"\n2. Tries to read a file \"missing.txt\" and catches the error\n3. Returns an object with both outcomes\n4. They also want me to reply \"DONE\" and stop after\n\nLet me write this program."},{"type":"tool-call","id":"call_00_6VNoF1gDSerTBKoCfYSH3765","name":"run_code","arguments":"{\"description\": \"Run bash echo and catch missing file read\", \"code\": \"\\nconst bashResult = await tools.bash({\\n  command: \\\"echo CODE_ROUND_OK\\\",\\n  description: \\\"Echo CODE_ROUND_OK\\\"\\n});\\n\\nlet readError = null;\\ntry {\\n  await tools.read({ file_path: \\\"missing.txt\\\" });\\n} catch (e) {\\n  readError = {\\n    toolName: e.toolName,\\n    message: e.message\\n  };\\n}\\n\\nreturn { bash: bashResult.stdout.text.trim(), readError };\\n\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"0d217af9-3bb3-425b-b53e-7ab0366f0c66"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204],"surfaceOp":"append"}
 {"type":"tool/call","seq":206,"time":1785013633108,"data":{"turn":1,"step":1,"callId":"call_00_6VNoF1gDSerTBKoCfYSH3765","name":"run_code","arguments":"{\"description\": \"Run bash echo and catch missing file read\", \"code\": \"\\nconst bashResult = await tools.bash({\\n  command: \\\"echo CODE_ROUND_OK\\\",\\n  description: \\\"Echo CODE_ROUND_OK\\\"\\n});\\n\\nlet readError = null;\\ntry {\\n  await tools.read({ file_path: \\\"missing.txt\\\" });\\n} catch (e) {\\n  readError = {\\n    toolName: e.toolName,\\n    message: e.message\\n  };\\n}\\n\\nreturn { bash: bashResult.stdout.text.trim(), readError };\\n\"}"}}
 {"type":"tool/code-dispatch-start","seq":207,"time":1785013633173,"data":{"parentCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765","subCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765:code:1","name":"bash","arguments":{"command":"echo CODE_ROUND_OK","description":"Echo CODE_ROUND_OK"}}}
 {"type":"tool/code-dispatch","seq":208,"time":1785013633196,"data":{"parentCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765","subCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765:code:1","name":"bash","arguments":{"command":"echo CODE_ROUND_OK","description":"Echo CODE_ROUND_OK"},"isError":false,"content":[{"type":"text","text":"CODE_ROUND_OK\n"}]}}
 {"type":"tool/code-dispatch-start","seq":209,"time":1785013633197,"data":{"parentCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765","subCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765:code:2","name":"read","arguments":{"file_path":"missing.txt"}}}
 {"type":"tool/code-dispatch","seq":210,"time":1785013633198,"data":{"parentCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765","subCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765:code:2","name":"read","arguments":{"file_path":"missing.txt"},"isError":true,"content":[{"type":"text","text":"Error: cannot read \"{{cwd}}/workspace/missing.txt\": not found"}]}}
-{"type":"tool/result","seq":211,"time":1785013633201,"data":{"turn":1,"step":1,"callId":"call_00_6VNoF1gDSerTBKoCfYSH3765","content":[{"type":"text","text":"{\n  \"bash\": \"CODE_ROUND_OK\",\n  \"readError\": {\n    \"toolName\": \"read\",\n    \"message\": \"cannot read \\\"{{cwd}}/workspace/missing.txt\\\": not found\"\n  }\n}"}],"isError":false},"sourceEventSeqs":[206],"surfaceOp":"append"}
+{"type":"tool/result","seq":211,"time":1785013633201,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_6VNoF1gDSerTBKoCfYSH3765"},"content":[{"type":"tool-result","toolCallId":"call_00_6VNoF1gDSerTBKoCfYSH3765","content":[{"type":"text","text":"{\n  \"bash\": \"CODE_ROUND_OK\",\n  \"readError\": {\n    \"toolName\": \"read\",\n    \"message\": \"cannot read \\\"{{cwd}}/workspace/missing.txt\\\": not found\"\n  }\n}"}],"isError":false}],"role":"user","id":"1580d4a8-5760-415f-984f-f93927271d3f"}},"sourceEventSeqs":[206],"surfaceOp":"append"}
 {"type":"step/end","seq":212,"time":1785013633204,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":213,"time":1785013633207,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":214,"time":1785013633985,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -30,6 +30,6 @@
 {"type":"assistant/chunk","seq":233,"time":1785013634223,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"DONE"}}}}
 {"type":"assistant/chunk","seq":234,"time":1785013634223,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":196,"outputTokens":17,"cacheReadTokens":8576,"reasoningTokens":14}}}}
 {"type":"assistant/chunk","seq":235,"time":1785013634223,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":236,"time":1785013634224,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"The program ran successfully. Let me now reply DONE as instructed."},{"type":"text","text":"DONE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":196,"outputTokens":17,"cacheReadTokens":8576,"reasoningTokens":14}},"sourceEventSeqs":[214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235],"surfaceOp":"append"}
+{"type":"assistant/message","seq":236,"time":1785013634224,"data":{"turn":1,"step":2,"usage":{"inputTokens":196,"outputTokens":17,"cacheReadTokens":8576,"reasoningTokens":14},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The program ran successfully. Let me now reply DONE as instructed."},{"type":"text","text":"DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"3037ec34-6c2a-4e01-91eb-a0ef89ce0d15"}},"sourceEventSeqs":[214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235],"surfaceOp":"append"}
 {"type":"step/end","seq":237,"time":1785013634225,"data":{"turn":1,"step":2}}
 {"type":"turn/end","seq":238,"time":1785013634225,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/code-mode-round/ui.expected.md b/apps/web/tests/snapshots/code-mode-round/ui.expected.md
index 4847924619..ccbf531e96 100644
--- a/apps/web/tests/snapshots/code-mode-round/ui.expected.md
+++ b/apps/web/tests/snapshots/code-mode-round/ui.expected.md
@@ -4,18 +4,23 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: "Using ONE run_code program: run bash `echo CODE_ROUND_OK`, then read the file missing.txt catching its error in the program. Return an object with both outcomes. Then reply DONE and stop."
+- text: "Using ONE run_code program: run bash `echo CODE_ROUND_OK`, then read the file missing.txt catching its error in the program. Return an object with both outcomes. Then reply DONE and stop. {{clock}}"
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - 'button "Think The user wants me to write a single `run_code` program that:"':
   - img
   - img
   - text: "Think The user wants me to write a single `run_code` program that:"
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button:
   - img
   - img
@@ -28,7 +33,11 @@
   - img
   - text: Think The program ran successfully. Let me now reply DONE as instructed.
 - paragraph: DONE
-- text: cache hit 52% · 17,490 tokens · 1 turns · 2 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 52% · 17,490 tokens · 1 turns · 2 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/cordis-tool-round/session.jsonl b/apps/web/tests/snapshots/cordis-tool-round/session.jsonl
index ab6c86089d..558f194795 100644
--- a/apps/web/tests/snapshots/cordis-tool-round/session.jsonl
+++ b/apps/web/tests/snapshots/cordis-tool-round/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1785157562825,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1785157562881,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1785157562882,"data":{"content":[{"type":"text","text":"Use only Cordis tools. First call cordis_inspect with what \"temporary\". Then call cordis_mount with this exact code: \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\". Read its returned id and call cordis_unmount with that exact id. After all three calls succeed, reply exactly CORDIS_UI_DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1785157562882,"data":{"content":[{"type":"text","text":"Use only Cordis tools. First call cordis_inspect with what \"temporary\". Then call cordis_mount with this exact code: \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\". Read its returned id and call cordis_unmount with that exact id. After all three calls succeed, reply exactly CORDIS_UI_DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"27e1fc72-20d0-4875-bf5b-d34750441f30"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1785157562883,"data":{"title":"Use only Cordis tools. First","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1785157562937,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1785157562938,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash","reasoningEffort":"high"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -12,9 +12,9 @@
 {"type":"assistant/chunk","seq":100,"time":1785157565491,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_KZk918WtlKan9pHMULIT8794","name":"cordis_inspect","arguments":"{\"what\": \"temporary\"}"}}}}
 {"type":"assistant/chunk","seq":101,"time":1785157565491,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":15137,"outputTokens":128,"cacheReadTokens":1280,"reasoningTokens":81}}}}
 {"type":"assistant/chunk","seq":102,"time":1785157565491,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":103,"time":1785157565495,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to:\n1. Call `cordis_inspect` with `what: \"temporary\"`\n2. Call `cordis_mount` with the exact code provided\n3. Read the returned id and call `cordis_unmount` with that exact id\n4. Reply exactly \"CORDIS_UI_DONE\" and stop\n\nLet me start with step 1."},{"type":"tool-call","id":"call_00_KZk918WtlKan9pHMULIT8794","name":"cordis_inspect","arguments":"{\"what\": \"temporary\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":15137,"outputTokens":128,"cacheReadTokens":1280,"reasoningTokens":81}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102],"surfaceOp":"append"}
+{"type":"assistant/message","seq":103,"time":1785157565495,"data":{"turn":1,"step":1,"usage":{"inputTokens":15137,"outputTokens":128,"cacheReadTokens":1280,"reasoningTokens":81},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to:\n1. Call `cordis_inspect` with `what: \"temporary\"`\n2. Call `cordis_mount` with the exact code provided\n3. Read the returned id and call `cordis_unmount` with that exact id\n4. Reply exactly \"CORDIS_UI_DONE\" and stop\n\nLet me start with step 1."},{"type":"tool-call","id":"call_00_KZk918WtlKan9pHMULIT8794","name":"cordis_inspect","arguments":"{\"what\": \"temporary\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"a3b56157-c6fd-43f8-bc8a-0e2a0c99a8b5"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102],"surfaceOp":"append"}
 {"type":"tool/call","seq":104,"time":1785157565496,"data":{"turn":1,"step":1,"callId":"call_00_KZk918WtlKan9pHMULIT8794","name":"cordis_inspect","arguments":"{\"what\": \"temporary\"}"}}
-{"type":"tool/result","seq":105,"time":1785157565500,"data":{"turn":1,"step":1,"callId":"call_00_KZk918WtlKan9pHMULIT8794","content":[{"type":"text","text":"## Temporary Plugins\nNo temporary Plugins are running. Temporary Plugins created with cordis_mount disappear when DSH restarts."}],"isError":false},"sourceEventSeqs":[104],"surfaceOp":"append"}
+{"type":"tool/result","seq":105,"time":1785157565500,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_KZk918WtlKan9pHMULIT8794"},"content":[{"type":"tool-result","toolCallId":"call_00_KZk918WtlKan9pHMULIT8794","content":[{"type":"text","text":"## Temporary Plugins\nNo temporary Plugins are running. Temporary Plugins created with cordis_mount disappear when DSH restarts."}],"isError":false}],"role":"user","id":"95ae124e-f065-4aab-a9c5-bd33c7eafff1"}},"sourceEventSeqs":[104],"surfaceOp":"append"}
 {"type":"step/end","seq":106,"time":1785157565503,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":107,"time":1785157565503,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":108,"time":1785157566524,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -25,9 +25,9 @@
 {"type":"assistant/chunk","seq":157,"time":1785157567041,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_OyrlxqqpnvzTe11sN5Oy2361","name":"cordis_mount","arguments":"{\"code\": \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\"}"}}}}
 {"type":"assistant/chunk","seq":158,"time":1785157567042,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":72,"outputTokens":82,"cacheReadTokens":16512,"reasoningTokens":22}}}}
 {"type":"assistant/chunk","seq":159,"time":1785157567042,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":160,"time":1785157567043,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"Good, no temporary plugins running. Now step 2: call cordis_mount with the exact code."},{"type":"tool-call","id":"call_00_OyrlxqqpnvzTe11sN5Oy2361","name":"cordis_mount","arguments":"{\"code\": \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":72,"outputTokens":82,"cacheReadTokens":16512,"reasoningTokens":22}},"sourceEventSeqs":[108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159],"surfaceOp":"append"}
+{"type":"assistant/message","seq":160,"time":1785157567043,"data":{"turn":1,"step":2,"usage":{"inputTokens":72,"outputTokens":82,"cacheReadTokens":16512,"reasoningTokens":22},"message":{"role":"assistant","content":[{"type":"reasoning","text":"Good, no temporary plugins running. Now step 2: call cordis_mount with the exact code."},{"type":"tool-call","id":"call_00_OyrlxqqpnvzTe11sN5Oy2361","name":"cordis_mount","arguments":"{\"code\": \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"b90af675-f604-4fb4-9c27-89c91b27f4ce"}},"sourceEventSeqs":[108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159],"surfaceOp":"append"}
 {"type":"tool/call","seq":161,"time":1785157567043,"data":{"turn":1,"step":2,"callId":"call_00_OyrlxqqpnvzTe11sN5Oy2361","name":"cordis_mount","arguments":"{\"code\": \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\"}"}}
-{"type":"tool/result","seq":162,"time":1785157567049,"data":{"turn":1,"step":2,"callId":"call_00_OyrlxqqpnvzTe11sN5Oy2361","content":[{"type":"text","text":"Temporary Plugin dyn-1 is running (plugin \"snapshot-noop\"; available until unmounted or DSH restarts)."}],"isError":false},"sourceEventSeqs":[161],"surfaceOp":"append"}
+{"type":"tool/result","seq":162,"time":1785157567049,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_OyrlxqqpnvzTe11sN5Oy2361"},"content":[{"type":"tool-result","toolCallId":"call_00_OyrlxqqpnvzTe11sN5Oy2361","content":[{"type":"text","text":"Temporary Plugin dyn-1 is running (plugin \"snapshot-noop\"; available until unmounted or DSH restarts)."}],"isError":false}],"role":"user","id":"3148fc0a-f511-4b72-abd5-d184e974cd49"}},"sourceEventSeqs":[161],"surfaceOp":"append"}
 {"type":"step/end","seq":163,"time":1785157567050,"data":{"turn":1,"step":2}}
 {"type":"step/start","seq":164,"time":1785157567050,"data":{"turn":1,"step":3}}
 {"type":"assistant/chunk","seq":165,"time":1785157567835,"data":{"turn":1,"step":3,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -38,9 +38,9 @@
 {"type":"assistant/chunk","seq":204,"time":1785157568279,"data":{"turn":1,"step":3,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_e38S6zeYdZGvbhecUCil6659","name":"cordis_unmount","arguments":"{\"id\": \"dyn-1\"}"}}}}
 {"type":"assistant/chunk","seq":205,"time":1785157568279,"data":{"turn":1,"step":3,"chunk":{"type":"usage","usage":{"inputTokens":63,"outputTokens":72,"cacheReadTokens":16640,"reasoningTokens":23}}}}
 {"type":"assistant/chunk","seq":206,"time":1785157568279,"data":{"turn":1,"step":3,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":207,"time":1785157568280,"data":{"turn":1,"step":3,"content":[{"type":"reasoning","text":"The id is \"dyn-1\". Now step 3: call cordis_unmount with that id."},{"type":"tool-call","id":"call_00_e38S6zeYdZGvbhecUCil6659","name":"cordis_unmount","arguments":"{\"id\": \"dyn-1\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":63,"outputTokens":72,"cacheReadTokens":16640,"reasoningTokens":23}},"sourceEventSeqs":[165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206],"surfaceOp":"append"}
+{"type":"assistant/message","seq":207,"time":1785157568280,"data":{"turn":1,"step":3,"usage":{"inputTokens":63,"outputTokens":72,"cacheReadTokens":16640,"reasoningTokens":23},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The id is \"dyn-1\". Now step 3: call cordis_unmount with that id."},{"type":"tool-call","id":"call_00_e38S6zeYdZGvbhecUCil6659","name":"cordis_unmount","arguments":"{\"id\": \"dyn-1\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"cdc5b840-4302-486a-8065-fa59bf13f2d9"}},"sourceEventSeqs":[165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206],"surfaceOp":"append"}
 {"type":"tool/call","seq":208,"time":1785157568280,"data":{"turn":1,"step":3,"callId":"call_00_e38S6zeYdZGvbhecUCil6659","name":"cordis_unmount","arguments":"{\"id\": \"dyn-1\"}"}}
-{"type":"tool/result","seq":209,"time":1785157568281,"data":{"turn":1,"step":3,"callId":"call_00_e38S6zeYdZGvbhecUCil6659","content":[{"type":"text","text":"Temporary Plugin dyn-1 was unmounted and removed."}],"isError":false},"sourceEventSeqs":[208],"surfaceOp":"append"}
+{"type":"tool/result","seq":209,"time":1785157568281,"data":{"turn":1,"step":3,"message":{"source":{"kind":"tool","callId":"call_00_e38S6zeYdZGvbhecUCil6659"},"content":[{"type":"tool-result","toolCallId":"call_00_e38S6zeYdZGvbhecUCil6659","content":[{"type":"text","text":"Temporary Plugin dyn-1 was unmounted and removed."}],"isError":false}],"role":"user","id":"758a7b27-56e3-4997-8c4d-effc699cc5e6"}},"sourceEventSeqs":[208],"surfaceOp":"append"}
 {"type":"step/end","seq":210,"time":1785157568282,"data":{"turn":1,"step":3}}
 {"type":"step/start","seq":211,"time":1785157568282,"data":{"turn":1,"step":4}}
 {"type":"assistant/chunk","seq":212,"time":1785157569185,"data":{"turn":1,"step":4,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -51,6 +51,6 @@
 {"type":"assistant/chunk","seq":244,"time":1785157569551,"data":{"turn":1,"step":4,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"CORDIS_UI_DONE"}}}}
 {"type":"assistant/chunk","seq":245,"time":1785157569551,"data":{"turn":1,"step":4,"chunk":{"type":"usage","usage":{"inputTokens":29,"outputTokens":30,"cacheReadTokens":16768,"reasoningTokens":22}}}}
 {"type":"assistant/chunk","seq":246,"time":1785157569551,"data":{"turn":1,"step":4,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":247,"time":1785157569553,"data":{"turn":1,"step":4,"content":[{"type":"reasoning","text":"All three calls succeeded. I should now reply exactly \"CORDIS_UI_DONE\" and stop."},{"type":"text","text":"CORDIS_UI_DONE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":29,"outputTokens":30,"cacheReadTokens":16768,"reasoningTokens":22}},"sourceEventSeqs":[212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246],"surfaceOp":"append"}
+{"type":"assistant/message","seq":247,"time":1785157569553,"data":{"turn":1,"step":4,"usage":{"inputTokens":29,"outputTokens":30,"cacheReadTokens":16768,"reasoningTokens":22},"message":{"role":"assistant","content":[{"type":"reasoning","text":"All three calls succeeded. I should now reply exactly \"CORDIS_UI_DONE\" and stop."},{"type":"text","text":"CORDIS_UI_DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"71540121-7bf6-410e-975c-f5ea99d8175a"}},"sourceEventSeqs":[212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246],"surfaceOp":"append"}
 {"type":"step/end","seq":248,"time":1785157569554,"data":{"turn":1,"step":4}}
 {"type":"turn/end","seq":249,"time":1785157569554,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md b/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md
index e5e5626be3..5e1a148f9e 100644
--- a/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md
+++ b/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md
@@ -4,18 +4,23 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: "Use only Cordis tools. First call cordis_inspect with what \"temporary\". Then call cordis_mount with this exact code: \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\". Read its returned id and call cordis_unmount with that exact id. After all three calls succeed, reply exactly CORDIS_UI_DONE and stop."
+- text: "Use only Cordis tools. First call cordis_inspect with what \"temporary\". Then call cordis_mount with this exact code: \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\". Read its returned id and call cordis_unmount with that exact id. After all three calls succeed, reply exactly CORDIS_UI_DONE and stop. {{clock}}"
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user wants me to:":
   - img
   - img
   - text: "Think The user wants me to:"
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button:
   - img
   - img
@@ -24,6 +29,11 @@
   - img
   - img
   - text: "Think Good, no temporary plugins running. Now step 2: call cordis_mount with the exact code."
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button [expanded]:
   - img
 - text: Mount temporary Plugin typescript
@@ -33,6 +43,11 @@
   - img
   - img
   - text: "Think The id is \"dyn-1\". Now step 3: call cordis_unmount with that id."
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button:
   - img
   - img
@@ -42,7 +57,11 @@
   - img
   - text: Think All three calls succeeded. I should now reply exactly "CORDIS_UI_DONE" and stop.
 - paragraph: CORDIS_UI_DONE
-- text: cache hit 77% · 66,813 tokens · 1 turns · 4 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 77% · 66,813 tokens · 1 turns · 4 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/fresh-round-trip/session.jsonl b/apps/web/tests/snapshots/fresh-round-trip/session.jsonl
index 53a75267e5..202b73f83a 100644
--- a/apps/web/tests/snapshots/fresh-round-trip/session.jsonl
+++ b/apps/web/tests/snapshots/fresh-round-trip/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1784973850091,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1784973850102,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1784973850103,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo WEB_E2E_OK. Then reply with the single word DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1784973850103,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo WEB_E2E_OK. Then reply with the single word DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"2c594e5a-bbcc-4c64-b5ee-8e84eb5dd949"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1784973850105,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1784973850164,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1784973850165,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -12,9 +12,9 @@
 {"type":"assistant/chunk","seq":52,"time":1784973851493,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_BYXlxjFaalMg95YVqEeF2495","name":"bash","arguments":"{\"command\": \"echo WEB_E2E_OK\", \"description\": \"Echo the test string\"}"}}}}
 {"type":"assistant/chunk","seq":53,"time":1784973851493,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":122,"outputTokens":85,"cacheReadTokens":7680,"reasoningTokens":17}}}}
 {"type":"assistant/chunk","seq":54,"time":1784973851493,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":55,"time":1784973851498,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to run a simple bash command and reply with \"DONE\"."},{"type":"tool-call","id":"call_00_BYXlxjFaalMg95YVqEeF2495","name":"bash","arguments":"{\"command\": \"echo WEB_E2E_OK\", \"description\": \"Echo the test string\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":122,"outputTokens":85,"cacheReadTokens":7680,"reasoningTokens":17}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54],"surfaceOp":"append"}
+{"type":"assistant/message","seq":55,"time":1784973851498,"data":{"turn":1,"step":1,"usage":{"inputTokens":122,"outputTokens":85,"cacheReadTokens":7680,"reasoningTokens":17},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to run a simple bash command and reply with \"DONE\"."},{"type":"tool-call","id":"call_00_BYXlxjFaalMg95YVqEeF2495","name":"bash","arguments":"{\"command\": \"echo WEB_E2E_OK\", \"description\": \"Echo the test string\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"e5d8d5fb-f555-4fda-948f-9ab178145929"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54],"surfaceOp":"append"}
 {"type":"tool/call","seq":56,"time":1784973851499,"data":{"turn":1,"step":1,"callId":"call_00_BYXlxjFaalMg95YVqEeF2495","name":"bash","arguments":"{\"command\": \"echo WEB_E2E_OK\", \"description\": \"Echo the test string\"}"}}
-{"type":"tool/result","seq":57,"time":1784973851515,"data":{"turn":1,"step":1,"callId":"call_00_BYXlxjFaalMg95YVqEeF2495","content":[{"type":"text","text":"WEB_E2E_OK\n"}],"isError":false},"sourceEventSeqs":[56],"surfaceOp":"append"}
+{"type":"tool/result","seq":57,"time":1784973851515,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_BYXlxjFaalMg95YVqEeF2495"},"content":[{"type":"tool-result","toolCallId":"call_00_BYXlxjFaalMg95YVqEeF2495","content":[{"type":"text","text":"WEB_E2E_OK\n"}],"isError":false}],"role":"user","id":"215febf9-ff3c-42b8-93b7-27ce1505c1c2"}},"sourceEventSeqs":[56],"surfaceOp":"append"}
 {"type":"step/end","seq":58,"time":1784973851517,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":59,"time":1784973851518,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":60,"time":1784973852194,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -26,6 +26,6 @@
 {"type":"assistant/chunk","seq":88,"time":1784973852460,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"DONE"}}}}
 {"type":"assistant/chunk","seq":89,"time":1784973852460,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":97,"outputTokens":26,"cacheReadTokens":7808,"reasoningTokens":23}}}}
 {"type":"assistant/chunk","seq":90,"time":1784973852460,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":91,"time":1784973852461,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"The command executed successfully and output \"WEB_E2E_OK\". I just need to reply with \"DONE\"."},{"type":"text","text":"DONE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":97,"outputTokens":26,"cacheReadTokens":7808,"reasoningTokens":23}},"sourceEventSeqs":[60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90],"surfaceOp":"append"}
+{"type":"assistant/message","seq":91,"time":1784973852461,"data":{"turn":1,"step":2,"usage":{"inputTokens":97,"outputTokens":26,"cacheReadTokens":7808,"reasoningTokens":23},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The command executed successfully and output \"WEB_E2E_OK\". I just need to reply with \"DONE\"."},{"type":"text","text":"DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"61e83a98-18a9-4f1f-9151-c8783cc39901"}},"sourceEventSeqs":[60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90],"surfaceOp":"append"}
 {"type":"step/end","seq":92,"time":1784973852461,"data":{"turn":1,"step":2}}
 {"type":"turn/end","seq":93,"time":1784973852462,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md b/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md
index 6a827420c4..d0810ea315 100644
--- a/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md
+++ b/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md
@@ -4,18 +4,23 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: "Use the bash tool to run exactly: echo WEB_E2E_OK. Then reply with the single word DONE and stop."
+- text: "Use the bash tool to run exactly: echo WEB_E2E_OK. Then reply with the single word DONE and stop. {{clock}}"
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user wants me to run a simple bash command and reply with \"DONE\".":
   - img
   - img
   - text: Think The user wants me to run a simple bash command and reply with "DONE".
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - img
 - text: Bash Echo the test string
 - button "Think The command executed successfully and output \"WEB_E2E_OK\". I just need to reply with \"DONE\".":
@@ -23,7 +28,11 @@
   - img
   - text: Think The command executed successfully and output "WEB_E2E_OK". I just need to reply with "DONE".
 - paragraph: DONE
-- text: cache hit 99% · 15,818 tokens · 1 turns · 2 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 99% · 15,818 tokens · 1 turns · 2 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md b/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md
index 33d1f7e6bf..6d601aad87 100644
--- a/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md
+++ b/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md
@@ -4,20 +4,24 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: Reply with the single word LIGHTHOUSE and stop.
+- text: Reply with the single word LIGHTHOUSE and stop. {{clock}}
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user wants me to reply with a single word. Let me comply.":
   - img
   - img
   - text: Think The user wants me to reply with a single word. Let me comply.
 - paragraph: LIGHTHOUSE
-- text: cache hit 99% · 7,810 tokens · 1 turns · 1 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 99% · 7,810 tokens · 1 turns · 1 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/lifecycle-chrome/session.jsonl b/apps/web/tests/snapshots/lifecycle-chrome/session.jsonl
index 4d7caa325d..a168f82d9e 100644
--- a/apps/web/tests/snapshots/lifecycle-chrome/session.jsonl
+++ b/apps/web/tests/snapshots/lifecycle-chrome/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1785015039278,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1785015039291,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1785015039292,"data":{"content":[{"type":"text","text":"Reply with the single word LIGHTHOUSE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1785015039292,"data":{"content":[{"type":"text","text":"Reply with the single word LIGHTHOUSE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"ee28d2ee-08b2-4ed7-a1e9-84865f55e2af"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1785015039294,"data":{"title":"Reply with the single word","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1785015039362,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1785015039363,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -12,6 +12,6 @@
 {"type":"assistant/chunk","seq":28,"time":1785015040242,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"LIGHTHOUSE"}}}}
 {"type":"assistant/chunk","seq":29,"time":1785015040242,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":109,"outputTokens":21,"cacheReadTokens":7680,"reasoningTokens":15}}}}
 {"type":"assistant/chunk","seq":30,"time":1785015040242,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":31,"time":1785015040244,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to reply with a single word. Let me comply."},{"type":"text","text":"LIGHTHOUSE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":109,"outputTokens":21,"cacheReadTokens":7680,"reasoningTokens":15}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30],"surfaceOp":"append"}
+{"type":"assistant/message","seq":31,"time":1785015040244,"data":{"turn":1,"step":1,"usage":{"inputTokens":109,"outputTokens":21,"cacheReadTokens":7680,"reasoningTokens":15},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to reply with a single word. Let me comply."},{"type":"text","text":"LIGHTHOUSE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"2101eadc-3475-4c77-ae0a-0107b12c34bc"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30],"surfaceOp":"append"}
 {"type":"step/end","seq":32,"time":1785015040246,"data":{"turn":1,"step":1}}
 {"type":"turn/end","seq":33,"time":1785015040247,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/live-interactions/cancel.expected.md b/apps/web/tests/snapshots/live-interactions/cancel.expected.md
index 3d092b17ec..a141ad147c 100644
--- a/apps/web/tests/snapshots/live-interactions/cancel.expected.md
+++ b/apps/web/tests/snapshots/live-interactions/cancel.expected.md
@@ -4,16 +4,21 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: Reply with a one-sentence description of event sourcing, then stop.
+- text: Reply with a one-sentence description of event sourcing, then stop. {{clock}}
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - paragraph: partial
-- text: 已停止 0 tokens · 1 turns · 1 steps
+- text: 已停止
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} 0 tokens · 1 turns · 1 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/live-interactions/error-auth.expected.md b/apps/web/tests/snapshots/live-interactions/error-auth.expected.md
index 5272bcf2d1..7fd444dc9d 100644
--- a/apps/web/tests/snapshots/live-interactions/error-auth.expected.md
+++ b/apps/web/tests/snapshots/live-interactions/error-auth.expected.md
@@ -4,14 +4,14 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: Reply with a one-sentence description of event sourcing, then stop.
+- text: Reply with a one-sentence description of event sourcing, then stop. {{clock}}
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/live-interactions/retry.expected.md b/apps/web/tests/snapshots/live-interactions/retry.expected.md
index 5935872557..d857b6663c 100644
--- a/apps/web/tests/snapshots/live-interactions/retry.expected.md
+++ b/apps/web/tests/snapshots/live-interactions/retry.expected.md
@@ -4,20 +4,24 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: Reply with a one-sentence description of event sourcing, then stop.
+- text: Reply with a one-sentence description of event sourcing, then stop. {{clock}}
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user is asking for a one-sentence description of event sourcing. This is a straightforward knowledge question that doesn't require any skill loading or tool calls.":
   - img
   - img
   - text: Think The user is asking for a one-sentence description of event sourcing. This is a straightforward knowledge question that doesn't require any skill loading or tool calls.
 - paragraph: Event sourcing is a pattern where all changes to an application's state are stored as an immutable, append-only sequence of events, rather than persisting only the current state, enabling full auditability, temporal queries, and event-driven architectures.
-- text: cache hit 99% · 7,869 tokens · 1 turns · 1 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 99% · 7,869 tokens · 1 turns · 1 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/live-interactions/session.jsonl b/apps/web/tests/snapshots/live-interactions/session.jsonl
index e002ec48ee..91b0c730a8 100644
--- a/apps/web/tests/snapshots/live-interactions/session.jsonl
+++ b/apps/web/tests/snapshots/live-interactions/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1784998084441,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1784998084454,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1784998084454,"data":{"content":[{"type":"text","text":"Reply with a one-sentence description of event sourcing, then stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1784998084454,"data":{"content":[{"type":"text","text":"Reply with a one-sentence description of event sourcing, then stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"dc9a43b2-63a3-49bc-97e9-9dfb6465f6c1"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1784998084457,"data":{"title":"Reply with a one-sentence description","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1784998084519,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1784998084520,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -12,6 +12,6 @@
 {"type":"assistant/chunk","seq":86,"time":1784998085814,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"Event sourcing is a pattern where all changes to an application's state are stored as an immutable, append-only sequence of events, rather than persisting only the current state, enabling full auditability, temporal queries, and event-driven architectures."}}}}
 {"type":"assistant/chunk","seq":87,"time":1784998085814,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":110,"outputTokens":79,"cacheReadTokens":7680,"reasoningTokens":31}}}}
 {"type":"assistant/chunk","seq":88,"time":1784998085814,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":89,"time":1784998085818,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user is asking for a one-sentence description of event sourcing. This is a straightforward knowledge question that doesn't require any skill loading or tool calls."},{"type":"text","text":"Event sourcing is a pattern where all changes to an application's state are stored as an immutable, append-only sequence of events, rather than persisting only the current state, enabling full auditability, temporal queries, and event-driven architectures."}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":110,"outputTokens":79,"cacheReadTokens":7680,"reasoningTokens":31}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88],"surfaceOp":"append"}
+{"type":"assistant/message","seq":89,"time":1784998085818,"data":{"turn":1,"step":1,"usage":{"inputTokens":110,"outputTokens":79,"cacheReadTokens":7680,"reasoningTokens":31},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user is asking for a one-sentence description of event sourcing. This is a straightforward knowledge question that doesn't require any skill loading or tool calls."},{"type":"text","text":"Event sourcing is a pattern where all changes to an application's state are stored as an immutable, append-only sequence of events, rather than persisting only the current state, enabling full auditability, temporal queries, and event-driven architectures."}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"3396618c-56a2-4f97-a9c7-2a3cbb16a3c8"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88],"surfaceOp":"append"}
 {"type":"step/end","seq":90,"time":1784998085820,"data":{"turn":1,"step":1}}
 {"type":"turn/end","seq":91,"time":1784998085821,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/message-actions/ui.expected.md b/apps/web/tests/snapshots/message-actions/ui.expected.md
new file mode 100644
index 0000000000..bd833b6c3c
--- /dev/null
+++ b/apps/web/tests/snapshots/message-actions/ui.expected.md
@@ -0,0 +1,51 @@
+- banner:
+  - navigation "Session hierarchy":
+    - button "Use the read tool twice" [disabled]
+  - tablist:
+    - tab "Chat" [selected]
+    - tab "Trajectory"
+- text: "Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop. {{clock}}"
+- button "复制":
+  - img
+- tooltip "复制"
+- button "在新对话中分支":
+  - img
+- button "编辑":
+  - img
+- button "Think The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel.":
+  - img
+  - img
+  - text: Think The user wants me to read a.txt and b.txt, then reply with "DONE". Let me do both reads in parallel.
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
+- img
+- text: Read
+- button "a.txt"
+- img
+- text: Read
+- button "b.txt"
+- button "Think Both files have been read. a.txt contains \"alpha\" and b.txt contains \"beta\". I'll now reply with DONE as instructed.":
+  - img
+  - img
+  - text: Think Both files have been read. a.txt contains "alpha" and b.txt contains "beta". I'll now reply with DONE as instructed.
+- paragraph: DONE
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 98% · 15,962 tokens · 1 turns · 2 steps
+- textbox "Message the agent"
+- button "Add attachment":
+  - img
+- text: Danger Full Access
+- combobox "Access mode":
+  - option "Read Only"
+  - option "Workspace Write"
+  - option "Danger Full Access" [selected]
+- button "选择模型，当前 deepseek-v4-flash":
+  - text: deepseek-v4-flash
+  - img
+- button "Send message" [disabled]
diff --git a/apps/web/tests/snapshots/navigation-panes/seed.jsonl b/apps/web/tests/snapshots/navigation-panes/seed.jsonl
index 72df45daac..3623cd749a 100644
--- a/apps/web/tests/snapshots/navigation-panes/seed.jsonl
+++ b/apps/web/tests/snapshots/navigation-panes/seed.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1785011380476,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1785011380489,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1785011380490,"data":{"content":[{"type":"text","text":"NavScenario: first run bash to print exactly NAVIGATION_OK, then read nav-a.md and nav-b.md using two read calls in ONE assistant message, then reply with the single word FIRST_DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1785011380490,"data":{"content":[{"type":"text","text":"NavScenario: first run bash to print exactly NAVIGATION_OK, then read nav-a.md and nav-b.md using two read calls in ONE assistant message, then reply with the single word FIRST_DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"7dcfe547-5555-4e5d-b2a0-3f9d2e0836a9"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1785011380492,"data":{"title":"NavScenario: first run bash to","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1785011380549,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1785011380550,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -18,13 +18,13 @@
 {"type":"assistant/chunk","seq":130,"time":1785011382087,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":3,"block":{"type":"tool-call","id":"call_02_k8Z6wGirxfnW96Iv8mkz9224","name":"read","arguments":"{\"file_path\": \"nav-b.md\"}"}}}}
 {"type":"assistant/chunk","seq":131,"time":1785011382087,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":140,"outputTokens":197,"cacheReadTokens":7680,"reasoningTokens":66}}}}
 {"type":"assistant/chunk","seq":132,"time":1785011382087,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":133,"time":1785011382091,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to follow a specific navigation scenario. Let me:\n\n1. Run bash to print \"NAVIGATION_OK\"\n2. Read nav-a.md and nav-b.md in two read calls in ONE message\n3. Reply with \"FIRST_DONE\"\n\nLet me start with the bash command and the reads."},{"type":"tool-call","id":"call_00_kFKHaEXcTYEex0iDZw0C2432","name":"bash","arguments":"{\"command\": \"echo NAVIGATION_OK\", \"description\": \"Print NAVIGATION_OK\"}"},{"type":"tool-call","id":"call_01_tK4hIIRVTMgAvdzs7m9j6212","name":"read","arguments":"{\"file_path\": \"nav-a.md\"}"},{"type":"tool-call","id":"call_02_k8Z6wGirxfnW96Iv8mkz9224","name":"read","arguments":"{\"file_path\": \"nav-b.md\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":140,"outputTokens":197,"cacheReadTokens":7680,"reasoningTokens":66}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132],"surfaceOp":"append"}
+{"type":"assistant/message","seq":133,"time":1785011382091,"data":{"turn":1,"step":1,"usage":{"inputTokens":140,"outputTokens":197,"cacheReadTokens":7680,"reasoningTokens":66},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to follow a specific navigation scenario. Let me:\n\n1. Run bash to print \"NAVIGATION_OK\"\n2. Read nav-a.md and nav-b.md in two read calls in ONE message\n3. Reply with \"FIRST_DONE\"\n\nLet me start with the bash command and the reads."},{"type":"tool-call","id":"call_00_kFKHaEXcTYEex0iDZw0C2432","name":"bash","arguments":"{\"command\": \"echo NAVIGATION_OK\", \"description\": \"Print NAVIGATION_OK\"}"},{"type":"tool-call","id":"call_01_tK4hIIRVTMgAvdzs7m9j6212","name":"read","arguments":"{\"file_path\": \"nav-a.md\"}"},{"type":"tool-call","id":"call_02_k8Z6wGirxfnW96Iv8mkz9224","name":"read","arguments":"{\"file_path\": \"nav-b.md\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"83b27ad0-4286-478f-92ef-3bc82bf0589b"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132],"surfaceOp":"append"}
 {"type":"tool/call","seq":134,"time":1785011382092,"data":{"turn":1,"step":1,"callId":"call_00_kFKHaEXcTYEex0iDZw0C2432","name":"bash","arguments":"{\"command\": \"echo NAVIGATION_OK\", \"description\": \"Print NAVIGATION_OK\"}"}}
-{"type":"tool/result","seq":135,"time":1785011382105,"data":{"turn":1,"step":1,"callId":"call_00_kFKHaEXcTYEex0iDZw0C2432","content":[{"type":"text","text":"NAVIGATION_OK\n"}],"isError":false},"sourceEventSeqs":[134],"surfaceOp":"append"}
+{"type":"tool/result","seq":135,"time":1785011382105,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_kFKHaEXcTYEex0iDZw0C2432"},"content":[{"type":"tool-result","toolCallId":"call_00_kFKHaEXcTYEex0iDZw0C2432","content":[{"type":"text","text":"NAVIGATION_OK\n"}],"isError":false}],"role":"user","id":"d6e29f9e-8177-4f12-b1b6-2eda4c8f5c31"}},"sourceEventSeqs":[134],"surfaceOp":"append"}
 {"type":"tool/call","seq":136,"time":1785011382105,"data":{"turn":1,"step":1,"callId":"call_01_tK4hIIRVTMgAvdzs7m9j6212","name":"read","arguments":"{\"file_path\": \"nav-a.md\"}"}}
 {"type":"tool/call","seq":137,"time":1785011382106,"data":{"turn":1,"step":1,"callId":"call_02_k8Z6wGirxfnW96Iv8mkz9224","name":"read","arguments":"{\"file_path\": \"nav-b.md\"}"}}
-{"type":"tool/result","seq":138,"time":1785011382113,"data":{"turn":1,"step":1,"callId":"call_01_tK4hIIRVTMgAvdzs7m9j6212","content":[{"type":"text","text":"<path>{{cwd}}/workspace/nav-a.md</path>\n<type>file</type>\n<content>\n1: # alpha nav\n\n(End of file - total 1 lines)\n</content>"}],"isError":false},"sourceEventSeqs":[136],"surfaceOp":"append"}
-{"type":"tool/result","seq":139,"time":1785011382114,"data":{"turn":1,"step":1,"callId":"call_02_k8Z6wGirxfnW96Iv8mkz9224","content":[{"type":"text","text":"<path>{{cwd}}/workspace/nav-b.md</path>\n<type>file</type>\n<content>\n1: # beta nav\n\n(End of file - total 1 lines)\n</content>"}],"isError":false},"sourceEventSeqs":[137],"surfaceOp":"append"}
+{"type":"tool/result","seq":138,"time":1785011382113,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_01_tK4hIIRVTMgAvdzs7m9j6212"},"content":[{"type":"tool-result","toolCallId":"call_01_tK4hIIRVTMgAvdzs7m9j6212","content":[{"type":"text","text":"<path>{{cwd}}/workspace/nav-a.md</path>\n<type>file</type>\n<content>\n1: # alpha nav\n\n(End of file - total 1 lines)\n</content>"}],"isError":false}],"role":"user","id":"efca7baf-cbb8-462f-ad75-580357b58676"}},"sourceEventSeqs":[136],"surfaceOp":"append"}
+{"type":"tool/result","seq":139,"time":1785011382114,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_02_k8Z6wGirxfnW96Iv8mkz9224"},"content":[{"type":"tool-result","toolCallId":"call_02_k8Z6wGirxfnW96Iv8mkz9224","content":[{"type":"text","text":"<path>{{cwd}}/workspace/nav-b.md</path>\n<type>file</type>\n<content>\n1: # beta nav\n\n(End of file - total 1 lines)\n</content>"}],"isError":false}],"role":"user","id":"e1c70190-9e68-429d-963d-8a30ab779ddd"}},"sourceEventSeqs":[137],"surfaceOp":"append"}
 {"type":"step/end","seq":140,"time":1785011382117,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":141,"time":1785011382118,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":142,"time":1785011382673,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -35,11 +35,11 @@
 {"type":"assistant/chunk","seq":200,"time":1785011383090,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"FIRST_DONE"}}}}
 {"type":"assistant/chunk","seq":201,"time":1785011383090,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":349,"outputTokens":56,"cacheReadTokens":7808,"reasoningTokens":51}}}}
 {"type":"assistant/chunk","seq":202,"time":1785011383090,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":203,"time":1785011383091,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"All three calls succeeded:\n1. bash printed \"NAVIGATION_OK\"\n2. nav-a.md contains \"# alpha nav\"\n3. nav-b.md contains \"# beta nav\"\n\nNow I need to reply with the single word \"FIRST_DONE\"."},{"type":"text","text":"FIRST_DONE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":349,"outputTokens":56,"cacheReadTokens":7808,"reasoningTokens":51}},"sourceEventSeqs":[142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202],"surfaceOp":"append"}
+{"type":"assistant/message","seq":203,"time":1785011383091,"data":{"turn":1,"step":2,"usage":{"inputTokens":349,"outputTokens":56,"cacheReadTokens":7808,"reasoningTokens":51},"message":{"role":"assistant","content":[{"type":"reasoning","text":"All three calls succeeded:\n1. bash printed \"NAVIGATION_OK\"\n2. nav-a.md contains \"# alpha nav\"\n3. nav-b.md contains \"# beta nav\"\n\nNow I need to reply with the single word \"FIRST_DONE\"."},{"type":"text","text":"FIRST_DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"0115844e-fac0-4985-9ba7-dcf1dde63b83"}},"sourceEventSeqs":[142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202],"surfaceOp":"append"}
 {"type":"step/end","seq":204,"time":1785011383091,"data":{"turn":1,"step":2}}
 {"type":"turn/end","seq":205,"time":1785011383092,"data":{"turn":1,"reason":{"kind":"completed"}}}
 {"type":"turn/start","seq":206,"time":1785011383106,"data":{"turn":2,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":207,"time":1785011383107,"data":{"content":[{"type":"text","text":"Reply in markdown with: a level-2 heading \"Navigation Summary\", a bulleted list of exactly two items, and a fenced code block containing echo WATERFALL. Then stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":207,"time":1785011383107,"data":{"content":[{"type":"text","text":"Reply in markdown with: a level-2 heading \"Navigation Summary\", a bulleted list of exactly two items, and a fenced code block containing echo WATERFALL. Then stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"1d7a8563-4ae6-49f3-9a65-41f3df300a75"},"surfaceOp":"append"}
 {"type":"step/start","seq":208,"time":1785011383107,"data":{"turn":2,"step":1}}
 {"type":"assistant/chunk","seq":209,"time":1785011383497,"data":{"turn":2,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
 {"type":"reasoning-chunks","seq0":210,"time0":1785011383497,"data":{"turn":2,"step":1,"index":0,"dt":[125,23,1,0,0,88,0,0,5,0,1,0,0,7,0],"texts":["The"," user"," wants"," me"," to"," reply"," with"," a"," specific"," format","."," Let"," me"," do"," that","."]}}
@@ -49,6 +49,6 @@
 {"type":"assistant/chunk","seq":247,"time":1785011383903,"data":{"turn":2,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"## Navigation Summary\n\n- alpha nav\n- beta nav\n\n```\necho WATERFALL\n```"}}}}
 {"type":"assistant/chunk","seq":248,"time":1785011383903,"data":{"turn":2,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":141,"outputTokens":36,"cacheReadTokens":8064,"reasoningTokens":16}}}}
 {"type":"assistant/chunk","seq":249,"time":1785011383903,"data":{"turn":2,"step":1,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":250,"time":1785011383904,"data":{"turn":2,"step":1,"content":[{"type":"reasoning","text":"The user wants me to reply with a specific format. Let me do that."},{"type":"text","text":"## Navigation Summary\n\n- alpha nav\n- beta nav\n\n```\necho WATERFALL\n```"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":141,"outputTokens":36,"cacheReadTokens":8064,"reasoningTokens":16}},"sourceEventSeqs":[209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249],"surfaceOp":"append"}
+{"type":"assistant/message","seq":250,"time":1785011383904,"data":{"turn":2,"step":1,"usage":{"inputTokens":141,"outputTokens":36,"cacheReadTokens":8064,"reasoningTokens":16},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to reply with a specific format. Let me do that."},{"type":"text","text":"## Navigation Summary\n\n- alpha nav\n- beta nav\n\n```\necho WATERFALL\n```"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"70b38db9-38a2-4f9f-8094-1bd799f5f270"}},"sourceEventSeqs":[209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249],"surfaceOp":"append"}
 {"type":"step/end","seq":251,"time":1785011383904,"data":{"turn":2,"step":1}}
 {"type":"turn/end","seq":252,"time":1785011383904,"data":{"turn":2,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/navigation-panes/terminal-card.expected.md b/apps/web/tests/snapshots/navigation-panes/terminal-card.expected.md
new file mode 100644
index 0000000000..298bf764b6
--- /dev/null
+++ b/apps/web/tests/snapshots/navigation-panes/terminal-card.expected.md
@@ -0,0 +1,3 @@
+- text: 已完成 {{workspace}} echo NAVIGATION_OK
+- button "复制"
+- text: NAVIGATION_OK
diff --git a/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md b/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md
index 80d6f161ca..0bdf87d67a 100644
--- a/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md
+++ b/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md
@@ -1 +1,53 @@
-- text: "Turn 1 Message {{duration}} #1 User NavScenario: first run bash to print exactly NAVIGATION_OK, then read nav-a.md and nav-b.md using two read calls in ONE assistant message, then reply with the single word FIRST_DONE and stop. +{{duration}} Step 1 {{duration}} bash read×2 #2 Tool bash · {\"command\": \"echo NAVIGATION_OK\", \"description\": \"Print NAVIGATION_OK\"} +{{duration}} #3 Tool read · {\"file_path\": \"nav-a.md\"} +{{duration}} #4 Tool read · {\"file_path\": \"nav-b.md\"} +{{duration}} Step 2 {{duration}} #5 Message FIRST_DONE 349 56 51 +{{duration}} Turn 2 Message {{duration}} #6 User Reply in markdown with: a level-2 heading \"Navigation Summary\", a bulleted list of exactly two items, and a fenced code block containing echo WATERFALL. Then stop. +{{duration}} Step 1 {{duration}} #7 Message ## Navigation Summary - alpha nav - beta nav ``` echo WATERFALL ``` 141 36 16 +{{duration}}"
+- toolbar "Trajectory toolbar":
+  - button "Use actual duration": Duration
+  - button "Collapse turns": Turns
+  - button "Collapse calls": Calls
+  - img
+  - searchbox "Search trajectory"
+- region "Trajectory timeline"
+- table:
+  - rowgroup:
+    - row "SYSTEM, Initial System Prompt":
+      - cell "SYSTEM"
+      - cell "Initial System Prompt"
+    - 'row "USER, NavScenario: first run bash to print exactly NAVIGATION_OK, then read nav-a.md and nav-b.md using two read calls in ONE assistant message, then reply with the single word FIRST_DONE and stop."':
+      - cell "Turn 1 USER"
+      - 'cell "NavScenario: first run bash to print exactly NAVIGATION_OK, then read nav-a.md and nav-b.md using two read calls in ONE assistant message, then reply with the single word FIRST_DONE and stop."'
+    - 'row "Request 1, ASSISTANT, The user wants me to follow a specific navigation scenario. Let me: Run bash to print \"NAVIGATION_OK\" Read nav-a.md and nav-b.md in two read calls in ONE message Reply with \"FIRST_DONE\" Let me start with the bash command and the reads."':
+      - 'cell "Request #1 ASSISTANT"':
+        - 'button "Request #1"'
+        - text: ASSISTANT
+      - 'cell "The user wants me to follow a specific navigation scenario. Let me: Run bash to print \"NAVIGATION_OK\" Read nav-a.md and nav-b.md in two read calls in ONE message Reply with \"FIRST_DONE\" Let me start with the bash command and the reads."'
+    - 'row "TOOL, bash {\"command\": \"echo NAVIGATION_OK\", \"description\": \"Print NAVIGATION_OK\"}" [selected]':
+      - cell "TOOL"
+      - 'cell "bash{\"command\": \"echo NAVIGATION_OK\", \"description\": \"Print NAVIGATION_OK\"} → NAVIGATION_OK"'
+    - 'row "TOOL, read {\"file_path\": \"nav-a.md\"}"':
+      - cell "TOOL"
+      - 'cell "read{\"file_path\": \"nav-a.md\"} → <path>{{cwd}}/nav-a.md</path> <type>file</type> <content> 1: # alpha nav (End of file - total 1 lines) </content>"'
+    - 'row "TOOL, read {\"file_path\": \"nav-b.md\"}"':
+      - cell "TOOL"
+      - 'cell "read{\"file_path\": \"nav-b.md\"} → <path>{{cwd}}/nav-b.md</path> <type>file</type> <content> 1: # beta nav (End of file - total 1 lines) </content>"'
+    - row "Request 2, ASSISTANT, FIRST_DONE":
+      - 'cell "Request #2 ASSISTANT"':
+        - 'button "Request #2"'
+        - text: ASSISTANT
+      - cell "FIRST_DONE"
+    - 'row "USER, Reply in markdown with: a level-2 heading \"Navigation Summary\", a bulleted list of exactly two items, and a fenced code block containing echo WATERFALL. Then stop."':
+      - cell "Turn 2 USER"
+      - 'cell "Reply in markdown with: a level-2 heading \"Navigation Summary\", a bulleted list of exactly two items, and a fenced code block containing echo WATERFALL. Then stop."'
+    - row "Request 3, ASSISTANT, Navigation Summary alpha nav beta nav echo WATERFALL":
+      - 'cell "Request #3 ASSISTANT"':
+        - 'button "Request #3"'
+        - text: ASSISTANT
+      - cell "Navigation Summary alpha nav beta nav echo WATERFALL"
+- complementary "Event details":
+  - separator "Resize event details"
+  - text: TOOL Turn 1 · Step 1
+  - button "Close details"
+  - tablist "Event details":
+    - tab "Summary"
+    - tab "Payload"
+    - tab "Result" [selected]
+    - tab "Schema"
+    - tab "Timing"
+  - tabpanel "Result": NAVIGATION_OK
diff --git a/apps/web/tests/snapshots/navigation-panes/waterfall.expected.md b/apps/web/tests/snapshots/navigation-panes/waterfall.expected.md
deleted file mode 100644
index 6c5ab1a046..0000000000
--- a/apps/web/tests/snapshots/navigation-panes/waterfall.expected.md
+++ /dev/null
@@ -1 +0,0 @@
-- text: 3 turns · 3 steps · 3 tool calls turn 0 turn 1 turn 2
diff --git a/apps/web/tests/snapshots/question-composer/answered.expected.md b/apps/web/tests/snapshots/question-composer/answered.expected.md
index 91ff2cdf88..bd03429898 100644
--- a/apps/web/tests/snapshots/question-composer/answered.expected.md
+++ b/apps/web/tests/snapshots/question-composer/answered.expected.md
@@ -4,18 +4,23 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: "Use the ask_user_question tool to ask me exactly one question with id \"color\", question \"Which color do you prefer?\", header \"Pick one\", and two options: label \"Blue\" with description \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\", and label \"Green\" with description \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\" After I answer, reply with the single word DONE and stop."
+- text: "Use the ask_user_question tool to ask me exactly one question with id \"color\", question \"Which color do you prefer?\", header \"Pick one\", and two options: label \"Blue\" with description \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\", and label \"Green\" with description \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\" After I answer, reply with the single word DONE and stop. {{clock}}"
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user wants me to use the ask_user_question tool with specific parameters. Let me do exactly that.":
   - img
   - img
   - text: Think The user wants me to use the ask_user_question tool with specific parameters. Let me do exactly that.
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button:
   - img
   - img
@@ -25,7 +30,11 @@
   - img
   - text: Think The user answered "Blue". I should now reply with the single word DONE and stop.
 - paragraph: DONE
-- text: cache hit 95% · 8,769 tokens · 1 turns · 2 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 95% · 8,769 tokens · 1 turns · 2 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/snapshots/question-composer/session.jsonl b/apps/web/tests/snapshots/question-composer/session.jsonl
index a98f2a92f2..b13a84e22c 100644
--- a/apps/web/tests/snapshots/question-composer/session.jsonl
+++ b/apps/web/tests/snapshots/question-composer/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1785150167878,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1785150167924,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1785150167925,"data":{"content":[{"type":"text","text":"Use the ask_user_question tool to ask me exactly one question with id \"color\", question \"Which color do you prefer?\", header \"Pick one\", and two options: label \"Blue\" with description \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\", and label \"Green\" with description \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\" After I answer, reply with the single word DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1785150167925,"data":{"content":[{"type":"text","text":"Use the ask_user_question tool to ask me exactly one question with id \"color\", question \"Which color do you prefer?\", header \"Pick one\", and two options: label \"Blue\" with description \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\", and label \"Green\" with description \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\" After I answer, reply with the single word DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"6deba879-8787-4853-a5f2-0d108a08eb2d"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1785150167927,"data":{"title":"Use the ask_user_question tool to","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1785150167928,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1785150167929,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}}
@@ -12,9 +12,9 @@
 {"type":"assistant/chunk","seq":128,"time":1785150169308,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_Cijldc88LYmVPCXYUsRq1617","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"color\", \"question\": \"Which color do you prefer?\", \"header\": \"Pick one\", \"options\": [{\"label\": \"Blue\", \"description\": \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\"}, {\"label\": \"Green\", \"description\": \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\"}]}]}"}}}}
 {"type":"assistant/chunk","seq":129,"time":1785150169308,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":113,"outputTokens":158,"cacheReadTokens":4096,"reasoningTokens":22}}}}
 {"type":"assistant/chunk","seq":130,"time":1785150169308,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":131,"time":1785150169311,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to use the ask_user_question tool with specific parameters. Let me do exactly that."},{"type":"tool-call","id":"call_00_Cijldc88LYmVPCXYUsRq1617","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"color\", \"question\": \"Which color do you prefer?\", \"header\": \"Pick one\", \"options\": [{\"label\": \"Blue\", \"description\": \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\"}, {\"label\": \"Green\", \"description\": \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\"}]}]}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":113,"outputTokens":158,"cacheReadTokens":4096,"reasoningTokens":22}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130],"surfaceOp":"append"}
+{"type":"assistant/message","seq":131,"time":1785150169311,"data":{"turn":1,"step":1,"usage":{"inputTokens":113,"outputTokens":158,"cacheReadTokens":4096,"reasoningTokens":22},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to use the ask_user_question tool with specific parameters. Let me do exactly that."},{"type":"tool-call","id":"call_00_Cijldc88LYmVPCXYUsRq1617","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"color\", \"question\": \"Which color do you prefer?\", \"header\": \"Pick one\", \"options\": [{\"label\": \"Blue\", \"description\": \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\"}, {\"label\": \"Green\", \"description\": \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\"}]}]}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"cdb1676c-e781-41ee-8f28-a3595371d729"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130],"surfaceOp":"append"}
 {"type":"tool/call","seq":132,"time":1785150169312,"data":{"turn":1,"step":1,"callId":"call_00_Cijldc88LYmVPCXYUsRq1617","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"color\", \"question\": \"Which color do you prefer?\", \"header\": \"Pick one\", \"options\": [{\"label\": \"Blue\", \"description\": \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\"}, {\"label\": \"Green\", \"description\": \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\"}]}]}"}}
-{"type":"tool/result","seq":133,"time":1785150169787,"data":{"turn":1,"step":1,"callId":"call_00_Cijldc88LYmVPCXYUsRq1617","content":[{"type":"text","text":"{\"answers\":[{\"id\":\"color\",\"selected\":[\"Blue\"]}]}"}],"isError":false},"sourceEventSeqs":[132],"surfaceOp":"append"}
+{"type":"tool/result","seq":133,"time":1785150169787,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_Cijldc88LYmVPCXYUsRq1617"},"content":[{"type":"tool-result","toolCallId":"call_00_Cijldc88LYmVPCXYUsRq1617","content":[{"type":"text","text":"{\"answers\":[{\"id\":\"color\",\"selected\":[\"Blue\"]}]}"}],"isError":false}],"role":"user","id":"c69ef39f-6f62-439f-b3f8-e8d10fba572f"}},"sourceEventSeqs":[132],"surfaceOp":"append"}
 {"type":"step/end","seq":134,"time":1785150169790,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":135,"time":1785150169790,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":136,"time":1785150170605,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -26,6 +26,6 @@
 {"type":"assistant/chunk","seq":160,"time":1785150170856,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"DONE"}}}}
 {"type":"assistant/chunk","seq":161,"time":1785150170856,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":284,"outputTokens":22,"cacheReadTokens":4096,"reasoningTokens":19}}}}
 {"type":"assistant/chunk","seq":162,"time":1785150170856,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":163,"time":1785150170857,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"The user answered \"Blue\". I should now reply with the single word DONE and stop."},{"type":"text","text":"DONE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":284,"outputTokens":22,"cacheReadTokens":4096,"reasoningTokens":19}},"sourceEventSeqs":[136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162],"surfaceOp":"append"}
+{"type":"assistant/message","seq":163,"time":1785150170857,"data":{"turn":1,"step":2,"usage":{"inputTokens":284,"outputTokens":22,"cacheReadTokens":4096,"reasoningTokens":19},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user answered \"Blue\". I should now reply with the single word DONE and stop."},{"type":"text","text":"DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"b5f572cb-f2a7-4fa0-8097-a5551700a920"}},"sourceEventSeqs":[136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162],"surfaceOp":"append"}
 {"type":"step/end","seq":164,"time":1785150170858,"data":{"turn":1,"step":2}}
 {"type":"turn/end","seq":165,"time":1785150170858,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/seeded-history/seed.jsonl b/apps/web/tests/snapshots/seeded-history/seed.jsonl
index abf7a61162..bf69bcf152 100644
--- a/apps/web/tests/snapshots/seeded-history/seed.jsonl
+++ b/apps/web/tests/snapshots/seeded-history/seed.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1784974100747,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1784974100758,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1784974100759,"data":{"content":[{"type":"text","text":"Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1784974100759,"data":{"content":[{"type":"text","text":"Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"38f072be-5254-4cb7-b76e-d612b2ae3b3a"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1784974100761,"data":{"title":"Use the read tool twice","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1784974100827,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1784974100828,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -15,11 +15,11 @@
 {"type":"assistant/chunk","seq":61,"time":1784974101974,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":2,"block":{"type":"tool-call","id":"call_01_Hw6AQjhf9gjxnOtppcGx0725","name":"read","arguments":"{\"file_path\": \"b.txt\"}"}}}}
 {"type":"assistant/chunk","seq":62,"time":1784974101974,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":124,"outputTokens":103,"cacheReadTokens":7680,"reasoningTokens":27}}}}
 {"type":"assistant/chunk","seq":63,"time":1784974101974,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":64,"time":1784974101978,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel."},{"type":"tool-call","id":"call_00_OsndvlcKnCcUmae7QXal8633","name":"read","arguments":"{\"file_path\": \"a.txt\"}"},{"type":"tool-call","id":"call_01_Hw6AQjhf9gjxnOtppcGx0725","name":"read","arguments":"{\"file_path\": \"b.txt\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":124,"outputTokens":103,"cacheReadTokens":7680,"reasoningTokens":27}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63],"surfaceOp":"append"}
+{"type":"assistant/message","seq":64,"time":1784974101978,"data":{"turn":1,"step":1,"usage":{"inputTokens":124,"outputTokens":103,"cacheReadTokens":7680,"reasoningTokens":27},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel."},{"type":"tool-call","id":"call_00_OsndvlcKnCcUmae7QXal8633","name":"read","arguments":"{\"file_path\": \"a.txt\"}"},{"type":"tool-call","id":"call_01_Hw6AQjhf9gjxnOtppcGx0725","name":"read","arguments":"{\"file_path\": \"b.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"ec076738-f75d-4525-ba99-c8fc16acf955"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63],"surfaceOp":"append"}
 {"type":"tool/call","seq":65,"time":1784974101979,"data":{"turn":1,"step":1,"callId":"call_00_OsndvlcKnCcUmae7QXal8633","name":"read","arguments":"{\"file_path\": \"a.txt\"}"}}
 {"type":"tool/call","seq":66,"time":1784974101981,"data":{"turn":1,"step":1,"callId":"call_01_Hw6AQjhf9gjxnOtppcGx0725","name":"read","arguments":"{\"file_path\": \"b.txt\"}"}}
-{"type":"tool/result","seq":67,"time":1784974101985,"data":{"turn":1,"step":1,"callId":"call_00_OsndvlcKnCcUmae7QXal8633","content":[{"type":"text","text":"<path>{{cwd}}/workspace/a.txt</path>\n<type>file</type>\n<content>\n1: alpha\n\n(End of file - total 1 lines)\n</content>"}],"isError":false},"sourceEventSeqs":[65],"surfaceOp":"append"}
-{"type":"tool/result","seq":68,"time":1784974101986,"data":{"turn":1,"step":1,"callId":"call_01_Hw6AQjhf9gjxnOtppcGx0725","content":[{"type":"text","text":"<path>{{cwd}}/workspace/b.txt</path>\n<type>file</type>\n<content>\n1: beta\n\n(End of file - total 1 lines)\n</content>"}],"isError":false},"sourceEventSeqs":[66],"surfaceOp":"append"}
+{"type":"tool/result","seq":67,"time":1784974101985,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_OsndvlcKnCcUmae7QXal8633"},"content":[{"type":"tool-result","toolCallId":"call_00_OsndvlcKnCcUmae7QXal8633","content":[{"type":"text","text":"<path>{{cwd}}/workspace/a.txt</path>\n<type>file</type>\n<content>\n1: alpha\n\n(End of file - total 1 lines)\n</content>"}],"isError":false}],"role":"user","id":"ea2294eb-8652-4492-8a08-9c24d3f8a60f"}},"sourceEventSeqs":[65],"surfaceOp":"append"}
+{"type":"tool/result","seq":68,"time":1784974101986,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_01_Hw6AQjhf9gjxnOtppcGx0725"},"content":[{"type":"tool-result","toolCallId":"call_01_Hw6AQjhf9gjxnOtppcGx0725","content":[{"type":"text","text":"<path>{{cwd}}/workspace/b.txt</path>\n<type>file</type>\n<content>\n1: beta\n\n(End of file - total 1 lines)\n</content>"}],"isError":false}],"role":"user","id":"f02b3acf-4e03-4b4d-beeb-1a564c9c6d61"}},"sourceEventSeqs":[66],"surfaceOp":"append"}
 {"type":"step/end","seq":69,"time":1784974101988,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":70,"time":1784974101988,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":71,"time":1784974102397,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -31,6 +31,6 @@
 {"type":"assistant/chunk","seq":105,"time":1784974102749,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"DONE"}}}}
 {"type":"assistant/chunk","seq":106,"time":1784974102750,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":215,"outputTokens":32,"cacheReadTokens":7808,"reasoningTokens":29}}}}
 {"type":"assistant/chunk","seq":107,"time":1784974102750,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":108,"time":1784974102750,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"Both files have been read. a.txt contains \"alpha\" and b.txt contains \"beta\". I'll now reply with DONE as instructed."},{"type":"text","text":"DONE"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":215,"outputTokens":32,"cacheReadTokens":7808,"reasoningTokens":29}},"sourceEventSeqs":[71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107],"surfaceOp":"append"}
+{"type":"assistant/message","seq":108,"time":1784974102750,"data":{"turn":1,"step":2,"usage":{"inputTokens":215,"outputTokens":32,"cacheReadTokens":7808,"reasoningTokens":29},"message":{"role":"assistant","content":[{"type":"reasoning","text":"Both files have been read. a.txt contains \"alpha\" and b.txt contains \"beta\". I'll now reply with DONE as instructed."},{"type":"text","text":"DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"07627a5e-4cb2-47ef-9b50-88893aac7406"}},"sourceEventSeqs":[71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107],"surfaceOp":"append"}
 {"type":"step/end","seq":109,"time":1784974102751,"data":{"turn":1,"step":2}}
 {"type":"turn/end","seq":110,"time":1784974102751,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/seeded-history/ui.expected.md b/apps/web/tests/snapshots/seeded-history/ui.expected.md
index 4f9181f702..c65bf7e712 100644
--- a/apps/web/tests/snapshots/seeded-history/ui.expected.md
+++ b/apps/web/tests/snapshots/seeded-history/ui.expected.md
@@ -1,32 +1,49 @@
 - banner:
   - navigation "Session hierarchy":
     - button "Use the read tool twice" [disabled]
-    - text: · 1 turns
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: "Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop."
+- text: "Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop. {{clock}}"
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- button "编辑":
+  - img
 - button "Think The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel.":
+  - img
   - img
   - text: Think The user wants me to read a.txt and b.txt, then reply with "DONE". Let me do both reads in parallel.
-- button:
+- button "复制":
   - img
-- text: Read a.txt
-- button:
+- button "在新对话中分支":
   - img
-- text: Read b.txt
+- text: {{clock}}
+- img
+- text: Read
+- button "a.txt"
+- img
+- text: Read
+- button "b.txt"
 - button "Think Both files have been read. a.txt contains \"alpha\" and b.txt contains \"beta\". I'll now reply with DONE as instructed.":
+  - img
   - img
   - text: Think Both files have been read. a.txt contains "alpha" and b.txt contains "beta". I'll now reply with DONE as instructed.
 - paragraph: DONE
-- text: cache hit 98% · 15,962 tokens · 1 turns · 2 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 98% · 15,962 tokens · 1 turns · 2 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
+- text: Danger Full Access
 - combobox "Access mode":
-  - option "Read-only" [selected]
-  - option "Read-write"
+  - option "Read Only"
+  - option "Workspace Write"
+  - option "Danger Full Access" [selected]
 - button "选择模型，当前 deepseek-v4-flash":
   - text: deepseek-v4-flash
   - img
diff --git a/apps/web/tests/snapshots/session-actions/rename-rows.json b/apps/web/tests/snapshots/session-actions/rename-rows.json
new file mode 100644
index 0000000000..db30b0d121
--- /dev/null
+++ b/apps/web/tests/snapshots/session-actions/rename-rows.json
@@ -0,0 +1,14 @@
+[
+  {
+    "label": "fixture4 sessions"
+  },
+  {
+    "label": "New Sessionnow"
+  },
+  {
+    "label": "分叉 实验记录now"
+  },
+  {
+    "label": "fixture2min"
+  }
+]
diff --git a/apps/web/tests/snapshots/steering/mid-steer.expected.md b/apps/web/tests/snapshots/steering/mid-steer.expected.md
index 8d33ea6283..b70aa38d2c 100644
--- a/apps/web/tests/snapshots/steering/mid-steer.expected.md
+++ b/apps/web/tests/snapshots/steering/mid-steer.expected.md
@@ -4,24 +4,27 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: Use the ask_user_question tool to ask me exactly one question with id "checkpoint", question "Ready to continue?", header "Checkpoint", and options labeled "Yes" and "No". After I answer, reply with one short sentence acknowledging my answer and stop.
+- text: Use the ask_user_question tool to ask me exactly one question with id "checkpoint", question "Ready to continue?", header "Checkpoint", and options labeled "Yes" and "No". After I answer, reply with one short sentence acknowledging my answer and stop. {{clock}}
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user wants me to use the ask_user_question tool to ask them a specific question with the given parameters. Let me do exactly that.":
   - img
   - img
   - text: Think The user wants me to use the ask_user_question tool to ask them a specific question with the given parameters. Let me do exactly that.
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button:
   - img
   - img
-- text: "Tool call ask_user_question · {\"questions\": [{\"id\": \"checkpoint\", \"question\": \"Ready to continue?\", \"header\": \"Checkpoint\", \"options\": [{\"label\": \"Yes\"}, {\"label\": \"No\"}]}]} 等待回答（1 题）"
-- button "▸ 问题内容"
-- text: cache hit 98% · 7,946 tokens · 1 turns · 1 steps
+- text: "Tool call ask_user_question · {\"questions\": [{\"id\": \"checkpoint\", \"question\": \"Ready to continue?\", \"header\": \"Checkpoint\", \"options\": [{\"label\": \"Yes\"}, {\"label\": \"No\"}]}]} cache hit 98% · 7,946 tokens · 1 turns · 1 steps"
 - region "Ready to continue?":
   - text: Checkpoint
   - heading "Ready to continue?" [level=2]
diff --git a/apps/web/tests/snapshots/steering/session.jsonl b/apps/web/tests/snapshots/steering/session.jsonl
index 4015fa4ab8..eb21251997 100644
--- a/apps/web/tests/snapshots/steering/session.jsonl
+++ b/apps/web/tests/snapshots/steering/session.jsonl
@@ -1,6 +1,6 @@
 {"type":"session","version":0,"id":"{{sessionId}}","createdAt":1785004180013,"cwd":"{{cwd}}/workspace"}
 {"type":"turn/start","seq":0,"time":1785004180030,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user","rpcId":"{{rpcId}}"}}}}
-{"type":"user/message","seq":1,"time":1785004180030,"data":{"content":[{"type":"text","text":"Use the ask_user_question tool to ask me exactly one question with id \"checkpoint\", question \"Ready to continue?\", header \"Checkpoint\", and options labeled \"Yes\" and \"No\". After I answer, reply with one short sentence acknowledging my answer and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"user/message","seq":1,"time":1785004180030,"data":{"content":[{"type":"text","text":"Use the ask_user_question tool to ask me exactly one question with id \"checkpoint\", question \"Ready to continue?\", header \"Checkpoint\", and options labeled \"Yes\" and \"No\". After I answer, reply with one short sentence acknowledging my answer and stop."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"eb77fa3d-5d60-4028-a592-e1f07a288f35"},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1785004180033,"data":{"title":"Use the ask_user_question tool to","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1785004180105,"data":{"turn":1,"step":1}}
 {"type":"request/header","seq":4,"time":1785004180106,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}","messagePrefix":["{{messagePrefix}}"]},"reason":"initial"}}
@@ -12,10 +12,10 @@
 {"type":"assistant/chunk","seq":85,"time":1785004181401,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"tool-call","id":"call_00_sAvjivLShvnWVk0sPQPV7661","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"checkpoint\", \"question\": \"Ready to continue?\", \"header\": \"Checkpoint\", \"options\": [{\"label\": \"Yes\"}, {\"label\": \"No\"}]}]}"}}}}
 {"type":"assistant/chunk","seq":86,"time":1785004181401,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":151,"outputTokens":115,"cacheReadTokens":7680,"reasoningTokens":29}}}}
 {"type":"assistant/chunk","seq":87,"time":1785004181402,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
-{"type":"assistant/message","seq":88,"time":1785004181406,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to use the ask_user_question tool to ask them a specific question with the given parameters. Let me do exactly that."},{"type":"tool-call","id":"call_00_sAvjivLShvnWVk0sPQPV7661","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"checkpoint\", \"question\": \"Ready to continue?\", \"header\": \"Checkpoint\", \"options\": [{\"label\": \"Yes\"}, {\"label\": \"No\"}]}]}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":151,"outputTokens":115,"cacheReadTokens":7680,"reasoningTokens":29}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87],"surfaceOp":"append"}
+{"type":"assistant/message","seq":88,"time":1785004181406,"data":{"turn":1,"step":1,"usage":{"inputTokens":151,"outputTokens":115,"cacheReadTokens":7680,"reasoningTokens":29},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to use the ask_user_question tool to ask them a specific question with the given parameters. Let me do exactly that."},{"type":"tool-call","id":"call_00_sAvjivLShvnWVk0sPQPV7661","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"checkpoint\", \"question\": \"Ready to continue?\", \"header\": \"Checkpoint\", \"options\": [{\"label\": \"Yes\"}, {\"label\": \"No\"}]}]}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"ac0cfafd-795f-4bc3-9dc9-a8df33538a38"}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87],"surfaceOp":"append"}
 {"type":"tool/call","seq":89,"time":1785004181407,"data":{"turn":1,"step":1,"callId":"call_00_sAvjivLShvnWVk0sPQPV7661","name":"ask_user_question","arguments":"{\"questions\": [{\"id\": \"checkpoint\", \"question\": \"Ready to continue?\", \"header\": \"Checkpoint\", \"options\": [{\"label\": \"Yes\"}, {\"label\": \"No\"}]}]}"}}
-{"type":"tool/result","seq":90,"time":1785004181867,"data":{"turn":1,"step":1,"callId":"call_00_sAvjivLShvnWVk0sPQPV7661","content":[{"type":"text","text":"{\"answers\":[{\"id\":\"checkpoint\",\"selected\":[\"Yes\"]}]}"}],"isError":false},"sourceEventSeqs":[89],"surfaceOp":"append"}
-{"type":"steering/message","seq":91,"time":1785004181867,"data":{"turn":1,"content":[{"type":"text","text":"Interjection: include the word BANANA in your final reply."}],"source":{"kind":"user","rpcId":"{{rpcId}}"}},"surfaceOp":"append"}
+{"type":"tool/result","seq":90,"time":1785004181867,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_sAvjivLShvnWVk0sPQPV7661"},"content":[{"type":"tool-result","toolCallId":"call_00_sAvjivLShvnWVk0sPQPV7661","content":[{"type":"text","text":"{\"answers\":[{\"id\":\"checkpoint\",\"selected\":[\"Yes\"]}]}"}],"isError":false}],"role":"user","id":"9c2175f8-3cd4-4d6c-9320-498d25f83342"}},"sourceEventSeqs":[89],"surfaceOp":"append"}
+{"type":"steering/message","seq":91,"time":1785004181867,"data":{"turn":1,"message":{"content":[{"type":"text","text":"Interjection: include the word BANANA in your final reply."}],"source":{"kind":"user","rpcId":"{{rpcId}}"},"role":"user","id":"291e02ad-cf8a-459f-a388-f81ed334e629"}},"surfaceOp":"append"}
 {"type":"step/end","seq":92,"time":1785004181870,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":93,"time":1785004181870,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":94,"time":1785004182322,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
@@ -26,6 +26,6 @@
 {"type":"assistant/chunk","seq":137,"time":1785004182893,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"Great, let's move forward. BANANA!"}}}}
 {"type":"assistant/chunk","seq":138,"time":1785004182893,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":172,"outputTokens":41,"cacheReadTokens":7808,"reasoningTokens":29}}}}
 {"type":"assistant/chunk","seq":139,"time":1785004182893,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
-{"type":"assistant/message","seq":140,"time":1785004182894,"data":{"turn":1,"step":2,"content":[{"type":"reasoning","text":"The user selected \"Yes\" and wants me to include the word \"BANANA\" in my final reply. Let me acknowledge their answer."},{"type":"text","text":"Great, let's move forward. BANANA!"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":172,"outputTokens":41,"cacheReadTokens":7808,"reasoningTokens":29}},"sourceEventSeqs":[94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139],"surfaceOp":"append"}
+{"type":"assistant/message","seq":140,"time":1785004182894,"data":{"turn":1,"step":2,"usage":{"inputTokens":172,"outputTokens":41,"cacheReadTokens":7808,"reasoningTokens":29},"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user selected \"Yes\" and wants me to include the word \"BANANA\" in my final reply. Let me acknowledge their answer."},{"type":"text","text":"Great, let's move forward. BANANA!"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"b49cca3a-fa40-4f45-91c9-e7c3f15fa233"}},"sourceEventSeqs":[94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139],"surfaceOp":"append"}
 {"type":"step/end","seq":141,"time":1785004182895,"data":{"turn":1,"step":2}}
 {"type":"turn/end","seq":142,"time":1785004182895,"data":{"turn":1,"reason":{"kind":"completed"}}}
diff --git a/apps/web/tests/snapshots/steering/settled.expected.md b/apps/web/tests/snapshots/steering/settled.expected.md
index f08fc518e8..790bd022f2 100644
--- a/apps/web/tests/snapshots/steering/settled.expected.md
+++ b/apps/web/tests/snapshots/steering/settled.expected.md
@@ -4,18 +4,23 @@
   - tablist:
     - tab "Chat" [selected]
     - tab "Trajectory"
-    - tab "Waterfall"
-- text: Use the ask_user_question tool to ask me exactly one question with id "checkpoint", question "Ready to continue?", header "Checkpoint", and options labeled "Yes" and "No". After I answer, reply with one short sentence acknowledging my answer and stop.
+- text: Use the ask_user_question tool to ask me exactly one question with id "checkpoint", question "Ready to continue?", header "Checkpoint", and options labeled "Yes" and "No". After I answer, reply with one short sentence acknowledging my answer and stop. {{clock}}
 - button "复制":
   - img
 - button "在新对话中分支":
   - img
 - button "编辑":
   - img
+- button "▸ 上下文注入"
 - button "Think The user wants me to use the ask_user_question tool to ask them a specific question with the given parameters. Let me do exactly that.":
   - img
   - img
   - text: Think The user wants me to use the ask_user_question tool to ask them a specific question with the given parameters. Let me do exactly that.
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}}
 - button:
   - img
   - img
@@ -25,7 +30,11 @@
   - img
   - text: Think The user selected "Yes" and wants me to include the word "BANANA" in my final reply. Let me acknowledge their answer.
 - paragraph: Great, let's move forward. BANANA!
-- text: cache hit 98% · 15,967 tokens · 1 turns · 2 steps
+- button "复制":
+  - img
+- button "在新对话中分支":
+  - img
+- text: {{clock}} cache hit 98% · 15,967 tokens · 1 turns · 2 steps
 - textbox "Message the agent"
 - button "Add attachment":
   - img
diff --git a/apps/web/tests/terminal-card.snapshot.ts b/apps/web/tests/terminal-card.snapshot.ts
new file mode 100644
index 0000000000..088a6e8326
--- /dev/null
+++ b/apps/web/tests/terminal-card.snapshot.ts
@@ -0,0 +1,306 @@
+// @vitest-environment jsdom
+// Terminal card snapshot over the BUILT client graph (the code-mode-fixture
+// idiom: real bundles via AppWebEntry, keyless FixtureApiClient transport).
+// Opens the fixture history session and pins the `card: 'terminal'` render
+// intent at both of its conversation render sites, for both chat-row shapes:
+// turn 60's `fx-bash` on the render-site fallback row (expand-gated body) and
+// turn 65's `bash` on the keyed BashRow registration (resident body). Turn 65
+// carries what turn 60's two clean prompt rows cannot — SGR runs resolved to
+// --dsw-* tokens, output past the chat cap, a nested cwd, and a non-zero exit
+// pill; turn 60 carries the multi-line command's per-line prompt rows.
+//
+// The details panel's Output section is NOT covered here: tool rows stopped
+// being details-panel click targets, and nothing else in the assembled
+// application opens that panel, so the surface cannot be driven end to end.
+// Its terminal rendering stays pinned in ui-conversation's
+// tests/terminal-card.spec.tsx, which mounts DetailsPanel with a selection
+// directly.
+import { readFileSync } from 'node:fs'
+import { join } from 'node:path'
+import { act, cleanup, fireEvent, screen, waitFor, within } from '@testing-library/react'
+import { afterEach, beforeEach, expect, it, vi } from 'vitest'
+import type { WebBootEntry } from '@deepseek-ai/dsh-client-modules/client'
+import { AppWebEntry } from '@deepseek-ai/dsh-client-web'
+
+const PLUGINS: readonly (WebBootEntry & { dir: string })[] = [
+  { id: '@deepseek-ai/dsh-client-connection', dir: 'connection', url: '/plugins/connection.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-runtime', dir: 'runtime', url: '/plugins/runtime.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-connection'], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-theme', dir: 'ui-theme', url: '/plugins/ui-theme.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-locale', dir: 'locale', url: '/plugins/locale.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-layout', dir: 'ui-layout', url: '/plugins/ui-layout.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-runtime'] },
+  { id: '@deepseek-ai/dsh-client-ui-sidebar', dir: 'ui-sidebar', url: '/plugins/ui-sidebar.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  { id: '@deepseek-ai/dsh-client-ui-conversation', dir: 'ui-conversation', url: '/plugins/ui-conversation.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  {
+    id: '@deepseek-ai/dsh-client-ui-workspace',
+    dir: 'ui-workspace',
+    url: '/plugins/ui-workspace.js',
+    rev: 'fx',
+    inject: [
+      '@deepseek-ai/dsh-client-runtime',
+      '@deepseek-ai/dsh-client-ui-conversation',
+      '@deepseek-ai/dsh-client-ui-sidebar',
+    ],
+  },
+]
+
+const bundles = new Map(PLUGINS.map(plugin => [
+  plugin.url,
+  readFileSync(join(process.cwd(), 'packages/client', plugin.dir, 'lib/client.js'), 'utf8'),
+]))
+
+interface FixtureWindow extends Window {
+  __DSH_BOOT__?: { rev: string; entries: WebBootEntry[] }
+  __ModuleLoader__?: unknown
+}
+
+class ResizeObserverStub {
+  observe(): void {}
+  disconnect(): void {}
+  unobserve(): void {}
+}
+
+const win = window as FixtureWindow
+let unmount: (() => void) | undefined
+
+beforeEach(() => {
+  localStorage.clear()
+  document.title = 'DeepSeek Harness'
+  vi.stubGlobal('ResizeObserver', ResizeObserverStub)
+  vi.stubGlobal('requestAnimationFrame', (callback: FrameRequestCallback) =>
+    setTimeout(() => { callback(0) }, 0) as unknown as number)
+  vi.stubGlobal('cancelAnimationFrame', (id: number) => { clearTimeout(id) })
+})
+
+afterEach(() => {
+  act(() => { unmount?.() })
+  unmount = undefined
+  cleanup()
+  delete win.__DSH_BOOT__
+  delete win.__ModuleLoader__
+  document.body.innerHTML = ''
+  document.head.querySelectorAll('style[data-plugin]').forEach((style) => { style.remove() })
+  document.title = ''
+  history.replaceState(null, '', '/')
+  vi.unstubAllGlobals()
+})
+
+/** Boot the complete built client graph against the populated fixture branch. */
+function boot(): void {
+  history.replaceState(null, '', '/?fixture')
+  const root = document.createElement('div')
+  root.id = 'root'
+  document.body.appendChild(root)
+  win.__DSH_BOOT__ = { rev: 'fx', entries: PLUGINS.map(({ dir: _dir, ...plugin }) => plugin) }
+  act(() => {
+    const entry = new AppWebEntry(root, {
+      fetchBundle: (url) => {
+        const code = bundles.get(url)
+        return code === undefined ? Promise.reject(new Error(`missing built bundle ${url}`)) : Promise.resolve(code)
+      },
+      executeBundle: (code) => { (0, eval)(code) },
+    })
+    void entry.run()
+    unmount = () => { entry.dispose() }
+  })
+}
+
+/** Collapse decorative whitespace while preserving the text a user sees. */
+function visibleText(element: Element): string {
+  return (element.textContent ?? '').replace(/\s+/g, ' ').trim()
+}
+
+/**
+ * Read one terminal card's user-visible state. Output lines keep their interior
+ * whitespace: holding column alignment is what this card exists for, so
+ * collapsing runs of spaces would hide the behavior under test.
+ */
+function readCard(card: Element) {
+  const status = card.querySelector('[class*="_status_"]')
+  const expander = card.querySelector('button[aria-expanded]')
+  return {
+    // One entry per command line: a multi-line command is one row per line.
+    prompt: [...card.querySelectorAll('[class*="_promptLine_"]')].map(row =>
+      `${row.querySelector('[class*="_cwd_"]')?.textContent ?? ''} ${row.querySelector('[class*="_command_"]')?.textContent ?? ''}`),
+    // Dots per prompt row: exactly one, on the first row — the exit status the
+    // view carries is the whole call's, so a dot per line would assert a
+    // per-line outcome bash does not report.
+    dotsPerPromptRow: [...card.querySelectorAll('[class*="_promptLine_"]')].map(row =>
+      row.querySelectorAll('[data-state]').length),
+    status: status === null ? null : status.textContent,
+    copy: card.querySelector('[class*="_copyButton_"]')?.textContent ?? null,
+    lines: [...card.querySelectorAll('[class*="_line_"]')].map(line => line.textContent),
+    expander: expander === null ? null : {
+      label: expander.getAttribute('aria-label'),
+      text: expander.textContent,
+      expanded: expander.getAttribute('aria-expanded'),
+    },
+    // The run-state dot at the head of the prompt line, by its StateDot state.
+    runState: card.querySelector('[class*="_runState_"][data-state]')?.getAttribute('data-state') ?? null,
+    runStateLabel: card.querySelector('[class*="_runStateLabel_"]')?.textContent ?? null,
+    // Every color the ANSI parser emits resolves through a --dsw-* token, so
+    // the card follows the theme instead of painting literal terminal rgb.
+    // Scoped to the output lines: the run-state dot is an inline-styled span
+    // too, and its geometry is not an ANSI-resolved color.
+    colors: [...new Set([...card.querySelectorAll('[class*="_line_"] span[style]')]
+      .map(span => span.getAttribute('style')))],
+  }
+}
+
+/** Open the fixture history session (the alpha log carrying both bash turns) and wait for its tail. */
+async function openFixtureSession(): Promise<void> {
+  const tree = await screen.findByRole('tree', { name: 'Sessions' }, { timeout: 10_000 })
+  // Anchor on the expandable Workspace group row: the title and the blank
+  // session row can both read "fixture".
+  const group = (await within(tree).findAllByText('fixture'))
+    .map(el => el.closest<HTMLElement>('[role="treeitem"]'))
+    .find(el => el?.getAttribute('aria-expanded') !== null)
+  if (group === null || group === undefined) throw new Error('fixture Workspace group missing')
+  if (group.getAttribute('aria-expanded') === 'false') {
+    fireEvent.click(within(group).getByText('fixture'))
+    await waitFor(() => {
+      expect(group.getAttribute('aria-expanded')).toBe('true')
+    })
+  }
+  fireEvent.click(await within(tree).findByText('Fixture 历史会话'))
+  await waitFor(() => {
+    expect(document.querySelector('[data-sample="bash-global"]')).not.toBeNull()
+  }, { timeout: 10_000 })
+}
+
+/** The keyed BashRow of fixture turn 65 (the one carrying the ANSI sample). */
+function keyedBashRow(): Element {
+  // Anchored on the BashRow wrapper (summary row + resident card), not on the
+  // summary row itself: the summary now shows the presenter's description (the
+  // contract's above-card text), so the command lives only in the card below it.
+  const row = [...document.querySelectorAll('[data-sample="bash-global"]')]
+    .map(node => node.parentElement)
+    .find((node): node is HTMLElement => node !== null && visibleText(node).includes('pnpm run check'))
+  if (row === undefined) throw new Error('keyed bash row for turn 65 missing')
+  return row
+}
+
+/** The turn-60 fallback row, which reaches the terminal card through GenericToolCard/ToolRow. */
+function fallbackBashRow(): Element {
+  const row = document.querySelector('[data-tool="fx-bash"]')
+  if (row === null) throw new Error('fx-bash fallback row missing')
+  return row
+}
+
+it('renders the keyed bash row with a resident terminal card', async () => {
+  boot()
+  await openFixtureSession()
+
+  const row = keyedBashRow()
+  const card = row.parentElement?.querySelector('[data-terminal]')
+  if (card === null || card === undefined) throw new Error('keyed bash row has no resident terminal card')
+  // The prompt shortens the nested cwd to its last segment, the exit pill comes
+  // from the sample's authored exit status (its body deliberately carries no
+  // `[exit code: N]` marker, since the real presenter consumes that one), ANSI
+  // runs land on theme tokens, and the chat cap (8) collapses the middle into a
+  // head/tail split with an expander between them.
+  expect(readCard(card)).toMatchInlineSnapshot(`
+    {
+      "colors": [
+        "font-weight: 700;",
+        "color: var(--dsw-alias-state-success-primary);",
+        "color: var(--dsw-alias-state-error-primary);",
+      ],
+      "copy": "复制",
+      "dotsPerPromptRow": [
+        1,
+      ],
+      "expander": {
+        "expanded": "false",
+        "label": "展开其余 13 行输出",
+        "text": "… 其余 13 行",
+      },
+      "lines": [
+        "Running 4 checks",
+        "✓ typecheck                                          1.82s",
+        "✓ lint                                               0.94s",
+        "✓ duplication                                        2.10s",
+        "StateDot.tsx                100%     100%        100%         -",
+        "markdown/Markdown.tsx       100%     100%        100%         -",
+        "",
+        "1 of 4 checks failed",
+      ],
+      "prompt": [
+        "nested pnpm run check",
+      ],
+      "runState": "error",
+      "runStateLabel": "失败",
+      "status": "退出码 1",
+    }
+  `)
+})
+
+it('the fallback row reaches the same card through its expand control', async () => {
+  boot()
+  await openFixtureSession()
+
+  const row = fallbackBashRow()
+  expect(row.querySelector('[data-terminal]')).toBeNull()
+  const toggle = row.querySelector('button[aria-expanded]')
+  if (toggle === null) throw new Error('fallback row expand control missing')
+  fireEvent.click(toggle)
+  const card = await waitFor(() => {
+    const found = row.querySelector('[data-terminal]')
+    if (found === null) throw new Error('terminal card missing after expanding the fallback row')
+    return found
+  })
+  // Three plain lines under the cap: no ANSI spans, no exit pill, no expander.
+  expect(readCard(card)).toMatchInlineSnapshot(`
+    {
+      "colors": [],
+      "copy": "复制",
+      "dotsPerPromptRow": [
+        1,
+        0,
+      ],
+      "expander": null,
+      "lines": [
+        "total 2",
+        "drwxr-xr-x fixture",
+        "-rw-r--r-- demo.txt",
+      ],
+      "prompt": [
+        "fixture ls -la",
+        "$ echo done",
+      ],
+      "runState": "done",
+      "runStateLabel": "已完成",
+      "status": null,
+    }
+  `)
+})
+
+it('the chat card expands the collapsed middle in place, without opening the details panel', async () => {
+  boot()
+  await openFixtureSession()
+
+  const card = keyedBashRow().parentElement?.querySelector('[data-terminal]')
+  if (card === null || card === undefined) throw new Error('resident terminal card missing')
+  const expander = card.querySelector('button[aria-expanded]')
+  if (expander === null) throw new Error('height-cap expander missing')
+  const capped = card.querySelectorAll('[class*="_line_"]').length
+
+  fireEvent.click(expander)
+  await waitFor(() => {
+    expect(card.querySelector('button[aria-expanded]')?.getAttribute('aria-expanded')).toBe('true')
+  })
+  expect({
+    cappedLines: capped,
+    expandedLines: card.querySelectorAll('[class*="_line_"]').length,
+    expanderLabel: card.querySelector('button[aria-expanded]')?.getAttribute('aria-label'),
+    // The card sits outside the summary row's click target, so toggling it
+    // left the details panel shut.
+    detailsOpen: screen.queryByText('Input') !== null,
+  }).toMatchInlineSnapshot(`
+    {
+      "cappedLines": 8,
+      "detailsOpen": false,
+      "expandedLines": 21,
+      "expanderLabel": "收起输出",
+    }
+  `)
+})
diff --git a/apps/web/tests/todo-display.snapshot.ts b/apps/web/tests/todo-display.snapshot.ts
index 8d6f7833b2..da621b6e35 100644
--- a/apps/web/tests/todo-display.snapshot.ts
+++ b/apps/web/tests/todo-display.snapshot.ts
@@ -189,7 +189,7 @@ it('hides the plan strip when the next turn starts', async () => {
   await openFixtureSession()
   expect(document.querySelector('[data-testid="todo-panel"]')).not.toBeNull()
 
-  const composer = await screen.findByPlaceholderText('Message the agent', {}, { timeout: 10_000 })
+  const composer = await screen.findByPlaceholderText('给智能体发消息', {}, { timeout: 10_000 })
   fireEvent.change(composer, { target: { value: '下一轮清空计划' } })
   fireEvent.keyDown(composer, { key: 'Enter' })
 
diff --git a/apps/web/tsconfig.json b/apps/web/tsconfig.json
index 1b0f807d5f..276251910c 100644
--- a/apps/web/tsconfig.json
+++ b/apps/web/tsconfig.json
@@ -28,13 +28,15 @@
     "tests/steering.e2e.ts",
     "tests/navigation-panes.e2e.ts",
     "tests/lifecycle-chrome.e2e.ts",
+    "tests/details-session-lifecycle.e2e.ts",
     "tests/settings-chrome.e2e.ts",
     "tests/workspace-management.e2e.ts",
     "tests/replay-round-trip.e2e.ts",
     "tests/seeded-history.e2e.ts",
     "tests/sidebar-scrollbar.e2e.ts",
     "tests/code-mode-round.e2e.ts",
-    "tests/cordis-tool-round.e2e.ts"
+    "tests/cordis-tool-round.e2e.ts",
+    "tests/message-actions.e2e.ts"
   ],
   "references": [
     {
diff --git a/bin/dsh b/bin/dsh
index 319d915b18..f85f28a5cd 100755
--- a/bin/dsh
+++ b/bin/dsh
@@ -1,7 +1,7 @@
 #!/bin/sh
-# dsh launcher: runs the apps/cli `dsh` bin FROM SOURCE through Node's native
-# TypeScript transform, so a symlink from anywhere (e.g. ~/.local/bin/dsh)
-# always executes the current working tree without a build step.
+# dsh launcher: runs the apps/cli `dsh` bin FROM SOURCE through the tsx ESM
+# hook, so a symlink from anywhere (e.g. ~/.local/bin/dsh) always executes the
+# current working tree without a build step.
 set -eu
 
 # Resolve symlink chains without readlink -f (not on every macOS).
@@ -15,8 +15,11 @@ while [ -L "$script" ]; do
 done
 root=$(CDPATH='' cd -- "$(dirname -- "$script")/.." && pwd)
 
-# The preloader projects this checkout's tsconfig paths into Node resolution;
-# TypeScript transformation itself remains Node-owned (no tsx/esbuild hook).
-exec node --experimental-transform-types \
-  --import "$root/scripts/tspath-loader.ts" \
+# The ESM-only tsx hook transforms TypeScript and projects this checkout's
+# tsconfig paths into Node resolution (the CJS hook stays off: the graph is
+# ESM-only and the CJS resolver costs ~0.4s of startup). Absolute paths keep
+# both the hook and the tsconfig anchored to this checkout when the launcher
+# runs from any cwd, where bare `tsx/esm` would not resolve.
+TSX_TSCONFIG_PATH="$root/tsconfig.json" \
+  exec node --import "$root/node_modules/tsx/dist/esm/index.mjs" \
   "$root/apps/cli/src/bin.ts" "$@"
diff --git a/docs/architecture.i18n.yaml b/docs/architecture.i18n.yaml
index c296e10fb1..4bc87d1ee1 100644
--- a/docs/architecture.i18n.yaml
+++ b/docs/architecture.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write docs/architecture.md
-architecture.md: 2ae982eba49b6dbd2365496915f9917071167813
-architecture.zh.md: abaef961504ff64dbcd1e8e8ba9bd002406fa7f4
+architecture.md: bb5414d6bb108056bf2ff25366e5afe261e1803a
+architecture.zh.md: 6d39a320019a1bf87141be0874a5a20a51fc3fbb
diff --git a/docs/architecture.md b/docs/architecture.md
index 2ae982eba4..bb5414d6bb 100644
--- a/docs/architecture.md
+++ b/docs/architecture.md
@@ -97,10 +97,10 @@ forever:
       'assistant/chunk'
       'assistant/message'
       schedule tool calls by ctx.tools.executionMode:
-        exclusive -> one-call barrier
-        parallel -> rolling pool, <= maxParallelToolCalls in flight; reclassify before start
-        each start -> 'tool/call' -> ordered tools/pre-execute -> concurrent tools/execute
-        each model-order result -> ordered tools/post-execute -> 'tool/result'
+        exclusive -> barrier
+        parallel -> rolling pool, <= maxParallelToolCalls; reclassify-at-start; scheduler failure -> stop starts, drain dispatches
+        start -> 'tool/call' -> ordered tools/pre-execute -> concurrent tools/execute
+        model-order result -> ordered tools/post-execute -> 'tool/result'
       drain accepted tool context and steering
       'step/end'
       continue for tools or steering unless a result concluded the turn
diff --git a/docs/architecture.zh.md b/docs/architecture.zh.md
index abaef96150..6d39a32001 100644
--- a/docs/architecture.zh.md
+++ b/docs/architecture.zh.md
@@ -97,10 +97,10 @@ forever:
       'assistant/chunk'
       'assistant/message'
       schedule tool calls by ctx.tools.executionMode:
-        exclusive -> one-call barrier
-        parallel -> rolling pool, <= maxParallelToolCalls in flight; reclassify before start
-        each start -> 'tool/call' -> ordered tools/pre-execute -> concurrent tools/execute
-        each model-order result -> ordered tools/post-execute -> 'tool/result'
+        exclusive -> barrier
+        parallel -> rolling pool, <= maxParallelToolCalls; reclassify-at-start; scheduler failure -> stop starts, drain dispatches
+        start -> 'tool/call' -> ordered tools/pre-execute -> concurrent tools/execute
+        model-order result -> ordered tools/post-execute -> 'tool/result'
       drain accepted tool context and steering
       'step/end'
       continue for tools or steering unless a result concluded the turn
diff --git a/docs/config-catalog.md b/docs/config-catalog.md
index 3c0543e4ca..514cda962b 100644
--- a/docs/config-catalog.md
+++ b/docs/config-catalog.md
@@ -1203,7 +1203,7 @@ export interface Config {
 }
 ```
 
-Source: [`packages/session-title/session-title/src/index.ts:75`](../packages/session-title/session-title/src/index.ts)
+Source: [`packages/session-title/session-title/src/index.ts:79`](../packages/session-title/session-title/src/index.ts)
 
 ## `@deepseek-ai/dsh-session-title-all-messages-llm`
 
@@ -1541,6 +1541,20 @@ export interface Config {
 
 Source: [`packages/context/time-context/src/index.ts:20`](../packages/context/time-context/src/index.ts)
 
+## `@deepseek-ai/dsh-tmux-context`
+
+Requires: `agents`
+
+```ts config-catalog
+/** Per-turn tmux-location scheduling. Invalid values fail plugin load. */
+export interface Config {
+  /** Minimum milliseconds between durable injections in one session. Omit or set to 0 to inject on every eligible change. */
+  refreshIntervalMs?: number
+}
+```
+
+Source: [`packages/context/tmux-context/src/index.ts:34`](../packages/context/tmux-context/src/index.ts)
+
 ## `@deepseek-ai/dsh-token-meter`
 
 ```ts config-catalog
diff --git a/docs/cordis-catalog/events.md b/docs/cordis-catalog/events.md
index 967241fbf6..3a60f441e0 100644
--- a/docs/cordis-catalog/events.md
+++ b/docs/cordis-catalog/events.md
@@ -660,7 +660,7 @@ Applies one command claim to the scoped Input. Dispatched with the session's sco
 'slash/input-begin-command'(request: BeginCommandRequest): true | undefined
 ```
 
-Source: [`packages/client/ui-slash/src/types.ts:230`](../../packages/client/ui-slash/src/types.ts)
+Source: [`packages/client/ui-slash/src/types.ts:232`](../../packages/client/ui-slash/src/types.ts)
 
 ### `slash/input-consume-token` — bail
 
@@ -676,7 +676,7 @@ Consumes one command token after business success (popup settle / menu-pick exec
 'slash/input-consume-token'(request: ConsumeTokenRequest): true | undefined
 ```
 
-Source: [`packages/client/ui-slash/src/types.ts:244`](../../packages/client/ui-slash/src/types.ts)
+Source: [`packages/client/ui-slash/src/types.ts:246`](../../packages/client/ui-slash/src/types.ts)
 
 ### `slash/input-insert-reference` — bail
 
@@ -692,7 +692,7 @@ Inserts one reference into the scoped Input (same carrier routing and applied-tr
 'slash/input-insert-reference'(request: InsertReferenceRequest): true | undefined
 ```
 
-Source: [`packages/client/ui-slash/src/types.ts:237`](../../packages/client/ui-slash/src/types.ts)
+Source: [`packages/client/ui-slash/src/types.ts:239`](../../packages/client/ui-slash/src/types.ts)
 
 ### `slash/input-insert-text` — bail
 
@@ -709,7 +709,7 @@ Replaces the trigger token span with literal text — the plain-text reference p
 'slash/input-insert-text'(request: InsertTextRequest): true | undefined
 ```
 
-Source: [`packages/client/ui-slash/src/types.ts:252`](../../packages/client/ui-slash/src/types.ts)
+Source: [`packages/client/ui-slash/src/types.ts:254`](../../packages/client/ui-slash/src/types.ts)
 
 ## `subagent/*`
 
diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md
index db31daaf06..20d3fbcd3b 100644
--- a/docs/cordis-catalog/services.md
+++ b/docs/cordis-catalog/services.md
@@ -1604,6 +1604,19 @@ Log-backed title fold plus asynchronous fallback generation.
  */
 get(session: Session): SessionTitleSnapshot | undefined
 
+/**
+ * Accept an explicit user title. Appends a `session/title` event with the
+ * `user` source, which pins the title: in-flight automatic generation is
+ * superseded and later user messages schedule none (an explicit
+ * {@link SessionTitleService.refresh} remains the deliberate unpin).
+ * @param session - exact live session to rename.
+ * @param title - raw user input; normalized before acceptance.
+ * @returns the accepted title snapshot.
+ * @throws {SessionTitleInvalidError} when the title normalizes to empty.
+ * @throws {Error} when the session is not live or the service is disposed.
+ */
+rename(session: Session, title: string): SessionTitleSnapshot
+
 /**
  * Explicitly retry the registered provider, or materialize the built-in
  * fallback when no provider is registered.
@@ -1624,7 +1637,7 @@ register(provider: SessionTitleProvider): () => Promise<void>
 
 Types: [Session](../core-data-structures/session.md) · [SessionTitleProvider](../core-data-structures/session-title.md) · [SessionTitleSnapshot](../core-data-structures/session-title.md)
 
-Source: [`packages/session-title/session-title/src/index.ts:240`](../../packages/session-title/session-title/src/index.ts)
+Source: [`packages/session-title/session-title/src/index.ts:261`](../../packages/session-title/session-title/src/index.ts)
 
 ## `ctx.skills` — `SkillService`
 
diff --git a/docs/core-data-structures/compaction.i18n.yaml b/docs/core-data-structures/compaction.i18n.yaml
index 94b6bd4c5d..89783b5d16 100644
--- a/docs/core-data-structures/compaction.i18n.yaml
+++ b/docs/core-data-structures/compaction.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write docs/core-data-structures/compaction.md
-compaction.md: 3ba5edd96c509e064ac7033b175b7ddd3c972452
-compaction.zh.md: d082b0d0545802500278e96ca41a273bce275f53
+compaction.md: 911b71d00fa4b42e9cdfa67f67d4e9b29e354a4a
+compaction.zh.md: 643a116ff2edbbb53d300b4f5ff0ad36d401130b
diff --git a/docs/core-data-structures/compaction.md b/docs/core-data-structures/compaction.md
index 3ba5edd96c..911b71d00f 100644
--- a/docs/core-data-structures/compaction.md
+++ b/docs/core-data-structures/compaction.md
@@ -13,7 +13,7 @@ Compaction extends [`SessionEventMap`](session.md) with three event types via de
 | Event | Payload | Role |
 |---|---|---|
 | `compact/start` | `{ turn }` | acquires the log-recorded lock |
-| `compact/summary` | `{ summary, shadowedRange, shadowedSeqs, shadowedTokenCount, provider, model, maxTokens? }` | provenance: the summary blocks, the shadowed surface-boundary pair (`start`/`end` seqs — a position span, not a numeric interval), the shadowed seqs in surface order, the estimated token count, and the summarize call's envelope (`provider`, `model`, plus its generation cap when one applied) — logged so the one-shot request is reconstructable from log + code (the reconstructability Agent Note) |
+| `compact/summary` | `{ summary, rawOutput?, shadowedRange, shadowedSeqs, shadowedTokenCount, provider, model, maxTokens?, usage? }` | provenance: the safe summary projection, optional complete provider output and usage, the shadowed surface-boundary pair (`start`/`end` seqs — a position span, not a numeric interval), the shadowed seqs in surface order, the estimated token count, and the summarize call's envelope (`provider`, `model`, plus its generation cap when one applied) — logged so the one-shot request is reconstructable from log + code (the reconstructability Agent Note) |
 | `compact/end` | `{ turn, error? }` | releases the lock (`error` set when summarization threw) |
 
 The lock brackets the **whole** operation: `compact/start` is appended first, then summarization, the `compact/summary` provenance record, and the `user/message` replacement all land, and only then `compact/end`. Releasing the lock last turns a crash mid-operation into a detectable orphaned lock (a `compact/start` with no matching `compact/end`) rather than a `compact/end` that falsely claims compaction finished.
@@ -22,7 +22,7 @@ These variants are merged inside a `declare module '@deepseek-ai/dsh-session'` b
 
 ## `CompactionResult`
 
-What a successful compaction returns to its caller: the bookkeeping-event seqs, raw summary, shadowed range and seqs, and estimated token count.
+What a successful compaction returns to its caller: the bookkeeping-event seqs, safe summary projection, shadowed range and seqs, and estimated token count.
 
 ```ts type-equiv
 /** Result of a successful compaction operation. */
diff --git a/docs/core-data-structures/compaction.zh.md b/docs/core-data-structures/compaction.zh.md
index d082b0d054..643a116ff2 100644
--- a/docs/core-data-structures/compaction.zh.md
+++ b/docs/core-data-structures/compaction.zh.md
@@ -13,7 +13,7 @@
 | 事件 | 载荷 | 作用 |
 |---|---|---|
 | `compact/start` | `{ turn }` | 获取日志记录的锁 |
-| `compact/summary` | `{ summary, shadowedRange, shadowedSeqs, shadowedTokenCount, provider, model, maxTokens? }` | provenance：摘要块、被遮蔽的 surface 边界对（`start`/`end` seq——位置跨度，而非数值区间）、按 surface 顺序排列的被遮蔽 seq、估算 token 数，以及摘要调用的 envelope（`provider`、`model`，若有生成上限则还包括该上限）——写入日志后，该一次性请求可由日志 + 代码重建（见可重建性 Agent Note） |
+| `compact/summary` | `{ summary, rawOutput?, shadowedRange, shadowedSeqs, shadowedTokenCount, provider, model, maxTokens?, usage? }` | provenance：安全摘要投影、可选的完整 provider 输出与 usage、被遮蔽的 surface 边界对（`start`/`end` seq——位置跨度，而非数值区间）、按 surface 顺序排列的被遮蔽 seq、估算 token 数，以及摘要调用的 envelope（`provider`、`model`，若有生成上限则还包括该上限）——写入日志后，该一次性请求可由日志 + 代码重建（见可重建性 Agent Note） |
 | `compact/end` | `{ turn, error? }` | 释放锁（摘要调用抛出异常时设置 `error`） |
 
 锁括住**整个**操作：先追加 `compact/start`，然后执行摘要生成、写入 `compact/summary` 来源记录与 `user/message` 替换，最后才追加 `compact/end`。最后释放锁意味着操作中途崩溃会表现为可检测的遗留锁（有 `compact/start` 而无匹配的 `compact/end`），而非一个虚假声称压缩已完成的 `compact/end`。
@@ -22,7 +22,7 @@
 
 ## `CompactionResult`
 
-成功压缩向调用方返回：记账事件 seq、原始摘要、被遮蔽的范围与 seq，以及估算 token 数。
+成功压缩向调用方返回：记账事件 seq、安全摘要投影、被遮蔽的范围与 seq，以及估算 token 数。
 
 ```ts type-equiv
 /** Result of a successful compaction operation. */
diff --git a/docs/core-data-structures/session-title.i18n.yaml b/docs/core-data-structures/session-title.i18n.yaml
index 64b3d8f342..ab368c4c37 100644
--- a/docs/core-data-structures/session-title.i18n.yaml
+++ b/docs/core-data-structures/session-title.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write docs/core-data-structures/session-title.md
-session-title.md: 33efc911c0ca1ae94dc4ded74676e5c32a73bdd5
-session-title.zh.md: a4b95a726d2bc89a13d14f1daa2f825cd5aa91b1
+session-title.md: fff1aa1f6be45d0cfc4d7f6a9527ccb93561618f
+session-title.zh.md: 73821b07c6be40d10d0961dd79b7c06bcadb7d0b
diff --git a/docs/core-data-structures/session-title.md b/docs/core-data-structures/session-title.md
index 33efc911c0..fff1aa1f6b 100644
--- a/docs/core-data-structures/session-title.md
+++ b/docs/core-data-structures/session-title.md
@@ -34,6 +34,10 @@ type SessionTitleSource =
     readonly provider: SessionTitleProviderId
     readonly model?: SessionTitleModelProvenance
   }
+  | {
+    /** Explicit user rename: pins the title — automatic generation stops scheduling. */
+    readonly kind: 'user'
+  }
 ```
 
 ```ts type-equiv
@@ -41,9 +45,9 @@ type SessionTitleSource =
 interface SessionTitleEventData {
   /** Normalized non-empty title text. */
   readonly title: string
-  /** Exact human `user/message` seqs used to derive this title. */
+  /** Exact human `user/message` seqs used to derive this title; empty for an explicit user rename. */
   readonly messageSeqs: number[]
-  /** Built-in fallback or registered-provider provenance. */
+  /** Built-in fallback, registered-provider, or explicit-user provenance. */
   readonly source: SessionTitleSource
 }
 ```
diff --git a/docs/core-data-structures/session-title.zh.md b/docs/core-data-structures/session-title.zh.md
index a4b95a726d..73821b07c6 100644
--- a/docs/core-data-structures/session-title.zh.md
+++ b/docs/core-data-structures/session-title.zh.md
@@ -34,6 +34,10 @@ type SessionTitleSource =
     readonly provider: SessionTitleProviderId
     readonly model?: SessionTitleModelProvenance
   }
+  | {
+    /** Explicit user rename: pins the title — automatic generation stops scheduling. */
+    readonly kind: 'user'
+  }
 ```
 
 ```ts type-equiv
@@ -41,9 +45,9 @@ type SessionTitleSource =
 interface SessionTitleEventData {
   /** Normalized non-empty title text. */
   readonly title: string
-  /** Exact human `user/message` seqs used to derive this title. */
+  /** Exact human `user/message` seqs used to derive this title; empty for an explicit user rename. */
   readonly messageSeqs: number[]
-  /** Built-in fallback or registered-provider provenance. */
+  /** Built-in fallback, registered-provider, or explicit-user provenance. */
   readonly source: SessionTitleSource
 }
 ```
diff --git a/docs/defensive-patterns.i18n.yaml b/docs/defensive-patterns.i18n.yaml
index 96c938a74f..674cd63efe 100644
--- a/docs/defensive-patterns.i18n.yaml
+++ b/docs/defensive-patterns.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-defensive-patterns.md: c69094db461048f5dbca5f8bdd1fb5581b08a962
-defensive-patterns.zh.md: eb57f035ad0bd67e62e285d451502d41e4efc2bc
+#   pnpm run verify-translation-pairing --write docs/defensive-patterns.md
+defensive-patterns.md: cc34877fb0d6a2e1740d8fa138f879363c8e69a3
+defensive-patterns.zh.md: 21b0977d8167ffecc21cdfab3c778efceefd8f03
diff --git a/docs/defensive-patterns.md b/docs/defensive-patterns.md
index c69094db46..cc34877fb0 100644
--- a/docs/defensive-patterns.md
+++ b/docs/defensive-patterns.md
@@ -26,4 +26,4 @@ A user-supplied listener that throws must not reject the promise it runs inside
 
 ## Never hand untrusted output the ambient environment or predictable paths
 
-Spawned commands get a scrubbed env (drop `*KEY*`/`*SECRET*`/`*TOKEN*`) so harness credentials cannot leak into output, `env`, or spill files. Temp/spill files use a private (0700) dir, random names, and exclusive owner-only opens (`'wx'`, `0o600`) — predictable world-readable paths invite symlink races and disclosure.
+Spawned commands get a scrubbed env (drop `*KEY*`/`*SECRET*`/`*TOKEN*`/`*PASSWORD*`) so harness credentials cannot leak into output, `env`, or spill files. Temp/spill files use a private (0700) dir, random names, and exclusive owner-only opens (`'wx'`, `0o600`) — predictable world-readable paths invite symlink races and disclosure.
diff --git a/docs/defensive-patterns.zh.md b/docs/defensive-patterns.zh.md
index eb57f035ad..21b0977d81 100644
--- a/docs/defensive-patterns.zh.md
+++ b/docs/defensive-patterns.zh.md
@@ -26,4 +26,4 @@
 
 ## 绝不将环境变量或可预测路径暴露给不可信输出
 
-spawn 的命令应获得一份经过清洗的 env（去除 `*KEY*`/`*SECRET*`/`*TOKEN*`），使 harness 凭证无法泄漏到输出、`env` 或溢出文件中。临时/溢出文件应使用私有（0700）目录、随机文件名和排他的仅所有者可访问打开方式（`'wx'`、`0o600`）——可预测的全局可读路径会招致符号链接竞争和信息泄露。
+spawn 的命令应获得一份经过清洗的 env（去除 `*KEY*`/`*SECRET*`/`*TOKEN*`/`*PASSWORD*`），使 harness 凭证无法泄漏到输出、`env` 或溢出文件中。临时/溢出文件应使用私有（0700）目录、随机文件名和排他的仅所有者可访问打开方式（`'wx'`、`0o600`）——可预测的全局可读路径会招致符号链接竞争和信息泄露。
diff --git a/docs/development.i18n.yaml b/docs/development.i18n.yaml
index 74d9dfbea5..311f872b4a 100644
--- a/docs/development.i18n.yaml
+++ b/docs/development.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write docs/development.md
-development.md: 32339fa2af8c1b6005d9e0b8165d57966a4145ca
-development.zh.md: c74a81346639c6f95568cbd86b401d134d5eb7fc
+development.md: 0a18e29d3da4f694707521e230017e6b22cad740
+development.zh.md: 885b51c701267215cc50d31ecd1694ae2c9af9ca
diff --git a/docs/development.md b/docs/development.md
index 32339fa2af..0a18e29d3d 100644
--- a/docs/development.md
+++ b/docs/development.md
@@ -27,7 +27,7 @@ If hooks are missing because dependencies were restored from cache or `postinsta
 node scripts/install-lefthook.mjs
 ```
 
-The wrapper refuses user-owned `core.hooksPath` values. An inherited system, global, or common-repository path requires `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`; command-scoped and worktree-scoped custom paths must be integrated or removed explicitly.
+The wrapper refuses user-owned `core.hooksPath` values. An inherited system, global, or common-repository path requires `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`. When Git seeds a new worktree with another registered worktree's marker-backed hook path, the wrapper replaces that copied value with the new worktree's own path; command-scoped and other worktree-scoped paths must be integrated or removed explicitly.
 
 Before enabling worktree config, migrate direct `extensions.*` in a format-0 common config, direct `core.worktree` or `core.bare=true`, and any non-empty dormant `config.worktree`. The common config and every worktree config must be regular files, while the owned hook directory may contain only unaliased regular files.
 
diff --git a/docs/development.zh.md b/docs/development.zh.md
index c74a813466..885b51c701 100644
--- a/docs/development.zh.md
+++ b/docs/development.zh.md
@@ -27,7 +27,7 @@ pnpm install
 node scripts/install-lefthook.mjs
 ```
 
-包装层会拒绝用户自有的 `core.hooksPath` 值。继承自系统、全局或共用仓库配置的路径必须设置 `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`；命令作用域和 worktree 作用域的自定义路径必须显式集成或移除。
+包装层会拒绝用户自有的 `core.hooksPath` 值。继承自系统、全局或共用仓库配置的路径必须设置 `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`。当 Git 使用另一个已注册 worktree 中由所有权标记佐证的钩子路径初始化新 worktree 时，包装层会将这个复制值替换为新 worktree 自有的路径；命令作用域和其他 worktree 作用域的路径必须显式集成或移除。
 
 启用 worktree 配置之前，请迁移格式 0 共用配置中直接设置的 `extensions.*`，并迁移直接设置的 `core.worktree` 或 `core.bare=true`，以及任何非空且尚未生效的 `config.worktree`。共用配置和每个 worktree 配置都必须是常规文件，而自有钩子目录只能包含不带别名的常规文件。
 
diff --git a/docs/event-producer-consumer.md b/docs/event-producer-consumer.md
index 71238305e5..6e7e9ed486 100644
--- a/docs/event-producer-consumer.md
+++ b/docs/event-producer-consumer.md
@@ -21,7 +21,7 @@ This matrix shows which packages dispatch each harness-owned event and which pac
 | `agent/session-start` | `emit` | [`packages/core/agent/src/types.ts:299`](../packages/core/agent/src/types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emitAgentEvent`) | [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hooks-claude`](../packages/hooks/hooks-claude), [`hooks-codex`](../packages/hooks/hooks-codex), [`workspace-context`](../packages/context/workspace-context) |
 | `agent/settled` | `emit` | [`packages/core/agent/src/types.ts:387`](../packages/core/agent/src/types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emitAgentEvent`) | [`compact-basic`](../packages/compact/compact-basic) |
 | `agent/status` | `emit` | [`packages/core/agent/src/types.ts:236`](../packages/core/agent/src/types.ts) | [`agent-loop`](../packages/core/agent-loop) (`emitAgentEvent`) | [`agent`](../packages/core/agent), `apiproxy`, [`goal-session`](../packages/goal/goal-session), [`tui`](../packages/ui/tui) |
-| `agent/step` | `serial` | [`packages/core/agent/src/types.ts:326`](../packages/core/agent/src/types.ts) | [`agent-loop`](../packages/core/agent-loop) (`serial`) | [`compact-basic`](../packages/compact/compact-basic), [`plan-mode`](../packages/plan/plan-mode), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`time-context`](../packages/context/time-context), [`tool-skill`](../packages/skill/tool-skill), [`user-approval`](../packages/ui/user-approval), [`workspace-context`](../packages/context/workspace-context) |
+| `agent/step` | `serial` | [`packages/core/agent/src/types.ts:326`](../packages/core/agent/src/types.ts) | [`agent-loop`](../packages/core/agent-loop) (`serial`) | [`compact-basic`](../packages/compact/compact-basic), [`plan-mode`](../packages/plan/plan-mode), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`time-context`](../packages/context/time-context), [`tmux-context`](../packages/context/tmux-context), [`tool-skill`](../packages/skill/tool-skill), [`user-approval`](../packages/ui/user-approval), [`workspace-context`](../packages/context/workspace-context) |
 | `agent/turn-stopping` | `serial` | [`packages/core/agent/src/types.ts:373`](../packages/core/agent/src/types.ts) | [`agent-loop`](../packages/core/agent-loop) (`serial`) | [`hooks-claude`](../packages/hooks/hooks-claude), [`hooks-codex`](../packages/hooks/hooks-codex) |
 | `approval/request` | `waterfall` | [`packages/ui/user-approval/src/index.ts:30`](../packages/ui/user-approval/src/index.ts) | [`user-approval`](../packages/ui/user-approval) (`waterfall`) | [`acp`](../packages/acp/acp), `apiproxy` |
 | `commands/change` | `emit` | [`packages/ui/commands/src/index.ts:154`](../packages/ui/commands/src/index.ts) | [`commands`](../packages/ui/commands) (`events.dispatch`) | `apiproxy`, [`tui`](../packages/ui/tui) |
@@ -35,10 +35,10 @@ This matrix shows which packages dispatch each harness-owned event and which pac
 | `session/disposed` | `emit` | [`packages/core/session/src/index.ts:81`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`agent-loop`](../packages/core/agent-loop), `apiproxy`, [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection-cache`](../packages/session-projection/session-projection-cache), [`session-telemetry`](../packages/telemetry/session-telemetry), [`session-title`](../packages/session-title/session-title) |
 | `session/event` | `emit` | [`packages/core/session/src/index.ts:93`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`acp`](../packages/acp/acp), `apiproxy`, [`cli-demo`](../packages/examples/cli-demo), [`compact`](../packages/compact/compact), [`compact-basic`](../packages/compact/compact-basic), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection`](../packages/session-projection/session-projection), [`session-projection-cache`](../packages/session-projection/session-projection-cache), [`session-telemetry`](../packages/telemetry/session-telemetry), [`session-title`](../packages/session-title/session-title), [`token-meter`](../packages/llm/token-meter), [`tools`](../packages/core/tools), [`tui`](../packages/ui/tui), [`user-approval`](../packages/ui/user-approval), [`workspace-context`](../packages/context/workspace-context) |
 | `session/flush` | `parallel` | [`packages/core/session/src/index.ts:103`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`session-persistence`](../packages/session-persistence/session-persistence), [`session-telemetry`](../packages/telemetry/session-telemetry) |
-| `slash/input-begin-command` | `bail` | [`packages/client/ui-slash/src/types.ts:230`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
-| `slash/input-consume-token` | `bail` | [`packages/client/ui-slash/src/types.ts:244`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
-| `slash/input-insert-reference` | `bail` | [`packages/client/ui-slash/src/types.ts:237`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
-| `slash/input-insert-text` | `bail` | [`packages/client/ui-slash/src/types.ts:252`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
+| `slash/input-begin-command` | `bail` | [`packages/client/ui-slash/src/types.ts:232`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
+| `slash/input-consume-token` | `bail` | [`packages/client/ui-slash/src/types.ts:246`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
+| `slash/input-insert-reference` | `bail` | [`packages/client/ui-slash/src/types.ts:239`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
+| `slash/input-insert-text` | `bail` | [`packages/client/ui-slash/src/types.ts:254`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` |
 | `subagent/end` | `emit` | [`packages/subagent/subagent/src/index.ts:140`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`hooks-claude`](../packages/hooks/hooks-claude), [`jsonrpc`](../packages/ui/jsonrpc), [`subagent`](../packages/subagent/subagent) |
 | `subagent/provider-added` | `emit` | [`packages/subagent/subagent/src/index.ts:114`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`emit`) | [`subagent`](../packages/subagent/subagent), [`tool-subagent`](../packages/subagent/tool-subagent) |
 | `subagent/provider-removed` | `emit` | [`packages/subagent/subagent/src/index.ts:120`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`subagent`](../packages/subagent/subagent), [`tool-subagent`](../packages/subagent/tool-subagent) |
@@ -65,7 +65,7 @@ This matrix shows which packages dispatch each harness-owned event and which pac
 | --- | --- | --- |
 | `commands/changed` | `runtime` (`emit`) | `ui-command` |
 | `connection/reset` | `runtime` (`emit`) | `ui-command` |
-| `internal/dispatch` | - | [`commands`](../packages/ui/commands), [`compact`](../packages/compact/compact), [`fs`](../packages/fs/fs), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`llm-retry`](../packages/llm/llm-retry), [`permission`](../packages/ui/permission), [`plan-mode`](../packages/plan/plan-mode), [`pty-local`](../packages/pty/pty-local), `runtime`, [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent), [`time-context`](../packages/context/time-context), [`tool-todo`](../packages/todo/tool-todo), [`tools`](../packages/core/tools), [`user-approval`](../packages/ui/user-approval), [`workflow`](../packages/workflow/workflow) |
+| `internal/dispatch` | - | [`commands`](../packages/ui/commands), [`compact`](../packages/compact/compact), [`fs`](../packages/fs/fs), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`llm-retry`](../packages/llm/llm-retry), [`permission`](../packages/ui/permission), [`plan-mode`](../packages/plan/plan-mode), [`pty-local`](../packages/pty/pty-local), `runtime`, [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-title`](../packages/session-title/session-title), [`subagent`](../packages/subagent/subagent), [`time-context`](../packages/context/time-context), [`tool-todo`](../packages/todo/tool-todo), [`tools`](../packages/core/tools), [`user-approval`](../packages/ui/user-approval), [`workflow`](../packages/workflow/workflow) |
 | `internal/plugin` | - | `hmr`, `modules`, `webserver` |
 | `internal/status` | - | [`agent`](../packages/core/agent) |
 | `locale/change` | `locale` (`emit`) | `locale`, `ui-models`, `ui-settings-general` |
diff --git a/docs/i18n/README.i18n.yaml b/docs/i18n/README.i18n.yaml
index e4fb87a322..5ad0a7b7dd 100644
--- a/docs/i18n/README.i18n.yaml
+++ b/docs/i18n/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write docs/i18n/README.md
-README.md: 9fe01174909f0400b94c70b95759526aec783d2e
-README.zh.md: 2a9a20ae4702e205879f0fbfb6d76d3a761d2446
+README.md: d1fff61a1d3db7fbe93d3c95b144598c2ca41c8b
+README.zh.md: 9f1d61f1b854ead0ef5851157cce2e9a6de83d89
diff --git a/docs/i18n/README.md b/docs/i18n/README.md
index 9fe0117490..d1fff61a1d 100644
--- a/docs/i18n/README.md
+++ b/docs/i18n/README.md
@@ -15,7 +15,7 @@ This repo's documentation is read by people and agents both inside and outside t
   foo.zh.md: 89e6c98d92887913cadf06b2adb97f26cde4849b
   ```
 
-  Blob hashes, not commit hashes, so the record is computable for files edited in the same PR (`git hash-object foo.md`) and consistency is a pure content comparison. The recorded hashes also recover the exact last-confirmed text of either side, so an out-of-sync pair is updated by patching the counterpart minimally against the edited side's diff — never by re-translating whole files. `pnpm run gen-translation-brief <pair>` assembles that update's working set mechanically at the narrowest safely aligned granularity — changed Markdown units, then heading sections, then whole document — with the edited side's diff since last confirmation, each changed span's three-way text, the terminology rows the change touches, and the binding update rules; a change confined to the pair's byte-identical code fences is computed outright, and `--apply` splices it into the counterpart after structural validation ([briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)). After bringing the pair back in line, `pnpm run verify-translation-pairing --write <pair>` re-records both hashes; that yaml diff is the reviewable act of confirming consistency, which is why `--write` requires naming the pairs you confirmed (`--write --all` is the explicit corpus-wide form).
+  Blob hashes, not commit hashes, so the record is computable for files edited in the same PR (`git hash-object foo.md`) and consistency is a pure content comparison. `--write` stores those snapshots in the local Git object database before recording them, including uncommitted working-tree contents, and pins every distinct stored blob under a content-addressed `refs/dsh/translation-pairing/snapshots/` ref so garbage collection cannot invalidate a recorded recovery pointer. The recorded hashes therefore recover the exact last-confirmed text of either side, so an out-of-sync pair is updated by patching the counterpart minimally against the edited side's diff — never by re-translating whole files. `pnpm run gen-translation-brief <pair>` assembles that update's working set mechanically at the narrowest safely aligned granularity — changed Markdown units, then heading sections, then whole document — with the edited side's diff since last confirmation, each changed span's three-way text, the terminology rows the change touches, and the binding update rules; a change confined to the pair's byte-identical code fences is computed outright, and `--apply` splices it into the counterpart after structural validation ([briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)). After bringing the pair back in line, `pnpm run verify-translation-pairing --write <pair>` re-records both hashes; that yaml diff is the reviewable act of confirming consistency, which is why `--write` requires naming the pairs you confirmed (`--write --all` is the explicit corpus-wide form).
 - **Language switcher.** Both files link to each other immediately after their H1 heading: the English file carries `English | [中文](foo.zh.md)` and the Chinese file carries `[English](foo.md) | 中文`.
 - **Structure mirrors the counterpart.** Heading depths and order, list kinds, ordered-list starts, list item counts, table row and column counts, link targets, and verbatim code blocks match one to one across the pair — see [translation-rules.md](translation-rules.md) for the full preservation rules. Existing Markdown gates apply to `.zh.md` files unchanged (`verify-md-wrap`, `verify-md-links`).
 
diff --git a/docs/i18n/README.zh.md b/docs/i18n/README.zh.md
index 2a9a20ae47..9f1d61f1b8 100644
--- a/docs/i18n/README.zh.md
+++ b/docs/i18n/README.zh.md
@@ -15,7 +15,7 @@
   foo.zh.md: 89e6c98d92887913cadf06b2adb97f26cde4849b
   ```
 
-  用 blob hash 而不是 commit hash，这样同一个 PR 里改动的文件也能算出记录（`git hash-object foo.md`），一致性是纯内容比较。记录的 hash 还能还原任一侧上次确认时的确切文本，所以失去同步的配对是「按被改一侧的 diff 最小化地修补另一侧」，从不整篇重译。`pnpm run gen-translation-brief <pair>` 会以能安全对齐的最窄粒度——先是有改动的 Markdown 单元，再是标题小节，最后是整篇文档——机械地汇集这次更新的工作集：被改一侧自上次确认以来的 diff、每个改动块的三方文本、改动触及的术语表行，以及有约束力的更新规则；仅落在配对中逐字节一致的围栏代码块内的改动可以直接算出，`--apply` 则经结构签名校验后把它拼接进对侧文件（[briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)）。两侧对齐后，`pnpm run verify-translation-pairing --write <pair>` 重新记录两个 hash；那份 yaml diff 就是「确认一致」这个动作本身，可以被评审，也正因如此，`--write` 要求点名你确认过的配对（`--write --all` 是显式的全语料形式）。
+  用 blob hash 而不是 commit hash，这样同一个 PR 里改动的文件也能算出记录（`git hash-object foo.md`），一致性是纯内容比较。`--write` 会先把这些快照存入本地 Git 对象库再写下记录，未提交的工作树内容也不例外；它还会在内容寻址的 `refs/dsh/translation-pairing/snapshots/` ref 下固定每个不同的已存 blob，使垃圾回收无法让已记录的恢复指针失效。因此记录的 hash 能还原任一侧上次确认时的确切文本，所以失去同步的配对是「按被改一侧的 diff 最小化地修补另一侧」，从不整篇重译。`pnpm run gen-translation-brief <pair>` 会以能安全对齐的最窄粒度——先是有改动的 Markdown 单元，再是标题小节，最后是整篇文档——机械地汇集这次更新的工作集：被改一侧自上次确认以来的 diff、每个改动块的三方文本、改动触及的术语表行，以及有约束力的更新规则；仅落在配对中逐字节一致的围栏代码块内的改动可以直接算出，`--apply` 则经结构签名校验后把它拼接进对侧文件（[briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)）。两侧对齐后，`pnpm run verify-translation-pairing --write <pair>` 重新记录两个 hash；那份 yaml diff 就是「确认一致」这个动作本身，可以被评审，也正因如此，`--write` 要求点名你确认过的配对（`--write --all` 是显式的全语料形式）。
 - **语言切换行。** 两个文件在各自 H1 标题之后立即互链：英文文件带 `English | [中文](foo.zh.md)`，中文文件带 `[English](foo.md) | 中文`。
 - **结构与另一侧一一对应。** 标题深度与顺序、列表类型、有序列表起始编号、列表项数量、表格行列数、链接目标与逐字节一致的代码块在配对两侧一一对应；完整保持规则见 [translation-rules.md](translation-rules.md)。既有 Markdown 门禁对 `.zh.md` 文件原样生效（`verify-md-wrap`、`verify-md-links`）。
 
diff --git a/docs/i18n/terminology.md b/docs/i18n/terminology.md
index 3a40629372..5072c2684d 100644
--- a/docs/i18n/terminology.md
+++ b/docs/i18n/terminology.md
@@ -54,7 +54,7 @@
 | Round | Round | | 回合、目标回合、Ralph 回合 | 外层策略使用 Round 时，领域层级为 Session > Round > Turn（轮次） > Step（步骤）；Round 是可选的外层策略迭代，并非每个会话轮次都具有的通用层级。Goal Round 与 Ralph Round 均保留英文。一个 Round 承载一个轮次，步骤隶属于该轮次；明确的零步骤轮次仍保持原义。 |
 | schema | schema | | | |
 | schema DSL | schema DSL | | | |
-| seam | seam | | | 与 `extension point` 是不同概念；根据具体语境，可译为`服务边界`或`可替换点` |
+| seam | seam | | 接缝 | 与 `extension point` 是不同概念；根据具体语境，可译为`服务边界`或`可替换点` |
 | skill | skill | skill（技能） | | |
 | spawn | spawn | | | |
 | steering | steering | steering（中途引导） | | |
@@ -74,21 +74,24 @@
 | adapter | 适配器 | | | |
 | adapter contract | 适配器契约 | 适配器契约（adapter contract） | | |
 | append-only | 仅追加 | | | |
-| artifact | 产物 | | | |
+| artifact | 产物 | | 制品 | |
 | backend | 后端 | | | |
 | background task | 后台任务 | | | |
 | block | 块 | | | |
 | build target | 构建目标 | | | |
 | cancel | 取消 | | | |
+| canary test | canary 测试 | | 金丝雀测试 | 本仓库保留 `canary` |
+| capability seam | 能力 seam | | 功能 seam、能力接缝 | 本仓库接口、实现与消费方分离的命名架构概念；普通 `seam` 仍按其词条处理 |
 | feature | 功能 | | 能力 | SDK 产品与工程模型中的可管理产品单元 |
 | feature option | 功能选项 | | variant | 一项 SDK 功能内有限、可选择的实现或配置 |
 | checkpoint | 检查点 | | | |
 | chunk | 分片 | | | |
 | compaction | 压缩 | 压缩（compaction） | | |
 | companion tool | 配套工具 | | | |
+| composition bundle | 组合包 | | | 只约束应用或插件的组合语境，不约束所有 `bundle` |
 | Cordis plugin config | Cordis 插件配置 | | | Cordis 插件公开的 `Config` 对象或配置结构 |
 | config key | 配置键 | | | Cordis 插件配置中的单个字段 |
-| consumer | 消费方 | | | |
+| consumer | 消费方 | | 消费者 | |
 | content block | 内容块 | | | |
 | Cookbook | 实操手册 | | | 文档标题用语 |
 | context | 上下文 | | | |
@@ -145,9 +148,10 @@
 | persistence | 持久化 | | | |
 | pipeline | 流水线 | | | |
 | plugin | 插件 | | | |
+| postmortem | 事故复盘 | 事故复盘（postmortem） | 事后分析、事故记录 | 事故记录与分析文档；目录或路径中的 `postmortem` 保持代码形式 |
 | prompt | 提示词 | | | |
 | provider | 提供方 | | | |
-| provider-neutral | 提供方无关 | | | |
+| provider-neutral | 提供方无关 | | 提供方中立 | |
 | quality gate | 质量门禁 | | | |
 | quiescence | 完全停稳 | | 静默、静止状态 | 指生命周期工作全部结算后的状态 |
 | reasoning | 推理 | 推理（reasoning） | | 需要和 `inference` 区分时保留英文括注 |
@@ -165,7 +169,7 @@
 | sidecar record | 伴随记录 | | 旁挂记录 | 指与文档同目录的伴随记录文件 |
 | smoke test | 冒烟测试 | | | |
 | snapshot | 快照 | | | |
-| source of truth | 真源 | | | |
+| source of truth | 真源 | | 事实来源、唯一来源 | |
 | spine | 主干 | | | |
 | staged | 暂存 | | | 沿用 git 官方中文翻译 |
 | stale | 陈旧 | | 过期 | 与 `fresh`（`新鲜`）成对；门禁输出中保留英文 `stale` 不翻译；`expired` 才译为`过期` |
diff --git a/docs/module-graph.md b/docs/module-graph.md
index e83228b00d..2b873c1ed3 100644
--- a/docs/module-graph.md
+++ b/docs/module-graph.md
@@ -173,6 +173,7 @@ flowchart TD
   subgraph group_context["packages/context"]
     pkg_session_reference["session-reference"]
     pkg_time_context["time-context"]
+    pkg_tmux_context["tmux-context"]
     pkg_workspace_context["workspace-context"]
   end
   subgraph group_examples["packages/examples"]
@@ -264,7 +265,6 @@ flowchart TD
   pkg_client_ui_primitives --> pkg_invariants
   pkg_client_ui_question --> pkg_invariants
   pkg_client_ui_slots --> pkg_invariants
-  pkg_client_ui_trajectory --> pkg_invariants
   pkg_client_web --> pkg_invariants
   pkg_client_web_react --> pkg_invariants
   pkg_code_runtime --> pkg_invariants
@@ -301,9 +301,8 @@ flowchart TD
   pkg_client_ui_sidebar --> pkg_client_ui_primitives
   pkg_client_ui_sidebar --> pkg_client_ui_slots
   pkg_client_ui_sidebar --> pkg_invariants
-  pkg_client_ui_slash --> pkg_client_runtime
-  pkg_client_ui_slash --> pkg_client_ui_slots
-  pkg_client_ui_slash --> pkg_invariants
+  pkg_client_ui_trajectory --> pkg_client_ui_primitives
+  pkg_client_ui_trajectory --> pkg_invariants
   pkg_client_ui_workspace --> pkg_client_runtime
   pkg_client_ui_workspace --> pkg_client_ui_primitives
   pkg_client_ui_workspace --> pkg_client_ui_slots
@@ -337,26 +336,17 @@ flowchart TD
   pkg_system_prompt --> pkg_scope
   pkg_web --> pkg_invariants
   pkg_web --> pkg_llm
-  pkg_client_ui_conversation --> pkg_client_runtime
-  pkg_client_ui_conversation --> pkg_client_ui_primitives
-  pkg_client_ui_conversation --> pkg_client_ui_slash
-  pkg_client_ui_conversation --> pkg_client_ui_slots
-  pkg_client_ui_conversation --> pkg_invariants
   pkg_client_ui_settings_general --> pkg_client_locale
   pkg_client_ui_settings_general --> pkg_client_runtime
   pkg_client_ui_settings_general --> pkg_client_ui_primitives
   pkg_client_ui_settings_general --> pkg_client_ui_settings
   pkg_client_ui_settings_general --> pkg_client_ui_slots
   pkg_client_ui_settings_general --> pkg_invariants
-  pkg_client_ui_skill --> pkg_client_connection
-  pkg_client_ui_skill --> pkg_client_runtime
-  pkg_client_ui_skill --> pkg_client_ui_slash
-  pkg_client_ui_skill --> pkg_client_ui_slots
-  pkg_client_ui_skill --> pkg_invariants
-  pkg_client_ui_subagent --> pkg_client_runtime
-  pkg_client_ui_subagent --> pkg_client_ui_slash
-  pkg_client_ui_subagent --> pkg_client_ui_slots
-  pkg_client_ui_subagent --> pkg_invariants
+  pkg_client_ui_slash --> pkg_client_locale
+  pkg_client_ui_slash --> pkg_client_runtime
+  pkg_client_ui_slash --> pkg_client_ui_primitives
+  pkg_client_ui_slash --> pkg_client_ui_slots
+  pkg_client_ui_slash --> pkg_invariants
   pkg_client_ui_theme --> pkg_client_locale
   pkg_client_ui_theme --> pkg_client_runtime
   pkg_client_ui_theme --> pkg_client_ui_primitives
@@ -421,17 +411,25 @@ flowchart TD
   pkg_app_boot --> pkg_invariants
   pkg_app_boot --> pkg_paths
   pkg_app_boot --> pkg_system_prompt
-  pkg_client_ui_command --> pkg_client_connection
-  pkg_client_ui_command --> pkg_client_runtime
-  pkg_client_ui_command --> pkg_client_ui_conversation
-  pkg_client_ui_command --> pkg_client_ui_primitives
-  pkg_client_ui_command --> pkg_client_ui_slash
-  pkg_client_ui_command --> pkg_client_ui_slots
-  pkg_client_ui_command --> pkg_invariants
+  pkg_client_ui_conversation --> pkg_client_locale
+  pkg_client_ui_conversation --> pkg_client_runtime
+  pkg_client_ui_conversation --> pkg_client_ui_primitives
+  pkg_client_ui_conversation --> pkg_client_ui_slash
+  pkg_client_ui_conversation --> pkg_client_ui_slots
+  pkg_client_ui_conversation --> pkg_invariants
   pkg_client_ui_layout --> pkg_client_runtime
   pkg_client_ui_layout --> pkg_client_ui_slots
   pkg_client_ui_layout --> pkg_client_ui_theme
   pkg_client_ui_layout --> pkg_invariants
+  pkg_client_ui_skill --> pkg_client_connection
+  pkg_client_ui_skill --> pkg_client_runtime
+  pkg_client_ui_skill --> pkg_client_ui_slash
+  pkg_client_ui_skill --> pkg_client_ui_slots
+  pkg_client_ui_skill --> pkg_invariants
+  pkg_client_ui_subagent --> pkg_client_runtime
+  pkg_client_ui_subagent --> pkg_client_ui_slash
+  pkg_client_ui_subagent --> pkg_client_ui_slots
+  pkg_client_ui_subagent --> pkg_invariants
   pkg_code_runtime_worker --> pkg_code_runtime
   pkg_code_runtime_worker --> pkg_invariants
   pkg_code_runtime_worker --> pkg_session
@@ -512,17 +510,20 @@ flowchart TD
   pkg_user_interaction --> pkg_agent
   pkg_user_interaction --> pkg_invariants
   pkg_user_interaction --> pkg_llm
-  pkg_client_ui_model --> pkg_client_connection
-  pkg_client_ui_model --> pkg_client_runtime
-  pkg_client_ui_model --> pkg_client_ui_command
-  pkg_client_ui_model --> pkg_client_ui_conversation
-  pkg_client_ui_model --> pkg_client_ui_primitives
-  pkg_client_ui_model --> pkg_client_ui_slash
-  pkg_client_ui_model --> pkg_client_ui_slots
-  pkg_client_ui_model --> pkg_invariants
+  pkg_client_ui_command --> pkg_client_connection
+  pkg_client_ui_command --> pkg_client_runtime
+  pkg_client_ui_command --> pkg_client_ui_conversation
+  pkg_client_ui_command --> pkg_client_ui_primitives
+  pkg_client_ui_command --> pkg_client_ui_slash
+  pkg_client_ui_command --> pkg_client_ui_slots
+  pkg_client_ui_command --> pkg_invariants
   pkg_time_context --> pkg_agent
   pkg_time_context --> pkg_invariants
   pkg_time_context --> pkg_session
+  pkg_tmux_context --> pkg_agent
+  pkg_tmux_context --> pkg_bash
+  pkg_tmux_context --> pkg_invariants
+  pkg_tmux_context --> pkg_session
   pkg_pty --> pkg_agent
   pkg_pty --> pkg_brand
   pkg_pty --> pkg_invariants
@@ -607,6 +608,14 @@ flowchart TD
   pkg_client_ui_goal --> pkg_client_ui_slots
   pkg_client_ui_goal --> pkg_goal
   pkg_client_ui_goal --> pkg_invariants
+  pkg_client_ui_model --> pkg_client_connection
+  pkg_client_ui_model --> pkg_client_runtime
+  pkg_client_ui_model --> pkg_client_ui_command
+  pkg_client_ui_model --> pkg_client_ui_conversation
+  pkg_client_ui_model --> pkg_client_ui_primitives
+  pkg_client_ui_model --> pkg_client_ui_slash
+  pkg_client_ui_model --> pkg_client_ui_slots
+  pkg_client_ui_model --> pkg_invariants
   pkg_pty_local --> pkg_agent
   pkg_pty_local --> pkg_invariants
   pkg_pty_local --> pkg_pty
@@ -701,6 +710,7 @@ flowchart TD
   pkg_plan_mode --> pkg_agent
   pkg_plan_mode --> pkg_commands
   pkg_plan_mode --> pkg_invariants
+  pkg_plan_mode --> pkg_llm
   pkg_plan_mode --> pkg_session
   pkg_plan_mode --> pkg_session_projection
   pkg_plan_mode --> pkg_system_prompt
@@ -795,6 +805,7 @@ flowchart TD
   pkg_tool_pty --> pkg_tools
   pkg_tool_tasks --> pkg_agent
   pkg_tool_tasks --> pkg_invariants
+  pkg_tool_tasks --> pkg_llm
   pkg_tool_tasks --> pkg_retention
   pkg_tool_tasks --> pkg_system_prompt
   pkg_tool_tasks --> pkg_tasks
@@ -847,6 +858,7 @@ flowchart TD
   pkg_tui --> pkg_session_reference
   pkg_tui --> pkg_session_title
   pkg_tui --> pkg_skill
+  pkg_tui --> pkg_subprocess
   pkg_tui --> pkg_system_prompt
   pkg_tui --> pkg_token_meter
   pkg_tui --> pkg_tools
@@ -982,7 +994,6 @@ flowchart TD
 | [`client-ui-primitives`](../packages/client/ui-primitives) | `client` | [`invariants`](../packages/support/invariants) |
 | [`client-ui-question`](../packages/client/ui-question) | `client` | [`invariants`](../packages/support/invariants) |
 | [`client-ui-slots`](../packages/client/ui-slots) | `client` | [`invariants`](../packages/support/invariants) |
-| [`client-ui-trajectory`](../packages/client/ui-trajectory) | `client` | [`invariants`](../packages/support/invariants) |
 | [`client-web`](../packages/client/web) | `client` | [`invariants`](../packages/support/invariants) |
 | [`client-web-react`](../packages/client/web-react) | `client` | [`invariants`](../packages/support/invariants) |
 | [`code-runtime`](../packages/code-runtime/code-runtime) | `code-runtime` | [`invariants`](../packages/support/invariants) |
@@ -1000,7 +1011,7 @@ flowchart TD
 | [`client-ui-models`](../packages/client/ui-models) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`client-ui-settings`](../packages/client/ui-settings) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`client-ui-sidebar`](../packages/client/ui-sidebar) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
-| [`client-ui-slash`](../packages/client/ui-slash) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
+| [`client-ui-trajectory`](../packages/client/ui-trajectory) | `client` | [`client-ui-primitives`](../packages/client/ui-primitives), [`invariants`](../packages/support/invariants) |
 | [`client-ui-workspace`](../packages/client/ui-workspace) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`helper`](../packages/sdk/helper) | `sdk` | [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`subprocess`](../packages/subprocess/subprocess) |
 | [`telemetry`](../packages/sdk/telemetry) | `sdk` | [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`paths`](../packages/util/paths) |
@@ -1013,10 +1024,8 @@ flowchart TD
 | [`session`](../packages/core/session) | `core` | [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope) |
 | [`system-prompt`](../packages/core/system-prompt) | `core` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope) |
 | [`web`](../packages/web/web) | `web` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm) |
-| [`client-ui-conversation`](../packages/client/ui-conversation) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`client-ui-settings-general`](../packages/client/ui-settings-general) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-settings`](../packages/client/ui-settings), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
-| [`client-ui-skill`](../packages/client/ui-skill) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
-| [`client-ui-subagent`](../packages/client/ui-subagent) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
+| [`client-ui-slash`](../packages/client/ui-slash) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`client-ui-theme`](../packages/client/ui-theme) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`host-directory-picker-browse`](../packages/host/directory-picker-browse) | `host` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`client-ui-workspace`](../packages/client/ui-workspace), [`invariants`](../packages/support/invariants) |
 | [`host-directory-picker-native`](../packages/host/directory-picker-native) | `host` | [`client-runtime`](../packages/client/runtime), [`client-ui-slots`](../packages/client/ui-slots), [`client-ui-workspace`](../packages/client/ui-workspace), [`invariants`](../packages/support/invariants) |
@@ -1036,8 +1045,10 @@ flowchart TD
 | [`session-persistence`](../packages/session-persistence/session-persistence) | `session-persistence` | [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session) |
 | [`llm-replay`](../packages/support/llm-replay) | `support` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session) |
 | [`app-boot`](../packages/ui/app-boot) | `ui` | [`invariants`](../packages/support/invariants), [`paths`](../packages/util/paths), [`system-prompt`](../packages/core/system-prompt) |
-| [`client-ui-command`](../packages/client/ui-command) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
+| [`client-ui-conversation`](../packages/client/ui-conversation) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`client-ui-layout`](../packages/client/ui-layout) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slots`](../packages/client/ui-slots), [`client-ui-theme`](../packages/client/ui-theme), [`invariants`](../packages/support/invariants) |
+| [`client-ui-skill`](../packages/client/ui-skill) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
+| [`client-ui-subagent`](../packages/client/ui-subagent) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`code-runtime-worker`](../packages/code-runtime/code-runtime-worker) | `code-runtime` | [`code-runtime`](../packages/code-runtime/code-runtime), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`timeout`](../packages/util/timeout) |
 | [`lsp-local`](../packages/lsp/lsp-local) | `lsp` | [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`lsp`](../packages/lsp/lsp), [`subprocess`](../packages/subprocess/subprocess), [`timeout`](../packages/util/timeout) |
 | [`sandbox-local`](../packages/sandbox/sandbox-local) | `sandbox` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`sandbox`](../packages/sandbox/sandbox) |
@@ -1058,8 +1069,9 @@ flowchart TD
 | [`commands`](../packages/ui/commands) | `ui` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`scope`](../packages/core/scope), [`session`](../packages/core/session) |
 | [`user-approval`](../packages/ui/user-approval) | `ui` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`system-prompt`](../packages/core/system-prompt) |
 | [`user-interaction`](../packages/ui/user-interaction) | `ui` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm) |
-| [`client-ui-model`](../packages/client/ui-model) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-command`](../packages/client/ui-command), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
+| [`client-ui-command`](../packages/client/ui-command) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`time-context`](../packages/context/time-context) | `context` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session) |
+| [`tmux-context`](../packages/context/tmux-context) | `context` | [`agent`](../packages/core/agent), [`bash`](../packages/bash/bash), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session) |
 | [`pty`](../packages/pty/pty) | `pty` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants) |
 | [`scripts`](../packages/sdk/scripts) | `sdk` | [`app-boot`](../packages/ui/app-boot), [`invariants`](../packages/support/invariants) |
 | [`session-projection-cache`](../packages/session-projection/session-projection-cache) | `session-projection` | [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection`](../packages/session-projection/session-projection), [`storage-domain`](../packages/storage/storage-domain) |
@@ -1077,6 +1089,7 @@ flowchart TD
 | [`acp`](../packages/acp/acp) | `acp` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`user-approval`](../packages/ui/user-approval) |
 | [`permission`](../packages/ui/permission) | `ui` | [`bash`](../packages/bash/bash), [`commands`](../packages/ui/commands), [`invariants`](../packages/support/invariants), [`sandbox`](../packages/sandbox/sandbox), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection), [`user-approval`](../packages/ui/user-approval) |
 | [`client-ui-goal`](../packages/client/ui-goal) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`goal`](../packages/goal/goal), [`invariants`](../packages/support/invariants) |
+| [`client-ui-model`](../packages/client/ui-model) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-command`](../packages/client/ui-command), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) |
 | [`pty-local`](../packages/pty/pty-local) | `pty` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`pty`](../packages/pty/pty), [`sandbox`](../packages/sandbox/sandbox), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`session`](../packages/core/session), [`subprocess`](../packages/subprocess/subprocess) |
 | [`tasks-local`](../packages/tasks/tasks-local) | `tasks` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`tasks`](../packages/tasks/tasks), [`timeout`](../packages/util/timeout) |
 | [`session-telemetry-otel`](../packages/telemetry/session-telemetry-otel) | `telemetry` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-telemetry`](../packages/telemetry/session-telemetry) |
@@ -1091,7 +1104,7 @@ flowchart TD
 | [`spill-policy`](../packages/spill/spill-policy) | `spill` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`retention`](../packages/util/retention), [`session`](../packages/core/session), [`spill`](../packages/spill/spill), [`tools`](../packages/core/tools) |
 | [`timeout-policy`](../packages/timeout/timeout-policy) | `timeout` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`timeout`](../packages/util/timeout), [`tools`](../packages/core/tools) |
 | [`tool-todo`](../packages/todo/tool-todo) | `todo` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection), [`tools`](../packages/core/tools) |
-| [`plan-mode`](../packages/plan/plan-mode) | `plan` | [`agent`](../packages/core/agent), [`commands`](../packages/ui/commands), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) |
+| [`plan-mode`](../packages/plan/plan-mode) | `plan` | [`agent`](../packages/core/agent), [`commands`](../packages/ui/commands), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) |
 | [`tool-cordis`](../packages/cordis/tool-cordis) | `cordis` | [`invariants`](../packages/support/invariants), [`scope`](../packages/core/scope), [`tools`](../packages/core/tools) |
 | [`hooks-codex`](../packages/hooks/hooks-codex) | `hooks` | [`agent`](../packages/core/agent), [`hook-protocol`](../packages/hooks/hook-protocol), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`tools`](../packages/core/tools) |
 | [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy) | `session-persistence` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`tools`](../packages/core/tools) |
@@ -1108,13 +1121,13 @@ flowchart TD
 | [`tool-lsp`](../packages/lsp/tool-lsp) | `lsp` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`lsp`](../packages/lsp/lsp), [`system-prompt`](../packages/core/system-prompt), [`timeout`](../packages/util/timeout), [`tools`](../packages/core/tools) |
 | [`mcp-client`](../packages/mcp/mcp-client) | `mcp` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`subprocess`](../packages/subprocess/subprocess), [`tools`](../packages/core/tools) |
 | [`tool-pty`](../packages/pty/tool-pty) | `pty` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`pty`](../packages/pty/pty), [`retention`](../packages/util/retention), [`system-prompt`](../packages/core/system-prompt), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) |
-| [`tool-tasks`](../packages/tasks/tool-tasks) | `tasks` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`retention`](../packages/util/retention), [`system-prompt`](../packages/core/system-prompt), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) |
+| [`tool-tasks`](../packages/tasks/tool-tasks) | `tasks` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`retention`](../packages/util/retention), [`system-prompt`](../packages/core/system-prompt), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) |
 | [`tool-workflow`](../packages/workflow/tool-workflow) | `workflow` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`workflow`](../packages/workflow/workflow) |
 | [`subagent-acp`](../packages/subagent/subagent-acp) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent), [`subprocess`](../packages/subprocess/subprocess) |
 | [`subagent-inprocess`](../packages/subagent/subagent-inprocess) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`user-approval`](../packages/ui/user-approval) |
 | [`tool-subagent`](../packages/subagent/tool-subagent) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`subagent`](../packages/subagent/subagent), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) |
 | [`hooks-claude`](../packages/hooks/hooks-claude) | `hooks` | [`agent`](../packages/core/agent), [`hook-protocol`](../packages/hooks/hook-protocol), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`subagent`](../packages/subagent/subagent), [`tools`](../packages/core/tools) |
-| [`tui`](../packages/ui/tui) | `ui` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`commands`](../packages/ui/commands), [`goal`](../packages/goal/goal), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-query`](../packages/session-query/session-query), [`session-reference`](../packages/context/session-reference), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`system-prompt`](../packages/core/system-prompt), [`token-meter`](../packages/llm/token-meter), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) |
+| [`tui`](../packages/ui/tui) | `ui` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`commands`](../packages/ui/commands), [`goal`](../packages/goal/goal), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-query`](../packages/session-query/session-query), [`session-reference`](../packages/context/session-reference), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`subprocess`](../packages/subprocess/subprocess), [`system-prompt`](../packages/core/system-prompt), [`token-meter`](../packages/llm/token-meter), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) |
 | [`client-ui-plan`](../packages/client/ui-plan) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`plan-mode`](../packages/plan/plan-mode) |
 | [`agent-spine-demo`](../packages/examples/agent-spine-demo) | `examples` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`paths`](../packages/util/paths), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`skill-local`](../packages/skill/skill-local), [`system-prompt`](../packages/core/system-prompt), [`tasks-local`](../packages/tasks/tasks-local), [`tool-bash`](../packages/bash/tool-bash), [`tool-goal`](../packages/goal/tool-goal), [`tool-skill`](../packages/skill/tool-skill), [`tool-tasks`](../packages/tasks/tool-tasks), [`tools`](../packages/core/tools), [`workspace-context`](../packages/context/workspace-context) |
 | [`sdk-protocol`](../packages/sdk/sdk-protocol) | `sdk` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent) |
diff --git a/docs/persistence-catalog.md b/docs/persistence-catalog.md
index a1535f36dc..dea3c4816b 100644
--- a/docs/persistence-catalog.md
+++ b/docs/persistence-catalog.md
@@ -213,7 +213,7 @@ Source: [`packages/ui/commands/src/index.ts:132`](../packages/ui/commands/src/in
 'compact/end': { turn: number; error?: string }
 ```
 
-Source: [`packages/compact/compact/src/types.ts:40`](../packages/compact/compact/src/types.ts)
+Source: [`packages/compact/compact/src/types.ts:44`](../packages/compact/compact/src/types.ts)
 
 #### `compact/start` — log-only
 
@@ -235,6 +235,8 @@ Source: [`packages/compact/compact/src/types.ts:15`](../packages/compact/compact
  */
 'compact/summary': {
   summary: ContentBlock[]
+  /** Complete provider output before the backend's safe summary projection. */
+  rawOutput?: ContentBlock[]
   shadowedRange: { start: number; end: number }
   shadowedSeqs: number[]
   shadowedTokenCount: number
@@ -249,10 +251,12 @@ Source: [`packages/compact/compact/src/types.ts:15`](../packages/compact/compact
   model: string
   /** The generation cap the summarize call sent, when one applied. */
   maxTokens?: number
+  /** Provider-reported token usage for the summarization request, when emitted. */
+  usage?: TokenUsage
 }
 ```
 
-Types: [ContentBlock](core-data-structures/core.md)
+Types: [ContentBlock](core-data-structures/core.md) · [TokenUsage](core-data-structures/llm-streaming.md)
 
 Source: [`packages/compact/compact/src/types.ts:22`](../packages/compact/compact/src/types.ts)
 
@@ -412,7 +416,7 @@ Source: [`packages/sandbox/sandbox-policy/src/session-mode.ts:33`](../packages/s
 
 Types: [SessionTitleEventData](core-data-structures/session-title.md)
 
-Source: [`packages/session-title/session-title/src/index.ts:96`](../packages/session-title/session-title/src/index.ts)
+Source: [`packages/session-title/session-title/src/index.ts:100`](../packages/session-title/session-title/src/index.ts)
 
 #### `session/title-llm-request` — log-only
 
diff --git a/docs/postmortem/README.i18n.yaml b/docs/postmortem/README.i18n.yaml
index e68d3a1a07..58b6521577 100644
--- a/docs/postmortem/README.i18n.yaml
+++ b/docs/postmortem/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write docs/postmortem/README.md
 README.md: df0e2fcb8540aeed005153dbecc451d781ca5ff1
-README.zh.md: 2ce6de475c705b02cd9dabfb2181929d81478e2c
+README.zh.md: e364ef30e342f8484a40f35e3970dea0f2f86ef3
diff --git a/docs/postmortem/README.zh.md b/docs/postmortem/README.zh.md
index 2ce6de475c..e364ef30e3 100644
--- a/docs/postmortem/README.zh.md
+++ b/docs/postmortem/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-事故复盘：一个 bug 到达了它不该到达的地方（真实用户、已合并的 PR（Pull Request）、已发布的版本），值得关注的是*为什么我们的流程放过了它*，而不仅仅是那一行修复。
+事故复盘记录的是：一个 bug 流入了不该流入的环节（真实用户、已合并的 PR（Pull Request）、已发布的版本），值得关注的是*为什么我们的流程放过了它*，而不仅仅是那一行修复。
 
-事故复盘不是 [Agent Note（agent 决策记录）](../../.agents/notes/README.md)（Agent Note 记录一个经过深思熟虑的设计决策及其被否决的替代方案，或提出未来工作）。它是一份回顾性的失败记录：什么坏了、机制是什么、为什么每道安全网都没拦住、以及加了哪些具体的防护措施使同类 bug 下次能被显式暴露。
+事故复盘不是 [Agent Note（agent 决策记录）](../../.agents/notes/README.md)（Agent Note 记录一个经过深思熟虑的设计决策及其被否决的替代方案，或提出未来工作）。它是一份回顾性的失败记录：什么坏了、机制是什么、为什么每道安全网都没拦住、以及为此新增了哪些具体防护措施，以确保同类 bug 下次出现时会明确报错。
 
 当一个 bug 满足以下条件时，请撰写事故复盘：**隐蔽**（机制不显而易见，即使是细心的工程师也得费力重新推导）、**系统性**（逃逸的原因是测试/工具/约定的缺口，而非一次性的笔误）、**重新发现的代价高**（它消耗了真实的调试时间，且下次还会如此）。请链接该事故复盘所推动建立的防护措施（测试、AGENTS.md 规则、ADR）。
 
-每篇事故复盘以一段**摘要**开头：一个简短段落，让忙碌的读者在三十秒内吸收要点——什么坏了、用直白的话说根因是什么、为什么逃逸了、持久的教训是什么——然后才是后续的详细「概述 / 时间线 / 根因 / 防护措施」各节。
+每篇事故复盘以一段**摘要**开头：一个简短段落，让忙碌的读者在三十秒内吸收要点——什么坏了、用直白的话说根因是什么、为什么逃逸了、可长期沿用的教训是什么——然后才是后续的详细「概述 / 时间线 / 根因 / 防护措施」各节。
 
 | # | 标题 |
 |---|---|
diff --git a/examples/README.i18n.yaml b/examples/README.i18n.yaml
index f5b73d9b6d..647d829b4f 100644
--- a/examples/README.i18n.yaml
+++ b/examples/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write examples/README.md
 README.md: 7f12178d1b67f1ebfac6f4f0e31403c54106e98f
-README.zh.md: 72ab92602d0a53cabdbfa8bc34838061df25d1c7
+README.zh.md: c7c1bf76593661616464558e554d57340d7c03b1
diff --git a/examples/README.zh.md b/examples/README.zh.md
index 72ab92602d..c7c1bf7659 100644
--- a/examples/README.zh.md
+++ b/examples/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-展示 harness 如何接线的可运行演示（不是 workspace）。每个示例都是一个 **轻量叶节点**：一份选择可替换后端、加载一个应用包（package）并可添加可选产品工具的 `cordis.yml`。组合和启动粘合代码位于 [`@deepseek-ai/dsh-tui-demo`](../packages/examples/tui-demo)、[`@deepseek-ai/dsh-cli-demo`](../packages/examples/cli-demo)、[`@deepseek-ai/dsh-acp-demo`](../packages/examples/acp-demo) 及它们共享的 [`@deepseek-ai/dsh-agent-spine-demo`](../packages/examples/agent-spine-demo) 组合包中。没有 `start.ts`；终端 `demo:*` 脚本通过 [`dsh`](../apps/cli/README.md) CLI（命令行界面）启动（该 CLI 挂载 `tui-demo` 组合包），无头／ACP（Agent Client Protocol）脚本则调用 `cli-demo`/`acp-demo` bin。
+展示 harness 如何组装的可运行演示（不是 workspace）。每个示例都是一个 **轻量叶节点**：一份选择可替换后端、加载一个应用包（package）并可添加可选产品工具的 `cordis.yml`。组合和启动粘合代码位于 [`@deepseek-ai/dsh-tui-demo`](../packages/examples/tui-demo)、[`@deepseek-ai/dsh-cli-demo`](../packages/examples/cli-demo)、[`@deepseek-ai/dsh-acp-demo`](../packages/examples/acp-demo) 及它们共享的 [`@deepseek-ai/dsh-agent-spine-demo`](../packages/examples/agent-spine-demo) 组合包中。没有 `start.ts`；终端 `demo:*` 脚本通过 [`dsh`](../apps/cli/README.md) CLI（命令行界面）启动（该 CLI 挂载 `tui-demo` 组合包），无头／ACP（Agent Client Protocol）脚本则调用 `cli-demo`/`acp-demo` bin。
 
 ## headless-agent
 
-非交互式 agent（智能体）演示：接受一个位置任务，在 `@deepseek-ai/dsh-cli-demo` 应用上运行一个完整模型／工具轮次，持久化新会话，打印 `text`、`json` 或 `stream-json`，然后退出。
+非交互式 agent（智能体）演示：接受一个位置参数形式的任务，在 `@deepseek-ai/dsh-cli-demo` 应用上运行一个完整模型／工具轮次，持久化新会话，打印 `text`、`json` 或 `stream-json`，然后退出。
 
 运行：`pnpm run demo:headless "task"`（需要 `DEEPSEEK_API_KEY`）。输出契约、安全边界和快照套件详见 [headless-agent/README.md](headless-agent/README.md)。
 
@@ -18,18 +18,18 @@
 
 ## jsonrpc-agent
 
-通过 Python SDK 驱动的无人值守编码 agent：JSON-RPC stdio、仅前台 `bash`、`read`/`write`/`edit`、一个前台 `subagent`、`todo_write`、JSONL 持久化和压缩。它不包含终端 UI、stdout 日志、批准、skill 和后台任务控制。详见 [jsonrpc-agent/README.md](jsonrpc-agent/README.md)。
+通过 Python SDK 驱动的无人值守编码 agent：JSON-RPC stdio、仅前台 `bash`、`read`/`write`/`edit`、一个前台 `subagent`、`todo_write`、JSONL 持久化和压缩。它不包含终端 UI、stdout 日志、批准、skill（技能）和后台任务控制。详见 [jsonrpc-agent/README.md](jsonrpc-agent/README.md)。
 
 ## cordis-agent
 
-**自指** 演示：编码主干加 [`@deepseek-ai/dsh-tool-cordis`](../packages/cordis/tool-cordis)，其三个工具（`cordis_inspect`/`cordis_mount`/`cordis_unmount`）使 agent 可以检查当前 DSH 进程、挂载模型编写的临时 Plugin（事件监听器、一个全新工具，或一个供另一临时 Plugin 注入的服务），并再次卸载它们。这些 Plugin 只存在于内存中，共享一个内部 `cordis-dynamic` fiber 子树；`ctx.fs`/`ctx.web` 仅作为它们可用的能力提供方。
+**自指** 演示：编码主干加 [`@deepseek-ai/dsh-tool-cordis`](../packages/cordis/tool-cordis)，其三个工具（`cordis_inspect`/`cordis_mount`/`cordis_unmount`）使 agent 可以检查当前 DSH 进程、挂载模型编写的临时插件（事件监听器、一个全新工具，或一个供另一个临时插件注入的服务），并再次卸载它们。这些插件只存在于内存中，共享一个内部 `cordis-dynamic` fiber 子树；`ctx.fs`/`ctx.web` 仅作为它们可用的能力提供方。
 
-使用 `pnpm run demo:cordis` 运行 TUI，使用 `pnpm run demo:cordis web` 在 `http://127.0.0.1:3081` 启动浏览器 UI，或使用 `pnpm run demo:cordis acp` 启动 ACP 服务器（三者均需 `DEEPSEEK_API_KEY`）。分阶段演示脚本详见 [cordis-agent/README.md](cordis-agent/README.md)，设计与沙箱注意事项详见[工具集 Agent Note](../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
+使用 `pnpm run demo:cordis` 运行 TUI，使用 `pnpm run demo:cordis web` 在 `http://127.0.0.1:3081` 启动浏览器 UI，或使用 `pnpm run demo:cordis acp` 启动 ACP 服务器（三者均需 `DEEPSEEK_API_KEY`）。分阶段演示脚本详见 [cordis-agent/README.md](cordis-agent/README.md)，设计与沙箱注意事项详见[工具集 Agent Note（agent 决策记录）](../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
 
 ## acp-agent
 
-作为 **Agent Client Protocol (ACP)** 自动化服务器通过 JSON-RPC stdio 公开的 agent，由 [`@deepseek-ai/dsh-acp-demo`](../packages/examples/acp-demo) 提供。程序化客户端可以创建新会话、发送文本提示词、消费已提交的 assistant 文本、回答一次性权限请求并取消工作。它拥有 ACP 无密钥快照套件。
+一个通过 JSON-RPC stdio 公开、作为 **Agent Client Protocol (ACP)** 自动化服务器运行的 agent，由 [`@deepseek-ai/dsh-acp-demo`](../packages/examples/acp-demo) 提供。程序化客户端可以创建新会话、发送文本提示词、消费已提交的 assistant 文本、回答一次性权限请求并取消工作。它拥有 ACP 无密钥快照套件。
 
 运行：`pnpm run demo:acp`（需要 `DEEPSEEK_API_KEY`）；`pnpm run demo:code-mode acp` 通过 `code-mode.cordis.yml` 覆盖以 Code Mode 启动同一服务器。协议与快照测试契约详见 [acp-agent/README.md](acp-agent/README.md)。
 
-默认 `cordis.yml` 组合 [`@deepseek-ai/dsh-sandbox-local`](../packages/sandbox/sandbox-local)、[`@deepseek-ai/dsh-bash-sandbox`](../packages/bash/bash-sandbox) 和 [`@deepseek-ai/dsh-user-approval`](../packages/ui/user-approval)。`workspace-write` 将 bash 和文件系统变更限制在每个会话 workspace 中；范围更广的重试会通过 ACP 成为一次性机器权限请求。
+默认 `cordis.yml` 组合 [`@deepseek-ai/dsh-sandbox-local`](../packages/sandbox/sandbox-local)、[`@deepseek-ai/dsh-bash-sandbox`](../packages/bash/bash-sandbox) 和 [`@deepseek-ai/dsh-user-approval`](../packages/ui/user-approval)。`workspace-write` 将 bash 和文件系统变更限制在每个会话 workspace 中；请求更广泛沙箱权限的重试会通过 ACP 触发一次性的机器权限请求。
diff --git a/examples/acp-agent/README.i18n.yaml b/examples/acp-agent/README.i18n.yaml
index 391967d91c..3b208eb9b5 100644
--- a/examples/acp-agent/README.i18n.yaml
+++ b/examples/acp-agent/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write examples/acp-agent/README.md
 README.md: 0d63ec1f2d9165b9faf0817bd94fbe15b97fa961
-README.zh.md: 0c5f8866ea640843513fd9a4c15a17ed4db59d3b
+README.zh.md: 84482aad8352ab38527dcf4d9e1bfefc8d496c91
diff --git a/examples/acp-agent/README.zh.md b/examples/acp-agent/README.zh.md
index 0c5f8866ea..84482aad83 100644
--- a/examples/acp-agent/README.zh.md
+++ b/examples/acp-agent/README.zh.md
@@ -2,29 +2,29 @@
 
 [English](README.md) | 中文
 
-通过 JSON-RPC stdio 提供的自动化导向 [Agent Client Protocol](https://agentclientprotocol.com) 服务器。它面向父 agent（智能体）、subagent 提供方和其他程序化客户端，而非产品 UI。
+通过 JSON-RPC stdio 提供的面向自动化的 [Agent Client Protocol（ACP）](https://agentclientprotocol.com) 服务器。它面向 parent agent（父智能体）、subagent 提供方和其他程序化客户端，而非产品 UI。
 
 ```sh
 pnpm run demo:acp             # needs DEEPSEEK_API_KEY (repo-root .env or env)
 pnpm run demo:code-mode acp   # same protocol with the Code Mode tool transport
 ```
 
-该叶节点加载 ACP 应用、DeepSeek 适配器、受沙箱限制的 bash 与文件系统栈、一次性批准策略、压缩（compaction）、subagent、工作流、钩子、派生会话查询索引和重复守卫。应用为每次 `session/new` 创建一个新 agent，将会话持久化到 JSONL，并保持 stdout 只含协议内容。[`session-query.cordis.yml`](session-query.cordis.yml) 为其专用快照显式选用 workspace 授权的查询工具和通用超时／溢出策略；[`fs.cordis.yml`](fs.cordis.yml) 为文件系统场景添加溢出存储，[`code-mode.cordis.yml`](code-mode.cordis.yml) 添加 `run_code` 及其生成的 TypeScript SDK，[`web.cordis.yml`](web.cordis.yml) 则为 web-fetch 快照添加 web seam、本地抓取提供方、`web_fetch` 与一个回环 HTML fixture 服务器。
+该叶节点加载 ACP 应用、DeepSeek 适配器、受沙箱限制的 bash 与文件系统栈、一次性批准策略、压缩（compaction）、subagent、工作流、钩子、派生会话查询索引和重复守卫。应用为每次 `session/new` 创建一个新 agent，将会话持久化到 JSONL，并保持 stdout 只含协议内容。[`session-query.cordis.yml`](session-query.cordis.yml) 为其专用快照显式选用 workspace 授权的查询工具和通用超时／溢出策略；[`fs.cordis.yml`](fs.cordis.yml) 为文件系统场景添加溢出存储，[`code-mode.cordis.yml`](code-mode.cordis.yml) 添加 `run_code` 及其生成的 TypeScript SDK，[`web.cordis.yml`](web.cordis.yml) 则为 web-fetch 快照添加 web seam、本地抓取提供方、`web_fetch` 与一个回环 HTML fixture（测试前置数据）服务器。
 
 ## 协议通道
 
-Stdout 只携带以换行分隔的 ACP JSON-RPC。`@deepseek-ai/dsh-acp-demo` 不安装 stdout logger；叶节点的附加项必须使用 stderr 输出诊断信息。
+Stdout 只携带以换行分隔的 ACP JSON-RPC。`@deepseek-ai/dsh-acp-demo` 不安装 stdout logger；该叶节点新增的组件必须使用 stderr 输出诊断信息。
 
 自动化契约（支持的方法、基线提示词内容、已提交文本输出，以及有意缺少的 UI 界面）位于 [`@deepseek-ai/dsh-acp`](../../packages/acp/acp/README.md)。
 
 ## 会话 workspace 与权限
 
-每次 `session/new` 都提供一个绝对 `cwd`。受沙箱限制的 bash 与文件系统变更会根据该会话 cwd 解析 `workspace-write`，因此并发会话可以使用不同的项目根目录；平台临时根目录仍是共享可写暂存空间（参见[沙箱契约](../../packages/sandbox/sandbox/README.md)）。`DSH_PERMISSION_MODE` 在部署和测试中选择 `workspace-write` 或 `danger-full-access`。
+每次 `session/new` 都提供一个绝对 `cwd`。受沙箱限制的 bash 和文件系统修改会以该会话 cwd 为基准应用 `workspace-write`，因此并发会话可以使用不同的项目根目录；平台临时根目录仍是共享可写暂存空间（参见[沙箱契约](../../packages/sandbox/sandbox/README.md)）。`DSH_PERMISSION_MODE` 在部署和测试中选择 `workspace-write` 或 `danger-full-access`。
 
-在 `workspace-write` 下，模型请求扩大沙箱权限的重试会触发 `session/request_permission`，选项为 `allow_once` 和 `reject_once`。客户端以程序方式决策；解除对话框或答案不可用时会失败闭合。选定结果仅适用于该次重试，并通过常规工具结果／审计路径记录。服务器绝不公开权限选择器，也不持久化客户端策略。
+在 `workspace-write` 下，如果模型重试请求更广泛的沙箱访问权限，就会触发 `session/request_permission`，选项为 `allow_once` 和 `reject_once`。客户端以程序方式决策；客户端放弃选择或无法给出答复时，系统会按拒绝处理。选定结果仅适用于该次重试，并通过常规工具结果／审计路径记录。服务器绝不公开权限选择器，也不持久化客户端策略。
 
 ## 快照测试
 
-此示例拥有 ACP 快照套件。它会启动真实自动化服务器，通过 `dsh-llm-replay` 回放已提交的模型流，并比较规范化后的协议输出与重新持久化的会话日志。录制使用真实模型；刷新会复用已提交的回放输入。覆盖场景包括抛出／挂起行为，可选 `workspace/` fixture（测试前置数据）则为外部状态检查预置环境。
+此示例拥有 ACP 快照套件。它会启动真实自动化服务器，通过 `dsh-llm-replay` 回放已提交的模型流，并比较规范化后的协议输出与重新持久化的会话日志。录制使用真实模型；刷新会复用已提交的回放输入。覆盖配置涵盖抛错／挂起行为，可选的 `workspace/` fixture 则为环境状态检查预置状态。
 
 大多数场景锁定后端行为，而非 ACP 专用行为；[仅面向自动化的 ACP 决策](../../.agents/notes/implemented/simplification/2026-07-23-acp-automation-only-protocol.md#snapshot-boundary)说明了为何该覆盖仍与传输层耦合。
diff --git a/examples/cordis-agent/README.i18n.yaml b/examples/cordis-agent/README.i18n.yaml
index f9a03c46fa..4cbfbb67cb 100644
--- a/examples/cordis-agent/README.i18n.yaml
+++ b/examples/cordis-agent/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write examples/cordis-agent/README.md
 README.md: 55970e932bc16d8361932daa9ea55af83ef73d33
-README.zh.md: c2873b6de96a8b47ad8ea4fb2cf03a7501406300
+README.zh.md: a8ec332d8b3673d6656663eb3bfd7d37a4e328f6
diff --git a/examples/cordis-agent/README.zh.md b/examples/cordis-agent/README.zh.md
index c2873b6de9..a8ec332d8b 100644
--- a/examples/cordis-agent/README.zh.md
+++ b/examples/cordis-agent/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-自指 harness 演示：在全屏 TUI 上运行 DeepSeek V4 编码主干，并加载 [`@deepseek-ai/dsh-tool-cordis`](../../packages/cordis/tool-cordis/README.md)。后者让模型检查当前 DSH 进程、挂载仅存于内存的临时 Plugin，并再次卸载它们。临时 Plugin 可跨 turn 保持活跃，但会在卸载、工具集卸载或 DSH 重启后消失；它们不创建文件或配置，也可能影响同一进程中的其他 session。`ctx.fs` 和 `ctx.web` 是这些 Plugin 可用的 provider-only 能力。设计详见[工具集 Agent Note](../../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
+自指 harness 演示：在全屏 TUI 上运行 DeepSeek V4 编码主干，并加载 [`@deepseek-ai/dsh-tool-cordis`](../../packages/cordis/tool-cordis/README.md)。后者让模型检查当前 DSH 进程、挂载仅存于内存的临时插件，并卸载它们。临时插件可跨轮次保持活跃，但会在卸载、工具集卸载或 DSH 重启后消失；它们不创建文件或配置，也可能影响同一进程中的其他会话。`ctx.fs` 和 `ctx.web` 仅以能力提供方形式加载，供这些插件使用。设计详见[工具集 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
 
 ## 运行
 
@@ -15,7 +15,7 @@ pnpm run demo:cordis web  # browser UI at http://127.0.0.1:3081
 pnpm run demo:cordis acp  # ACP server
 ```
 
-预期演示分阶段进行：先验证监听器链接，再让 agent 扩展自身：
+预期演示分阶段进行：先验证监听器链路，再让 agent（智能体）扩展自身：
 
 ```
 > Mount a temporary Plugin that listens to the 'agent/status' event and logs every status change, then run `echo hi` with bash.
@@ -30,8 +30,8 @@ pnpm run demo:cordis acp  # ACP server
   [tool call] cordis_unmount({"id": "dyn-1"})
 ```
 
-请求 `cordis_inspect` 并使用 `what: "api"` 或 `what: "events"`，即可查看编写 Plugin 代码所用的生成服务／事件资料。还可挂载两个协作临时 Plugin（一个中调用 `ctx.provide`，另一个中使用 `inject`），观察 Cordis 如何暂停并恢复消费方。
+请求 `cordis_inspect` 并使用 `what: "api"` 或 `what: "events"`，即可查看编写插件代码所用的生成服务／事件资料。还可挂载两个协作临时插件（一个中调用 `ctx.provide`，另一个中使用 `inject`），观察 Cordis 如何暂停并恢复消费方。
 
 ## 端到端测试
 
-`tests/keyless-smoke.e2e.ts` 使用虚拟密钥通过 Loader 启动真实 `cordis.yml`，并断言横幅、包名解析和 EOF 后干净退出。`tests/cordis-tools.e2e.ts` 是带密钥的冒烟测试：真实模型挂载一个临时状态 listener，测试验证其带标记的 console 行；然后创建并使用 `reverse_text` 工具，再通过 provide/inject 组合两个临时 Plugin。[`packages/cordis/tool-cordis`](../../packages/cordis/tool-cordis) 在每文件 100% 覆盖率门禁下承载单元覆盖。
+`tests/keyless-smoke.e2e.ts` 使用虚拟密钥通过 Loader 启动真实 `cordis.yml`，并断言横幅、包名解析，以及收到 EOF 后正常退出。`tests/cordis-tools.e2e.ts` 是带密钥的冒烟测试：真实模型挂载一个临时状态监听器，测试验证其带标记的控制台输出行；然后创建并使用 `reverse_text` 工具，再通过 provide/inject 组合两个临时插件。[`packages/cordis/tool-cordis`](../../packages/cordis/tool-cordis) 包含相关单元测试，并受逐文件 100% 覆盖率门禁约束。
diff --git a/examples/headless-agent/README.i18n.yaml b/examples/headless-agent/README.i18n.yaml
index 7c4dbaf463..c4cfb23631 100644
--- a/examples/headless-agent/README.i18n.yaml
+++ b/examples/headless-agent/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write examples/headless-agent/README.md
 README.md: 445804a2611e5e8093eadf345ad10a2a7984c012
-README.zh.md: 68ec718afe0b2aca276be2689cbae74167ee1c7b
+README.zh.md: 956bc82e77f79c3f05e4b51297fd5365e6e89be1
diff --git a/examples/headless-agent/README.zh.md b/examples/headless-agent/README.zh.md
index 68ec718afe..956bc82e77 100644
--- a/examples/headless-agent/README.zh.md
+++ b/examples/headless-agent/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-无头单次 agent（智能体）接线：DeepSeek V4 + 本地 bash 与文件系统工具 + subagent 委托 + 工作流与新 agent Ralph 迭代 + `todo_write` + JSONL 持久化，并以 [`@deepseek-ai/dsh-cli-demo`](../../packages/examples/cli-demo) 作为应用入口。
+无头单次 agent（智能体）接线：DeepSeek V4 + 本地 bash 与文件系统工具 + subagent 委托 + 工作流与全新 agent Ralph 迭代 + `todo_write` + JSONL 持久化，并以 [`@deepseek-ai/dsh-cli-demo`](../../packages/examples/cli-demo) 作为应用入口。
 
 ## 运行
 
@@ -15,12 +15,12 @@ pnpm run demo:headless --output-format json -- "summarize the implementation"
 pnpm run demo:headless --output-format stream-json -- "run the focused tests"
 ```
 
-必须提供且只能提供一个非空位置任务；含空格的任务需要加引号。没有 `-p` 标志。`text` 打印最后一条包含文本的 assistant 消息，`json` 打印一条 DSH 原生结果记录，`stream-json` 则在该记录之前发出顶层会话的规范任务轮次事件。子会话只通过父工具事件和结果对外显示。
+必须提供一个且仅一个非空的任务位置参数；含空格的任务需要加引号。没有 `-p` 标志。`text` 打印最后一条包含文本的 assistant 消息，`json` 打印一条 DSH 原生结果记录，`stream-json` 则在该记录之前发出顶层会话的规范任务轮次事件。子会话只通过父会话的工具事件和结果对外显示。
 
-每次调用都会创建并持久化新会话，在一个轮次中运行所有模型和工具步骤，然后刷新、释放并退出。这是非交互式自动化：没有提示符、批准、恢复、第二轮次或 stdin 上下文。已配置工具可以修改启动 workspace、运行命令、spawn 子 agent，并消耗提供方 token。
+每次调用都会创建并持久化新会话，在一个轮次中运行所有模型和工具步骤，然后刷写持久化数据、执行 dispose（资源释放），再退出。这是非交互式自动化：没有提示符、批准、恢复、第二轮次或 stdin 上下文。已配置工具可以修改启动时所在的工作区、运行命令、spawn 子 agent，并消耗提供方 token。
 
 ## 高级与快照接线
 
-[`advanced.cordis.yml`](advanced.cordis.yml) 在已交付叶节点上添加 Code Mode 和 Cordis 工具。[`advanced.cordis.snapshot.yml`](advanced.cordis.snapshot.yml) 只将实时 LLM（大语言模型）替换为回放。[`tests/`](tests/) 下的测试拥有无密钥真实 Loader 冒烟测试、密钥门控的外部状态验证冒烟测试，以及带父子会话 fixture（测试前置数据）的 `stream-json` 回放快照。
+[`advanced.cordis.yml`](advanced.cordis.yml) 在已交付叶节点上添加 Code Mode 和 Cordis 工具。[`advanced.cordis.snapshot.yml`](advanced.cordis.snapshot.yml) 只将实时 LLM（大语言模型）替换为回放。[`tests/`](tests/) 下涵盖无密钥真实 Loader 冒烟测试、密钥门控的外部状态验证冒烟测试，以及带父子会话 fixture（测试前置数据）的 `stream-json` 回放快照。
 
-包级 [CLI 契约](../../packages/examples/cli-demo/README.md)记录输出记录、退出状态、取消、持久化以及模型／token 影响。
+这份包（package）级 [CLI（命令行界面）契约](../../packages/examples/cli-demo/README.md) 说明输出记录、退出状态、取消、持久化以及模型／token 影响。
diff --git a/examples/jsonrpc-agent/README.i18n.yaml b/examples/jsonrpc-agent/README.i18n.yaml
index 59c18e0131..a64415bf2a 100644
--- a/examples/jsonrpc-agent/README.i18n.yaml
+++ b/examples/jsonrpc-agent/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write examples/jsonrpc-agent/README.md
 README.md: 6ee4e9d824315bde76b7a534679f018df9a6d3e8
-README.zh.md: dc9b6233e7074e7a9b13bf10bcd2f310b0ad7bf3
+README.zh.md: 43290fc3659750724679a370be5b33ffdca2a5cb
diff --git a/examples/jsonrpc-agent/README.zh.md b/examples/jsonrpc-agent/README.zh.md
index dc9b6233e7..43290fc365 100644
--- a/examples/jsonrpc-agent/README.zh.md
+++ b/examples/jsonrpc-agent/README.zh.md
@@ -2,16 +2,16 @@
 
 [English](README.md) | 中文
 
-面向 Python SDK 内置 JSON-RPC 运行时的无人值守编码 agent（智能体）组合。它有意不加载终端 UI、console logger、批准界面或用户交互工具，因为 stdout 属于 SDK 协议，轮次由 SDK 驱动。
+面向 Python SDK 内置 JSON-RPC 运行时的无人值守编码 agent（智能体）组合。它有意不加载终端 UI、控制台日志记录器、批准界面或用户交互工具，因为 stdout 属于 SDK 协议，轮次由 SDK 驱动。
 
 面向模型的工具为：
 
 - `bash`，仅前台
 - `read`、`write` 和 `edit`
-- `subagent`，使用一个前台进程内 spawn 提供方
+- `subagent`，使用一个在进程内以前台方式运行的 spawn 提供方
 - `todo_write`
 
-周边运行时还加载 JSONL 会话持久化和自动上下文压缩（compaction）。`maxTokensAsSuccess` 将受 token 上限限制的模型轮次保留为已接受的评估结果，同时保留其 `max-tokens` 原因。
+周边运行时还加载 JSONL 会话持久化和自动上下文压缩（context compaction）。`maxTokensAsSuccess` 将受 token 上限限制的模型轮次保留为已接受的评估结果，同时保留其 `max-tokens` 原因。
 
 ## 运行时环境
 
@@ -24,4 +24,4 @@
 | `DSH_SESSION_ROOT` | JSONL 轨迹目录 |
 | `DSH_SYSTEM_PROMPT` | 由部署提供的编码人格 |
 
-通过 Python SDK 的 `cordis` 选项或 `DSH_CORDIS_CONFIG` 传入配置路径。内置可执行文件已携带此文件命名的每个插件；目标机器无需 Node.js。
+通过 Python SDK 的 `cordis` 选项或 `DSH_CORDIS_CONFIG` 传入配置路径。内置可执行文件已携带此文件中指定的每个插件；目标机器无需 Node.js。
diff --git a/examples/tui-agent/README.i18n.yaml b/examples/tui-agent/README.i18n.yaml
index 3d29530411..863092ffb8 100644
--- a/examples/tui-agent/README.i18n.yaml
+++ b/examples/tui-agent/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write examples/tui-agent/README.md
 README.md: ea8695d37ea247a38644392a4572c1ea9855fd44
-README.zh.md: b3f6dc18536b159379eac7433367ccf2cd8fcc53
+README.zh.md: c6acd39d8713816d870c00fa8597754d0d09880a
diff --git a/examples/tui-agent/README.zh.md b/examples/tui-agent/README.zh.md
index b3f6dc1853..c6acd39d87 100644
--- a/examples/tui-agent/README.zh.md
+++ b/examples/tui-agent/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-全屏交互式编码 agent（智能体）：DeepSeek V4、本地 bash 与文件系统工具、压缩（compaction）、subagent、工作流与新 agent Ralph 迭代、plan mode（`/plan` 进入，`exit_plan_mode` 评审退出）、超时／溢出策略，以及通过 [`@deepseek-ai/dsh-tui-demo`](../../packages/examples/tui-demo) 提供的 JSONL 持久化；该应用从 `cordis.yml` 加载。同级 [`headless-agent`](../headless-agent/README.md) 以适合单次管道的任务形式运行同一能力类，[`acp-agent`](../acp-agent/README.md) 则通过 JSON-RPC 提供该能力。
+全屏交互式编码 agent（智能体）：DeepSeek V4、本地 bash 与文件系统工具、压缩（compaction）、subagent、工作流与全新 agent Ralph 迭代、plan mode（`/plan` 进入，`exit_plan_mode` 评审退出）、超时／溢出策略，以及通过 [`@deepseek-ai/dsh-tui-demo`](../../packages/examples/tui-demo) 提供的 JSONL 持久化；该应用从 `cordis.yml` 加载。同级 [`headless-agent`](../headless-agent/README.md) 以适合管道调用的单次任务形式运行同一能力类，[`acp-agent`](../acp-agent/README.md) 则通过 JSON-RPC 提供该能力。
 
 ## 运行
 
@@ -13,13 +13,13 @@
 pnpm run demo:tui
 ```
 
-演示脚本和可安装的 `dsh` CLI（[`apps/cli`](../../apps/cli/README.md)）都会作为已交付的默认配置启动此示例的 `cordis.yml`；`dsh` 还会应用 `~/.dsh` 中的个人覆盖，并将调用目录作为 workspace。
+演示脚本和可安装的 `dsh` CLI（命令行界面，见 [`apps/cli`](../../apps/cli/README.md)）都会以此示例的 `cordis.yml` 作为已交付的默认配置启动；`dsh` 还会应用 `~/.dsh` 中的个人覆盖，并将调用目录作为工作区。
 
-输入一项编码任务。agent 使用 `read`/`write`/`edit` 文件系统工具处理常规文件操作，使用 `bash`（加上面向后台任务的通用 `task_output`/`task_list`/`task_kill`）执行 shell 命令、搜索和测试。每次操作都在新的 `bash -c` 中运行（系统提示词要求模型传递 `workdir`，而不是使用 `cd`）。fs 工具和 bash 都会根据会话 workspace 解析相对路径。agent 还可以通过 `subagent`/`subagent_fork` 委托。
+输入一项编码任务。agent 使用 `read`/`write`/`edit` 文件系统工具处理常规文件操作，使用 `bash`（加上面向后台任务的通用 `task_output`/`task_list`/`task_kill`）执行 shell 命令、搜索和测试。每次 bash 调用都在新的 `bash -c` 中运行（系统提示词要求模型传递 `workdir`，而不是使用 `cd`）。文件系统工具和 bash 都会相对于会话工作区解析相对路径。agent 还可以通过 `subagent`/`subagent_fork` 委托。
 
 `todo_write` 任务跟踪器是选用的，不在已交付配置中：请将 `@deepseek-ai/dsh-tool-todo` 添加到 `cordis.yml`（或在 `~/.dsh` 下使用个人配置覆盖）以公开该工具。加载后，模型会把整表计划记录到会话日志，TUI 则渲染它。
 
-TUI 渲染 Markdown 历史、推理、工具所有的终端／diff／通用卡片、token 总量，以及加载 `todo_write` 时的最新计划。较长的工具正文保留首尾预览；Ctrl+O 展开或折叠所有卡片。Enter 用于提交，或在 agent 运行时进行 steering（中途引导）；Ctrl+R 切换推理，Escape 取消，`/help` 列出命令。`/plan` 为下一步骤选择 plan mode；`/plan <message>` 还会将消息提交到该步骤，`/plan off` 则在没有模型输入的情况下选择默认 mode。`/status` 会展开当前会话的标识、活动计数、精确 token／缓存 bucket、上下文用量和时间戳，而不中断正在运行的轮次。`/model` 打开当前提供方目录的键盘选择器；使用 Up/Down 聚焦模型，使用 Shift+Tab 循环切换为该模型公布的推理强度，再用 Enter 选择；也可以使用 `/model <model>` 和 `/model <provider>/<model>` 直接选择。`ask_user_question` 会打开一个位于左下方的宽键盘面板，包含批次进度和编号选项。
+TUI 渲染 Markdown 历史、推理（reasoning）、工具自有的终端／diff／通用卡片、token 总量，以及加载 `todo_write` 时的最新计划。较长的工具正文保留首尾预览；Ctrl+O 展开或折叠所有卡片。Enter 用于提交，或在 agent 运行时进行 steering（中途引导）；Ctrl+R 切换推理，Escape 取消，`/help` 列出命令。`/plan` 为下一步骤选择 plan mode；`/plan <message>` 还会将消息提交到该步骤，`/plan off` 则在没有模型输入的情况下选择默认 mode。`/status` 会展开当前会话的标识、活动计数、精确 token／缓存 bucket、上下文用量和时间戳，而不中断正在运行的轮次。`/model` 打开当前提供方目录的键盘选择器；使用 Up/Down 聚焦模型，使用 Shift+Tab 循环切换为该模型公布的推理强度，再用 Enter 选择；也可以使用 `/model <model>` 和 `/model <provider>/<model>` 直接选择。`ask_user_question` 会打开一个位于左下方的宽键盘面板，包含批次进度和编号选项。
 
 ### 恢复早先的会话
 
@@ -29,11 +29,11 @@ TUI 渲染 Markdown 历史、推理、工具所有的终端／diff／通用卡
 dsh --resume <prior-session-id>
 ```
 
-`/resume` 打开可搜索键盘选择器，显示标题、活动、上一轮结果、模型路由、持久 goal 阶段和实时／已持久化状态。已安装的 `dsh` 宿主会刷新并释放当前应用，然后以 `dsh --resume <id>` 替换进程。TUI 仍会在退出时打印该命令，并在自定义宿主无法移交时显示它。`dsh --resume <id>` 在启动上下文中提供 id，`cordis.yml` 会读取它（`resumeSessionId: !!js "typeof resumeSessionId === 'string' ? resumeSessionId : undefined"`）；没有标志时，agent 会开始新会话。缺失或无法读取的 id 不会启动 agent，而会发出 `agent-loop/config-start-failed`：TUI 打印失败并以非零状态退出。选择器没有跨进程会话锁，因此拥有并发宿主的部署必须自行协调会话所有权。
+`/resume` 打开可搜索键盘选择器，显示标题、活动、上一轮结果、模型路由、持久化目标阶段和实时／已持久化状态。已安装的 `dsh` 宿主会等待刷写完成，对当前应用执行 dispose（资源释放），然后以 `dsh --resume <id>` 替换进程。TUI 仍会在退出时打印该命令，并在自定义宿主无法移交时显示它。`dsh --resume <id>` 在启动上下文中提供 id，`cordis.yml` 会读取它（`resumeSessionId: !!js "typeof resumeSessionId === 'string' ? resumeSessionId : undefined"`）；没有标志时，agent 会开始新会话。缺失或无法读取的 id 不会启动 agent，而会发出 `agent-loop/config-start-failed`：TUI 打印失败并以非零状态退出。选择器没有跨进程会话锁，因此拥有并发宿主的部署必须自行协调会话所有权。
 
 ## Code Mode
 
-[`code-mode.cordis.yml`](code-mode.cordis.yml) 在同一树上覆盖 worker 线程运行时和 `tools: { mode: code }`。模型会收到一个 `run_code` 传输工具，加上一份为可见工具生成的 TypeScript SDK；只有程序输出会返回模型上下文。使用 `mode: both` 可在 `run_code` 旁同时公开原生调用。执行契约详见 [Code Mode Agent Note](../../.agents/notes/implemented/feature/2026-06-15-code-mode.md)。
+[`code-mode.cordis.yml`](code-mode.cordis.yml) 在同一树上覆盖 worker 线程运行时和 `tools: { mode: code }`。模型会收到一个 `run_code` 传输工具，加上一份为可见工具生成的 TypeScript SDK；只有程序输出会返回模型上下文。使用 `mode: both` 可在 `run_code` 旁同时公开原生调用。执行契约详见 [Code Mode Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-06-15-code-mode.md)。
 
 ```sh
 pnpm run demo:code-mode        # this overlay under the TUI (default UI)
@@ -54,27 +54,27 @@ pnpm run demo:code-mode acp    # the acp-agent example's same-shaped overlay
 |---|---|
 | `hmr` (`@cordisjs/plugin-hmr`) | 开发／演示的编辑-重载循环：它是 **叶节点** 配置项（不内置到应用），因为它依赖 Loader 的内部模块访问 |
 | `llm-deepseek` | 默认原生适配器 |
-| `bash` (`dsh-bash-local`) | 执行器实现：bash seam 的可替换一半。面向模型的 `bash` schema（`tool-bash`）和通用 `task_*` 控制（`tool-tasks`）由 `dsh-agent-spine-demo` 提供，因此叶节点只选择执行器 |
+| `bash` (`dsh-bash-local`) | 执行器实现：bash seam 中可替换的实现侧。面向模型的 `bash` schema（`tool-bash`）和通用 `task_*` 控制（`tool-tasks`）由 `dsh-agent-spine-demo` 提供，因此叶节点只选择执行器 |
 | `tui-agent` (`@deepseek-ai/dsh-tui-demo`) | 应用组合包：agent-spine 演示 + JSONL 持久化 + pi-tui 通道 + 预创建的 `main` agent |
 | `subagent`, `subagent-spawn`, `subagent-fork` | subagent 提供方注册表加两个进程内后端：新子 agent，以及用父 agent 已完成轮次前缀播种的子 agent |
 | `tool-subagent`, `tool-subagent-fork` | 两次面向模型的 `dsh-tool-subagent` 加载，每次绑定不同提供方，并以不同工具名（`subagent`、`subagent_fork`）公开 |
 | `workflow-workerthread`, `tool-workflow` | worker 线程工作流引擎及其面向模型的 `workflow` 工具，子调用通过 spawn 后端路由 |
 | `plan-mode` | 插件拥有的 `/plan [message]` 进入命令和 `/plan off` 退出命令、plan-mode 提示词策略、工具限制，以及经评审的 `exit_plan_mode` 转换 |
-| `fs-local`, `fs-policy`, `tool-fs` | 文件系统栈：本地 `ctx.fs` 提供方、先读后写／编辑策略门禁（位于 `fs/*` 事件门禁），以及面向模型的 `read`/`write`/`edit` 工具。相对路径根据会话 workspace 解析 |
+| `fs-local`, `fs-policy`, `tool-fs` | 文件系统栈：本地 `ctx.fs` 提供方、先读后写／编辑策略门禁（位于 `fs/*` 事件门禁），以及面向模型的 `read`/`write`/`edit` 工具。相对路径相对于会话工作区解析 |
 
 ## 端到端测试（`pnpm run test:e2e`）
 
 与 UI 无关的带密钥套件通过 `tests/harness.ts` 以程序方式组装完整栈（无 PTY、无 Loader）：
 
 - `tests/full-loop.e2e.ts`：canary 测试：真实模型通过真实 bash 工具运行 `echo e2e-ok`；断言 `tool/call`/`tool/result` 会话事件和最终答案。
-- `tests/coding-task.e2e.ts`：类 swebench 冒烟测试：临时目录包含 `add.js`（其中 `a - b` 写在本应是 `a + b` 的位置）和失败的 `add.test.js`；agent 必须修复错误并验证。测试会自行重新运行 `node add.test.js` 并检查文件，不信任 agent 的声称。
-- `tests/resume.e2e.ts`：跨进程持久连续性：第一次运行告诉真实模型一个密码并将轮次持久化到临时 JSONL 根目录，然后释放整个上下文；第二次运行在同一根目录上创建新上下文，恢复会话 id 并要求模型回忆密码。只有重新水化的日志能够提供该回忆。
-- `tests/compaction.e2e.ts`：压缩冒烟测试：一项真实多步 bash 任务在故意设得很小的上下文窗口中运行，使自动压缩监听器在会话中途触发。测试验证外部状态：真实日志中出现 `compact/start…end` 对，表层缩减（替换节点遮蔽旧节点），且 agent 在压缩后仍给出正确最终答案。
-- `tests/todo-write.e2e.ts`：加载选用 `todo_write` 工具，由真实模型驱动，测试验证产生的 `todo/write` 会话事件。
-- `tests/code-mode.e2e.ts`：带密钥 Code Mode 证明：使用真实模型和双工具任务，断言线上工具列表精确为 `[run_code]`，`tool/code-dispatch` 事件位于父调用下，且筛选后的答案已返回。
+- `tests/coding-task.e2e.ts`：类 swebench 冒烟测试：临时目录包含 `add.js`（其中 `a - b` 写在本应是 `a + b` 的位置）和失败的 `add.test.js`；agent 必须修复错误并验证。测试会自行重新运行 `node add.test.js` 并检查文件，不信任 agent 的说法。
+- `tests/resume.e2e.ts`：跨进程持久连续性：第一次运行告诉真实模型一个密码并将轮次持久化到临时 JSONL 根目录，然后 dispose 整个上下文；第二次运行在同一根目录上创建新上下文，恢复会话 id 并要求模型回忆密码。只有重新水化的日志能够提供该回忆。
+- `tests/compaction.e2e.ts`：压缩冒烟测试：一项真实多步 bash 任务在故意设得很小的上下文窗口中运行，使自动压缩监听器在会话中途触发。测试验证外部状态：真实日志中出现 `compact/start…end` 对，模型可见内容缩减（一个替换节点遮蔽了较旧节点），且 agent 在压缩后仍给出正确最终答案。
+- `tests/todo-write.e2e.ts`：加载选用的 `todo_write` 工具，由真实模型驱动，测试验证产生的 `todo/write` 会话事件。
+- `tests/code-mode.e2e.ts`：带密钥 Code Mode 证明：使用真实模型和双工具任务，断言协议层工具列表精确为 `[run_code]`，`tool/code-dispatch` 事件位于父调用下，且筛选后的答案已返回。
 
-这些测试在没有 `DEEPSEEK_API_KEY` 时自行跳过。无密钥 `tests/tui-keyless-smoke.e2e.ts` 通过 PTY 启动真实 Loader 树（唯一获准的 PTY 界面）：基础启动 + `/plan` + `/exit`，一次带问题对话框和工具往返的脚本 LLM 对话，Code Mode 覆盖欢迎行，以及恢复失败退出路径。
+这些测试在没有 `DEEPSEEK_API_KEY` 时自行跳过。无密钥 `tests/tui-keyless-smoke.e2e.ts` 通过 PTY 启动真实 Loader 树（唯一获准的 PTY 界面）：基础启动 + `/plan` + `/exit`，一次带问题对话框和工具往返的脚本 LLM（大语言模型）对话，Code Mode 覆盖配置的欢迎行，以及恢复失败退出路径。
 
 ## 快照测试
 
-`tests/snapshots/<scenario>/session.jsonl` 提供已录制的用户提示词和模型分片；同级子日志驱动 subagent 和工作流。无密钥套件通过真实循环和工具实现执行这些脚本，然后比较可读的预期终端单元格／样式输出。使用 `pnpm run test:snapshot:refresh` 刷新仅展示变更；已录制模型旅程改变时，使用 DeepSeek 密钥运行 `pnpm run test:snapshot:record`。已实现的 [TUI 快照 Agent Note](../../.agents/notes/implemented/testing/2026-07-18-tui-terminal-state-snapshots.md) 拥有场景矩阵，以及已录制旅程、瞬时包快照与 PTY 覆盖之间的分工。
+`tests/snapshots/<scenario>/session.jsonl` 提供已录制的用户提示词和模型分片；同级子日志驱动 subagent 和工作流。无密钥套件通过真实循环和工具实现执行这些脚本，然后比较可读的预期终端单元格／样式输出。对于仅涉及展示的变更，使用 `pnpm run test:snapshot:refresh`；已录制的模型流程改变时，使用 DeepSeek 密钥运行 `pnpm run test:snapshot:record`。已实现的 [TUI 快照 Agent Note](../../.agents/notes/implemented/testing/2026-07-18-tui-terminal-state-snapshots.md) 规定了场景矩阵，以及已录制旅程、包级瞬态快照与 PTY 覆盖之间的分工。
diff --git a/knip.json b/knip.json
index 38eb32c223..e9c8f79e8f 100644
--- a/knip.json
+++ b/knip.json
@@ -174,6 +174,15 @@
         "tests/**/*.ts"
       ]
     },
+    "packages/context/tmux-context": {
+      "entry": [
+        "tests/**/*.spec.ts"
+      ],
+      "project": [
+        "src/**/*.ts",
+        "tests/**/*.ts"
+      ]
+    },
     "packages/lsp/lsp-local": {
       "entry": [
         "tests/**/*.spec.ts",
diff --git a/native/README.i18n.yaml b/native/README.i18n.yaml
index 4013fc656d..b58b0fe1cc 100644
--- a/native/README.i18n.yaml
+++ b/native/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write native/README.md
 README.md: 84808b2ee9dafa4f9f980c35a81ebe12480a4f5d
-README.zh.md: f73d4176454d9a577bf674a6bfe3f15cce3402b4
+README.zh.md: 276db0e655f2d632da9729c787b613cd231b2d40
diff --git a/native/README.zh.md b/native/README.zh.md
index f73d417645..276db0e655 100644
--- a/native/README.zh.md
+++ b/native/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-`node-addon-landlock-run` 的记录真源：这是 harness 从 npm 消费的 Landlock「先限制自身、再执行」启动器（`packages/sandbox/sandbox-local`、`packages/bash/bash-sandbox`）。启动器在此处开发，与消费方相邻；独立仓库是打包并发布 npm 包系列的发布镜像。
+`node-addon-landlock-run` 的权威源码位于此处：这是 harness 从 npm 引入并使用的 Landlock「先限制自身、再执行」启动器（`packages/sandbox/sandbox-local`、`packages/bash/bash-sandbox`）。启动器在此处开发，与消费方相邻；独立仓库是打包并发布 npm 包（package）系列的发布镜像。
 
 ## 发布镜像
 
@@ -17,6 +17,6 @@
 1. 先通过常规 harness PR 将启动器更改落地于此；触发 `Landlock Run` 工作流，并确保其所有任务通过。
 2. 在镜像 checkout 中替换 `.github/` 以外的所有内容：`git -C <mirror> rm -rq -- . ':!.github'`，然后执行 `git -C <harness> archive HEAD:native/landlock-run | tar -x -C <mirror>`，最后执行 `git -C <mirror> add -A` 并提交。
 3. 在镜像中按照其发布清单（`docs/release.md`）操作：`pnpm release:commit <version>` → 合并 → 标记 `vX.Y.Z` → 两阶段 `Release` 工作流（先以 `publish=false` 预演，再从标签以 `publish=true` 发布）。
-4. 使用已发布的标签／commit 更新上方 manifest（元数据清单）表，并在同一更改中提升 harness 消费方的依赖范围。
+4. 使用已发布的标签／commit 更新上方 manifest（元数据清单）表，并在同一更改中上调 harness 消费方的依赖版本范围。
 
-镜像不得分叉：如果更改直接提交到镜像中（例如发布期间的热修复），必须在下次导出前将其移植回此处。
+发布镜像不得与此处的权威源码产生分歧：如果更改直接提交到镜像中（例如发布期间的热修复），必须在下次导出前将其移植回此处。
diff --git a/native/landlock-run/README.i18n.yaml b/native/landlock-run/README.i18n.yaml
index 7212b6a317..e732ebdf31 100644
--- a/native/landlock-run/README.i18n.yaml
+++ b/native/landlock-run/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write native/landlock-run/README.md
 README.md: 284d5df764cf5a5205973696211aee2366d3b76e
-README.zh.md: 7163314abac0362afccee6fcc4506a84701cc27a
+README.zh.md: f369799cc8dcfb6de7c4b7b8c18857693d310418
diff --git a/native/landlock-run/README.zh.md b/native/landlock-run/README.zh.md
index 7163314aba..f369799cc8 100644
--- a/native/landlock-run/README.zh.md
+++ b/native/landlock-run/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-一个 [Landlock](https://landlock.io/)「先限制自身、再执行」启动器，用于在 Linux 上限制子进程。它以每平台预构建 npm 包加一个轻量 JS 入口包的形式发布；入口包负责解析二进制文件并遵循其 CLI（命令行界面）契约。该启动器面向需要在文件系统允许清单下运行不可信命令、但不能限制自身的 agent harness 和其他宿主。
+一个 [Landlock](https://landlock.io/)「先限制自身、再执行」启动器，用于在 Linux 上限制子进程。它以按平台预构建的 npm 包（package）以及一个轻量 JS 入口包的形式发布；入口包负责解析二进制文件并遵循其 CLI（命令行界面）契约。该启动器面向需要让不可信命令在文件系统允许清单约束下运行、同时保持自身不受限制的 agent harness（智能体框架）和其他宿主。
 
 第一个工具是 **`landlock-run`**：一个「先限制自身、再执行」的 [Landlock](https://landlock.io/) 启动器（基于原始内核 UAPI 编写，约 300 行 C11，并与 musl 静态链接）。它在自身上安装 Landlock 规则集，再 `exec` 被包装的命令；该规则集会跨 `execve` 继承，因此命令及其产生的每个进程都在限制下运行，调用进程仍不受限制。它采用失败闭合：如果内核无法强制执行，则不运行命令并直接退出。
 
@@ -34,12 +34,12 @@ if (probe(launcher) !== 'unusable') {
 }
 ```
 
-公开 API 有意保持简小：
+公开 API 有意保持精简：
 
 - `launcherPath()`：当前宿主启动器的绝对路径（有意不检查是否存在；探测结果才是可用性信号）。
 - `probe(launcher?, { timeoutMs? })`：功能性强制执行探测，返回 `'full' | 'partial' | 'unusable'`。
 - `grantArgs({ readOnly?, readWrite? })`：启动器的授权 argv；未授予的一切都被拒绝。
-- `LAUNCHER_BIN`、`LAUNCHER_FAILURE_EXIT` (125)：契约常量。
+- `LAUNCHER_BIN`、`LAUNCHER_FAILURE_EXIT`（125）：契约常量。
 
 完整的二进制契约（argv 语法、退出码、报告行）锁定在 [docs/cli-contract.md](docs/cli-contract.md) 中。
 
@@ -57,4 +57,4 @@ pnpm build:native    # this Linux architecture's binaries (apt-get install musl-
 pnpm test
 ```
 
-二进制文件被 git 忽略，并且按架构原生构建：本地只构建当前机器的版本，CI 的每架构 runner 则是记录中的构建者。发布流程详见 [docs/release.md](docs/release.md)。
+二进制文件被 git 忽略，并且按架构原生构建：本地只构建当前机器的版本，CI 各架构 runner 产出的构建则作为正式发布依据。发布流程详见 [docs/release.md](docs/release.md)。
diff --git a/native/landlock-run/packages/entry/README.i18n.yaml b/native/landlock-run/packages/entry/README.i18n.yaml
index 2985be0419..86fe43ff4d 100644
--- a/native/landlock-run/packages/entry/README.i18n.yaml
+++ b/native/landlock-run/packages/entry/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write native/landlock-run/packages/entry/README.md
 README.md: e402cdfe71c4eb81b977a21955fe3fff6bf55fd3
-README.zh.md: 03dd18969d8e5c94be31605201b74b126ae5c06d
+README.zh.md: 6f8136c33560515af891b8873d007eb3e9b013e0
diff --git a/native/landlock-run/packages/entry/README.zh.md b/native/landlock-run/packages/entry/README.zh.md
index 03dd18969d..6f8136c335 100644
--- a/native/landlock-run/packages/entry/README.zh.md
+++ b/native/landlock-run/packages/entry/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-用于在 Linux 上限制子进程的 Landlock「先限制自身、再执行」启动器：此入口包解析每平台预构建二进制文件，运行功能性强制执行探测，并构建其授权 argv。消费方无需自行拼写启动器标志或解析启动器输出。
+用于在 Linux 上限制子进程的 Landlock「先限制自身、再执行」启动器：此入口包（package）定位对应平台的预构建二进制文件，运行功能性强制执行探测，并构建其授权 argv。消费方无需自行拼写启动器标志或解析启动器输出。
 
 ```js
 import { grantArgs, launcherPath, probe } from 'node-addon-landlock-run';
@@ -13,6 +13,6 @@ if (probe(launcher) !== 'unusable') {
 }
 ```
 
-启动器在自身上安装 Landlock 规则集，再 `exec` 被包装的命令；该规则集会跨 `execve` 继承，因此整个进程树都在限制下运行。未授予的一切都被拒绝；启动器失败时以 `125` 退出且不运行命令：始终失败闭合，绝不失败开放。二进制契约锁定在仓库的 `docs/cli-contract.md` 中；C 源码作为 `src/main.c` 随该 tarball 分发，便于审计。
+启动器在自身上安装 Landlock 规则集，再 `exec` 被包装的命令；该规则集会跨 `execve` 继承，因此整个进程树都在限制下运行。未授予的一切都被拒绝；启动器失败时以 `125` 退出且不运行命令：采用失败闭合策略，绝不在失败时放行。二进制契约锁定在仓库的 `docs/cli-contract.md` 中；C 源码作为 `src/main.c` 随该 tarball 分发，便于审计。
 
-平台包（由 `os`/`cpu` 选择的可选依赖，内部不含 JavaScript）：`node-addon-landlock-run-linux-x64`、`node-addon-landlock-run-linux-arm64`。在缺少对应包的宿主上，`launcherPath()` 返回确定且不存在的路径，`probe()` 报告 `'unusable'`；系统有意不提供安装时编译回退。
+平台包（由 `os`/`cpu` 选择的可选依赖，内部不含 JavaScript）：`node-addon-landlock-run-linux-x64`、`node-addon-landlock-run-linux-arm64`。在缺少对应包的宿主上，`launcherPath()` 返回一个固定但不存在的路径，`probe()` 报告 `'unusable'`；系统有意不提供安装时编译回退。
diff --git a/native/landlock-run/packages/linux-arm64/README.i18n.yaml b/native/landlock-run/packages/linux-arm64/README.i18n.yaml
index b6b14f1eef..76d4b9884a 100644
--- a/native/landlock-run/packages/linux-arm64/README.i18n.yaml
+++ b/native/landlock-run/packages/linux-arm64/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write native/landlock-run/packages/linux-arm64/README.md
 README.md: e5117988cf0bae2227edaa041700c2f75753899c
-README.zh.md: 93fee68207a9f03a54f214c69904d44729ed71e5
+README.zh.md: abbd0d1040638ad4d64f3ab219bedcd845eb5a9b
diff --git a/native/landlock-run/packages/linux-arm64/README.zh.md b/native/landlock-run/packages/linux-arm64/README.zh.md
index 93fee68207..abbd0d1040 100644
--- a/native/landlock-run/packages/linux-arm64/README.zh.md
+++ b/native/landlock-run/packages/linux-arm64/README.zh.md
@@ -2,8 +2,8 @@
 
 [English](README.md) | 中文
 
-面向 linux-arm64 的预构建 `bin/landlock-run` Landlock 启动器：一个从 [`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run) 中随包发布的 C 源码原生编译而成的静态 musl 二进制文件（不使用交叉工具链）。npm 的 `os`/`cpu` 字段在安装时选择此包；入口包将其解析为文件路径。该包不包含 JavaScript，也绝不会被导入。
+面向 linux-arm64 的预构建 `bin/landlock-run` Landlock 启动器：一个由 [`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run) 包（package）所附的 C 源码原生编译而成的静态 musl 二进制文件（不使用交叉工具链）。npm 的 `os`/`cpu` 字段在安装时选择此包；入口包将其定位到文件路径。该包不包含 JavaScript，也绝不会被导入。
 
-该二进制文件被 git 忽略，并通过 `files` 列表进入 npm tarball；如果文件缺失或 ELF 架构错误，`prepack` 门禁会拒绝打包，发布流水线则会按字节将打包二进制文件锁定到其来源 CI 构建。静态 musl 链接使同一个二进制文件同时适用于 glibc 和 musl 发行版，因此名称中没有 libc 后缀。
+该二进制文件被 git 忽略，并通过 `files` 列表进入 npm tarball；如果文件缺失或 ELF 架构错误，`prepack` 门禁会拒绝打包，发布流水线则会按字节核验打包的二进制文件与其来源 CI 构建产物一致。静态 musl 链接使同一个二进制文件同时适用于 glibc 和 musl 发行版，因此名称中没有 libc 后缀。
 
 同级包：`node-addon-landlock-run-linux-x64`。
diff --git a/native/landlock-run/packages/linux-x64/README.i18n.yaml b/native/landlock-run/packages/linux-x64/README.i18n.yaml
index af3f916fb4..e24a9393c7 100644
--- a/native/landlock-run/packages/linux-x64/README.i18n.yaml
+++ b/native/landlock-run/packages/linux-x64/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write native/landlock-run/packages/linux-x64/README.md
 README.md: 68b5dfc9b6f437a387c3792ee047a1f11630aca0
-README.zh.md: b1fa2e3f16c20c4d7e287c17ab0ba946e6cadbea
+README.zh.md: e813bcef7143b46a756e5716234f3bc3850de712
diff --git a/native/landlock-run/packages/linux-x64/README.zh.md b/native/landlock-run/packages/linux-x64/README.zh.md
index b1fa2e3f16..e813bcef71 100644
--- a/native/landlock-run/packages/linux-x64/README.zh.md
+++ b/native/landlock-run/packages/linux-x64/README.zh.md
@@ -2,8 +2,8 @@
 
 [English](README.md) | 中文
 
-面向 linux-x64 的预构建 `bin/landlock-run` Landlock 启动器：一个从 [`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run) 中随包发布的 C 源码原生编译而成的静态 musl 二进制文件（不使用交叉工具链）。npm 的 `os`/`cpu` 字段在安装时选择此包；入口包将其解析为文件路径。该包不包含 JavaScript，也绝不会被导入。
+面向 linux-x64 的预构建 `bin/landlock-run` Landlock 启动器：一个由 [`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run) 包（package）所附的 C 源码原生编译而成的静态 musl 二进制文件（不使用交叉工具链）。npm 的 `os`/`cpu` 字段在安装时选择此包；入口包将其定位到文件路径。该包不包含 JavaScript，也绝不会被导入。
 
-该二进制文件被 git 忽略，并通过 `files` 列表进入 npm tarball；如果文件缺失或 ELF 架构错误，`prepack` 门禁会拒绝打包，发布流水线则会按字节将打包二进制文件锁定到其来源 CI 构建。静态 musl 链接使同一个二进制文件同时适用于 glibc 和 musl 发行版，因此名称中没有 libc 后缀。
+该二进制文件被 git 忽略，并通过 `files` 列表进入 npm tarball；如果文件缺失或 ELF 架构错误，`prepack` 门禁会拒绝打包，发布流水线则会按字节核验打包的二进制文件与其来源 CI 构建产物一致。静态 musl 链接使同一个二进制文件同时适用于 glibc 和 musl 发行版，因此名称中没有 libc 后缀。
 
 同级包：`node-addon-landlock-run-linux-arm64`。
diff --git a/package.json b/package.json
index afa74fdf1e..a51261a4c2 100644
--- a/package.json
+++ b/package.json
@@ -96,13 +96,13 @@
     "constraints": "tsx scripts/check-workspace-constraints.ts",
     "doc-sync": "tsx scripts/run-gates.ts doc-sync",
     "hygiene": "pnpm run knip && pnpm run publint && pnpm run constraints && pnpm run verify-package-invariants && pnpm run verify-built-package-invariants && pnpm run verify-cordis-config && pnpm run verify-node-next-types && pnpm run verify-runtime-closure",
-    "dsh": "node --experimental-transform-types --import ./scripts/tspath-loader.ts apps/cli/src/bin.ts",
+    "dsh": "node --import tsx/esm apps/cli/src/bin.ts",
     "demo:headless": "node --import tsx packages/examples/cli-demo/src/bin.ts --config examples/headless-agent/cordis.yml",
-    "demo:tui": "node --experimental-transform-types --import ./scripts/tspath-loader.ts apps/cli/src/bin.ts",
+    "demo:tui": "node --import tsx/esm apps/cli/src/bin.ts",
     "demo:code-mode": "node scripts/demo-code-mode.mjs",
     "demo:cordis": "node scripts/demo-cordis.mjs",
     "demo:acp": "node --import tsx packages/examples/acp-demo/src/bin.ts --config examples/acp-agent/cordis.yml",
-    "demo:web": "npm run build && node --experimental-transform-types --import ./scripts/tspath-loader.ts apps/cli/src/bin.ts web",
+    "demo:web": "npm run build && node --import tsx/esm apps/cli/src/bin.ts web",
     "mock:llm": "node --import tsx packages/support/llm-mock-server/src/bin.ts",
     "dev:web": "tsx scripts/dev-web.ts --poll",
     "postinstall": "node scripts/install-lefthook.mjs"
diff --git a/packages/acp/README.i18n.yaml b/packages/acp/README.i18n.yaml
index cf2cf1791a..08051fea69 100644
--- a/packages/acp/README.i18n.yaml
+++ b/packages/acp/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/acp/README.md
 README.md: 326615210e5cfc39004fc5ab7462623089ac4126
-README.zh.md: 9999ecdd019501c3f501a6c69fab5e0ccfaf555c
+README.zh.md: 8679f2428a9e82a81de69d7d1413132a946fcafa
diff --git a/packages/acp/README.zh.md b/packages/acp/README.zh.md
index 9999ecdd01..8679f2428a 100644
--- a/packages/acp/README.zh.md
+++ b/packages/acp/README.zh.md
@@ -6,6 +6,6 @@ ACP（Agent Client Protocol）组将 harness 中的 agent（智能体）公开
 
 | 包 | 职责 |
 |---|---|
-| [`acp/`](acp/README.md) | 仅面向自动化的 ACP 服务器：新文本会话、已提交的 assistant 输出、机器权限策略、取消和由连接拥有的清理。 |
+| [`acp/`](acp/README.md) | 仅面向自动化的 ACP 服务器：新文本会话、已提交的 assistant 输出、机器权限策略、取消和由连接负责的清理。 |
 
-与之匹配的进程外 subagent 客户端仍位于 [`subagent/subagent-acp`](../subagent/subagent-acp/README.md)，因为它实现 subagent 提供方接口；任意 ACP 客户端都可以驱动同一服务器契约。
+与之匹配的进程外 subagent 客户端仍位于 [`subagent/subagent-acp`](../subagent/subagent-acp/README.md)，因为它实现 subagent 提供方接口；任意 ACP 客户端都可以按照同一服务器契约驱动该服务器。
diff --git a/packages/acp/acp/README.i18n.yaml b/packages/acp/acp/README.i18n.yaml
index 5583a2b9a5..d359613cd6 100644
--- a/packages/acp/acp/README.i18n.yaml
+++ b/packages/acp/acp/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/acp/acp/README.md
 README.md: 1b188b994d17ce56e8d5df019ddef755338fcc88
-README.zh.md: f8abe9e45a5efffa436513f7d4a931c69c624b84
+README.zh.md: c1e7d045b55119b62ad44d81071188e1ed6110d5
diff --git a/packages/acp/acp/README.zh.md b/packages/acp/acp/README.zh.md
index f8abe9e45a..c1e7d045b5 100644
--- a/packages/acp/acp/README.zh.md
+++ b/packages/acp/acp/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-通过 JSON-RPC stdio 提供的仅面向自动化的 [Agent Client Protocol](https://agentclientprotocol.com) 服务器。程序化客户端可以创建新 harness agent（智能体）、发送文本提示词、收集已提交的 assistant 文本、通过策略解决一次性权限请求并取消工作。仓库中的主要客户端是 [`dsh-subagent-acp`](../../subagent/subagent-acp/README.md)。
+通过 JSON-RPC stdio 提供的仅面向自动化的 [ACP（Agent Client Protocol）](https://agentclientprotocol.com) 服务器。程序化客户端可以创建新 harness agent（智能体）、发送文本提示词、收集已提交的 assistant 文本、按策略响应一次性权限请求并取消工作。仓库中的主要客户端是 [`dsh-subagent-acp`](../../subagent/subagent-acp/README.md)。
 
-此包（package）是传输适配器，而非 UI 集成或能力 seam。它不公开编辑器导航、transcript（文本记录）回放、命令、mode、配置选择器、信息征集、推理、计划、标题或工具展示。交互渲染与人类问题属于 Web 和 TUI 模块。
+此包（package）是传输适配器，而非 UI 集成或能力 seam。它不公开编辑器导航、transcript（文本记录）回放、命令、模式、配置选择器、信息征集、推理、计划、标题或工具展示。交互式渲染与向用户提问属于 Web 和 TUI 模块。
 
 ## 插件
 
@@ -15,7 +15,7 @@
 | `provider` | 无 | 每个已创建 agent 的初始提供方路由。 |
 | `model` | 无 | 每个已创建 agent 的初始模型。 |
 
-两个字段都是可选的，以便由另一个 agent/request 监听器提供目标。可运行 ACP 组合同时要求两者。
+两个字段都是可选的，以便由另一个 agent/request 监听器提供目标。可运行的 ACP 组合同时要求两者。
 
 ## 协议契约
 
@@ -23,19 +23,19 @@
 |---|---|
 | `initialize` | 协商受支持的版本，并仅公布基线提示词（无图像、音频或嵌入上下文能力）。不公布会话、编辑器、终端、文件系统或 MCP 能力。 |
 | `authenticate` | 空操作，因为服务器不公布身份验证方法。 |
-| `session/new` | 使用绝对主 `cwd` 创建新 agent；接受空的 `additionalDirectories` 和 `mcpServers`，拒绝非空值。 |
-| `session/prompt` | 连接文本块，将基线资源链接渲染为带方括号的文本引用，拒绝空输入或超出基线的输入，每个会话只允许一个正在处理的请求，并从该请求拥有的持久 `turn/end` 结算。 |
-| `session/cancel` | 仅取消被定址的 agent，并将其待处理提示词结算为 `cancelled`；未知 id 为空操作。 |
+| `session/new` | 以绝对路径作为主 `cwd` 创建新 agent；接受空的 `additionalDirectories` 和 `mcpServers`，拒绝非空值。 |
+| `session/prompt` | 拼接文本块，将基线资源链接渲染为带方括号的文本引用，拒绝空输入或超出基线的输入，每个会话只允许一个正在处理的请求，并根据该请求所属的持久 `turn/end` 结算。 |
+| `session/cancel` | 仅取消指定的 agent，并将其待处理提示词结算为 `cancelled`；未知 id 为空操作。 |
 | `session/update` | 为每个非空文本块发出一个 `agent_message_chunk`；这些文本块来自已提交的 `assistant/message`。省略原始增量和非消息事件。 |
-| `session/request_permission` | 为携带工具调用 id 的桥接层所有批准请求提供一次性允许／拒绝选项。客户端可以自动回答。 |
+| `session/request_permission` | 为携带工具调用 id、由桥接层拥有的批准请求提供一次性允许／拒绝选项。客户端可以自动回答。 |
 
-一个连接可以拥有多个会话。桥接层使用带品牌的 session id 为记录建键，并在路由事件或权限请求前检查精确的 agent 标识。每个会话都有独立的提示词槽位、workspace、取消路径和 disposer。
+一个连接可以拥有多个会话。桥接层以带品牌的会话 id 作为记录键，并在路由事件或权限请求前检查 agent 是否为同一对象。每个会话都有独立的提示词槽位、工作区、取消路径和资源释放器。
 
-已提交消息输出有意以逐 token 延迟换取干净的自动化结果。未提交的提供方分片和重试尝试无法泄漏部分文本；推理与工具活动仍保留在会话日志中，以便其他界面观测。
+已提交消息输出有意牺牲逐 token 输出的低延迟，以换取干净的自动化结果。未提交的提供方分片和重试尝试无法泄漏部分文本；推理与工具活动仍保留在会话日志中，以便其他界面观测。
 
 ## 生命周期
 
-客户端断开与 Cordis 释放共用同一个记忆化清理流程。桥接层先拒绝新会话和提示词，结算待处理提示词，然后并行释放所有已拥有的 agent handle，并等待它们的循环／会话清理完成。因此，仅 ACP 的插件重载不会遗留 agent。
+客户端断开连接与 Cordis 的 dispose（资源释放）共用同一个记忆化清理流程。桥接层先拒绝新会话和提示词，结算待处理提示词，然后并行对其拥有的全部 agent 句柄执行 dispose，并等待它们的循环／会话清理完成。因此，单独重载 ACP 插件不会遗留孤儿 agent。
 
 ## 运行
 
@@ -45,13 +45,13 @@
 
 ### 提示词文本
 
-#### 模型所见内容
+#### 模型看到的内容
 
-`session/prompt` 文本块会原样连接为一条用户消息；基线资源链接会在该消息中表示为带方括号的 `[resource_link name=… uri=…]` 引用，模型可以使用自身工具打开它。协议元数据、客户端能力、权限选择和 session id 绝不进入模型请求。
+`session/prompt` 文本块会原样拼接为一条用户消息；基线资源链接会在该消息中表示为带方括号的 `[resource_link name=… uri=…]` 引用，模型可以使用自身工具打开它。协议元数据、客户端能力、权限选择和 session id 绝不进入模型请求。
 
 #### Token 影响
 
-提示词 token 取决于数据，并保留在该会话的历史中直到压缩。并发 ACP 会话保留独立上下文。
+提示词 token 取决于数据，并保留在该会话的历史中直到上下文压缩（context compaction）。并发 ACP 会话保留独立上下文。
 
 #### KV Cache 影响
 
@@ -59,21 +59,21 @@
 
 ### 权限决策
 
-#### 模型所见内容
+#### 模型看到的内容
 
-没有直接内容。拥有该决策的工具通过常规工具结果路径记录允许、拒绝、取消或不可用结果。
+不会直接看到任何内容。所属工具通过常规工具结果路径记录其结果：允许、拒绝、取消或不可用。
 
 #### Token 影响
 
-只有拥有该决策的工具结果会贡献 token。
+只有该工具的结果会贡献 token。
 
 #### KV Cache 影响
 
-通过所属工具结果仅追加。
+随该工具的结果仅追加。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - **仅新会话**：不支持加载、列出、恢复、删除和 fork。
-- **仅基线提示词和一个 workspace**：图像、音频、嵌入资源、非空附加目录和 MCP 服务器都会被拒绝；资源链接会被展平为文本引用，而不是已获取内容。
-- **仅已提交答案**：实时进度、推理、工具活动、计划、标题和用量不上线。
-- **连接拥有的生命期**：一个连接会释放其所有会话；尚未实现每会话关闭。
+- **仅基线提示词和一个 workspace**：图像、音频、嵌入资源、非空附加目录和 MCP 服务器都会被拒绝；资源链接只会展平为文本引用，不会获取其内容。
+- **仅已提交答案**：实时进度、推理、工具活动、计划、标题和用量不会通过协议传输。
+- **由连接管理的生命周期**：一个连接会释放其所有会话；尚未实现单个会话关闭功能。
diff --git a/packages/bash/README.i18n.yaml b/packages/bash/README.i18n.yaml
index 7c222fb6f5..0af14fda76 100644
--- a/packages/bash/README.i18n.yaml
+++ b/packages/bash/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/bash/README.md
 README.md: e60ad9b0e4c48cf35a2601e7dec4d2d50807707b
-README.zh.md: 57c28b45cf713aeaac725edb70d1fc24912c35db
+README.zh.md: deb23ea820de40c99f0affd3726d9a49857039ea
diff --git a/packages/bash/README.zh.md b/packages/bash/README.zh.md
index 57c28b45cf..deb23ea820 100644
--- a/packages/bash/README.zh.md
+++ b/packages/bash/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-规范的三包能力 seam（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：抽象执行器接口、具体实现，以及消费该接口的面向模型工具。这些全是**产品** 包。
+规范的三包能力 seam（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：抽象执行器接口、具体实现，以及消费该接口的面向模型工具。这些全是**产品**包。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
@@ -11,4 +11,4 @@
 | `bash-sandbox/` | 消费沙箱的 `BashExecutor`（通过 `ctx.sandbox` 包装每个命令 argv，标记拒绝／强制执行事实；扩展 `bash-local` 的机制） | （注册 `ctx.bash`） |
 | `tool-bash/` | 面向模型的 `bash` schema；后台进程注册到通用 [`tasks/`](../tasks/README.md) 运行时 | （注册到 `ctx.tools`） |
 
-接口位于 `bash/bash/`。以 `bash-sandbox` 替换 `bash-local`，同时不改动接口或工具，正是这种拆分存在的意义：叶级 `cordis.yml` 选择一个执行器配置项；受限实现还需选择一个 `ctx.sandbox` 提供方配置项（见 [acp-agent 示例的默认组合](../../examples/acp-agent/)）。
+接口位于 `bash/bash/`。以 `bash-sandbox` 替换 `bash-local`，同时不改动接口或工具，正是这种拆分存在的意义：叶级 `cordis.yml` 选择一个执行器插件条目；受限实现还需再选择一个 `ctx.sandbox` 提供方插件条目（见 [acp-agent 示例的默认组合](../../examples/acp-agent/)）。
diff --git a/packages/bash/bash-local/README.i18n.yaml b/packages/bash/bash-local/README.i18n.yaml
index f94d04f90e..e72432de87 100644
--- a/packages/bash/bash-local/README.i18n.yaml
+++ b/packages/bash/bash-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/bash/bash-local/README.md
 README.md: 694b7a7686ea6c38da5a354ff6b6e6d2c4520706
-README.zh.md: aa6de87df48ee943ccdd2c6227ad977f596b5516
+README.zh.md: c56543f26965effebaf020dd8d9d4ba130cd9b17
diff --git a/packages/bash/bash-local/README.zh.md b/packages/bash/bash-local/README.zh.md
index aa6de87df4..c56543f269 100644
--- a/packages/bash/bash-local/README.zh.md
+++ b/packages/bash/bash-local/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-`@deepseek-ai/dsh-bash` 执行器 seam 的本地实现，构建在 [`@deepseek-ai/dsh-subprocess`](../../subprocess/subprocess/README.md) 服务之上：`LocalBashExecutor` 每次调用都通过 `ctx.subprocess` 把 `bash -c <command>` 作为受管进程组 spawn，并拥有所有 bash 形态的职责（命令默认值补全与上限、超时与取消分类、适合模型的终端环境，以及后台读取时面向模型的 stdout/stderr 合并）。进程组机制（以 spill 文件兜底的有界输出、凭据清除、kill 升级、dispose（资源释放））归进程管理器服务所有。
+`@deepseek-ai/dsh-bash` 执行器 seam 的本地实现，构建在 [`@deepseek-ai/dsh-subprocess`](../../subprocess/subprocess/README.md) 服务之上：`LocalBashExecutor` 每次调用都通过 `ctx.subprocess` 把 `bash -c <command>` 作为受管进程组 spawn，并负责所有 Bash 层职责（命令默认值补全与上限、超时与取消分类、适合模型的终端环境，以及后台读取时面向模型的 stdout/stderr 合并）。以 spill 文件兜底的有界输出、凭据清除、kill 升级和 dispose（资源释放）等进程组机制则由 subprocess 服务负责。
 
 包根目录导出默认与具名的 `LocalBashExecutor` 插件及其 `Config`。
 
@@ -24,11 +24,11 @@
 
 设计时调研了 Claude Code、OpenCode、Codex 和 pi 的 bash 工具，主要取舍如下：
 
-- **每次调用都 spawn，不保留 shell 状态**：每次调用都启动新的非登录 `bash -c`（行为确定，不读取 rc 文件）。调研的四种工具均会每次调用单独 spawn。`XXX(stateful-shell)` 位于 `src/index.ts`，记录了两种已验证的有状态设计（Claude Code 仅持久化 cwd；Codex 使用 PTY exec 会话），供真实工作流程需要时采用。
+- **每次调用都 spawn，不保留 shell 状态**：每次调用都启动新的非登录 `bash -c`（行为确定，不读取 rc 文件）。调研的四种工具均会每次调用单独 spawn。`XXX(stateful-shell)` 位于 `src/index.ts`，记录了两种已验证的有状态设计（Claude Code 仅持久化 cwd；Codex 使用 PTY exec 会话），供真实工作流需要时采用。
 - **在受管进程组之上应用配置预算**：`resolve()` 从配置补全 `workdir`／`timeoutMs`／`stdoutMaxBytes`，每次 spawn 都向服务传入显式的字节上限、spill 上限与 `graceMs`（默认 3 秒，沿用 OpenCode 的升级策略）。进程组终止、退出后的管道排空宽限期、尾部保留截断与有界 spill 文件是 [`dsh-subprocess-local`](../../subprocess/subprocess-local/README.md) 的机制。前台 `BashExecRequest.stdoutMaxBytes` 可为某个受信任调用方提高单次 stdout 捕获预算；stderr 和后台运行仍使用 `maxOutputBytes`。
-- **超时与取消分类**：`run()` 通过同一个 deadline 把经配置钳位的超时与调用方的信号融合；只有执行器自身的超时报告 `timedOut`，上游取消报告 `aborted`，自行发出信号终止的命令两者皆不报告（见[超时库 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-06-timeout-deadline-library.md)）。
+- **超时与取消分类**：`run()` 通过同一个 deadline 把经配置钳位的超时与调用方的信号融合；只有执行器自身的超时报告 `timedOut`，上游取消报告 `aborted`，自身因信号终止的命令两者皆不报告（见[超时库 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-06-timeout-deadline-library.md)）。
 - **适合模型的终端环境**：设置 `NO_COLOR=1 TERM=dumb PAGER=cat GIT_PAGER=cat`（Codex 硬编码的集合），防止分页器与 ANSI 颜色破坏结果；这些条目作为普通 env 合并，遵循服务的凭据清除与 `DSH_*` 通道规则；调用方的显式条目依旧优先。详见 [stdin/env Agent Note](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md) 与 [受管环境 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md)。
-- **后台进程**：`start()` 会立即返回实时 `BashProcess` 句柄，不应用超时（Claude Code 在转为后台时会解除超时）；句柄的 `readOutput()` 把服务基于偏移量的 stdout/stderr 读取合并为一条带标记分节的增量，由一个消费游标驱动。仍在运行的进程归进程管理器服务所有，因此它能在执行器重载后存活，并随服务的 dispose 被终止且等待退出。所有具有任务形态的事项（id、所有权、轮询、通知）都属于通用 [`ctx.tasks` 运行时](../../tasks/tasks/README.md)，工具层会在其中注册该句柄；本执行器不会接触会话或注册表。
+- **后台进程**：`start()` 会立即返回活动的 `BashProcess` 句柄，不应用超时（Claude Code 在转为后台时会解除超时）；句柄的 `readOutput()` 把服务基于偏移量的 stdout/stderr 读取合并为一条带分节标记的增量，并以消费游标记录读取进度。仍在运行的进程则由 subprocess 服务负责，因此它能在执行器重载后存活，并随服务的 dispose 被终止且等待退出。所有具有任务形态的事项（id、所有权、轮询、通知）都属于通用 [`ctx.tasks` 运行时](../../tasks/tasks/README.md)，工具层会在其中注册该句柄；本执行器不会接触会话或注册表。
 
 ## 模型体验
 
@@ -36,12 +36,12 @@
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由具名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由具名消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **自身不受约束**：此执行器始终以 harness 进程的权限运行命令；需要限制的部署可以组合 [`dsh-bash-sandbox`](../bash-sandbox/README.md)，每次调用的 allow/deny/ask 策略则属于 `tools/pre-execute`。
-- **没有持久 shell 或 PTY**：每次调用都启动新的非登录 `bash -c`；仅持久化 cwd 与交互式终端会话均继续暂缓，直到真实工作流程需要它们。
+- **自身不提供隔离**：此执行器始终以 harness 进程的权限运行命令；需要限制的部署可以组合 [`dsh-bash-sandbox`](../bash-sandbox/README.md)，每次调用的 allow/deny/ask 策略则属于 `tools/pre-execute`。
+- **没有持久 shell 或 PTY**：每次调用都启动新的非登录 `bash -c`；仅持久化 cwd 与交互式终端会话均继续暂缓，直到真实工作流需要它们。
 - **仅支持 POSIX**：`bash` 二进制已硬编码，底层服务的进程组语义也是 POSIX 的；不支持 Windows。
 - **后台 spawn 失败提示只交付一次**：进程管理器不会为从未真正运行的进程缓冲任何输出，因此执行器把 `spawn failed: …` 注入恰好一个 `readOutput()` 增量；丢弃了该增量的读取方无法再恢复它。
 
diff --git a/packages/bash/bash-sandbox/README.i18n.yaml b/packages/bash/bash-sandbox/README.i18n.yaml
index 86168916fe..e4e294b001 100644
--- a/packages/bash/bash-sandbox/README.i18n.yaml
+++ b/packages/bash/bash-sandbox/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/bash/bash-sandbox/README.md
 README.md: ca77a9c626784b29145712535d69de4afbd3a697
-README.zh.md: c1a65ead539ef3930d70d27f2b176a5346daded3
+README.zh.md: 4ecc8d533f7af373bdacd133d44a8def6d265868
diff --git a/packages/bash/bash-sandbox/README.zh.md b/packages/bash/bash-sandbox/README.zh.md
index c1a65ead53..4ecc8d533f 100644
--- a/packages/bash/bash-sandbox/README.zh.md
+++ b/packages/bash/bash-sandbox/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-消费 [`@deepseek-ai/dsh-bash`](../bash/) 执行器 seam 的沙箱实现。加载它时，应**用它替代** `@deepseek-ai/dsh-bash-local`，并同时加载 [`ctx.sandbox`](../../sandbox/sandbox/) 提供方（例如 [`@deepseek-ai/dsh-sandbox-local`](../../sandbox/sandbox-local/)）及 [`ctx.sandboxPolicy`](../../sandbox/sandbox-policy/)；后者拥有默认模式 + 工作区根目录，并与受沙箱约束的文件系统共享这些设置。无需使用替代工具插件；`dsh-tool-bash` 会检测执行器的 `sandboxMode` 能力并添加升权字段。
+这是使用沙箱能力的 [`@deepseek-ai/dsh-bash`](../bash/) 执行器 seam 实现。加载它时，应**用它替代** `@deepseek-ai/dsh-bash-local`，并同时加载 [`ctx.sandbox`](../../sandbox/sandbox/) 提供方（例如 [`@deepseek-ai/dsh-sandbox-local`](../../sandbox/sandbox-local/)）及 [`ctx.sandboxPolicy`](../../sandbox/sandbox-policy/)；默认模式和工作区根目录由后者负责，并与受沙箱约束的文件系统共享这些设置。无需使用替代工具插件；`dsh-tool-bash` 会检测执行器的 `sandboxMode` 能力并添加升权字段。
 
 包根目录导出默认与具名的 `SandboxBashExecutor` 插件及其 `Config`；引号处理与结果分类 helper 保留在内部。
 
-每条命令的限制方式都是：把本执行器即将 spawn 的精确 `['bash', '-c', command]` argv 交给提供方，再 spawn 其返回的（已包装）argv。由哪种平台 runner 执行限制，以及是否有 runner 可用（必须快速失败并返回结构化 `SANDBOX_UNAVAILABLE` 错误，绝不能静默无约束运行），属于提供方职责；本包只拥有 bash 侧。
+每条命令的限制方式都是：把本执行器即将 spawn 的精确 `['bash', '-c', command]` argv 交给提供方，再 spawn 其返回的（已包装）argv。由哪种平台 runner 执行限制，以及是否有 runner 可用，属于提供方职责；若无可用 runner，则按失败关闭原则拒绝执行并返回结构化 `SANDBOX_UNAVAILABLE` 错误，绝不能静默地无约束运行。本包只负责 bash 侧。
 
 | 模式 | 文件影响 |
 |---|---|
@@ -17,12 +17,12 @@
 语义：
 
 - **拒绝是结果事实。** 如果一次失败运行的 stderr 包含所选后端自身的拒绝方言，即提供方在每次包装时加上的特征（bwrap 下的 EROFS 文本、Landlock 下的 EACCES、Seatbelt 下的 EPERM），则结果报告 `BashRunResult.sandbox.denied: true`（从已收集的 stderr 尾部进行保守分类）。每次受限制运行还会携带执行时模式（`result.sandbox.mode`）与提供方强制执行完整性（`result.sandbox.enforcement`：`full`，或在较旧 Landlock ABI 上为 `partial`）。
-- **Runner 失败是沙箱失败，绝不是命令失败。** 前台执行会抛出 `SANDBOX_UNAVAILABLE`；已结算的后台进程会标记 `process.sandbox.runnerFailed`，bash 产生方通过通用 `task_output` 渲染它。spawn 失败也会经过结算，因此受限制的后台句柄会保留自身的模式／强制执行事实，并释放每进程计数。
-- **部署回退，每次调用策略。** [`ctx.sandboxPolicy`](../../sandbox/sandbox-policy/) 为每次工具调用解析完整的 `SandboxExecutionPolicy`：调用会话提供自身的模式覆盖与不可变 cwd 根目录，部署配置则为无 agent 调用提供回退。已批准的升权只更改该策略的模式，会话根目录仍然附着其上。`resolve()` 把策略带入 spec，因此来自不同项目的重叠命令会在各自的根目录与模式下运行、分类和报告。能力事实 `ctx.bash.sandboxMode` 报告已配置的默认值，因此工具层只在装载该执行器时才公布升权。模型只能通过结果事实了解沙箱：静态 bash 工具描述会解释拒绝标记，系统提示词中不会声明当前模式。
+- **Runner 失败是沙箱失败，绝不是命令失败。** 前台执行会抛出 `SANDBOX_UNAVAILABLE`；已结算的后台进程会标记 `process.sandbox.runnerFailed`，Bash 结果生成方通过通用 `task_output` 渲染它。spawn 失败也会经过结算，因此受限制的后台句柄会保留自身的模式／强制执行事实，并释放每进程计数。
+- **部署回退，每次调用策略。** [`ctx.sandboxPolicy`](../../sandbox/sandbox-policy/) 为每次工具调用解析完整的 `SandboxExecutionPolicy`：调用会话提供自身的模式覆盖与不可变 cwd 根目录，部署配置则为无 agent（智能体）调用提供回退。已批准的升权只更改该策略的模式，会话根目录仍然附着其上。`resolve()` 把策略带入 spec，因此来自不同项目的重叠命令会在各自的根目录与模式下运行、分类和报告。能力事实 `ctx.bash.sandboxMode` 报告已配置的默认值，因此工具层只在装载该执行器时才公布升权。模型只能通过结果事实了解沙箱：静态 bash 工具描述会解释拒绝标记，系统提示词中不会声明当前模式。
 - **只限制文件影响。** 设计上不限制网络与进程可见性：模式词汇不会声称覆盖后端未强制执行的范围。
 - 进程机制（spawn、进程组终止、输出收集／spill、后台句柄、凭证清理）继承自 [`dsh-bash-local`](../bash-local/)；runner 选择位于 [`dsh-sandbox-local`](../../sandbox/sandbox-local/)。
 
-seam 上仅拒绝：拒绝是一项已报告事实，本执行器绝不自行协商权限。批准问题位于工具层（`dsh-tool-bash`），由它驱动本包遵守的覆盖。
+该 seam 只报告拒绝：拒绝是一项结果事实，本执行器绝不自行协商权限。批准问题位于工具层（`dsh-tool-bash`），由它设置本包所遵守的模式覆盖值。
 
 ```yaml
 - id: sandbox
@@ -36,7 +36,7 @@ seam 上仅拒绝：拒绝是一项已报告事实，本执行器绝不自行协
   name: '@deepseek-ai/dsh-bash-sandbox'
 ```
 
-无密钥消费方集成证明是 `tests/bwrap.e2e.ts`、`tests/landlock.e2e.ts` 和 `tests/seatbelt.e2e.ts`（通过 `ctx.bash` 驱动真实提供方 + 真实 runner，在真实世界验证，并在相应 runner 缺失时各自自行跳过）。agent-spine e2e 还会在一个 Cordis 上下文中驱动两个并发会话，并证明每个真实 bash 工具调用只能写入自身项目。可运行 demo 见 [acp-agent 示例的默认组合](../../../examples/acp-agent/)。
+无密钥消费方集成证明是 `tests/bwrap.e2e.ts`、`tests/landlock.e2e.ts` 和 `tests/seatbelt.e2e.ts`（通过 `ctx.bash` 驱动真实提供方 + 真实 runner，从外部验证实际文件效果，并在相应 runner 缺失时各自自行跳过）。agent-spine e2e 还会在一个 Cordis 上下文中驱动两个并发会话，并证明每个真实 bash 工具调用只能写入自身项目。可运行 demo 见 [acp-agent 示例的默认组合](../../../examples/acp-agent/)。
 
 ## 模型体验
 
@@ -44,11 +44,11 @@ seam 上仅拒绝：拒绝是一项已报告事实，本执行器绝不自行协
 
 #### 模型看到的内容
 
-基线是生成的 [`dsh-tool-bash` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-bash)。通过公布一个执行限制的 `sandboxMode`，此后端会为 `bash` 增加 `sandbox_permissions`，其 enum 为 `workspace-write` | `danger-full-access`，并增加 `justification`。后端不添加提示词文本，会话的有效模式仍不会声明。
+基线是生成的 [`dsh-tool-bash` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-bash)。通过公布表明启用隔离的 `sandboxMode` 能力，此后端会为 `bash` 增加 `sandbox_permissions`，其 enum 为 `workspace-write` | `danger-full-access`，并增加 `justification`。后端不添加提示词文本，会话的有效模式仍不会声明。
 
 #### Token 影响
 
-在 `bash` 可见的请求上增加少量固定 schema；模式切换不增加上下文 token。
+在 `bash` 可见的请求上，schema 固定增加少量内容；模式切换不增加上下文 token。
 
 #### KV Cache 影响
 
@@ -62,29 +62,29 @@ seam 上仅拒绝：拒绝是一项已报告事实，本执行器绝不自行协
 
 #### Token 影响
 
-除普通输出外，正常允许的运行不会增加 token。拒绝或失败会增加上述有条件标记，并保留到压缩。
+除普通输出外，正常允许的运行不会增加 token。拒绝或失败会增加上述有条件标记，并保留到上下文压缩（context compaction）。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 间接的 Bash 工具错误
 
 #### 模型看到的内容
 
-如果没有 runner 能强制执行受限模式，前台调用会传播 [`SANDBOX_UNAVAILABLE` 错误；它由 `dsh-sandbox` 持有](../../sandbox/sandbox/README.md#confinement-error-indirectly)。如果 runner 在执行时失败，此后端会提供第一行 stderr 作为详细信息。
+如果没有 runner 能强制执行受限模式，前台调用会传播 [`SANDBOX_UNAVAILABLE` 错误](../../sandbox/sandbox/README.md#confinement-error-indirectly)；该错误由 `dsh-sandbox` 定义。如果 runner 在执行时失败，此后端会提供第一行 stderr 作为详细信息。
 
 #### Token 影响
 
-该次调用可见的是有条件错误文本，并保留在历史记录中直到压缩。
+该次调用会在相应条件下显示错误文本，该文本会保留在历史记录中直到上下文压缩。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
 - **限制只覆盖文件影响**：网络访问与进程可见性不变，因此这些模式不是通用安全沙箱。
-- **拒绝从失败命令的 stderr 推断**：后端特征使该推断可跨平台使用，但匹配的应用错误可能被分类为拒绝，也可能遗漏未出现在保留尾部中的拒绝。
+- **拒绝从失败命令的 stderr 推断**：后端特征使该推断可跨平台使用，但包含相同后端特征的应用错误可能被分类为拒绝，也可能遗漏未出现在保留尾部中的拒绝。
 - **后台 runner 失败没有即时错误通道**：它记录在已结算进程上，并在调用方使用 `task_output` 读取通用任务时呈现。
 - **`danger-full-access` 有意绕过 `ctx.sandbox`**：它是显式无约束模式，不是更宽的沙箱 profile。
diff --git a/packages/bash/bash/README.i18n.yaml b/packages/bash/bash/README.i18n.yaml
index 520828acde..4a2c37ad91 100644
--- a/packages/bash/bash/README.i18n.yaml
+++ b/packages/bash/bash/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/bash/bash/README.md
 README.md: d7bf746969f52000fe298b65b995b7c631d8001c
-README.zh.md: 14476b770397e4ef850c7c867e3058e25085c339
+README.zh.md: a7c0cac0bce2154362c822c213a44f3c507d541c
diff --git a/packages/bash/bash/README.zh.md b/packages/bash/bash/README.zh.md
index 14476b7703..a7c0cac0bc 100644
--- a/packages/bash/bash/README.zh.md
+++ b/packages/bash/bash/README.zh.md
@@ -4,7 +4,7 @@
 
 **bash 执行器 seam**：抽象 `BashExecutor` 服务（`ctx.bash`）定义 bash 后端做什么，即运行前台命令与启动后台进程，但不规定如何实现。task id、所有权、收集、取消与通知属于通用 `ctx.tasks` 运行时。
 
-本包是 bash 能力中负责接口的四分之一，各项职责因此可以独立演进（和替换）：
+本包（package）是 bash 能力中负责接口的四分之一，各项职责因此可以独立演进（和替换）：
 
 | 包 | 职责 |
 |---|---|
@@ -13,19 +13,19 @@
 | `@deepseek-ai/dsh-bash-sandbox` | 实现：沿用 `dsh-bash-local` 的机制，但通过 [`ctx.sandbox`](../../sandbox/sandbox/) 限制每次 spawn，并将拒绝报告为结果事实 |
 | `@deepseek-ai/dsh-tool-bash` | 基于 `ctx.bash`、面向模型的工具 schema |
 
-该拆分与 LLM seam（`LlmService`／`LlmAdapter`）及 agent 工具调研结果一致：pi 将执行隐藏在 `BashOperations` 接口之后（本地 shell／SSH／VM 后端），Codex 则隐藏在 exec-server 协议之后。`dsh-bash-sandbox` 正是这种替换的实际应用：沙箱执行器位于同一接口之后；消费方检测其 `sandboxMode` 能力并添加升权字段，无需导入实现。容器化或远程执行器也可以同样接入。
+该拆分与 LLM（大语言模型） seam（`LlmService`／`LlmAdapter`）及 agent（智能体）工具调研结果一致：pi 将执行隐藏在 `BashOperations` 接口之后（本地 shell／SSH／VM 后端），Codex 则隐藏在 exec-server 协议之后。`dsh-bash-sandbox` 正是这种替换的实际应用：沙箱执行器位于同一接口之后；消费方检测其 `sandboxMode` 能力并添加升权字段，无需导入实现。容器化或远程执行器也可以同样接入。
 
 ## 服务 API（`ctx.bash`）
 
 | 成员 | 语义 |
 |---|---|
-| `run(spec)` | 前台执行。命令完成时 resolve。**只会因基础设施失败而 reject**（工作目录不可用、shell 缺失、信号已在调用前中止）；非零退出、超时终止和中止终止都会 resolve 为描述性 `BashRunResult`。 |
+| `run(spec)` | 前台执行。命令完成时 resolve。**只会因基础设施失败而 reject**（工作目录不可用、shell 缺失、信号已在调用前中止）；非零退出、超时终止和中止导致的终止都会 resolve 为描述性 `BashRunResult`。 |
 | `start(spec)` | 后台执行。立即返回不含任务语义的 `BashProcess` 句柄；**不应用超时**。调用方可以将其适配到 `ctx.tasks`。 |
 | `sandboxMode` | 工具层的能力事实：沙箱执行器用于限制执行的默认模式（基类中为 `undefined`，即「此执行器不使用沙箱」）。`dsh-tool-bash` 会在注册时读取它，仅当组合确实支持升权字段时才公布这些字段。 |
 | `BashProcess.readOutput()` | **增量** 读取输出：连续读取绝不会重复交付。因缓冲区边界丢失数据的读取会标记 `lossy`，并指向完整流 spill 文件。 |
 | `BashProcess.kill()` | 终止进程组。如果进程已结束，返回 `false`。 |
 
-实现会继承 `BashExecutor` 并实现抽象方法。dispose 必须终止每个运行中的进程并等待其退出，详见 HMR 安全测试。
+实现会继承 `BashExecutor` 并实现抽象方法。dispose（资源释放）必须终止每个运行中的进程并等待其退出，详见 HMR（热模块替换）安全测试。
 
 ## 词汇
 
@@ -33,7 +33,7 @@
 
 每会话沙箱模式覆盖词汇（`'sandbox/mode'` 事件、`effectiveSandboxMode(events)` fold 以及 `setSandboxMode(session, mode)` 写入路径）不位于此处。它是所有强制执行家族共享的策略状态，属于 [`@deepseek-ai/dsh-sandbox-policy`](../../sandbox/sandbox-policy/)。`run()` 返回 `BashRunResult`；`start()` 返回 `BashProcess`，其增量读取与终止方法由 `dsh-tool-bash` 适配为通用任务注册。沙箱执行器会在前台结果与已结算进程句柄上标记 `BashSandboxInfo`。详见 `src/types.ts` 与 [core-data-structures/bash.md](../../../docs/core-data-structures/bash.md)。
 
-`stdin` 与普通 `env` 由同进程插件（hooks 桥接、原生插件）设置，用于向 hook 命令提供其 JSON payload 和 `CLAUDE_PROJECT_DIR`／`CLAUDE_PLUGIN_ROOT` 值。`dshEnv` 是受类型限制、仅允许受管 key 的独立受信任 overlay；导出的 `DSH_ENV_PREFIX` 是该 namespace、其 `DshEnvironmentKey` 模板类型、执行器清理、注册表验证、派生内置名称与模型指引的单一真源。模型 bash 使用 `ctx.bashEnv` 收集的当前快照。实现会移除继承的受管 key，再在普通 `env` 之后合并 `dshEnv`，因此省略的当前事实不会回退到陈旧环境状态，`env` 条目也无法顶掉受管值。面向模型的工具不公开任何一个字段。这三者在已解析 spec 上仍然可选；缺失表示没有输入／overlay。详见 [bash-stdin-env Agent Note](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md) 与 [会话环境 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md)。
+`stdin` 与普通 `env` 由同进程插件（hooks 桥接、原生插件）设置，用于向 hook 命令提供其 JSON payload 和 `CLAUDE_PROJECT_DIR`／`CLAUDE_PLUGIN_ROOT` 值。`dshEnv` 是受类型限制、仅允许受管 key 的独立受信任 overlay；导出的 `DSH_ENV_PREFIX` 是该 namespace、其 `DshEnvironmentKey` 模板类型、执行器清理、注册表验证、派生内置名称与模型指引的统一来源。模型 bash 使用 `ctx.bashEnv` 收集的当前快照。实现会移除继承的受管 key，再在普通 `env` 之后合并 `dshEnv`，因此省略的当前事实不会回退到陈旧环境状态，`env` 条目也无法顶掉受管值。面向模型的工具不将这三者中的任何一个公开为参数。这三者在已解析 spec 上仍然可选；缺失表示没有输入／overlay。详见 [bash-stdin-env Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md) 与 [会话环境 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md)。
 
 ## 模型体验
 
@@ -41,9 +41,9 @@
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由具名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由具名消费方负责。
 
 ## 已知限制与暂缓事项
 
 - **没有交互式输入词汇**：`stdin` 只会在 spawn 时写入一次并关闭；seam 不提供向运行中任务继续输入的通道，也没有 PTY 会话概念。
-- **前台超时始终由执行器拥有**：seam 上的调用方拥有 deadline 模式已由 [工具调用超时策略 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-07-tool-call-timeout-policy.md) 明确暂缓。
+- **前台超时始终由执行器负责**：seam 上由调用方负责 deadline 的模式已由 [工具调用超时策略 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-07-tool-call-timeout-policy.md) 明确暂缓。
diff --git a/packages/client/connection/src/client/fixture.ts b/packages/client/connection/src/client/fixture.ts
index cab616aabd..3253efbb06 100644
--- a/packages/client/connection/src/client/fixture.ts
+++ b/packages/client/connection/src/client/fixture.ts
@@ -80,6 +80,62 @@ const MARKDOWN_FIXTURE = [
 
 const USER_MARKDOWN_LITERAL = '用户字面量：# 不渲染 `code` [link](https://example.com)'
 
+/**
+ * SGR wrapper for the terminal output sample below: authoring the escapes as
+ * `\u001b` keeps literal control bytes out of this source file.
+ * @param code - the SGR parameter (an ANSI color or attribute number).
+ * @param body - the text the attribute applies to.
+ * @returns the body wrapped in the attribute and a reset.
+ */
+function sgr(code: number, body: string): string {
+  return `\u001b[${code}m${body}\u001b[0m`
+}
+
+/**
+ * Terminal output sample for fixture turn 65, authored to carry every feature
+ * the terminal card draws that turn 60's two prompt rows cannot reach:
+ * basic-16 SGR foreground runs (green, red, bright-black) that must resolve to
+ * `--dsw-*` tokens, a bold run, column-aligned table rows that must scroll
+ * rather than fold, more than DEFAULT_TERMINAL_MAX_LINES (16) lines so the
+ * height cap collapses the middle. The exit status is authored separately in
+ * TERMINAL_EXIT_STATUS and deliberately absent from this text: the real bash
+ * presenter CONSUMES its `[exit code: N]` marker out of the body, because a
+ * terminal card shows the exit as its own pill and leaving the marker in would
+ * render it twice (packages/bash/tool-bash/src/render.ts).
+ */
+const TERMINAL_OUTPUT_FIXTURE = [
+  sgr(1, 'Running 4 checks'),
+  `${sgr(32, '\u2713')} typecheck                                          1.82s`,
+  `${sgr(32, '\u2713')} lint                                               0.94s`,
+  `${sgr(32, '\u2713')} duplication                                        2.10s`,
+  `${sgr(31, '\u2717')} unit                                               8.41s`,
+  '',
+  sgr(90, 'packages/client/ui-primitives/tests/terminal-block.spec.tsx'),
+  `  ${sgr(31, 'FAIL')} caps output at the configured line budget`,
+  '    expected 16 lines, received 24',
+  '',
+  'NAME                        LINES    BRANCHES    FUNCTIONS    UNCOVERED',
+  'TerminalBlock.tsx           100%     100%        100%         -',
+  'ansi.ts                     100%     100%        100%         -',
+  'clipboard.ts                100%     100%        100%         -',
+  'CodeBlock.tsx               98.4%    96.2%       100%         41-43',
+  'highlight.ts                100%     100%        100%         -',
+  'Pill.tsx                    100%     100%        100%         -',
+  'StateDot.tsx                100%     100%        100%         -',
+  'markdown/Markdown.tsx       100%     100%        100%         -',
+  '',
+  sgr(31, '1 of 4 checks failed'),
+].join('\n')
+
+/**
+ * Exit status for each terminal sample, keyed by its output text. Authored
+ * alongside the sample rather than parsed back out of its trailing marker,
+ * which is the bash tool's own job and not something to reimplement here.
+ */
+const TERMINAL_EXIT_STATUS: Record<string, { exitCode: number } | { signal: string }> = {
+  [TERMINAL_OUTPUT_FIXTURE]: { exitCode: 1 },
+}
+
 const DEEPSEEK_REASONING = {
   efforts: [
     { id: 'off', name: 'Off' },
@@ -170,7 +226,9 @@ function buildAlphaLog(): SessionEvent[] {
     push({ type: 'step/end', data: { turn, step: 0 } })
     push({ type: 'turn/end', data: { turn, reason: { kind: 'completed' } } })
   }
-  toolTurn(60, 'fx-bash', '{"command":"ls -la","cwd":"/tmp/fixture"}', 'total 2\ndrwxr-xr-x fixture\n-rw-r--r-- demo.txt')
+  // A two-line command, so the fixture covers the terminal card's one-row-per-
+  // command-line prompt (and that the card still marks the call exactly once).
+  toolTurn(60, 'fx-bash', '{"command":"ls -la\\necho done","cwd":"/tmp/fixture"}', 'total 2\ndrwxr-xr-x fixture\n-rw-r--r-- demo.txt')
   toolTurn(61, 'fx-write', '{"path":"notes/demo.txt","content":"hello fixture\\n"}', 'wrote notes/demo.txt')
   toolTurn(62, 'edit', '{"file_path":"notes/demo.txt","old_string":"hello","new_string":"hello fixture"}', '已编辑')
   toolTurn(63, 'write', '{"file_path":"notes/new-demo.txt","content":"hello fixture\\n"}', '已写入')
@@ -224,8 +282,22 @@ function buildAlphaLog(): SessionEvent[] {
     { content: '实现 fixture 样本', status: 'in_progress' },
     { content: '浏览器验收', status: 'pending' },
   ]
+  // Turn 65: the terminal sample turn 60's two clean prompt rows cannot cover —
+  // ANSI SGR coloring, output past the terminal card's height cap, a nested cwd
+  // whose prompt label is its last segment, and a non-zero exit authored beside
+  // the sample in TERMINAL_EXIT_STATUS — its body deliberately carries no
+  // `[exit code: N]` marker, since the real presenter consumes that one out of
+  // the body. Named `bash`, so it also covers
+  // the keyed toolview row (turn 60's `fx-bash` covers the render-site fallback
+  // row) — the two chat-row shapes the terminal card renders in.
+  //
+  // Ordered BEFORE the todo turn deliberately: the standing plan retires at the
+  // next `turn/start`, so a turn appended after it would leave the dock's plan
+  // strip empty and take the todo surfaces' own coverage with it.
+  toolTurn(65, 'bash', '{"command":"pnpm run check","cwd":"/tmp/fixture/deep/nested"}', TERMINAL_OUTPUT_FIXTURE)
+
   const todoArgs = JSON.stringify({ todos: fixtureTodos })
-  toolTurn(65, 'todo_write', todoArgs, 'Updated todo list: 1 pending, 1 in progress, 1 completed.')
+  toolTurn(66, 'todo_write', todoArgs, 'Updated todo list: 1 pending, 1 in progress, 1 completed.')
   // The real tool appends the snapshot mid-execution — between tool/call and
   // tool/result — so the fixture reproduces that exact ordering (the last
   // toolTurn events run ... tool/call, tool/result, step/end, turn/end).
@@ -250,7 +322,10 @@ function presentCall(name: string, argsRaw: string): ToolCallView | undefined {
     return undefined
   }
   switch (name) {
+    // Both names present the same terminal card: `fx-bash` lands on the
+    // render-site fallback row, `bash` on the keyed BashRow registration.
     case 'fx-bash':
+    case 'bash':
       return { card: 'terminal', title: str(args.command), cwd: str(args.cwd, '/tmp/fixture'), description: 'fixture 终端样本' }
     case 'fx-write':
       return {
@@ -271,7 +346,10 @@ function presentResult(name: string, argsRaw: string, resultText: string): ToolR
   if (call === undefined) return undefined
   switch (call.card) {
     case 'terminal':
-      return { card: 'terminal', output: resultText, exitCode: 0 }
+      // The sample's own exit status, authored beside it: re-parsing the
+      // trailing marker here would duplicate the bash tool's `parseExitStatus`,
+      // which this client-side fixture cannot import.
+      return { card: 'terminal', output: resultText, ...(TERMINAL_EXIT_STATUS[resultText] ?? { exitCode: 0 }) }
     case 'diff':
       return { card: 'diff', diffs: call.diffs }
     case 'generic':
@@ -943,6 +1021,27 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy {
         if (options.dropSessionCreateResponse) throw new Error('fixture: dropped session.create response after publication')
         return ok(request, { sessionId: created.sessionId })
       },
+      rename: (request) => {
+        const missing = requireSession(request)
+        if (missing !== undefined) return missing
+        const { sessionId, title } = request.payload
+        const normalized = title.trim().replace(/\s+/g, ' ')
+        if (normalized.length === 0) {
+          return err(request, {
+            code: 'title-invalid',
+            message: 'session title must contain visible characters',
+            details: { sessionId },
+          })
+        }
+        // The append emits the session/event and its session/projection frame
+        // (host parallel); the unary response settles the caller first.
+        append(sessionId, {
+          type: 'session/title',
+          data: { title: normalized, messageSeqs: [], source: { kind: 'user' } },
+        })
+        const appended = logOf(sessionId).at(-1) as SessionEvent
+        return ok(request, { title: normalized, seq: appended.seq })
+      },
       history: async (request) => {
         const log = logs.get(request.payload.sessionId) ?? []
         // Snapshot at request time, deliver after the transit delay (mirrors a real host under latency).
@@ -1486,6 +1585,7 @@ export class FixtureApiClient extends AbstractApiClient {
       case 'session.history': return this.api.sessions.history(request)
       case 'session.models': return this.api.sessions.models(request)
       case 'session.selectModel': return this.api.sessions.selectModel(request)
+      case 'session.rename': return this.api.sessions.rename(request)
       case 'session.prompt': return this.api.sessions.prompt(request)
       case 'session.cancel': return this.api.sessions.cancel(request)
       case 'host.describe': return this.api.host.describe(request)
diff --git a/packages/client/connection/tests/fake-api.ts b/packages/client/connection/tests/fake-api.ts
index 6a876c8ed4..650f982b11 100644
--- a/packages/client/connection/tests/fake-api.ts
+++ b/packages/client/connection/tests/fake-api.ts
@@ -45,6 +45,7 @@ export class FakeApiClient implements IApiClient {
   // Programmable slots (defaults answer OK-empty); reassign per case.
   onList: (payload: unknown) => Promise<RpcResponse<{ items: never[] }>> = () => Promise.resolve(ok({ items: [] }))
   onCreate: (payload: unknown) => Promise<RpcResponse<{ sessionId: SessionId }>> = () => Promise.resolve(ok({ sessionId: 'fk-new' as SessionId }))
+  onRename: (payload: unknown) => Promise<RpcResponse<{ title: string; seq: number }>> = () => Promise.resolve(ok({ title: 'fk-renamed', seq: 0 }))
   onHistory: (payload: { sessionId: SessionId; beforeSeq?: number; maxMessages?: number })
   => Promise<RpcResponse<{ events: never[]; hasMore: boolean; modelTarget: ModelTarget }>> =
     () => Promise.resolve(ok({
@@ -96,6 +97,7 @@ export class FakeApiClient implements IApiClient {
     models: (payload: unknown) => this.record('session.models', payload, this.onModels(payload)),
     selectModel: (payload: ModelTarget & { sessionId: SessionId }) =>
       this.record('session.selectModel', payload, this.onSelectModel(payload)),
+    rename: (payload: unknown) => this.record('session.rename', payload, this.onRename(payload)),
     prompt: (payload: unknown) => this.record('session.prompt', payload, this.onPrompt(payload)),
     cancel: (payload: unknown) => this.record('session.cancel', payload, this.onCancel(payload)),
   }
diff --git a/packages/client/connection/tests/fixture.spec.ts b/packages/client/connection/tests/fixture.spec.ts
index 09a3efecd3..ac27323eb9 100644
--- a/packages/client/connection/tests/fixture.spec.ts
+++ b/packages/client/connection/tests/fixture.spec.ts
@@ -464,6 +464,47 @@ describe('createFixtureApi', () => {
     expect(seen.map(f => f.type)).toEqual(['host/workspace-changed', 'host/workspace-changed'])
   })
 
+  it('session.rename covers not-found, blank title, and the accepted append + title frame', async () => {
+    const api = createFixtureApi()
+    const abort = new AbortController()
+    const framesPromise = (async () => {
+      const frames: MuxFrame[] = []
+      for await (const envelope of api.events.mux(req({}), abort.signal)) {
+        frames.push(envelope.payload)
+        if (frames.some(f => f.type === 'session/projection' && f.key === 'title' && f.value === '重命名')) abort.abort()
+      }
+      return frames
+    })()
+    await new Promise(resolve => setTimeout(resolve, 10))
+
+    const missing = await api.sessions.rename(req({ sessionId: sid('fx-void'), title: 'x' }))
+    expect(missing.result).toMatchObject({ ok: false, error: { code: 'session-not-found', details: { sessionId: 'fx-void' } } })
+
+    const blank = await api.sessions.rename(req({ sessionId: sid('fx-alpha'), title: '   ' }))
+    expect(blank.result).toMatchObject({ ok: false, error: { code: 'title-invalid', details: { sessionId: 'fx-alpha' } } })
+
+    const renamed = await api.sessions.rename(req({ sessionId: sid('fx-alpha'), title: '  重命名  ' }))
+    if (!renamed.result.ok) throw new Error('rename failed')
+    expect(renamed.result.value.title).toBe('重命名')
+    const acceptedSeq = renamed.result.value.seq
+    // The response seq addresses the appended title event (the client plane
+    // has no session/title in its event union — titles ride the projection —
+    // so the event is located by seq and its payload checked structurally).
+    const history = await api.sessions.history(req({ sessionId: sid('fx-alpha'), maxMessages: 100 }))
+    if (!history.result.ok) throw new Error('history failed')
+    const appended = history.result.value.events.find(entry => entry.event.seq === acceptedSeq)
+    expect(appended?.event).toMatchObject({
+      type: 'session/title',
+      data: { title: '重命名', messageSeqs: [], source: { kind: 'user' } },
+    })
+    // Beyond the subscribe-time baseline replay, the append emitted exactly
+    // one title projection frame carrying the new value at the response seq.
+    const frames = await framesPromise
+    const titleFrames = frames.filter(f => f.type === 'session/projection' && f.key === 'title' && f.sessionId === sid('fx-alpha') && f.value === '重命名')
+    expect(titleFrames).toHaveLength(1)
+    expect(titleFrames[0]).toMatchObject({ seq: acceptedSeq })
+  })
+
   it('workspace.insertSessionBefore moves, appends, no-ops, and rejects invalid ids', async () => {
     const api = createFixtureApi()
     const wsid = 'fx-ws-fixture' as WorkspaceId
diff --git a/packages/client/hmr/README.i18n.yaml b/packages/client/hmr/README.i18n.yaml
index ce05fdae22..4eef7fa3c2 100644
--- a/packages/client/hmr/README.i18n.yaml
+++ b/packages/client/hmr/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/hmr/README.md
 README.md: 2b2f63c25cbf3a46babef78a4dfb52f859156887
-README.zh.md: 6d94ca4a5e91f390e58575aa4ddf64fc18a509de
+README.zh.md: 58fbad900d9ab86a9d28979f691f24de29e9b6f4
diff --git a/packages/client/hmr/README.zh.md b/packages/client/hmr/README.zh.md
index 6d94ca4a5e..58fbad900d 100644
--- a/packages/client/hmr/README.zh.md
+++ b/packages/client/hmr/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-为通过 fetch 到达的客户端插件提供热重载。该静态到达配置项只组合进 `--dev` 图（`dsh web --dev`）；生产图省略此行，因此外壳打包的代码保持不活动。
+为通过 fetch 加载的客户端插件提供热重载。该静态加载配置项只组合进 `--dev` 图（`dsh web --dev`）；生产图省略该项，因此打包进 shell 的代码保持不活动。
 
-浏览器侧订阅系统 SSE 通道（`GET /plugins/events`），每个 `rebuilt` 帧重载一个插件，并通过队列串行执行（组合包交接 slot 只能容纳一个）。每帧的顺序是：`prefetch`（在触碰任何内容前抓取新组合包）、`invalidate`、`registry.delete`（在 fiber 之前执行：只释放 fiber 会触发 vendored Loader 的 self-dispose 分支，把配置项标为禁用）、排空旧 fiber、删除 `entry.fiber`、移除自身拥有的 `<style data-plugin>` 标签、通过 `entry.refresh()` 重新导入并挂载、以 `fiber.await()` 将启动失败高声重新抛出。依赖方由 cordis 自身重载：fiber 的激活 epoch 会串联其服务提供方的 uid，因此替换提供方 fiber 会级联所有依赖方，无需客户端图分析。node 侧使用一个 interval 检测重建：从同步基线开始 stat-poll 每个图组合包；新增一行后立即重新计算 hash；缺失行保持 dirty；只广播真实 rev 变更。因此，任何生成组合包的 tsdown watch 进程都能触发 HMR，无需 builder→host 通道。
+浏览器侧订阅系统 SSE（Server-Sent Events）通道（`GET /plugins/events`），每个 `rebuilt` 帧重载一个插件，并通过队列串行执行（组合包交接 slot 只能容纳一个）。每帧的顺序是：`prefetch`（在触碰任何内容前抓取新组合包）、`invalidate`、`registry.delete`（在 fiber dispose（资源释放）之前执行：仅 dispose fiber 会触发 vendored Loader 的 self-dispose 分支，把配置项标为禁用）、排空旧 fiber、删除 `entry.fiber`、移除自身拥有的 `<style data-plugin>` 标签、通过 `entry.refresh()` 重新导入并挂载、通过 `fiber.await()` 直接重新抛出启动失败。依赖方由 Cordis 自身重载：fiber 的激活 epoch 会串联其服务提供方的 uid，因此替换提供方 fiber 会级联所有依赖方，无需客户端图分析。node 侧使用一个 interval 检测重建：从同步基线开始 stat-poll 每个图组合包；新增一行后立即重新计算 hash；缺失行保持 dirty；只广播真实 rev 变更。因此，任何生成组合包的 tsdown watch 进程都能触发 HMR（热模块替换），无需 builder→host 通道。
 
 ## 模型体验
 
@@ -12,10 +12,10 @@
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
-- **重载有意保持粗粒度**：会创建全新的 fiber 和组件；重载插件中的 React 状态会丢失，数据层（connection/runtime fiber、Session 对象）不受影响。react-refresh 级状态保留与「重新执行组合包会重新运行 factory」冲突，因此有意排除。
-- **失败时不回滚**：失败的重载会使配置项处于 FAILED 状态，并在 loader 状态投影中高声报告；自动恢复先前组合包会等到实际需要出现后再实现。
-- **重建帧不会刷新图 rev**：陈旧 rev 无害（组合包端点以 no-cache 提供内容）；rev 刷新会随重新连接握手机制落地。
+- **重载有意保持粗粒度**：会创建全新的 fiber 和组件；重载插件中的 React 状态会丢失，数据层（连接 fiber、运行时 fiber 和 Session 对象）不受影响。react-refresh 级状态保留与「重新执行组合包会重新运行 factory」冲突，因此有意排除。
+- **失败时不回滚**：失败的重载会使配置项处于 FAILED 状态，并在 loader 状态投影中明确显示；自动恢复先前组合包会等到实际需要出现后再实现。
+- **重建帧不会刷新图 rev**：陈旧 rev 无害（组合包端点以 no-cache 提供内容）；rev 刷新将在重新连接握手机制中实现。
diff --git a/packages/client/locale/README.i18n.yaml b/packages/client/locale/README.i18n.yaml
index 699c9561cc..1a73e06872 100644
--- a/packages/client/locale/README.i18n.yaml
+++ b/packages/client/locale/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/locale/README.md
 README.md: 9015af2b44a33771b06863ace139fe97695df616
-README.zh.md: 6b129bcabbef5b5a00c5073ebc9142a0e406ddba
+README.zh.md: 12205e21bb75a4433902b8e85c1cf7bdb0147bbf
diff --git a/packages/client/locale/README.zh.md b/packages/client/locale/README.zh.md
index 6b129bcabb..12205e21bb 100644
--- a/packages/client/locale/README.zh.md
+++ b/packages/client/locale/README.zh.md
@@ -10,9 +10,9 @@ locale 插件：LocaleService 包含浏览器 locale 偏好（`zh`／`en`，以
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
 - **只有设置界面完成翻译**：其他页面仍保留内联文案；将全仓文案提取到字典的工作暂缓。
-- **切换 locale 只重新渲染已订阅的消费方**：未接入 `locale/change` 的分区会保留已渲染文本，直到重新挂载。
+- **切换 locale 只重新渲染已订阅的消费方**：未接入 `locale/change` 的界面区域会保留已渲染文本，直到重新挂载。
diff --git a/packages/client/modules/README.i18n.yaml b/packages/client/modules/README.i18n.yaml
index 9175950a50..80bf46a996 100644
--- a/packages/client/modules/README.i18n.yaml
+++ b/packages/client/modules/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/modules/README.md
 README.md: efba9e2eb0b148677fc7ac18bfad6333fb6f80da
-README.zh.md: 7d1aa8af08256c47c1ae65343e46c30e910128d0
+README.zh.md: b057bfdd8c0a269252496d0c6a0fc4184932fd72
diff --git a/packages/client/modules/README.zh.md b/packages/client/modules/README.zh.md
index 7d1aa8af08..b057bfdd8c 100644
--- a/packages/client/modules/README.zh.md
+++ b/packages/client/modules/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-客户端模块系统：Node 内部 ESM loader 的浏览器端对等实现，以惰性 CJS 表构建。web 外壳挂载 vendored cordis Loader 来治理配置项（fiber 生命周期、inject 等待、update/refresh），并把该包的 `ClientModuleLoader` 作为其 `internal` seam 注入；vendored 一侧唯一的消费点是 `EntryTree.import`，因此替换 `internal` 恰好只会替换「插件代码如何到达」，不会改变其他内容。
+客户端模块系统：Node 内部 ESM loader 的浏览器端对等实现，以惰性 CJS 表实现。web 外壳挂载 vendored cordis Loader 来治理配置项（fiber 生命周期、inject 等待、update/refresh），并把该包（package）的 `ClientModuleLoader` 作为其 `internal` seam 注入；vendored 一侧唯一的消费点是 `EntryTree.import`，因此替换 `internal` 恰好只会替换「插件代码如何到达」，不会改变其他内容。
 
-惰性 CJS 模型（web2）：执行插件组合包只会注册其 factory（`window.__ModuleLoader__.load({id, factory})`）；每个模块主体的副作用（包括 CSS 注入）都位于 factory 闭包中，在物化时运行（`factory(require)` → 导出表层，并在 `loadCache` 中记忆化），不会在脚本执行时运行。如果 factory 请求另一个已注册但尚未物化的模块，系统会递归物化它，因此加载顺序无需外部编排；require 循环会抛出异常（factory 形式的 CJS 无法交付部分导出）。`<id>/client` 与裸 id 指向同一表层（一个插件组合包就是其包的客户端侧）。
+惰性 CJS 模型（web2）：执行插件组合包只会注册其 factory（`window.__ModuleLoader__.load({id, factory})`）；每个模块主体的副作用（包括 CSS 注入）都位于 factory 闭包中，在物化时运行（`factory(require)` → 导出表层，并在 `loadCache` 中记忆化），不会在脚本执行时运行。如果 factory 依赖另一个已注册但尚未物化的模块，系统会递归物化它，因此加载顺序无需外部编排；require 循环会抛出异常（factory 形式的 CJS 无法提供部分导出）。`<id>/client` 与裸 id 指向同一表层（一个插件组合包就是其包的客户端侧）。
 
-解析分支顺序（`import(specifier)`）：平台种子词 → 外壳实例；记忆化记录 → 表层；外壳自身的静态注册表（`registerStatic`，app-shell）→ 模块；已注册 factory → 物化；图行（`window.__DSH_BOOT__`）→ 抓取 + 执行 + 物化；其他情况一律抛出异常。这是构建时组合包纯度门禁的运行时镜像。交给 factory 的同步 `require` 采用相同顺序，但不含抓取分支，并把观察到的边记录到模块记录中。`prefetch` 是第一阶段到达 hook（抓取 + 执行，只注册；并发调用共享一个进行中的 task）；`invalidate` 会丢弃 factory 与物化记录，使下一次 prefetch/import 重新抓取（HMR hook）。
+解析分支顺序（`import(specifier)`）：平台种子词 → 外壳实例；记忆化记录 → 表层；外壳自身的静态注册表（`registerStatic`，app-shell）→ 模块；已注册 factory → 物化；模块图记录（`window.__DSH_BOOT__`）→ 抓取 + 执行 + 物化；其他情况一律抛出异常。这是构建时组合包纯度门禁的运行时镜像。交给 factory 的同步 `require` 采用相同顺序，但不含抓取分支，并把观察到的边记录到模块记录中。`prefetch` 是第一阶段加载钩子（抓取 + 执行，只注册；并发调用共享一个进行中的任务）；`invalidate` 会丢弃 factory 与物化记录，使下一次 prefetch/import 重新抓取；它是 HMR（热模块替换）钩子。
 
 ## 模型体验
 
@@ -18,5 +18,5 @@
 
 ## 已知限制与暂缓事项
 
-- **有意采用扁平模块图**：每个组合包是一个模块节点，其边只指向表叶；接口（loadCache/edges/invalidate）按通用模块图塑形，因此可以改变 externalization 粒度而不更改接口。
-- **自身不记录卸载账目**：样式移除与 fiber 拆卸顺序属于 HMR 驱动器（`@deepseek-ai/dsh-client-hmr`）；loader 只逐记录清点自身拥有的样式标签 id。
+- **有意采用扁平模块图**：每个组合包是一个模块节点，其边只指向表中的叶节点；接口（loadCache/edges/invalidate）按通用模块图塑形，因此可以改变 externalization 粒度而不更改接口。
+- **自身不记录卸载账目**：样式移除与 fiber 拆卸顺序属于 HMR 驱动器（`@deepseek-ai/dsh-client-hmr`）；loader 只在每条记录中登记其拥有的样式标签 id。
diff --git a/packages/client/runtime/README.i18n.yaml b/packages/client/runtime/README.i18n.yaml
index 429ae8f0a9..157083018a 100644
--- a/packages/client/runtime/README.i18n.yaml
+++ b/packages/client/runtime/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/runtime/README.md
-README.md: 25eb60e2c95059ae918669c9f5169b6b8e9c6816
-README.zh.md: e3085f91750503aeaffda41d86c40c62943b4ba9
+README.md: b51cc0276d8635ea9faa506e30246a107c1c1418
+README.zh.md: 4b2248d875ae37f1b848c51a0009d2497c6b3e61
diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md
index 25eb60e2c9..b51cc0276d 100644
--- a/packages/client/runtime/README.md
+++ b/packages/client/runtime/README.md
@@ -2,7 +2,7 @@
 
 English | [中文](README.zh.md)
 
-Client cordis boot and React-free object services: SlotsService wraps SlotCore and supplies renderer data sources; SessionsService owns Session objects, list/scope/history state; WorkspacesService depends on SessionsService and owns Workspace objects, list/actions, default-target derivation, and the New Session blank-reuse entry (`connectWorkspace`). The runtime fans the shared Host stream into both managers. Client sessions are always Host-born (Session+Agent+cwd in one `session.create`); the client holds no pre-entity session state — a session's Agent scope (the client mirror of host dsh-scope, keyed by the shared agent/session id) is born when its row enters the list mirror and dies with the prune. Contract: api-contracts v3 §4. Each `Session` holds a generic `ProjectionValueStore` seeded from the history-tail `projections` block and updated by `session/projection` frames under higher-seq-wins; domain keys (including `todos`) are read via `projections.faceOf` / `useProjection`, not via `ConversationSnapshot`.
+Client cordis boot and React-free object services: SlotsService wraps SlotCore and supplies renderer data sources; SessionsService owns Session objects and the Chat-facing list, scope, and event-window state; SessionHistoryService lazily owns independent raw-history ledgers for inspection consumers; WorkspacesService depends on SessionsService and owns Workspace objects, list/actions, default-target derivation, and the New Session blank-reuse entry (`connectWorkspace`). The runtime fans the shared Host stream into the Session, Workspace, and activated history owners without routing inspection state through Session or SessionManager. Client sessions are always Host-born (Session+Agent+cwd in one `session.create`); the client holds no pre-entity session state — a session's Agent scope (the client mirror of host dsh-scope, keyed by the shared agent/session id) is born when its row enters the list mirror and dies with the prune. Contract: api-contracts v3 §4. Each `Session` holds a generic `ProjectionValueStore` seeded from the history-tail `projections` block and updated by `session/projection` frames under higher-seq-wins; domain keys (including `todos`) are read via `projections.faceOf` / `useProjection`, not via `ConversationSnapshot`.
 
 ## Workspace and Session lists
 
@@ -22,7 +22,7 @@ SlotsService gives the renderer separate bare observables for `useSessions` and
 
 ## Session title projection
 
-`SessionManager` retains the latest validated `session/title` control snapshot independently of list and session-instance arrival. Newer event seqs replace older snapshots, title timestamps contribute to list recency, and a subscription baseline discards any retained title beyond its `lastSeq` before the optional folded title arrives. Explicit session removal also clears the retained title. The client-facing `SessionSummary.title` is therefore only the actual durable title; `displayTitle` is always present and falls back through the cwd basename and session id. A cold persisted session keeps that fallback until opening or resuming it causes the host to fold and project its log-backed title.
+`SessionManager` retains the latest validated `session/title` control snapshot independently of list and session-instance arrival. Newer event seqs replace older snapshots, title timestamps contribute to list recency, and a subscription baseline discards any retained title beyond its `lastSeq` before the optional folded title arrives. Explicit session removal also clears the retained title. The client-facing `SessionSummary.title` is therefore only the actual durable title; `displayTitle` is always present and falls back through the cwd basename and session id. A cold persisted session keeps that fallback until opening or resuming it causes the host to fold and project its log-backed title. `ISession.rename` settles the `title` projection cell directly from the unary response's `{title, seq}` under the same higher-seq-wins rule — the list row and every `useProjection('title')` reader update ahead of the push frame, whose later replay of the same seq is a no-op.
 
 ## Session model selection
 
diff --git a/packages/client/runtime/README.zh.md b/packages/client/runtime/README.zh.md
index e3085f9175..4b2248d875 100644
--- a/packages/client/runtime/README.zh.md
+++ b/packages/client/runtime/README.zh.md
@@ -2,19 +2,19 @@
 
 [English](README.md) | 中文
 
-客户端 cordis 启动与不依赖 React 的对象服务：SlotsService 包装 SlotCore 并提供 renderer 数据源；SessionsService 拥有 Session 对象、列表／scope／history 状态；WorkspacesService 依赖 SessionsService，拥有 Workspace 对象、列表／操作、默认目标派生，以及 New Session 空会话复用入口（`connectWorkspace`）。运行时把共享 Host 流分发给两个 manager。客户端 Session 一律由 Host 出生（一次 `session.create` 同瞬产出 Session+Agent+cwd）；客户端不持有任何实体化之前的会话状态——Agent scope（host dsh-scope 的客户端镜像，以 agent/session 共用 id 为键）在会话行进入列表镜像时出生，随 prune 死亡。契约：api-contracts v3 §4。每个 `Session` 持有一个通用的 `ProjectionValueStore`，由历史尾页的 `projections` 块播种，并经 `session/projection` 帧按 seq 高者胜更新；领域键（含 `todos`）经 `projections.faceOf`／`useProjection` 读取，不经 `ConversationSnapshot`。
+客户端 cordis 启动与不依赖 React 的对象服务：SlotsService 包装 SlotCore 并提供 renderer 数据源；SessionsService 拥有 Session 对象以及 Chat 所需的列表、scope 和事件窗口状态；SessionHistoryService 为检查类消费方惰性拥有彼此独立的原始历史账本；WorkspacesService 依赖 SessionsService，拥有 Workspace 对象、列表／操作、默认目标派生，以及 New Session 空会话复用入口（`connectWorkspace`）。运行时把共享 Host 流分发给 Session、Workspace 和已激活的历史数据所有者，不让检查状态经过 Session 或 SessionManager。客户端会话一律由 Host 创建（一次 `session.create` 同时产生 Session、agent（智能体）和 cwd）；客户端不持有任何实体化之前的会话状态——agent scope（host dsh-scope 的客户端镜像，以 agent/session 共用 id 为键）在会话行进入列表镜像时创建，并随 prune 销毁。契约：api-contracts v3 §4。每个 `Session` 持有一个通用的 `ProjectionValueStore`，由历史记录尾部的 `projections` 块播种，并经 `session/projection` 帧按 seq 高者胜更新；领域键（含 `todos`）经 `projections.faceOf`／`useProjection` 读取，不经 `ConversationSnapshot`。
 
 ## Workspace 与 Session 列表
 
-Workspace 和 Session 列表各自具有单调的 `pending` → `ready` 基线阶段，也有各自的刷新活动／错误状态。列表请求期间到达的增量更新／移除帧与一元变更回显会在其响应之上回放。第一次成功的基线建立 Host 顺序；后续刷新更新行和成员关系，但不改变已经显示的标识之间的相对顺序。已移除的 Workspace id 会保留进程本地删除标记，避免延迟到达的 changed 帧将其复活；重连仍以 `workspace.list` 作为基线。Workspace 新近程度只在两条基线都 ready 后派生，且绝不改变 Workspace 列表顺序。
+Workspace 和 Session 列表各自具有单调的 `pending` → `ready` 基线阶段，也有各自的刷新活动／错误状态。列表请求期间到达的增量插入或更新／移除帧与一元变更回显会在其响应之上回放。第一次成功的基线建立 Host 顺序；后续刷新更新行和成员关系，但不改变已经显示的标识之间的相对顺序。已移除的 Workspace id 会保留进程本地删除标记，避免延迟到达的 changed 帧将其复活；重连仍以 `workspace.list` 作为基线。Workspace 新近程度只在两条基线都 ready 后派生，且绝不改变 Workspace 列表顺序。
 
-`WorkspacesService.delete(workspaceId)` 在一元响应成功后从客户端投影中移除注册记录；对应的 `host/workspace-removed` 帧具有幂等性，并负责同步其他标签页。Session 状态与当前 Session selection 相互独立，因此 Workspace 消失后，其已记账的 Session 会立即投影到 Ungrouped 下。
+`WorkspacesService.delete(workspaceId)` 在一元响应成功后从客户端投影中移除注册记录；对应的 `host/workspace-removed` 帧具有幂等性，并负责同步其他标签页。Session 状态与当前 Session selection 相互独立，因此 Workspace 消失后，其已纳入客户端投影的 Session 会立即投影到 Ungrouped 下。
 
-SlotsService 分别为 renderer 提供 `useSessions` 与 `useWorkspaces` 的裸 observable；web-react 创建 hook。Workspace 业务状态不会进入 `SessionListState` 或配置项 store。
+SlotsService 分别为 renderer 提供 `useSessions` 与 `useWorkspaces` 的裸 observable；web-react 创建钩子。Workspace 业务状态不会进入 `SessionListState` 或配置项 store。
 
 ## New Session 与 blank 镜像
 
-`WorkspacesService.connectWorkspace(workspaceId)` 解析 New Session 流程最终落入的会话：先在列表镜像中复用该 workspace 的既有空会话（`blank && cwd == workspace.path`），未命中则调用 `session.create({workspaceId})`，返回会话 id 由调用方 open。`SessionSummary.blank` 镜像主机派生的空日志位，在客户端只降不升：由 `session.list`／`host/session-added` 帧播种，本地首次**受理成功**的 `prompt()`（RPC 成功响应时——受理即证明用户消息已入主机日志；首讯被拒则会话保持 blank、保持可复用）与任何 `running: true` 状态帧翻为 false，每次列表重拉重新对齐。列表表面隐藏 blank 行；store 保留全部行。`SessionsService.create` 接受可选的、由调用方预先分配的 SessionId，失败时抛出 `SessionCreateError`（携带 `requestedSessionId`）。
+`WorkspacesService.connectWorkspace(workspaceId)` 解析 New Session 流程最终落入的会话：先在列表镜像中复用该 workspace 的既有空会话（`blank && cwd == workspace.path`），未命中则调用 `session.create({workspaceId})`，返回会话 id 由调用方 open。`SessionSummary.blank` 镜像主机派生的空日志位，在客户端只降不升：由 `session.list`／`host/session-added` 帧播种，本地首次获 Host 接受的 `prompt()`（RPC 成功响应时——受理即证明用户消息已入主机日志；首讯被拒则会话保持 blank、保持可复用）与任何 `running: true` 状态帧翻为 false，每次列表重拉重新对齐。列表界面隐藏 blank 行；store 保留全部行。`SessionsService.create` 接受可选的、由调用方预先分配的 SessionId，失败时抛出 `SessionCreateError`（携带 `requestedSessionId`）。
 
 ## Code Mode 子调用索引
 
@@ -22,7 +22,7 @@ SlotsService 分别为 renderer 提供 `useSessions` 与 `useWorkspaces` 的裸
 
 ## Session 标题投影
 
-`SessionManager` 独立于列表和 Session 实例到达情况，保留最近一次通过验证的 `session/title` 控制快照。seq 更新的事件会替换旧快照，标题时间戳计入列表新近程度；订阅基线会先丢弃 seq 超过其 `lastSeq` 的任何已保留标题，再接收可选的折叠标题。显式移除 Session 也会清除已保留标题。因此，面向客户端的 `SessionSummary.title` 只包含真实的持久标题；`displayTitle` 始终存在，并依次回退到 cwd basename 和 Session id。冷启动的持久会话会保持该回退值，直到打开或恢复会话，促使主机折叠并投影日志支持的标题。
+`SessionManager` 独立于列表和 Session 实例到达情况，保留最近一次通过验证的 `session/title` 控制快照。seq 更高的事件会替换旧快照，标题时间戳计入列表新近程度；订阅基线会先丢弃 seq 超过其 `lastSeq` 的任何已保留标题，再接收可选的折叠标题。显式移除 Session 也会清除已保留标题。因此，面向客户端的 `SessionSummary.title` 只包含实际的持久化标题；`displayTitle` 始终存在，并依次回退到 cwd basename 和 Session id。冷态持久化会话会保持该回退值，直到打开或恢复会话，促使主机折叠并投影由日志支撑的标题。`ISession.rename` 用 unary 响应中的 `{title, seq}` 直接结算 `title` 投影格，遵循同一 seq 高者胜规则——列表行和所有 `useProjection('title')` 读者在推送帧到达前即更新；推送帧随后重放同一 seq 时为无操作。
 
 ## 会话模型选择
 
@@ -30,14 +30,14 @@ SlotsService 分别为 renderer 提供 `useSessions` 与 `useWorkspaces` 的裸
 
 ## 模型体验
 
-无，因为 Session 对象层会选择后续 Host 请求使用的提供方/模型路由，但不添加任何模型可见内容。
+无，因为会话对象层会选择后续 Host 请求使用的提供方/模型路由，但不添加任何模型可见内容。
 
 #### KV Cache 影响
 
-更改目标可能改变提供方侧的缓存复用，或使其失效；该包本身不会改变提示词前缀。
+更改目标可能改变提供方侧的缓存复用，或使其失效；该包（package）本身不会改变提示词前缀。
 
 ## 已知限制与暂缓事项
 
-- **`loader.unload` 是 stub（抛出 not-implemented）**：完整链路（fiber 释放 → 注册级联 → 样式移除）随 HMR 项目落地。
-- **scope 拆卸由阶段驱动，目前只能有一个占用者**：已 staged 的 Session 精确跟随 `list.current`（staging 就是打开信号：事件窗口打开 ⟺ Session 位于 stage）；在 staged 状态下被移除的 Session，其 scope 会冻结保留，直到 stage 转向其他 Session，而非直到真实观察者数量降为零。解析（`binding()`／`scope()`）只是纯寻址，可安全用于渲染；渲染层经 `currentProvideInfo` observable 读取当前 bundle。并发 pane 落地时，staged 状态可以扩展为多 pane 列表。
-- **插件组合包从该包执行值导入时必须使用 `/client` 子路径**：裸包名不在 loader external 表中，会内联第二个模块实例；其私有 scope-tag Symbol 永远无法匹配（空状态 P0 事故复盘）。
+- **`loader.unload` 是 stub（抛出 not-implemented）**：完整链路（fiber dispose（资源释放） → 注册级联 → 样式移除）随 HMR（热模块替换）项目落地。
+- **scope 拆卸由阶段驱动，目前只能有一个占用者**：已 staged 的会话精确跟随 `list.current`（staging 就是打开信号：事件窗口打开 ⟺ 会话位于 stage）；在 staged 状态下被移除的会话，其 scope 会冻结保留，直到 stage 转向其他会话，而非直到真实观察者数量降为零。解析（`binding()`／`scope()`）只是纯寻址，可安全用于渲染；渲染层经 `currentProvideInfo` observable 读取当前 bundle。并发 pane 落地时，staged 状态可以扩展为多 pane 列表。
+- **插件组合包从该包导入值时必须使用 `/client` 子路径**：裸包名不在 loader externals 表中，会内联第二个模块实例；其私有 scope-tag Symbol 永远无法匹配。这是空状态 P0 的事故复盘（postmortem）所记录的问题。
diff --git a/packages/client/runtime/src/client/contract/session-history.ts b/packages/client/runtime/src/client/contract/session-history.ts
new file mode 100644
index 0000000000..2b6679fd05
--- /dev/null
+++ b/packages/client/runtime/src/client/contract/session-history.ts
@@ -0,0 +1,35 @@
+import type {
+  RpcError, SessionId,
+} from '@deepseek-ai/dsh-client-connection/client'
+import type { SessionHistoryInspection } from '../sessions/history.ts'
+import type { ObservableSnapshot } from './store.ts'
+
+/** Observable state of one independently loaded session history ledger. */
+export interface SessionHistorySnapshot {
+  state: 'cold' | 'loading' | 'ready' | 'error'
+  error: RpcError | null
+  hasMore: boolean
+  inspection: SessionHistoryInspection
+}
+
+/** Read-only history source addressed by session id. */
+export interface SessionHistoryFace
+  extends ObservableSnapshot<SessionHistorySnapshot> {
+  readonly sessionId: SessionId
+  /**
+   * Load the tail and exhaust every available older page.
+   * @param signal - Consumer lifetime; abort is observed between page requests.
+   * @returns When the available ledger is complete or stops advancing.
+   */
+  loadAll(signal?: AbortSignal): Promise<void>
+}
+
+/** Runtime service resolving independent history sources. */
+export interface ISessionHistory {
+  /**
+   * Resolve the identity-stable source for a session.
+   * @param sessionId - Host session identity.
+   * @returns The source owned outside Session and SessionManager.
+   */
+  source(sessionId: SessionId): SessionHistoryFace
+}
diff --git a/packages/client/runtime/src/client/contract/session.ts b/packages/client/runtime/src/client/contract/session.ts
index 31335c388c..2ba8725b02 100644
--- a/packages/client/runtime/src/client/contract/session.ts
+++ b/packages/client/runtime/src/client/contract/session.ts
@@ -41,6 +41,13 @@ export interface ISession {
    * @returns acceptance, or the business error.
    */
   cancel(): Promise<RpcResult<{ accepted: true }>>
+  /**
+   * Rename this session (explicit user title; pins it against automatic
+   * regeneration).
+   * @param title - raw title text (the host normalizes acceptance).
+   * @returns the normalized accepted title and its event seq, or the business error.
+   */
+  rename(title: string): Promise<RpcResult<{ title: string; seq: number }>>
   /**
    * Extend the history window backwards (older messages pagination).
    * @returns completion; failures land in snapshot.openState/loadingOlder.
diff --git a/packages/client/runtime/src/client/index.ts b/packages/client/runtime/src/client/index.ts
index 3b8e02b5a9..f359eb1eac 100644
--- a/packages/client/runtime/src/client/index.ts
+++ b/packages/client/runtime/src/client/index.ts
@@ -5,6 +5,7 @@ import type { MaybeSnapshotSelectorHook, SnapshotSelectorHook } from '@deepseek-
 import { SlotsService } from './slots.ts'
 import { SessionsService } from './sessions/service.ts'
 import type { SessionListState } from './sessions/service.ts'
+import { SessionHistoryService } from './session-history/service.ts'
 import { WorkspacesService } from './workspaces/service.ts'
 import type { ConversationSnapshot, RunningToolCall, ToolResultNode } from './sessions/conversation.ts'
 import type { UseProjection } from './sessions/projection-store.ts'
@@ -12,6 +13,7 @@ import type { UseProjection } from './sessions/projection-store.ts'
 export { SlotsService } from './slots.ts'
 export type { RootOwnerProps } from './slots.ts'
 export { SessionCreateError, SessionsService, scopeOf, workspaceTitleOf } from './sessions/service.ts'
+export { SessionHistoryService } from './session-history/service.ts'
 // The provide channel is shared with the client test runtime (one
 // materialization/projection implementation; no test-side mirror to drift).
 export { SessionProvideChannel } from './sessions/provide.ts'
@@ -21,6 +23,9 @@ export type { AgentScopeHandle } from './agents/scope.ts'
 export { DirectoryBrowseError, WorkspaceCreateError, WorkspacesService } from './workspaces/service.ts'
 export type { Session } from './sessions/session.ts'
 export type { ISession, ProjectionsFace, SessionFace } from './contract/session.ts'
+export type {
+  ISessionHistory, SessionHistoryFace, SessionHistorySnapshot,
+} from './contract/session-history.ts'
 export type { ISessions } from './contract/sessions.ts'
 export type { IWorkspaces } from './contract/workspaces.ts'
 export type {
@@ -38,10 +43,19 @@ export type {
   EngineStoreHandle, EngineStoreInstance, ObservableSnapshot, SnapshotStore,
 } from './contract/store.ts'
 export type {
-  AssistantBlock, AssistantMessageNode, CodeSubCall, CommandNode, ComposerPhase, ContextMessageNode, ConversationNode,
+  AssistantBlock, AssistantMessageNode, AssistantProvenanceView, AssistantRequestConfig,
+  AssistantTiming, CodeSubCall, CommandNode, ComposerPhase, ContextMessageNode, ConversationNode,
   ConversationSnapshot, QueuedMessage, RunningToolCall,
   SteeringMessageNode, TodoItem, ToolResultNode, UnknownSurfaceNode, UserMessageNode,
 } from './sessions/conversation.ts'
+export type {
+  ConversationContext, ConversationContextOriginKind,
+} from './sessions/conversation-context.ts'
+export type {
+  ConversationPromptSnapshot, RequestInspectionSnapshot, RequestPromptChange, RequestView,
+} from './sessions/request-inspection.ts'
+export type { ConversationHistoryProjection } from './session-history/history-fold.ts'
+export type { SessionHistoryInspection } from './sessions/history.ts'
 export { PendingWait } from './sessions/pending.ts'
 export type { PendingInteraction, PendingKind, PendingPayloads } from './sessions/pending.ts'
 // Projection value store (session-projection RFC, push model): host-computed
@@ -120,6 +134,8 @@ declare module 'cordis' {
     slots: import('./slots.ts').SlotsService
     /** The outward face only; the concrete service stays inside the runtime. */
     sessions: import('./contract/sessions.ts').ISessions
+    /** Read-only history sources isolated from Chat sessions and workspace state. */
+    sessionHistory: import('./contract/session-history.ts').ISessionHistory
     /** The outward face only; the concrete service stays inside the runtime. */
     workspaces: import('./contract/workspaces.ts').IWorkspaces
   }
@@ -135,30 +151,55 @@ export function apply(ctx: Context): void {
   ctx.plugin(SlotsService)
   const connection = ctx.get('connection') as ConnectionHandle
   const sessions = new SessionsService(ctx, connection.api)
+  const sessionHistory = new SessionHistoryService(ctx, connection.api)
   const workspaces = new WorkspacesService(ctx, connection.api, sessions)
   ctx.effect(
     () => workspaces.startInitialSelection(),
     'runtime: initial Workspace selection',
   )
   const loop = connection.start({
-    onMuxEnvelope: (envelope) => { sessions.handleMuxEnvelope(envelope) },
+    onMuxEnvelope: (envelope) => {
+      sessions.handleMuxEnvelope(envelope)
+      try {
+        sessionHistory.handleMuxEnvelope(envelope)
+      } catch (error) {
+        console.error('[web-runtime] history frame routing failed:', error)
+      }
+    },
     onHostEnvelope: (envelope) => {
       sessions.handleHostEnvelope(envelope)
       workspaces.handleHostEnvelope(envelope)
       // Typed-event bridge: the session layer ignores registry frames (no
       // session routing); consumers (command directory caches) subscribe on ctx.
       if (envelope.payload.type === 'host/commands-changed') ctx.emit('commands/changed')
+      try {
+        sessionHistory.handleHostEnvelope(envelope)
+      } catch (error) {
+        console.error('[web-runtime] history host-frame routing failed:', error)
+      }
     },
     onConnected: () => {
       sessions.handleConnected()
       workspaces.handleConnected()
       ctx.emit('connection/reset')
+      try {
+        sessionHistory.handleConnected()
+      } catch (error) {
+        console.error('[web-runtime] history reconnect failed:', error)
+      }
     },
     onStateChange: (state) => {
       // Generation death fires before any next-generation frame can arrive
       // (reconnect replays flow from stream open, ahead of onConnected):
       // the only safe moment to drop generation-scoped interaction state.
-      if (state === 'reconnecting') sessions.handleDisconnected()
+      if (state === 'reconnecting') {
+        sessions.handleDisconnected()
+        try {
+          sessionHistory.handleDisconnected()
+        } catch (error) {
+          console.error('[web-runtime] history disconnect failed:', error)
+        }
+      }
     },
   })
   ctx.effect(() => () => { loop.stop() }, 'runtime: connection stream loop')
diff --git a/packages/client/runtime/src/client/session-history/history-fold.ts b/packages/client/runtime/src/client/session-history/history-fold.ts
new file mode 100644
index 0000000000..e7f9922425
--- /dev/null
+++ b/packages/client/runtime/src/client/session-history/history-fold.ts
@@ -0,0 +1,472 @@
+import type { ContentBlock } from '@deepseek-ai/dsh-llm/types'
+import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
+import {
+  SurfaceManager, isSurfaceEligibleType, isSurfaceEvent,
+} from '@deepseek-ai/dsh-session/surface'
+import type {
+  HistoryEntry, ToolCallView, ToolResultView,
+} from '@deepseek-ai/dsh-client-connection/client'
+import type {
+  AssistantRequestConfig, AssistantTiming, CodeSubCall, ConversationNode,
+  PartialAssistant, RunningToolCall,
+} from '../sessions/conversation.ts'
+import { toAssistantBlocks } from '../sessions/conversation.ts'
+import type {
+  ConversationContext, ConversationContextOriginKind,
+} from '../sessions/conversation-context.ts'
+import type { ConversationPromptSnapshot } from '../sessions/request-inspection.ts'
+import { PartialAccumulator } from '../sessions/partial.ts'
+
+interface CallIndexEntry {
+  name: string
+  argsRaw: string
+  time: number
+  callView: ToolCallView | null
+}
+
+interface FoldedContext {
+  generation: number
+  nodes: readonly number[]
+  originSeq?: number
+}
+
+interface AssistantStepMetadata {
+  stepStartTime: number | null
+  firstTokenTime: number | null
+}
+
+/** Immutable conversation projections derived only from the history source. */
+export interface ConversationHistoryProjection {
+  eventNodes: readonly ConversationNode[]
+  contexts: readonly ConversationContext[]
+  interruptedNodes: readonly ConversationNode[]
+  partial: PartialAssistant | null
+  runningCalls: readonly RunningToolCall[]
+  codeDispatches: ReadonlyMap<string, readonly CodeSubCall[]>
+}
+
+function assistantStepKey(turn: number, step: number): string {
+  return `${turn}\u0000${step}`
+}
+
+// Trajectory owns surface-window reconstruction so its immutable ledger does
+// not depend on Chat's live fold adapter or Session's mutable state.
+/* jscpd:ignore-start */
+function paddingEvent(seq: number): SessionEvent {
+  return { type: 'noop/padding', seq, time: 0, data: {} } as unknown as SessionEvent
+}
+
+function replacementCrossesWindowHead(event: SessionEvent, baseSeq: number): boolean {
+  if (!isSurfaceEvent(event) || event.surfaceOp === 'append') return false
+  return event.surfaceOp.start < baseSeq || event.surfaceOp.end < baseSeq
+}
+/* jscpd:ignore-end */
+
+function contextOriginKind(event: SessionEvent | undefined): ConversationContextOriginKind {
+  if (event?.type !== 'user/message') return 'rewrite'
+  const source = event.data.source
+  if (typeof source === 'object' && 'kind' in source && 'plugin' in source) {
+    if (source.plugin === 'compact') return 'compaction'
+    if (source.plugin === 'rewind') return 'rewind'
+  }
+  return 'rewrite'
+}
+
+function isTokenDelta(chunk: SessionEvent<'assistant/chunk'>['data']['chunk']): boolean {
+  switch (chunk.type) {
+    case 'text-delta':
+    case 'reasoning-delta':
+      return chunk.text !== ''
+    case 'tool-call-delta':
+      return chunk.argumentsDelta !== '' || chunk.name !== undefined
+    default:
+      return false
+  }
+}
+
+function foldContexts(events: readonly SessionEvent[]): readonly FoldedContext[] {
+  const replay: SessionEvent[] = []
+  const surface = new SurfaceManager(replay)
+  const contexts: FoldedContext[] = []
+  let generation = 0
+  let originSeq: number | undefined
+  for (const event of events) {
+    if (isSurfaceEvent(event) && event.surfaceOp !== 'append') {
+      contexts.push({
+        generation,
+        nodes: [...surface.nodes],
+        ...(originSeq === undefined ? {} : { originSeq }),
+      })
+      generation++
+      originSeq = event.seq
+    }
+    replay.push(event)
+  }
+  contexts.push({
+    generation,
+    nodes: [...surface.nodes],
+    ...(originSeq === undefined ? {} : { originSeq }),
+  })
+  return contexts
+}
+
+// History projection owns its node mapping so Chat's live adapter remains free
+// of inspection metadata and lifecycle coupling.
+/* jscpd:ignore-start */
+function materializeNode(
+  event: SessionEvent,
+  callIndex: ReadonlyMap<string, CallIndexEntry>,
+  resultView: ToolResultView | null,
+  assistantTiming: AssistantTiming | undefined,
+  requestConfig: AssistantRequestConfig | undefined,
+): ConversationNode {
+  switch (event.type) {
+    case 'user/message':
+      if (event.data.source.kind !== 'user') {
+        return {
+          kind: 'context', seq: event.seq, time: event.time,
+          content: event.data.content, source: event.data.source,
+        }
+      }
+      return {
+        kind: 'user', seq: event.seq, time: event.time,
+        content: event.data.content, source: event.data.source,
+      }
+    case 'assistant/message':
+      return {
+        kind: 'assistant', seq: event.seq, time: event.time,
+        turn: event.data.turn, step: event.data.step,
+        blocks: toAssistantBlocks(event.data.message.content), usage: event.data.usage,
+        provenance: {
+          provider: event.data.message.source.provider,
+          model: event.data.message.source.model,
+        },
+        ...(requestConfig === undefined ? {} : { requestConfig }),
+        ...(assistantTiming === undefined ? {} : { timing: assistantTiming }),
+      }
+    case 'steering/message':
+      return {
+        kind: 'steering', seq: event.seq, time: event.time, turn: event.data.turn,
+        content: event.data.message.content, source: event.data.message.source,
+      }
+    case 'tool/result': {
+      const result = event.data.message.content[0]
+      const callId = String(event.data.message.source.callId)
+      const call = callIndex.get(callId)
+      return {
+        kind: 'tool-result', seq: event.seq, time: event.time,
+        callId,
+        call: call === undefined ? null : { name: call.name, argsRaw: call.argsRaw },
+        callTime: call?.time ?? null,
+        content: result.content, isError: result.isError === true,
+        ...(event.data.error === undefined ? {} : { error: event.data.error }),
+        meta: event.data.meta,
+        callView: call?.callView ?? null,
+        resultView,
+      }
+    }
+    default:
+      return {
+        kind: 'unknown', seq: event.seq, time: event.time,
+        type: event.type, data: (event as { data?: unknown }).data,
+      }
+  }
+}
+/* jscpd:ignore-end */
+
+function projectTransient(entries: readonly HistoryEntry[]): Pick<
+  ConversationHistoryProjection,
+  'interruptedNodes' | 'partial' | 'runningCalls' | 'codeDispatches'
+> {
+  let partial: PartialAccumulator | null = null
+  const openCalls = new Map<string, RunningToolCall>()
+  const interruptedNodes: ConversationNode[] = []
+  const codeDispatches = new Map<string, readonly CodeSubCall[]>()
+
+  for (const entry of entries) {
+    const { event } = entry
+    if ((event.type as string) === 'tool/code-dispatch-start') {
+      const data = event.data as unknown as {
+        parentCallId: string
+        subCallId: string
+        name: string
+        arguments: unknown
+      }
+      const siblings = codeDispatches.get(data.parentCallId) ?? []
+      // The independent replay emits the same public running-call shape as
+      // Chat without reading or mutating Session's live index.
+      /* jscpd:ignore-start */
+      codeDispatches.set(data.parentCallId, [...siblings, {
+        callId: data.subCallId,
+        name: data.name,
+        argsRaw: JSON.stringify(data.arguments),
+        turn: 0,
+        step: 0,
+        time: event.time,
+        callView: null,
+      }])
+      /* jscpd:ignore-end */
+      continue
+    }
+    if ((event.type as string) === 'tool/code-dispatch') {
+      const data = event.data as unknown as {
+        parentCallId: string
+        subCallId: string
+        name: string
+        arguments: unknown
+        isError: boolean
+        content: ContentBlock[]
+      }
+      const siblings = codeDispatches.get(data.parentCallId) ?? []
+      const at = siblings.findIndex(sub => sub.callId === data.subCallId)
+      const started = at === -1 ? undefined : siblings[at]
+      // History independently reproduces the public settled-call shape instead
+      // of consuming Session's live code-dispatch projection.
+      /* jscpd:ignore-start */
+      const settled: CodeSubCall = {
+        kind: 'tool-result', seq: event.seq, time: event.time,
+        callId: data.subCallId,
+        call: { name: data.name, argsRaw: JSON.stringify(data.arguments) },
+        callTime: started?.time ?? null,
+        content: data.content,
+        isError: data.isError,
+        callView: null,
+        resultView: null,
+      }
+      codeDispatches.set(
+        data.parentCallId,
+        at === -1
+          ? [...siblings, settled]
+          : siblings.map((sub, index) => index === at ? settled : sub),
+      )
+      /* jscpd:ignore-end */
+      continue
+    }
+    switch (event.type) {
+      case 'assistant/chunk': {
+        const { turn, step, chunk } = event.data
+        if (partial === null || partial.turn !== turn || partial.step !== step) {
+          partial = new PartialAccumulator(turn, step)
+        }
+        partial.push(chunk)
+        break
+      }
+      case 'assistant/message':
+        if (partial?.turn === event.data.turn && partial.step === event.data.step) partial = null
+        break
+      case 'tool/call':
+        // History reconstructs its own in-flight index; this intentionally
+        // mirrors the published Chat node shape, not Chat's mutable state.
+        /* jscpd:ignore-start */
+        openCalls.set(String(event.data.callId), {
+          callId: String(event.data.callId),
+          name: event.data.name,
+          argsRaw: event.data.arguments,
+          turn: event.data.turn,
+          step: event.data.step,
+          time: event.time,
+          callView: entry.view?.for === 'call' ? entry.view.view : null,
+        })
+        /* jscpd:ignore-end */
+        break
+      case 'tool/result':
+        openCalls.delete(String(event.data.message.source.callId))
+        break
+      case 'turn/end': {
+        if (partial !== null && partial.turn === event.data.turn) {
+          const { blocks } = partial.toPartial()
+          const visible = blocks.some(block =>
+            block.kind === 'text' || block.kind === 'reasoning' ? block.text !== '' : true)
+          if (visible) {
+            interruptedNodes.push({
+              kind: 'assistant', seq: event.seq - 0.9, time: event.time,
+              turn: partial.turn, step: partial.step, blocks, interrupted: true,
+            })
+          }
+          partial = null
+        }
+        let callOffset = 0
+        for (const [callId, call] of openCalls) {
+          if (call.turn !== event.data.turn) continue
+          openCalls.delete(callId)
+          // Interrupted terminal nodes are reconstructed independently so a
+          // Trajectory replay cannot observe Session's frozen-node lifecycle.
+          /* jscpd:ignore-start */
+          interruptedNodes.push({
+            kind: 'tool-result', seq: event.seq - 0.8 + callOffset++ * 0.01,
+            time: event.time,
+            callId,
+            call: { name: call.name, argsRaw: call.argsRaw },
+            callTime: call.time,
+            content: [],
+            isError: true,
+            error: { name: 'Interrupted', code: 'interrupted' },
+            callView: call.callView,
+            resultView: null,
+          })
+          /* jscpd:ignore-end */
+        }
+        break
+      }
+      default:
+        break
+    }
+  }
+
+  return {
+    interruptedNodes,
+    partial: partial?.toPartial() ?? null,
+    runningCalls: [...openCalls.values()],
+    codeDispatches,
+  }
+}
+
+/**
+ * Project one immutable history ledger without reading or mutating Chat state.
+ * @param entries - Contiguous history entries in sequence order.
+ * @returns Event order, context lineage, and transient tail state.
+ */
+export function projectConversationHistory(
+  entries: readonly HistoryEntry[],
+): ConversationHistoryProjection {
+  const events = entries.map(entry => entry.event)
+  const baseSeq = events[0]?.seq ?? 0
+  const padded = [
+    ...Array.from({ length: baseSeq }, (_, seq) => paddingEvent(seq)),
+    ...events,
+  ]
+  const callIndex = new Map<string, CallIndexEntry>()
+  const resultViews = new Map<number, ToolResultView>()
+  const assistantSteps = new Map<string, AssistantStepMetadata>()
+  const assistantTimings = new Map<number, AssistantTiming>()
+  const assistantRequestConfigs = new Map<number, AssistantRequestConfig>()
+  const promptsByContext = new Map<number, ConversationPromptSnapshot>()
+  let activeRequestConfig: AssistantRequestConfig | undefined
+  let activePrompt: ConversationPromptSnapshot | undefined
+  let contextGeneration = 0
+
+  for (const [index, event] of events.entries()) {
+    const view = entries[index]?.view
+    if (event.type === 'tool/call') {
+      callIndex.set(String(event.data.callId), {
+        name: event.data.name,
+        argsRaw: event.data.arguments,
+        time: event.time,
+        callView: view?.for === 'call' ? view.view : null,
+      })
+    } else if (event.type === 'tool/result' && view?.for === 'result') {
+      resultViews.set(event.seq, view.view)
+    }
+    if (isSurfaceEvent(event) && event.surfaceOp !== 'append') {
+      contextGeneration++
+      if (activePrompt !== undefined) promptsByContext.set(contextGeneration, activePrompt)
+    }
+    if (event.type === 'request/header') {
+      activeRequestConfig = event.data.header.config
+      activePrompt = {
+        config: event.data.header.config,
+        system: event.data.header.system ?? '',
+        tools: event.data.header.tools ?? [],
+      }
+      promptsByContext.set(contextGeneration, activePrompt)
+    } else if (event.type === 'step/start') {
+      assistantSteps.set(
+        assistantStepKey(event.data.turn, event.data.step),
+        { stepStartTime: event.time, firstTokenTime: null },
+      )
+    } else if (event.type === 'assistant/chunk' && isTokenDelta(event.data.chunk)) {
+      const key = assistantStepKey(event.data.turn, event.data.step)
+      const current = assistantSteps.get(key) ?? {
+        stepStartTime: null,
+        firstTokenTime: null,
+      }
+      if (current.firstTokenTime === null) {
+        assistantSteps.set(key, { ...current, firstTokenTime: event.time })
+      }
+    } else if (event.type === 'assistant/message') {
+      assistantTimings.set(
+        event.seq,
+        {
+          ...(assistantSteps.get(assistantStepKey(event.data.turn, event.data.step)) ?? {
+            stepStartTime: null,
+            firstTokenTime: null,
+          }),
+          completedTime: event.time,
+        },
+      )
+      if (activeRequestConfig !== undefined) {
+        assistantRequestConfigs.set(event.seq, activeRequestConfig)
+      }
+    }
+  }
+
+  const nodeCache = new Map<number, ConversationNode>()
+  const materialize = (seq: number): ConversationNode | undefined => {
+    const cached = nodeCache.get(seq)
+    if (cached !== undefined) return cached
+    const event = padded[seq]
+    if (event === undefined || !isSurfaceEligibleType(event.type)) return
+    const node = materializeNode(
+      event,
+      callIndex,
+      resultViews.get(seq) ?? null,
+      assistantTimings.get(seq),
+      assistantRequestConfigs.get(seq),
+    )
+    nodeCache.set(seq, node)
+    return node
+  }
+  const eventNodes = events.flatMap((event) => {
+    const node = materialize(event.seq)
+    return node === undefined ? [] : [node]
+  })
+
+  let contexts: readonly ConversationContext[]
+  if (events.some(event => replacementCrossesWindowHead(event, baseSeq))) {
+    contexts = [{
+      id: 0,
+      ...(activePrompt === undefined ? {} : { prompt: activePrompt }),
+      nodes: eventNodes,
+    }]
+  } else {
+    try {
+      contexts = foldContexts(padded).map((context): ConversationContext => {
+        const nodes = context.nodes.flatMap((seq) => {
+          const node = materialize(seq)
+          return node === undefined ? [] : [node]
+        })
+        const prompt = promptsByContext.get(context.generation)
+        if (context.originSeq === undefined) {
+          return {
+            id: context.generation,
+            ...(prompt === undefined ? {} : { prompt }),
+            nodes,
+          }
+        }
+        const originEvent = padded[context.originSeq]
+        return {
+          id: context.generation,
+          parentId: context.generation - 1,
+          origin: contextOriginKind(originEvent),
+          originSeq: context.originSeq,
+          ...(originEvent === undefined ? {} : { createdAt: originEvent.time }),
+          ...(prompt === undefined ? {} : { prompt }),
+          nodes,
+        }
+      })
+    } catch (error) {
+      console.error('[web-runtime] history surface fold failed, using event order:', error)
+      contexts = [{
+        id: 0,
+        ...(activePrompt === undefined ? {} : { prompt: activePrompt }),
+        nodes: eventNodes,
+      }]
+    }
+  }
+
+  return {
+    eventNodes,
+    contexts,
+    ...projectTransient(entries),
+  }
+}
diff --git a/packages/client/runtime/src/client/session-history/service.ts b/packages/client/runtime/src/client/session-history/service.ts
new file mode 100644
index 0000000000..b5aba32bce
--- /dev/null
+++ b/packages/client/runtime/src/client/session-history/service.ts
@@ -0,0 +1,66 @@
+import type { Context } from 'cordis'
+import type {
+  HostFrame, IApiClient, MuxFrame, RpcRequest, SessionId,
+} from '@deepseek-ai/dsh-client-connection/client'
+import type {
+  ISessionHistory, SessionHistoryFace,
+} from '../contract/session-history.ts'
+import { SessionHistorySource } from './source.ts'
+
+/** Root registry and frame router for independent inspection histories. */
+export class SessionHistoryService implements ISessionHistory {
+  private readonly sources = new Map<SessionId, SessionHistorySource>()
+
+  /**
+   * @param ctx - Client root context.
+   * @param api - Shared wire client.
+   */
+  constructor(ctx: Context, private readonly api: IApiClient) {
+    ctx.reflect.provide('sessionHistory', this, undefined)
+  }
+
+  /**
+   * Resolve one identity-stable history source.
+   * @param sessionId - Host session identity.
+   * @returns Source independent from SessionManager.
+   */
+  source(sessionId: SessionId): SessionHistoryFace {
+    let source = this.sources.get(sessionId)
+    if (source === undefined) {
+      source = new SessionHistorySource(sessionId, this.api)
+      this.sources.set(sessionId, source)
+    }
+    return source
+  }
+
+  /**
+   * Route history-relevant mux frames only to an existing source.
+   * @param envelope - Validated mux envelope.
+   */
+  handleMuxEnvelope(envelope: RpcRequest<MuxFrame>): void {
+    const frame = envelope.payload
+    if (frame.type === 'stream/error') return
+    this.sources.get(frame.sessionId)?.handleMuxFrame(frame)
+  }
+
+  /**
+   * Drop a removed session's independent history source.
+   * @param envelope - Validated host envelope.
+   */
+  handleHostEnvelope(envelope: RpcRequest<HostFrame>): void {
+    const frame = envelope.payload
+    if (frame.type !== 'host/session-removed') return
+    this.sources.get(frame.sessionId)?.dispose()
+    this.sources.delete(frame.sessionId)
+  }
+
+  /** Invalidate requests from the dead connection generation. */
+  handleDisconnected(): void {
+    for (const source of this.sources.values()) source.handleDisconnected()
+  }
+
+  /** Rebuild every previously activated source from the new generation. */
+  handleConnected(): void {
+    for (const source of this.sources.values()) source.resync()
+  }
+}
diff --git a/packages/client/runtime/src/client/session-history/source.ts b/packages/client/runtime/src/client/session-history/source.ts
new file mode 100644
index 0000000000..8de11b9da9
--- /dev/null
+++ b/packages/client/runtime/src/client/session-history/source.ts
@@ -0,0 +1,352 @@
+import type {
+  HistoryEntry, IApiClient, MuxFrame, RpcError, SessionId,
+} from '@deepseek-ai/dsh-client-connection/client'
+import { transportError } from '@deepseek-ai/dsh-host-apiproxy/api'
+import type {
+  SessionHistoryFace, SessionHistorySnapshot,
+} from '../contract/session-history.ts'
+import { createHistoryInspection } from '../sessions/history.ts'
+import { Notifier } from '../sessions/notifier.ts'
+
+const HISTORY_PAGE_MESSAGES = 50
+
+function isAborted(signal: AbortSignal | undefined): boolean {
+  return signal?.aborted === true
+}
+
+/** Independent raw-history owner used only by inspection consumers. */
+export class SessionHistorySource implements SessionHistoryFace {
+  private entries: readonly HistoryEntry[] = []
+  private baseSeq = 0
+  private hasMore = false
+  private state: SessionHistorySnapshot['state'] = 'cold'
+  private error: RpcError | null = null
+  private generation = 0
+  private persistentConsumer = false
+  private readonly consumerSignals = new Set<AbortSignal>()
+  private openPromise: Promise<void> | null = null
+  private olderPromise: Promise<void> | null = null
+  private stitching = false
+  private liveBuffer: HistoryEntry[] = []
+  private subscribedLastSeq: number | null = null
+  private inspectionCache: {
+    entries: readonly HistoryEntry[]
+    value: SessionHistorySnapshot['inspection']
+  } | null = null
+  private snapshotCache: SessionHistorySnapshot
+  private readonly notifier = new Notifier(() => {
+    this.snapshotCache = this.buildSnapshot()
+  })
+
+  /**
+   * @param sessionId - Host session identity.
+   * @param api - Shared wire client.
+   */
+  constructor(
+    readonly sessionId: SessionId,
+    private readonly api: IApiClient,
+  ) {
+    this.snapshotCache = this.buildSnapshot()
+  }
+
+  /**
+   * Subscribe to ledger changes.
+   * @param listener - Change callback.
+   * @returns Unsubscribe function.
+   */
+  subscribe(listener: () => void): () => void {
+    return this.notifier.subscribe(listener)
+  }
+
+  /**
+   * Read the cached ledger snapshot.
+   * @returns Stable snapshot until the source changes.
+   */
+  getSnapshot(): SessionHistorySnapshot {
+    this.notifier.ensureFresh()
+    return this.snapshotCache
+  }
+
+  /**
+   * Load the tail and exhaust all available older pages.
+   * @param signal - Consumer lifetime.
+   * @returns When paging completes, fails to advance, or is aborted.
+   */
+  async loadAll(signal?: AbortSignal): Promise<void> {
+    if (signal?.aborted === true) return
+    this.trackConsumer(signal)
+    await this.open()
+    while (
+      !isAborted(signal)
+      && this.state === 'ready'
+      && this.hasMore
+    ) {
+      const previousBaseSeq = this.baseSeq
+      await this.loadOlder()
+      if (isAborted(signal) || this.baseSeq === previousBaseSeq) return
+    }
+  }
+
+  /** Rebuild and page for whichever mounted consumers survive a reconnect. */
+  private async loadForConsumers(): Promise<void> {
+    await this.open()
+    while (
+      this.hasConsumer()
+      && this.state === 'ready'
+      && this.hasMore
+    ) {
+      const previousBaseSeq = this.baseSeq
+      await this.loadOlder()
+      if (!this.hasConsumer() || this.baseSeq === previousBaseSeq) return
+    }
+  }
+
+  /**
+   * Route a relevant mux frame without involving the Chat session.
+   * @param frame - Session-addressed frame.
+   */
+  handleMuxFrame(frame: MuxFrame): void {
+    if (frame.type === 'session/subscribed') {
+      this.subscribedLastSeq = frame.lastSeq
+      return
+    }
+    if (frame.type !== 'session/event') return
+    this.acceptLive({ event: frame.event, ...(frame.view === undefined ? {} : { view: frame.view }) })
+  }
+
+  /** Invalidate dead-generation requests while retaining the last readable snapshot. */
+  handleDisconnected(): void {
+    this.generation++
+    this.openPromise = null
+    this.olderPromise = null
+    this.stitching = false
+    this.liveBuffer = []
+    this.subscribedLastSeq = null
+    if (this.state !== 'cold') {
+      this.state = 'cold'
+      this.error = null
+      this.notifier.markDirty()
+    }
+  }
+
+  /** Rebuild an activated ledger from the new connection generation. */
+  resync(): void {
+    if (!this.hasConsumer()) return
+    this.generation++
+    this.openPromise = null
+    this.olderPromise = null
+    this.stitching = false
+    this.liveBuffer = []
+    this.subscribedLastSeq = null
+    this.entries = []
+    this.baseSeq = 0
+    this.hasMore = false
+    this.state = 'cold'
+    this.error = null
+    this.notifier.markDirty()
+    void this.loadForConsumers()
+  }
+
+  /** Stop future refresh work after the host removes the session. */
+  dispose(): void {
+    this.persistentConsumer = false
+    this.consumerSignals.clear()
+    this.generation++
+    this.openPromise = null
+    this.olderPromise = null
+    this.liveBuffer = []
+  }
+
+  private open(): Promise<void> {
+    if (this.state === 'ready') return Promise.resolve()
+    if (this.openPromise !== null) return this.openPromise
+    const generation = this.generation
+    const operation = this.doOpen(generation)
+    const settled = operation.finally(() => {
+      if (this.openPromise === settled) this.openPromise = null
+    })
+    this.openPromise = settled
+    return settled
+  }
+
+  private trackConsumer(signal: AbortSignal | undefined): void {
+    if (signal === undefined) {
+      this.persistentConsumer = true
+      return
+    }
+    if (this.consumerSignals.has(signal)) return
+    this.consumerSignals.add(signal)
+    signal.addEventListener('abort', () => {
+      this.consumerSignals.delete(signal)
+    }, { once: true })
+  }
+
+  private hasConsumer(): boolean {
+    return this.persistentConsumer || this.consumerSignals.size > 0
+  }
+
+  private async doOpen(generation: number): Promise<void> {
+    this.state = 'loading'
+    this.error = null
+    this.notifier.markDirty()
+    try {
+      let { result } = await this.api.sessions.history({
+        sessionId: this.sessionId,
+        maxMessages: HISTORY_PAGE_MESSAGES,
+      })
+      if (generation !== this.generation) return
+      if (!result.ok) {
+        this.state = 'error'
+        this.error = result.error
+        return
+      }
+      this.installTail(result.value.events, result.value.hasMore, true)
+      const tailSeq = this.tailSeq()
+      if (
+        this.subscribedLastSeq !== null
+        && tailSeq !== null
+        && this.subscribedLastSeq > tailSeq
+      ) {
+        result = (await this.api.sessions.history({
+          sessionId: this.sessionId,
+          maxMessages: HISTORY_PAGE_MESSAGES,
+        })).result
+        if (generation !== this.generation) return
+        if (result.ok) this.installTail(result.value.events, result.value.hasMore, true)
+      }
+      this.state = 'ready'
+    } catch (error) {
+      if (generation !== this.generation) return
+      this.state = 'error'
+      const folded = transportError<never>(error)
+      /* v8 ignore next -- transportError always returns the error branch. */
+      this.error = folded.ok ? null : folded.error
+    } finally {
+      if (generation === this.generation) this.notifier.markDirty()
+    }
+  }
+
+  private loadOlder(): Promise<void> {
+    if (this.olderPromise !== null) return this.olderPromise
+    if (this.state !== 'ready' || !this.hasMore) return Promise.resolve()
+    const generation = this.generation
+    const operation = (async () => {
+      try {
+        const { result } = await this.api.sessions.history({
+          sessionId: this.sessionId,
+          beforeSeq: this.baseSeq,
+          maxMessages: HISTORY_PAGE_MESSAGES,
+        })
+        if (generation !== this.generation || this.state !== 'ready' || !result.ok) return
+        const older = result.value.events
+        if (older.length === 0) {
+          this.hasMore = result.value.hasMore
+          return
+        }
+        const tail = older.at(-1)
+        if (tail === undefined || tail.event.seq + 1 !== this.baseSeq) {
+          console.error(
+            `[web-runtime] inspection history page discontinuous: tail seq ${tail?.event.seq} vs baseSeq ${this.baseSeq}`,
+          )
+          this.hasMore = false
+          return
+        }
+        this.entries = [...older, ...this.entries]
+        this.baseSeq = older[0]?.event.seq ?? this.baseSeq
+        this.hasMore = result.value.hasMore
+      } catch (error) {
+        console.error('[web-runtime] inspection history paging failed:', error)
+      }
+    })()
+    const settled = operation.finally(() => {
+      if (this.olderPromise !== settled) return
+      this.olderPromise = null
+      this.notifier.markDirty()
+    })
+    this.olderPromise = settled
+    return settled
+  }
+
+  private installTail(
+    tail: readonly HistoryEntry[],
+    hasMore: boolean,
+    replace: boolean,
+  ): void {
+    if (replace) {
+      this.entries = [...tail]
+      this.hasMore = hasMore
+    } else {
+      const firstSeq = tail[0]?.event.seq
+      const prefix = firstSeq === undefined
+        ? this.entries
+        : this.entries.filter(entry => entry.event.seq < firstSeq)
+      this.entries = [...prefix, ...tail]
+    }
+    this.baseSeq = this.entries[0]?.event.seq ?? 0
+    const buffered = this.liveBuffer
+    this.liveBuffer = []
+    for (const entry of buffered) this.appendLive(entry)
+    this.notifier.markDirty()
+  }
+
+  private acceptLive(entry: HistoryEntry): void {
+    if (this.state === 'loading' || this.stitching) {
+      this.liveBuffer.push(entry)
+      return
+    }
+    if (this.state !== 'ready') return
+    const tailSeq = this.tailSeq()
+    if (tailSeq !== null && entry.event.seq > tailSeq + 1) {
+      this.liveBuffer.push(entry)
+      void this.repairGap()
+      return
+    }
+    this.appendLive(entry)
+    this.notifier.markDirty()
+  }
+
+  private appendLive(entry: HistoryEntry): void {
+    const tailSeq = this.tailSeq()
+    if (tailSeq !== null && entry.event.seq <= tailSeq) return
+    this.entries = [...this.entries, entry]
+  }
+
+  private async repairGap(): Promise<void> {
+    if (this.stitching) return
+    this.stitching = true
+    const generation = this.generation
+    try {
+      const { result } = await this.api.sessions.history({
+        sessionId: this.sessionId,
+        maxMessages: HISTORY_PAGE_MESSAGES,
+      })
+      if (result.ok && generation === this.generation && this.state === 'ready') {
+        this.installTail(result.value.events, result.value.hasMore, false)
+      }
+    } catch (error) {
+      console.error('[web-runtime] inspection history gap repair failed:', error)
+    } finally {
+      if (generation === this.generation) this.stitching = false
+    }
+  }
+
+  private tailSeq(): number | null {
+    return this.entries.at(-1)?.event.seq ?? null
+  }
+
+  private buildSnapshot(): SessionHistorySnapshot {
+    if (this.inspectionCache?.entries !== this.entries) {
+      const entries = this.entries
+      this.inspectionCache = {
+        entries,
+        value: createHistoryInspection(() => entries),
+      }
+    }
+    return {
+      state: this.state,
+      error: this.error,
+      hasMore: this.hasMore,
+      inspection: this.inspectionCache.value,
+    }
+  }
+}
diff --git a/packages/client/runtime/src/client/sessions/conversation-context.ts b/packages/client/runtime/src/client/sessions/conversation-context.ts
new file mode 100644
index 0000000000..20a8bbcadd
--- /dev/null
+++ b/packages/client/runtime/src/client/sessions/conversation-context.ts
@@ -0,0 +1,23 @@
+import type { ConversationNode } from './conversation.ts'
+import type { ConversationPromptSnapshot } from './request-inspection.ts'
+
+/** Operation that started a new append-only model context. */
+export type ConversationContextOriginKind = 'compaction' | 'rewind' | 'rewrite'
+
+/** One immutable model-context generation reconstructed from surface replacements. */
+export interface ConversationContext {
+  /** Zero-based generation within the session; stable across later appends. */
+  id: number
+  /** Previous generation in this session; absent for the initial context. */
+  parentId?: number
+  /** Why this generation exists; absent for the initial context. */
+  origin?: ConversationContextOriginKind
+  /** Event seq of the replacement that created this generation. */
+  originSeq?: number
+  /** Unix epoch ms of the replacement that created this generation. */
+  createdAt?: number
+  /** Latest request header observed in this generation, inherited until a later header replaces it. */
+  prompt?: ConversationPromptSnapshot
+  /** Final frozen nodes for historical generations, or current folded nodes for the tail. */
+  nodes: readonly ConversationNode[]
+}
diff --git a/packages/client/runtime/src/client/sessions/conversation.ts b/packages/client/runtime/src/client/sessions/conversation.ts
index f5f0717236..cd03c92485 100644
--- a/packages/client/runtime/src/client/sessions/conversation.ts
+++ b/packages/client/runtime/src/client/sessions/conversation.ts
@@ -10,9 +10,26 @@ import type {
   RpcError, SessionId, ToolCallView, ToolResultView,
 } from '@deepseek-ai/dsh-client-connection/client'
 import type { PendingInteraction } from './pending.ts'
-
 export type { TodoItem }
 
+/** Request configuration recorded for one provider call. */
+export interface AssistantRequestConfig {
+  provider: string
+  model: string
+  purpose?: string
+  thinking?: string
+  reasoningEffort?: string
+  temperature?: number
+  maxTokens?: number
+  stop?: readonly string[]
+}
+
+/** Stable provider/model identity reported for one completed request. */
+export interface AssistantProvenanceView {
+  provider: string
+  model: string
+}
+
 /** Assistant content blocks sorted by what the UI cares about
  *  (text body / collapsible reasoning / tool-call card head / other fallback). */
 export type AssistantBlock =
@@ -54,6 +71,16 @@ export interface UserMessageNode {
   source: unknown
 }
 
+/** Recorded boundaries used to derive assistant latency and throughput. */
+export interface AssistantTiming {
+  /** Matching step/start timestamp, or null when it is outside the current event window. */
+  stepStartTime: number | null
+  /** First non-empty text/reasoning/tool delta timestamp, or null when no token delta was recorded. */
+  firstTokenTime: number | null
+  /** Final assistant/message timestamp. */
+  completedTime: number
+}
+
 /** A finalized (or interruption-frozen) assistant message. */
 export interface AssistantMessageNode {
   kind: 'assistant'
@@ -64,6 +91,10 @@ export interface AssistantMessageNode {
   step: number
   blocks: readonly AssistantBlock[]
   usage?: unknown
+  provenance?: AssistantProvenanceView
+  requestConfig?: AssistantRequestConfig
+  /** Timing derived from the recorded step/chunk/message event sequence. */
+  timing?: AssistantTiming
   /** Frozen partial of an aborted turn (no finalize ever arrives): rendered with a 已停止 marker.
    *  Synthetic seq (fractional, derived from the turn/end seq) keeps it ordered inside the flow. */
   interrupted?: true
diff --git a/packages/client/runtime/src/client/sessions/fold-adapter.ts b/packages/client/runtime/src/client/sessions/fold-adapter.ts
index 039c36056b..fc9efb70e4 100644
--- a/packages/client/runtime/src/client/sessions/fold-adapter.ts
+++ b/packages/client/runtime/src/client/sessions/fold-adapter.ts
@@ -7,7 +7,9 @@ import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
 // Subpath export (package.json exports "./surface", alias added for this): all value imports
 // go through it — the package root points at lib/index.js (needs a build) which the vite
 // browser bundle cannot resolve; surface.ts has no Node dependencies.
-import { SurfaceManager, isSurfaceEligibleType } from '@deepseek-ai/dsh-session/surface'
+import {
+  SurfaceManager, isSurfaceEligibleType, isSurfaceEvent,
+} from '@deepseek-ai/dsh-session/surface'
 import type { CommandId } from '@deepseek-ai/dsh-commands/brand'
 import type { ToolCallView, ToolEventView, ToolResultView } from '@deepseek-ai/dsh-client-connection/client'
 import type { CommandNode, ConversationNode } from './conversation.ts'
@@ -33,6 +35,11 @@ function paddingEvent(seq: number): SessionEvent {
   return { type: 'noop/padding', seq, time: 0, data: {} } as unknown as SessionEvent
 }
 
+function replacementCrossesWindowHead(event: SessionEvent, baseSeq: number): boolean {
+  if (!isSurfaceEvent(event) || event.surfaceOp === 'append') return false
+  return event.surfaceOp.start < baseSeq || event.surfaceOp.end < baseSeq
+}
+
 /** One event -> UI node (pure function; the six-variant ConversationNode union). */
 function materializeNode(
   event: SessionEvent,
@@ -137,7 +144,7 @@ export class FoldAdapter {
     for (const event of events) this.padded.push(event)
     this.surface = new SurfaceManager(this.padded)
     this.nodeCache.clear()
-    this.degraded = false
+    this.degraded = events.some(event => replacementCrossesWindowHead(event, baseSeq))
     this.callIdx = new Map()
     this.resultViews.clear()
     this.commandIdx = new Map()
@@ -160,6 +167,7 @@ export class FoldAdapter {
   append(event: SessionEvent, view?: ToolEventView): void {
     this.rev++
     this.padded.push(event)
+    if (replacementCrossesWindowHead(event, this.baseSeq)) this.degraded = true
     this.indexCall(event, view)
     this.indexCommand(event)
   }
diff --git a/packages/client/runtime/src/client/sessions/history.ts b/packages/client/runtime/src/client/sessions/history.ts
new file mode 100644
index 0000000000..520503a561
--- /dev/null
+++ b/packages/client/runtime/src/client/sessions/history.ts
@@ -0,0 +1,66 @@
+import type { ToolSchema } from '@deepseek-ai/dsh-llm/types'
+import type { HistoryEntry } from '@deepseek-ai/dsh-client-connection/client'
+import type {
+  CodeSubCall, ConversationNode, PartialAssistant, RunningToolCall,
+} from './conversation.ts'
+import type { ConversationContext } from './conversation-context.ts'
+import { projectConversationHistory } from '../session-history/history-fold.ts'
+import { inspectRequests, type RequestView } from './request-inspection.ts'
+
+/** Lazily derived inspection data for one immutable session-history window. */
+export interface SessionHistoryInspection {
+  eventNodes: readonly ConversationNode[]
+  contexts: readonly ConversationContext[]
+  requests: readonly RequestView[]
+  callSchemas: ReadonlyMap<string, ToolSchema>
+  interruptedNodes: readonly ConversationNode[]
+  partial: PartialAssistant | null
+  runningCalls: readonly RunningToolCall[]
+  codeDispatches: ReadonlyMap<string, readonly CodeSubCall[]>
+}
+
+/**
+ * Create a lazy inspection projection over an immutable history window.
+ * Conversation consumers retain the cheap wrapper; only Trajectory snapshots
+ * the entries and replays event order and request lifecycle state.
+ * @param loadEntries - Lazily snapshots contiguous raw entries in sequence order.
+ * @returns Lazy, memoized inspection fields for that exact window.
+ */
+export function createHistoryInspection(
+  loadEntries: () => readonly HistoryEntry[],
+): SessionHistoryInspection {
+  let entries: readonly HistoryEntry[] | undefined
+  let conversation: ReturnType<typeof projectConversationHistory> | undefined
+  let requests: ReturnType<typeof inspectRequests> | undefined
+  const historyEntries = () => entries ??= loadEntries()
+  const conversationProjection = () =>
+    conversation ??= projectConversationHistory(historyEntries())
+  const requestProjection = () =>
+    requests ??= inspectRequests(historyEntries())
+  return {
+    get eventNodes() {
+      return conversationProjection().eventNodes
+    },
+    get contexts() {
+      return conversationProjection().contexts
+    },
+    get interruptedNodes() {
+      return conversationProjection().interruptedNodes
+    },
+    get partial() {
+      return conversationProjection().partial
+    },
+    get runningCalls() {
+      return conversationProjection().runningCalls
+    },
+    get codeDispatches() {
+      return conversationProjection().codeDispatches
+    },
+    get requests() {
+      return requestProjection().requests
+    },
+    get callSchemas() {
+      return requestProjection().callSchemas
+    },
+  }
+}
diff --git a/packages/client/runtime/src/client/sessions/request-inspection.ts b/packages/client/runtime/src/client/sessions/request-inspection.ts
new file mode 100644
index 0000000000..e15ad55c8b
--- /dev/null
+++ b/packages/client/runtime/src/client/sessions/request-inspection.ts
@@ -0,0 +1,401 @@
+// Request-centric inspection read model. Ordinary generation and compaction
+// calls share one chronological projection; presentation-specific grouping
+// remains in the trajectory consumer.
+
+import type { ContentBlock, TokenUsage, ToolSchema } from '@deepseek-ai/dsh-llm/types'
+import type { HistoryEntry } from '@deepseek-ai/dsh-client-connection/client'
+import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
+import type {
+  AssistantProvenanceView, AssistantRequestConfig,
+} from './conversation.ts'
+
+export type {
+  AssistantProvenanceView, AssistantRequestConfig,
+} from './conversation.ts'
+
+/** Complete model-visible request header in force for an ordinary generation. */
+export interface ConversationPromptSnapshot {
+  /** Provider/model and sampling configuration from the effective request header. */
+  config: AssistantRequestConfig
+  /** Rendered system prompt text; empty when the request had no system prompt. */
+  system: string
+  /** Complete tool catalog sent with the request, including tools that were never called. */
+  tools: readonly ToolSchema[]
+}
+
+/** System/tool change introduced while preparing one ordinary request. */
+export interface RequestPromptChange {
+  /** Sequence of the request/header event that introduced this state. */
+  seq: number
+  /** Unix epoch ms from the request/header event. */
+  time: number
+  /** How the model-visible prompt differs from the previous recorded state. */
+  kind: 'initial' | 'system' | 'tools' | 'system-and-tools'
+  /** State immediately before this change; absent for the initial header. */
+  previous?: ConversationPromptSnapshot
+}
+
+/** One provider request reconstructed from durable request lifecycle events. */
+export interface RequestView {
+  /** Request category; compaction is a purpose, not a separate projection. */
+  purpose: 'assistant' | 'compaction'
+  /** Sequence that opened the operation represented by this request. */
+  startSeq: number
+  turn: number
+  /** Agent-loop step, or zero for a direct compaction request. */
+  step: number
+  startedAt: number
+  completedAt: number | null
+  status: 'running' | 'complete' | 'error'
+  error?: string
+  /** Effective ordinary request input, inherited until a later header changes it. */
+  prompt?: ConversationPromptSnapshot
+  /** Prompt change logged while preparing this request. */
+  promptChange?: RequestPromptChange
+  provenance?: AssistantProvenanceView
+  requestConfig?: AssistantRequestConfig
+  usage?: unknown
+  /** Assistant message or compaction summary sequence produced by this request. */
+  resultSeq?: number
+  /** Compaction replacement message sequence, when one was committed. */
+  replacementSeq?: number
+  /** Safe compaction summary projection. */
+  summary?: readonly ContentBlock[]
+  /** Complete compaction provider output before the safe projection. */
+  rawOutput?: readonly ContentBlock[]
+  /** Retry ordinal scheduled after a failed ordinary request. */
+  retry?: number
+  maxRetries?: number
+  retryDelayMs?: number
+}
+
+/** Immutable request-centric projection derived from one history window. */
+export interface RequestInspectionSnapshot {
+  requests: readonly RequestView[]
+  callSchemas: ReadonlyMap<string, ToolSchema>
+}
+
+/**
+ * Derive the request-centric read model from one immutable history window.
+ * Compaction participates as a request purpose rather than a parallel
+ * top-level collection.
+ * @param entries - Contiguous raw session history.
+ * @returns Requests and call-time schemas derived from that history.
+ */
+export function inspectRequests(
+  entries: readonly HistoryEntry[],
+): RequestInspectionSnapshot {
+  const events = entries.map(entry => entry.event)
+  return {
+    requests: deriveRequests(events),
+    callSchemas: deriveCallSchemas(events),
+  }
+}
+
+interface RetryEvent {
+  type: 'llm/retry'
+  seq: number
+  time: number
+  data: {
+    turn: number
+    step: number
+    retry: number
+    maxRetries: number
+    delayMs: number
+    failure: { message: string }
+  }
+}
+
+interface CompactionStartEvent {
+  type: 'compact/start'
+  seq: number
+  time: number
+  data: { turn: number }
+}
+
+interface CompactionSummaryEvent {
+  type: 'compact/summary'
+  seq: number
+  time: number
+  data: {
+    summary: readonly ContentBlock[]
+    rawOutput?: readonly ContentBlock[]
+    provider: string
+    model: string
+    maxTokens?: number
+    usage?: unknown
+  }
+}
+
+interface CompactionEndEvent {
+  type: 'compact/end'
+  seq: number
+  time: number
+  data: { turn: number; error?: string }
+}
+
+function requestKey(turn: number, step: number): string {
+  return `${turn}\u0000${step}`
+}
+
+function addTokenUsage(current: unknown, next: TokenUsage): TokenUsage {
+  const previous = current as TokenUsage | undefined
+  return {
+    inputTokens: (previous?.inputTokens ?? 0) + next.inputTokens,
+    outputTokens: (previous?.outputTokens ?? 0) + next.outputTokens,
+    ...(previous?.cacheReadTokens === undefined && next.cacheReadTokens === undefined
+      ? {}
+      : {
+        cacheReadTokens:
+          (previous?.cacheReadTokens ?? 0) + (next.cacheReadTokens ?? 0),
+      }),
+    ...(previous?.cacheWriteTokens === undefined && next.cacheWriteTokens === undefined
+      ? {}
+      : {
+        cacheWriteTokens:
+          (previous?.cacheWriteTokens ?? 0) + (next.cacheWriteTokens ?? 0),
+      }),
+    ...(previous?.reasoningTokens === undefined && next.reasoningTokens === undefined
+      ? {}
+      : {
+        reasoningTokens:
+          (previous?.reasoningTokens ?? 0) + (next.reasoningTokens ?? 0),
+      }),
+  }
+}
+
+function deriveCallSchemas(
+  events: readonly SessionEvent[],
+): ReadonlyMap<string, ToolSchema> {
+  let active = new Map<string, ToolSchema>()
+  const calls = new Map<string, ToolSchema>()
+  const capture = (callId: string, name: string): void => {
+    if (calls.has(callId)) return
+    const schema = active.get(name)
+    if (schema !== undefined) calls.set(callId, schema)
+  }
+  for (const event of events) {
+    if (event.type === 'request/header') {
+      const tools: unknown = event.data.header.tools
+      active = new Map(
+        Array.isArray(tools)
+          ? (tools as ToolSchema[]).map(schema => [schema.name, schema])
+          : [],
+      )
+      continue
+    }
+    if (event.type === 'tool/call') {
+      capture(String(event.data.callId), event.data.name)
+      continue
+    }
+    const type = event.type as string
+    if (type === 'tool/code-dispatch-start' || type === 'tool/code-dispatch') {
+      const data = event.data as unknown as { subCallId: string; name: string }
+      capture(data.subCallId, data.name)
+    }
+  }
+  return calls
+}
+
+function promptChange(
+  previous: ConversationPromptSnapshot | undefined,
+  prompt: ConversationPromptSnapshot,
+  event: SessionEvent<'request/header'>,
+): RequestPromptChange | undefined {
+  const systemChanged = previous !== undefined && previous.system !== prompt.system
+  const toolsChanged = previous !== undefined
+    && JSON.stringify(previous.tools) !== JSON.stringify(prompt.tools)
+  if (previous !== undefined && !systemChanged && !toolsChanged) return
+  return {
+    seq: event.seq,
+    time: event.time,
+    kind: previous === undefined
+      ? 'initial'
+      : systemChanged && toolsChanged
+        ? 'system-and-tools'
+        : systemChanged
+          ? 'system'
+          : 'tools',
+    ...(previous === undefined ? {} : { previous }),
+  }
+}
+
+/** Project ordinary and compaction provider calls into one chronological request stream. */
+function deriveRequests(events: readonly SessionEvent[]): readonly RequestView[] {
+  const requests: RequestView[] = []
+  const ordinaryByStep = new Map<string, number>()
+  let activeStep: string | undefined
+  let activePrompt: ConversationPromptSnapshot | undefined
+  let activeCompaction: number | undefined
+
+  const update = (index: number | undefined, change: Partial<RequestView>): void => {
+    if (index === undefined) return
+    const request = requests[index]
+    if (request !== undefined) requests[index] = { ...request, ...change }
+  }
+
+  for (const sourceEvent of events) {
+    if (sourceEvent.type === 'step/start') {
+      const { turn, step } = sourceEvent.data
+      const key = requestKey(turn, step)
+      ordinaryByStep.set(key, requests.length)
+      requests.push({
+        purpose: 'assistant',
+        startSeq: sourceEvent.seq,
+        turn,
+        step,
+        startedAt: sourceEvent.time,
+        completedAt: null,
+        status: 'running',
+        ...(activePrompt === undefined
+          ? {}
+          : { prompt: activePrompt, requestConfig: activePrompt.config }),
+      })
+      activeStep = key
+      continue
+    }
+    if (sourceEvent.type === 'request/header') {
+      const tools: unknown = sourceEvent.data.header.tools
+      const prompt: ConversationPromptSnapshot = {
+        config: sourceEvent.data.header.config,
+        system: sourceEvent.data.header.system ?? '',
+        tools: Array.isArray(tools) ? tools as ToolSchema[] : [],
+      }
+      const change = promptChange(activePrompt, prompt, sourceEvent)
+      activePrompt = prompt
+      update(activeStep === undefined ? undefined : ordinaryByStep.get(activeStep), {
+        prompt,
+        requestConfig: prompt.config,
+        ...(change === undefined ? {} : { promptChange: change }),
+      })
+      continue
+    }
+    if (
+      sourceEvent.type === 'assistant/chunk'
+      && sourceEvent.data.chunk.type === 'usage'
+    ) {
+      const index = ordinaryByStep.get(
+        requestKey(sourceEvent.data.turn, sourceEvent.data.step),
+      )
+      const request = index === undefined ? undefined : requests[index]
+      update(index, {
+        usage: addTokenUsage(request?.usage, sourceEvent.data.chunk.usage),
+      })
+      continue
+    }
+    if (sourceEvent.type === 'assistant/message') {
+      const index = ordinaryByStep.get(
+        requestKey(sourceEvent.data.turn, sourceEvent.data.step),
+      )
+      const request = index === undefined ? undefined : requests[index]
+      update(index, {
+        completedAt: sourceEvent.time,
+        status: 'complete',
+        resultSeq: sourceEvent.seq,
+        provenance: {
+          provider: sourceEvent.data.message.source.provider,
+          model: sourceEvent.data.message.source.model,
+        },
+        ...(request?.usage !== undefined || sourceEvent.data.usage === undefined
+          ? {}
+          : { usage: sourceEvent.data.usage }),
+      })
+      continue
+    }
+    if (sourceEvent.type === 'step/end') {
+      const key = requestKey(sourceEvent.data.turn, sourceEvent.data.step)
+      const index = ordinaryByStep.get(key)
+      const request = index === undefined ? undefined : requests[index]
+      if (request?.status === 'running') {
+        update(index, {
+          completedAt: sourceEvent.time,
+          status: 'error',
+        })
+      }
+      if (activeStep === key) activeStep = undefined
+      continue
+    }
+    if ((sourceEvent.type as string) === 'llm/retry') {
+      const event = sourceEvent as unknown as RetryEvent
+      update(ordinaryByStep.get(requestKey(event.data.turn, event.data.step)), {
+        status: 'error',
+        error: event.data.failure.message,
+        retry: event.data.retry,
+        maxRetries: event.data.maxRetries,
+        retryDelayMs: event.data.delayMs,
+      })
+      continue
+    }
+    if (sourceEvent.type === 'turn/end' && sourceEvent.data.reason.kind === 'error') {
+      const reason = sourceEvent.data.reason
+      update(ordinaryByStep.get(requestKey(sourceEvent.data.turn, reason.step)), {
+        status: 'error',
+        error: 'failure' in reason ? reason.failure.message : reason.message,
+      })
+      continue
+    }
+
+    const type = sourceEvent.type as string
+    if (type === 'compact/start') {
+      const event = sourceEvent as unknown as CompactionStartEvent
+      activeCompaction = requests.length
+      requests.push({
+        purpose: 'compaction',
+        startSeq: event.seq,
+        turn: event.data.turn,
+        step: 0,
+        startedAt: event.time,
+        completedAt: null,
+        status: 'running',
+      })
+      continue
+    }
+    if (type === 'compact/summary' && activeCompaction !== undefined) {
+      const event = sourceEvent as unknown as CompactionSummaryEvent
+      update(activeCompaction, {
+        resultSeq: event.seq,
+        summary: event.data.summary,
+        ...(event.data.rawOutput === undefined ? {} : { rawOutput: event.data.rawOutput }),
+        provenance: {
+          provider: event.data.provider,
+          model: event.data.model,
+        },
+        requestConfig: {
+          provider: event.data.provider,
+          model: event.data.model,
+          purpose: 'compaction',
+          ...(event.data.maxTokens === undefined ? {} : { maxTokens: event.data.maxTokens }),
+        },
+        ...(event.data.usage === undefined ? {} : { usage: event.data.usage }),
+      })
+      continue
+    }
+    if (
+      sourceEvent.type === 'user/message'
+      && activeCompaction !== undefined
+      && isCompactionSource(sourceEvent.data.source)
+    ) {
+      update(activeCompaction, { replacementSeq: sourceEvent.seq })
+      continue
+    }
+    if (type !== 'compact/end' || activeCompaction === undefined) continue
+    const event = sourceEvent as unknown as CompactionEndEvent
+    update(activeCompaction, {
+      completedAt: event.time,
+      status: event.data.error === undefined ? 'complete' : 'error',
+      ...(event.data.error === undefined ? {} : { error: event.data.error }),
+    })
+    activeCompaction = undefined
+  }
+
+  return requests.sort((left, right) => left.startSeq - right.startSeq)
+}
+
+function isCompactionSource(source: unknown): boolean {
+  return typeof source === 'object'
+    && source !== null
+    && 'kind' in source
+    && source.kind === 'plugin'
+    && 'plugin' in source
+    && source.plugin === 'compact'
+}
diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts
index 0f5d39ac8e..f18ced4dfe 100644
--- a/packages/client/runtime/src/client/sessions/session.ts
+++ b/packages/client/runtime/src/client/sessions/session.ts
@@ -252,6 +252,25 @@ export class Session implements SessionFace {
     return result
   }
 
+  /**
+   * Rename: contract session.rename 1:1. On success settle the 'title'
+   * projection cell from the response's `{title, seq}` under the store's
+   * higher-seq-wins rule (the push frame arriving later is a no-op replay),
+   * so the list row and any useProjection('title') reader update without
+   * waiting for the mux frame.
+   * @param title - raw title text (the host normalizes acceptance).
+   * @returns the rename result (normalized accepted title + title event seq).
+   */
+  async rename(title: string): Promise<RpcResult<{ title: string; seq: number }>> {
+    try {
+      const { result } = await this.api.sessions.rename({ sessionId: this.sessionId, title })
+      if (result.ok) this.projections.apply('title', result.value.title, result.value.seq)
+      return result
+    } catch (error) {
+      return transportError(error)
+    }
+  }
+
   /**
    * Execute one slash-command line against this session's agent — pure
    * admission semantics (the host executor durably logs the lifecycle;
diff --git a/packages/client/runtime/tests/fake-api.ts b/packages/client/runtime/tests/fake-api.ts
index 0a1de3f7e1..04ca13ee11 100644
--- a/packages/client/runtime/tests/fake-api.ts
+++ b/packages/client/runtime/tests/fake-api.ts
@@ -63,6 +63,7 @@ export class FakeApiClient implements IApiClient {
   onList: (payload: unknown) => Promise<RpcResponse<{ items: never[] }>> = () => Promise.resolve(ok({ items: [] }))
   onCreate: (payload: unknown) => Promise<RpcResponse<{ sessionId: SessionId }>> = () => Promise.resolve(ok({ sessionId: 'fk-new' as SessionId }))
   readonly defaultModel: ModelTarget = { provider: 'deepseek', model: 'deepseek-v4-flash' }
+  onRename: (payload: unknown) => Promise<RpcResponse<{ title: string; seq: number }>> = () => Promise.resolve(ok({ title: 'fk-renamed', seq: 0 }))
   onHistory: (payload: { sessionId: SessionId; beforeSeq?: number; maxMessages?: number })
   => Promise<RpcResponse<{ events: never[]; hasMore: boolean }>> =
     () => Promise.resolve(ok({ events: [], hasMore: false }))
@@ -115,6 +116,7 @@ export class FakeApiClient implements IApiClient {
     models: (payload: unknown) => this.record('session.models', payload, this.onModels(payload)),
     selectModel: (payload: { provider: string; model: string }) =>
       this.record('session.selectModel', payload, this.onSelectModel(payload)),
+    rename: (payload: unknown) => this.record('session.rename', payload, this.onRename(payload)),
     prompt: (payload: unknown) => this.record('session.prompt', payload, this.onPrompt(payload)),
     cancel: (payload: unknown) => this.record('session.cancel', payload, this.onCancel(payload)),
   }
diff --git a/packages/client/runtime/tests/fold-adapter.spec.ts b/packages/client/runtime/tests/fold-adapter.spec.ts
index b40bdb4111..0544d84e7d 100644
--- a/packages/client/runtime/tests/fold-adapter.spec.ts
+++ b/packages/client/runtime/tests/fold-adapter.spec.ts
@@ -8,6 +8,7 @@ import { createUserMessage, CallId, createMessage, createToolResultMessage } fro
 import { describe, expect, it, vi } from 'vitest'
 import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
 import { FoldAdapter } from '../src/client/sessions/fold-adapter.ts'
+import { projectConversationHistory } from '../src/client/session-history/history-fold.ts'
 import { ev, plainTurn } from './event-script.ts'
 
 const at = (seq: number, e: Record<string, unknown>): SessionEvent =>
@@ -27,6 +28,7 @@ describe('FoldAdapter', () => {
     const adapter = new FoldAdapter()
     adapter.reset(plainTurn(0, 0, 'a', 'b'), 0)
     const first = adapter.nodes()
+    expect(adapter.nodes()).toBe(first)
     adapter.append(ev.user(6, '追加'))
     const second = adapter.nodes()
     expect(second.nodes).toHaveLength(3)
@@ -35,6 +37,52 @@ describe('FoldAdapter', () => {
     expect(second.nodes).not.toBe(first.nodes) // array itself fresh per call
   })
 
+  it('projects frozen surface generations without widening the core live surface', () => {
+    const events = [
+      ev.user(0, 'a'),
+      ev.user(1, 'b'),
+      at(2, {
+        type: 'assistant/message',
+        surfaceOp: { op: 'replace', start: 0, end: 0 },
+        sourceEventSeqs: [0],
+        data: {
+          turn: 1,
+          step: 1,
+          message: createMessage({
+            role: 'assistant',
+            content: [{ type: 'text', text: 'summary' }],
+            source: { kind: 'model', provider: 'fake', model: 'fake' },
+          }),
+        },
+      }),
+      at(3, {
+        type: 'assistant/message',
+        surfaceOp: { op: 'replace', start: 2, end: 1 },
+        sourceEventSeqs: [2, 1],
+        data: {
+          turn: 1,
+          step: 2,
+          message: createMessage({
+            role: 'assistant',
+            content: [{ type: 'text', text: 'summary 2' }],
+            source: { kind: 'model', provider: 'fake', model: 'fake' },
+          }),
+        },
+      }),
+    ]
+
+    expect(projectConversationHistory(events.map(event => ({ event }))).contexts.map(context => ({
+      id: context.id,
+      parentId: context.parentId,
+      originSeq: context.originSeq,
+      nodes: context.nodes.map(node => node.seq),
+    }))).toEqual([
+      { id: 0, parentId: undefined, originSeq: undefined, nodes: [0, 1] },
+      { id: 1, parentId: 0, originSeq: 2, nodes: [2, 1] },
+      { id: 2, parentId: 1, originSeq: 3, nodes: [3] },
+    ])
+  })
+
   it('materializes all six node variants with field mapping', () => {
     const adapter = new FoldAdapter()
     const events = [
@@ -114,6 +162,68 @@ describe('FoldAdapter', () => {
     }
   })
 
+  it('silently degrades when a replacement needs an earlier history page', () => {
+    const adapter = new FoldAdapter()
+    const errorSpy = vi.spyOn(console, 'error').mockImplementation(() => undefined)
+    try {
+      adapter.reset([
+        at(10, {
+          type: 'assistant/message',
+          surfaceOp: { op: 'replace', start: 1, end: 3 },
+          sourceEventSeqs: [1, 3],
+          data: {
+            turn: 1,
+            step: 1,
+            message: createMessage({
+              role: 'assistant',
+              content: [{ type: 'text', text: 'partial summary' }],
+              source: { kind: 'model', provider: 'fake', model: 'fake' },
+            }),
+          },
+        }),
+        ev.user(11, 'newer message'),
+      ], 10)
+
+      expect(adapter.nodes()).toMatchObject({
+        degraded: true,
+        nodes: [{ seq: 10 }, { seq: 11 }],
+      })
+      expect(errorSpy).not.toHaveBeenCalled()
+    } finally {
+      errorSpy.mockRestore()
+    }
+  })
+
+  it('silently degrades when a live replacement needs an earlier history page', () => {
+    const adapter = new FoldAdapter()
+    const errorSpy = vi.spyOn(console, 'error').mockImplementation(() => undefined)
+    try {
+      adapter.reset([ev.user(10, 'window head')], 10)
+      adapter.append(at(11, {
+        type: 'assistant/message',
+        surfaceOp: { op: 'replace', start: 1, end: 1 },
+        sourceEventSeqs: [1],
+        data: {
+          turn: 1,
+          step: 1,
+          message: createMessage({
+            role: 'assistant',
+            content: [{ type: 'text', text: 'live summary' }],
+            source: { kind: 'model', provider: 'fake', model: 'fake' },
+          }),
+        },
+      }))
+
+      expect(adapter.nodes()).toMatchObject({
+        degraded: true,
+        nodes: [{ seq: 10 }, { seq: 11 }],
+      })
+      expect(errorSpy).not.toHaveBeenCalled()
+    } finally {
+      errorSpy.mockRestore()
+    }
+  })
+
   it('materializes a tool-result error field when present', () => {
     const adapter = new FoldAdapter()
     adapter.reset([
@@ -130,6 +240,44 @@ describe('FoldAdapter', () => {
     expect(adapter.nodes().nodes[0]).toMatchObject({ kind: 'tool-result', isError: true, error: { code: 'boom' } })
   })
 
+  it('projects assistant timing and the active request header from history', () => {
+    const projection = projectConversationHistory([
+      ev.stepStart(0, 1, 2),
+      at(1, { type: 'request/header', data: {
+        reason: 'initial',
+        header: {
+          config: { provider: 'fake', model: 'first' },
+          tools: [],
+        },
+      } }),
+      ev.chunkStart(2, 1, 2),
+      ev.chunkText(3, 1, 'token', 2),
+      ev.assistant(4, 1, 'done', 2),
+      ev.stepStart(5, 2, 1),
+      ev.chunkText(6, 2, 'next', 1),
+      ev.assistant(7, 2, 'next done', 1),
+    ].map(event => ({ event })))
+
+    expect(projection.eventNodes[0]).toMatchObject({
+      kind: 'assistant',
+      timing: {
+        stepStartTime: 1_700_000_000_000,
+        firstTokenTime: 1_700_000_000_003,
+        completedTime: 1_700_000_000_004,
+      },
+      requestConfig: { provider: 'fake', model: 'first' },
+    })
+
+    expect(projection.eventNodes.at(-1)).toMatchObject({
+      timing: {
+        stepStartTime: 1_700_000_000_005,
+        firstTokenTime: 1_700_000_000_006,
+        completedTime: 1_700_000_000_007,
+      },
+      requestConfig: { provider: 'fake', model: 'first' },
+    })
+  })
+
   it('exposes the in-window call index for runningCalls material', () => {
     const adapter = new FoldAdapter()
     adapter.reset([ev.toolCall(0, 1, 'c9', 'slow', '{}')], 0)
diff --git a/packages/client/runtime/tests/request-inspection.spec.ts b/packages/client/runtime/tests/request-inspection.spec.ts
new file mode 100644
index 0000000000..a76036c076
--- /dev/null
+++ b/packages/client/runtime/tests/request-inspection.spec.ts
@@ -0,0 +1,184 @@
+import { describe, expect, it } from 'vitest'
+import type { HistoryEntry } from '@deepseek-ai/dsh-client-connection/client'
+import { createAssistantMessage, createUserMessage } from '@deepseek-ai/dsh-llm'
+import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
+import { inspectRequests } from '../src/client/sessions/request-inspection.ts'
+
+const at = (seq: number, type: string, data: unknown): SessionEvent =>
+  ({ seq, time: 1_700_000_000_000 + seq, type, data }) as SessionEvent
+
+const entriesOf = (events: readonly SessionEvent[]): HistoryEntry[] =>
+  events.map(event => ({ event }))
+
+describe('inspectRequests', () => {
+  it('projects ordinary and compaction calls into one chronological request stream', () => {
+    const events = [
+      at(0, 'step/start', { turn: 1, step: 1 }),
+      at(1, 'request/header', {
+        reason: 'initial',
+        header: {
+          config: { provider: 'fake', model: 'model' },
+          system: 'system',
+          tools: [{
+            name: 'read',
+            description: 'Read a file.',
+            parameters: { type: 'object' },
+          }],
+        },
+      }),
+      at(2, 'tool/call', {
+        turn: 1,
+        step: 1,
+        callId: 'call-1',
+        name: 'read',
+        arguments: '{}',
+      }),
+      at(3, 'assistant/message', {
+        turn: 1,
+        step: 1,
+        message: createAssistantMessage({
+          content: [{ type: 'text', text: 'done' }],
+          source: { provider: 'fake', model: 'model' },
+        }),
+        usage: { inputTokens: 5, outputTokens: 2 },
+      }),
+      at(4, 'step/end', { turn: 1, step: 1 }),
+      at(5, 'compact/start', { turn: 1 }),
+      at(6, 'compact/summary', {
+        summary: [{ type: 'text', text: 'summary' }],
+        rawOutput: [
+          { type: 'reasoning', text: 'thought' },
+          { type: 'text', text: 'summary' },
+        ],
+        provider: 'fake',
+        model: 'compact-model',
+        usage: { inputTokens: 8, outputTokens: 3 },
+      }),
+      at(7, 'user/message', createUserMessage({
+        content: [{ type: 'text', text: 'checkpoint' }],
+        source: { kind: 'plugin', plugin: 'compact' },
+      })),
+      at(8, 'compact/end', { turn: 1 }),
+    ]
+    const snapshot = inspectRequests(entriesOf(events))
+    expect(snapshot.requests).toMatchObject([
+      {
+        purpose: 'assistant',
+        startSeq: 0,
+        resultSeq: 3,
+        status: 'complete',
+        prompt: {
+          config: { provider: 'fake', model: 'model' },
+          system: 'system',
+        },
+        promptChange: { seq: 1, kind: 'initial' },
+      },
+      {
+        purpose: 'compaction',
+        startSeq: 5,
+        resultSeq: 6,
+        replacementSeq: 7,
+        status: 'complete',
+        summary: [{ type: 'text', text: 'summary' }],
+      },
+    ])
+    expect(snapshot.callSchemas.get('call-1')?.name).toBe('read')
+  })
+
+  it('captures schemas for nested tool dispatches from the active request header', () => {
+    const snapshot = inspectRequests(entriesOf([
+      at(0, 'request/header', {
+        reason: 'initial',
+        header: {
+          config: { provider: 'fake', model: 'model' },
+          tools: [{
+            name: 'read',
+            description: 'Read a file.',
+            parameters: { type: 'object' },
+          }],
+        },
+      }),
+      at(1, 'tool/code-dispatch-start', {
+        parentCallId: 'parent',
+        subCallId: 'nested',
+        name: 'read',
+        arguments: {},
+      }),
+    ]))
+
+    expect(snapshot.callSchemas.get('nested')?.name).toBe('read')
+  })
+
+  it('keeps chunk-reported usage through request failure and prefers it to message fallback', () => {
+    const chunkUsage = { inputTokens: 21, outputTokens: 3 }
+    const retryUsage = {
+      inputTokens: 5,
+      outputTokens: 2,
+      cacheReadTokens: 8,
+      reasoningTokens: 1,
+    }
+    const snapshot = inspectRequests(entriesOf([
+      at(0, 'step/start', { turn: 1, step: 1 }),
+      at(1, 'assistant/chunk', {
+        turn: 1,
+        step: 1,
+        chunk: { type: 'usage', usage: chunkUsage },
+      }),
+      at(2, 'llm/retry', {
+        turn: 1,
+        step: 1,
+        retry: 1,
+        maxRetries: 2,
+        delayMs: 100,
+        failure: { message: 'rate limited' },
+      }),
+      at(3, 'assistant/chunk', {
+        turn: 1,
+        step: 1,
+        chunk: { type: 'usage', usage: retryUsage },
+      }),
+      at(4, 'assistant/message', {
+        turn: 1,
+        step: 1,
+        message: createAssistantMessage({
+          content: [{ type: 'text', text: 'recovered' }],
+          source: { provider: 'fake', model: 'model' },
+        }),
+        usage: { inputTokens: 1, outputTokens: 1 },
+      }),
+    ]))
+
+    expect(snapshot.requests[0]).toMatchObject({
+      status: 'complete',
+      usage: {
+        inputTokens: 26,
+        outputTokens: 5,
+        cacheReadTokens: 8,
+        reasoningTokens: 1,
+      },
+    })
+  })
+
+  it('treats a scrubbed durable-fixture tool catalog as unavailable', () => {
+    const snapshot = inspectRequests(entriesOf([
+      at(0, 'step/start', { turn: 1, step: 1 }),
+      at(1, 'request/header', {
+        reason: 'initial',
+        header: {
+          config: { provider: 'fake', model: 'model' },
+          tools: '{{tools}}',
+        },
+      }),
+      at(2, 'tool/call', {
+        turn: 1,
+        step: 1,
+        callId: 'call-1',
+        name: 'read',
+        arguments: '{}',
+      }),
+    ]))
+
+    expect(snapshot.callSchemas).toEqual(new Map())
+    expect(snapshot.requests[0]?.prompt?.tools).toEqual([])
+  })
+})
diff --git a/packages/client/runtime/tests/session-history-source.spec.ts b/packages/client/runtime/tests/session-history-source.spec.ts
new file mode 100644
index 0000000000..1375338e90
--- /dev/null
+++ b/packages/client/runtime/tests/session-history-source.spec.ts
@@ -0,0 +1,98 @@
+import { describe, expect, it } from 'vitest'
+import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
+import type { SessionId } from '@deepseek-ai/dsh-client-connection/client'
+import { SessionHistorySource } from '../src/client/session-history/source.ts'
+import { FakeApiClient, deferred, err, ok } from './fake-api.ts'
+import { entries, ev, plainTurn } from './event-script.ts'
+
+const SID = 'history-s1' as SessionId
+
+function histResponse(events: SessionEvent[], hasMore = false) {
+  return Promise.resolve(ok({ events: entries(events) as never[], hasMore }))
+}
+
+describe('SessionHistorySource', () => {
+  it('loads every older page without changing a Chat session', async () => {
+    const pages = [
+      plainTurn(0, 0, '最早问', '最早答'),
+      plainTurn(6, 1, '中间问', '中间答'),
+      plainTurn(12, 2, '最新问', '最新答'),
+    ]
+    const api = new FakeApiClient()
+    api.onHistory = (payload) => {
+      if (payload.beforeSeq === undefined) return histResponse(pages[2]!, true)
+      if (payload.beforeSeq === 12) return histResponse(pages[1]!, true)
+      return histResponse(pages[0]!, false)
+    }
+    const source = new SessionHistorySource(SID, api)
+
+    await source.loadAll()
+
+    expect(api.callsOf('session.history')).toHaveLength(3)
+    expect(source.getSnapshot().hasMore).toBe(false)
+    expect(source.getSnapshot().inspection.eventNodes.map(node => node.seq))
+      .toEqual([1, 3, 7, 9, 13, 15])
+  })
+
+  it('pins a lazy inspection to the entries in its source snapshot', async () => {
+    const api = new FakeApiClient()
+    api.onHistory = () => histResponse(plainTurn(0, 0, '问', '答'))
+    const source = new SessionHistorySource(SID, api)
+    await source.loadAll()
+    const before = source.getSnapshot()
+
+    source.handleMuxFrame({
+      type: 'session/event',
+      sessionId: SID,
+      event: ev.user(6, 'later'),
+    })
+
+    expect(before.inspection.eventNodes.map(node => node.seq)).toEqual([1, 3])
+    expect(source.getSnapshot().inspection.eventNodes.map(node => node.seq))
+      .toEqual([1, 3, 6])
+  })
+
+  it('stops loading when an older page fails to advance', async () => {
+    const api = new FakeApiClient()
+    api.onHistory = payload => payload.beforeSeq === undefined
+      ? histResponse(plainTurn(6, 1, '新问', '新答'), true)
+      : Promise.resolve(err({
+        code: 'internal',
+        message: 'page unavailable',
+        details: {},
+      }))
+    const source = new SessionHistorySource(SID, api)
+
+    await source.loadAll()
+
+    expect(api.callsOf('session.history')).toHaveLength(2)
+    expect(source.getSnapshot().hasMore).toBe(true)
+  })
+
+  it('observes consumer cancellation between older pages', async () => {
+    const middle = deferred<Awaited<ReturnType<FakeApiClient['onHistory']>>>()
+    const olderStarted = deferred<undefined>()
+    const api = new FakeApiClient()
+    api.onHistory = (payload) => {
+      if (payload.beforeSeq === undefined) {
+        return histResponse(plainTurn(12, 2, '最新问', '最新答'), true)
+      }
+      olderStarted.resolve(undefined)
+      return middle.promise
+    }
+    const source = new SessionHistorySource(SID, api)
+    const controller = new AbortController()
+    const complete = source.loadAll(controller.signal)
+    await olderStarted.promise
+    controller.abort()
+    middle.resolve(ok({
+      events: entries(plainTurn(6, 1, '中间问', '中间答')) as never[],
+      hasMore: true,
+    }))
+
+    await complete
+
+    expect(api.callsOf('session.history')).toHaveLength(2)
+    expect(source.getSnapshot().hasMore).toBe(true)
+  })
+})
diff --git a/packages/client/runtime/tests/session.spec.ts b/packages/client/runtime/tests/session.spec.ts
index c7be330d55..c80f046ceb 100644
--- a/packages/client/runtime/tests/session.spec.ts
+++ b/packages/client/runtime/tests/session.spec.ts
@@ -301,6 +301,32 @@ describe('prompt and cancel errors', () => {
   })
 })
 
+describe('rename', () => {
+  it('settles the title projection cell from the unary response (higher-seq-wins vs the push frame)', async () => {
+    const { api, session } = makeSession()
+    api.onRename = () => Promise.resolve(ok({ title: '正名', seq: 7 }))
+    const result = await session.rename('  正名  ')
+    expect(result).toMatchObject({ ok: true, value: { title: '正名', seq: 7 } })
+    expect(api.callsOf('session.rename')).toMatchObject([{ sessionId: SID, title: '  正名  ' }])
+    expect(session.projections.faceOf('title').getSnapshot()).toBe('正名')
+    // A stale lower-seq apply (the push-frame path routes into this same
+    // store) must not roll the settled value back.
+    session.projections.apply('title', '旧名', 3)
+    expect(session.projections.faceOf('title').getSnapshot()).toBe('正名')
+  })
+
+  it('returns the business error untouched and folds a transport throw to internal', async () => {
+    const { api, session } = makeSession()
+    api.onRename = () => Promise.resolve(err({ code: 'title-invalid', message: 'empty', details: { sessionId: SID } }))
+    const rejected = await session.rename('   ')
+    expect(rejected).toMatchObject({ ok: false, error: { code: 'title-invalid' } })
+    expect(session.projections.faceOf('title').getSnapshot()).toBeUndefined()
+    api.onRename = () => Promise.reject(new Error('rename transport down'))
+    const folded = await session.rename('x')
+    expect(folded).toMatchObject({ ok: false, error: { code: 'internal' } })
+  })
+})
+
 describe('pending interactions', () => {
   it('adds approval/question on requested and removes them on resolved', async () => {
     const { session } = makeSession()
@@ -701,6 +727,7 @@ describe('resync', () => {
     expect(snapshot.openState).toBe('open') // stale failure did not settle the fresh generation into error
     expect(snapshot.nodes.map(n => n.seq)).toEqual([7, 9])
   })
+
 })
 
 describe('run_code sub-dispatch indexing', () => {
@@ -814,20 +841,23 @@ describe('reference stability (the memo contract)', () => {
     await session.open()
     const feed = (event: SessionEvent) => { session.handleMuxEnvelope('r' as never, { type: 'session/event', sessionId: SID, event }) }
     feed(ev.turnStart(6, 1))
-    feed(ev.toolCall(7, 1, 'c1', 'echo', '{}'))
+    feed(ev.stepStart(7, 1))
+    feed(ev.toolCall(8, 1, 'c1', 'echo', '{}'))
     session.handleMuxEnvelope('ra' as never, { type: 'approval/requested', sessionId: SID, approvalId: 'ap1' as never, toolName: 'rm' })
     const before = session.getSnapshot()
-    // A chunk storm touches partial/nodes only: runningCalls and pending must keep identity.
-    feed(ev.chunkStart(8, 1))
-    feed(ev.chunkText(9, 1, '与工具无关的流式'))
+    // A chunk storm touches partial/nodes only: unrelated projections keep identity.
+    feed(ev.chunkStart(9, 1))
+    feed(ev.chunkText(10, 1, '与工具无关的流式'))
     const after = session.getSnapshot()
     expect(after).not.toBe(before)
     expect(after.runningCalls).toBe(before.runningCalls)
     expect(after.pending).toBe(before.pending)
     // And a mutation on the tracked domain swaps that array.
-    feed(ev.toolResult(10, 1, 'c1', 'ECHO'))
+    feed(ev.toolResult(11, 1, 'c1', 'ECHO'))
     const resolved = session.getSnapshot()
     expect(resolved.runningCalls).not.toBe(after.runningCalls)
     expect(resolved.pending).toBe(after.pending)
+    feed(ev.assistant(12, 1, '完成'))
+    expect(session.getSnapshot()).not.toBe(resolved)
   })
 })
diff --git a/packages/client/test-runtime/src/sessions.ts b/packages/client/test-runtime/src/sessions.ts
index 5ec4652aef..59dd7d9186 100644
--- a/packages/client/test-runtime/src/sessions.ts
+++ b/packages/client/test-runtime/src/sessions.ts
@@ -107,6 +107,14 @@ export class FixtureSession implements SessionFace {
   loadOlder(): never {
     throw new Error(`test session "${this.sessionId}": loadOlder is not stubbed — supply it on the fixture's session face`)
   }
+
+  /**
+   * Fail-loud stub; supply `rename` on the fixture's session face to exercise it.
+   * @returns never — always throws.
+   */
+  rename(): never {
+    throw new Error(`test session "${this.sessionId}": rename is not stubbed — supply it on the fixture's session face`)
+  }
 }
 
 /** One live test session: fixture-derived stores plus its minted scope state. */
diff --git a/packages/client/test-runtime/tests/runtime.spec.tsx b/packages/client/test-runtime/tests/runtime.spec.tsx
index 22d3bb5cfc..00faae9905 100644
--- a/packages/client/test-runtime/tests/runtime.spec.tsx
+++ b/packages/client/test-runtime/tests/runtime.spec.tsx
@@ -470,6 +470,7 @@ describe('fixture session face', () => {
     expect(() => bare.cancel()).toThrow(/cancel is not stubbed/)
     expect(() => bare.command()).toThrow(/command is not stubbed/)
     expect(() => bare.loadOlder()).toThrow(/loadOlder is not stubbed/)
+    expect(() => bare.rename()).toThrow(/rename is not stubbed/)
     await runtime.dispose()
   })
 
diff --git a/packages/client/tsdown.client.ts b/packages/client/tsdown.client.ts
index 9b93feae8b..bf560d4d53 100644
--- a/packages/client/tsdown.client.ts
+++ b/packages/client/tsdown.client.ts
@@ -6,6 +6,7 @@
  * lightningcss inside the bundle: importing `x.module.css` yields the
  * hashed class map, and the css text auto-injects a <style data-plugin="<id>">
  * tag at factory execution (the loader removes plugin-owned tags on unload).
+ * The virtual loader registers each real stylesheet as a watch dependency.
  */
 import { readFile } from 'node:fs/promises'
 import { basename, dirname, resolve as resolvePath } from 'node:path'
@@ -127,6 +128,7 @@ export function clientBundle(id: string, libEntry: readonly string[]): UserConfi
       async load(virtualId: string) {
         if (!virtualId.startsWith(CSS_VIRTUAL_PREFIX)) return null
         const fileId = virtualId.slice(CSS_VIRTUAL_PREFIX.length, -CSS_VIRTUAL_SUFFIX.length)
+        this.addWatchFile(fileId)
         const source = await readFile(fileId)
         const { code, exports: cssExports } = transform({
           filename: fileId,
diff --git a/packages/client/ui-command/src/client/PopupSelectView.module.css b/packages/client/ui-command/src/client/PopupSelectView.module.css
index 14cf581e13..16e1e5c11d 100644
--- a/packages/client/ui-command/src/client/PopupSelectView.module.css
+++ b/packages/client/ui-command/src/client/PopupSelectView.module.css
@@ -13,8 +13,10 @@
   display: flex;
   flex-direction: column;
   min-width: 220px;
+  /* Height cap: the 320px design maximum, clamped at runtime to the space
+   * above the composer (inline max-height set in PopupSelectView.tsx). */
   max-height: 320px;
-  overflow-y: auto;
+  overflow: hidden;
   /* Elevated surface: the scrollbar thumb takes the l2 elevation tokens
      (see ui-theme styles/scrollbar.css for the rebinding contract). */
   --dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2);
@@ -26,6 +28,13 @@
   outline: none;
 }
 
+.viewport {
+  display: flex;
+  flex-direction: column;
+  min-height: 0;
+  overflow-y: auto;
+}
+
 .row {
   display: flex;
   align-items: center;
@@ -34,11 +43,11 @@
   border-radius: 8px;
   cursor: pointer;
   font-size: 13px;
-  color: var(--dsw-alias-text-primary);
+  color: var(--dsw-alias-label-primary);
 }
 
 .rowActive {
-  background: var(--dsw-alias-fill-hover);
+  background: var(--dsw-alias-interactive-bg-hover);
 }
 
 .label {
@@ -50,19 +59,20 @@
 
 .detail {
   font-size: 12px;
-  color: var(--dsw-alias-text-tertiary);
+  color: var(--dsw-alias-label-tertiary);
   white-space: nowrap;
 }
 
 .check {
   display: inline-flex;
-  color: var(--dsw-alias-text-secondary);
+  flex: none;
+  color: var(--dsw-alias-label-primary);
 }
 
 .status {
-  padding: 8px;
-  font-size: 12px;
-  color: var(--dsw-alias-text-tertiary);
+  padding: 8px 10px;
+  font-size: 13px;
+  color: var(--dsw-alias-label-tertiary);
 }
 
 .search {
@@ -72,7 +82,7 @@
   border-radius: 8px;
   background: transparent;
   font-size: 13px;
-  color: var(--dsw-alias-text-primary);
+  color: var(--dsw-alias-label-primary);
   outline: none;
 }
 
@@ -97,6 +107,6 @@
   border-radius: 6px;
   background: transparent;
   font-size: 12px;
-  color: var(--dsw-alias-text-primary);
+  color: var(--dsw-alias-label-primary);
   cursor: pointer;
 }
diff --git a/packages/client/ui-command/src/client/PopupSelectView.tsx b/packages/client/ui-command/src/client/PopupSelectView.tsx
index 9d2807ded1..ec0bbdd2bb 100644
--- a/packages/client/ui-command/src/client/PopupSelectView.tsx
+++ b/packages/client/ui-command/src/client/PopupSelectView.tsx
@@ -3,19 +3,23 @@
  * store into the conversation.input.overlay anchor. Unlike the slash menu
  * (combobox — textarea keeps focus), this shell HOLDS focus while open: the
  * inner search input takes focus, plain typing filters the loaded options
- * locally, Enter/↑↓ drive the filtered highlight, Escape dismisses back to
- * the composer, and ←→ keep the search input's native caret. Any pointer
- * interaction outside the box dismisses (the click's own target takes
- * focus). Closed state renders null; the overlay slot stays mounted.
+ * locally, Enter/↑↓ drive the filtered highlight (scrolled into view), Escape
+ * dismisses back to the composer, and ←→ keep the search input's native
+ * caret. Any pointer interaction outside the box dismisses (the click's own
+ * target takes focus). Closed state renders null; the overlay slot stays
+ * mounted. The card height clamps to the space above the composer.
  */
 import { useEffect, useRef } from 'react'
 import { useSyncExternalStore } from 'react'
 import clsx from 'clsx'
-import { IconCheckOutline16 } from '@deepseek-ai/dsh-client-ui-primitives'
+import { IconCheckOutline16, useAnchoredMaxHeight } from '@deepseek-ai/dsh-client-ui-primitives'
 import { filterOptions } from './popup.ts'
 import type { PopupSelectController } from './popup.ts'
 import css from './PopupSelectView.module.css'
 
+/** Design cap on the card height (same MenuDropdown family as the slash menu). */
+const MAX_HEIGHT = 320
+
 /** Injected business face of the popupSelect overlay entry. */
 export interface PopupSelectInjected {
   /** The session's shell controller (state store + verbs; the view never touches the open-context type). */
@@ -34,6 +38,17 @@ export function PopupSelectView({ popup }: PopupSelectInjected) {
   )
   const cardRef = useRef<HTMLDivElement>(null)
   const searchRef = useRef<HTMLInputElement>(null)
+  // The card is bottom-anchored above the composer; clamp the design cap to
+  // the space above it, re-measured on every store update.
+  const maxHeight = useAnchoredMaxHeight(cardRef, MAX_HEIGHT, state)
+  const active = state.open ? state.active : null
+
+  // The search input keeps focus while arrows move a virtual highlight, so
+  // the browser never scrolls the active row into view — do it here.
+  useEffect(() => {
+    if (active === null) return
+    cardRef.current?.querySelector('[aria-selected="true"]')?.scrollIntoView({ block: 'nearest' })
+  }, [active])
 
   // Focus ownership: the search input grabs on open (the design's
   // transient-layer rule), and ANY outside pointer interaction dismisses —
@@ -42,7 +57,6 @@ export function PopupSelectView({ popup }: PopupSelectInjected) {
   // takes focus naturally, so no focusComposer here.
   useEffect(() => {
     if (!state.open) return
-    searchRef.current?.focus()
     const onPointerDown = (ev: PointerEvent): void => {
       if (cardRef.current !== null && ev.target instanceof Node && cardRef.current.contains(ev.target)) return
       popup.dismiss()
@@ -51,6 +65,11 @@ export function PopupSelectView({ popup }: PopupSelectInjected) {
     return () => { document.removeEventListener('pointerdown', onPointerDown, true) }
   }, [state.open, popup])
 
+  // Focus the search input after it mounts (separate effect so the ref is populated).
+  useEffect(() => {
+    if (state.open) searchRef.current?.focus()
+  }, [state.open])
+
   if (!state.open) return null
 
   const rows = filterOptions(state.options, state.search)
@@ -83,6 +102,7 @@ export function PopupSelectView({ popup }: PopupSelectInjected) {
     <div
       ref={cardRef}
       className={css.card}
+      style={{ maxHeight }}
       aria-label={`/${String(state.command)} options`}
       onKeyDown={onKeyDown}
     >
@@ -108,7 +128,7 @@ export function PopupSelectView({ popup }: PopupSelectInjected) {
       {state.submitting && <div className={css.status}>Applying…</div>}
       {state.status === 'ready' && rows.length === 0 && <div className={css.status}>No options</div>}
       {state.status === 'ready' && (
-        <div role="listbox" aria-label={`/${String(state.command)} matches`}>
+        <div role="listbox" aria-label={`/${String(state.command)} matches`} className={css.viewport}>
           {rows.map((option, index) => (
             <div
               key={option.id}
diff --git a/packages/client/ui-command/tests/popup-view.spec.tsx b/packages/client/ui-command/tests/popup-view.spec.tsx
index bda650e190..2cd9891478 100644
--- a/packages/client/ui-command/tests/popup-view.spec.tsx
+++ b/packages/client/ui-command/tests/popup-view.spec.tsx
@@ -4,17 +4,28 @@
  * focus on open and plain typing filters locally, ↑↓ move the filtered
  * highlight while ←→ stay native to the input, Enter selects single-flight,
  * Escape dismisses back through focusComposer, outside pointerdown dismisses
- * plainly, and the submitting/failed states render pending text and a
- * working retry button.
+ * plainly, the submitting/failed states render pending text and a working
+ * retry button, the highlighted row scrolls into view, and the card height
+ * clamps to the space above the composer.
  */
-import { afterEach, describe, expect, it, vi } from 'vitest'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
 import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
 import type { SelectOption } from '../src/client/contract.ts'
 import type { PopupSpec, TokenSegment } from '../src/client/popup.ts'
 import { PopupSelectController } from '../src/client/popup.ts'
 import { PopupSelectView } from '../src/client/PopupSelectView.tsx'
 
-afterEach(cleanup)
+// jsdom has no scrollIntoView; the view calls it on the highlighted row.
+const scrollIntoView = vi.fn()
+beforeEach(() => {
+  Element.prototype.scrollIntoView = scrollIntoView
+  scrollIntoView.mockClear()
+})
+
+afterEach(() => {
+  cleanup()
+  vi.restoreAllMocks()
+})
 
 const OPTIONS: SelectOption[] = [
   { id: 'dark', label: 'Dark' },
@@ -87,6 +98,27 @@ describe('PopupSelectView', () => {
     expect(fireEvent.keyDown(search, { key: 'ArrowRight' })).toBe(true)
   })
 
+  it('scrolls the highlighted row into view when the highlight moves', async () => {
+    const { search } = await mountOpen()
+    scrollIntoView.mockClear()
+    act(() => { fireEvent.keyDown(search, { key: 'ArrowDown' }) })
+    const options = screen.getAllByRole('option')
+    expect(scrollIntoView).toHaveBeenCalledWith({ block: 'nearest' })
+    expect(scrollIntoView.mock.instances.at(-1)).toBe(options[1])
+  })
+
+  it('caps the card height at the design maximum when the composer sits low enough', async () => {
+    vi.spyOn(Element.prototype, 'getBoundingClientRect').mockReturnValue({ bottom: 800 } as DOMRect)
+    await mountOpen()
+    expect(screen.getByLabelText('/theme options').style.maxHeight).toBe('320px')
+  })
+
+  it('clamps the card height to the space above the composer minus the safe margin', async () => {
+    vi.spyOn(Element.prototype, 'getBoundingClientRect').mockReturnValue({ bottom: 200 } as DOMRect)
+    await mountOpen()
+    expect(screen.getByLabelText('/theme options').style.maxHeight).toBe('188px')
+  })
+
   it('Enter selects the highlighted row: onSelect, consume, close, focusComposer', async () => {
     const seen: Array<{ option: SelectOption; context: string }> = []
     const { view, search, consume, focusComposer } = await mountOpen({
diff --git a/packages/client/ui-conversation/README.i18n.yaml b/packages/client/ui-conversation/README.i18n.yaml
index 9f7a544438..31a714a655 100644
--- a/packages/client/ui-conversation/README.i18n.yaml
+++ b/packages/client/ui-conversation/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-conversation/README.md
-README.md: ea86c6fd430d99a64615b2b01668ada801ffe729
-README.zh.md: 9336d3d36087c41e84dacabe80a697c2ebe3686b
+README.md: 5e24e4aad5154430fa48c80eb439694005df7c6f
+README.zh.md: 89a34041e156e137d966bdafbc92d86477df166e
diff --git a/packages/client/ui-conversation/README.md b/packages/client/ui-conversation/README.md
index ea86c6fd43..5e24e4aad5 100644
--- a/packages/client/ui-conversation/README.md
+++ b/packages/client/ui-conversation/README.md
@@ -2,23 +2,25 @@
 
 English | [中文](README.zh.md)
 
-Conversation domain: skeleton (header/tabs/composer/empty state), chat view (grouped step-summary flow, streaming tail isolation, stats line, per-tool row slot with a bash sample registrant and the todo row), input dock (queue rows plus the todo plan strip), minimal details panel, scope-addressed ConversationService. Contract: api-contracts v3 §7 plus the slot terminal design (store seat / props shares).
+Conversation domain: skeleton (header/tabs/composer/empty state), chat view (grouped step-summary flow, streaming tail isolation, per-tool row slot with a bash sample registrant and the todo row), composer dock (session stats sticky with the input), input dock (queue rows plus the todo plan strip), minimal details panel, scope-addressed ConversationService. Contract: api-contracts v3 §7 plus the slot terminal design (store seat / props shares).
 
-The resident conversation shell survives no-session and session transitions. Without a current session it renders a disabled input bar; its root-scoped `conversation.hero.workspace` slot hosts the Workspace picker. Selecting a Workspace connects or reuses its Host-owned blank session and opens that session without replacing the shell. Blank sessions render the same composer body as active sessions, while the InputHub carries drafts across Workspace switches and mirrors them into the session store.
+The resident conversation shell survives no-session and session transitions. Without a current session it renders a disabled input bar; its root-scoped `conversation.hero.workspace` slot hosts the Workspace picker. Selecting a Workspace connects or reuses its Host-owned blank session and opens that session without replacing the shell. Blank sessions render the same composer body as active sessions, while the InputHub carries drafts across Workspace switches and mirrors them into the session store. In the active phase the session header occupies the top as ordinary column chrome; beneath it a scrollport (`data-conversation-scroll`) holds the flowing views and the sticky composer stack (stats dock + input docks + bar). Wheel over the textarea chains: the capped draft scrolls locally until its edge, then forwards to that host.
 
 The view ring IS a slot: the conversation registration declares the `'conversation.view'` list slot (session scope) in its `children` table, ConversationRoot renders the active entry through its renderSlot share (`only: <active id>`), and view tabs project from the ring ledger's registration options (`id`/`order`/`label`). The chat view is this package's own ring entry; other plugins (ui-trajectory) contribute tabs through plain `ctx.slots.register` — the former package-local view registry (`registerView`/`ViewEntry`/`ConversationViewMap` and the chrome attachment table) is retired, with per-view chrome dissolved into the view components themselves.
 
-Approvals take over the composer through the chain this package declares: `ApprovalPanel` registers as a selector-routed `'conversation.composer'` entry (the ui-question pattern) and occupies the composer in place of the InputBar while an approval wait is pending (amber strip, justification headline, paired command line from the running call's args, one-shot refuse/allow). The `PendingApproval` domain face in `contract/slots.ts` owns the wire encoding — the `ApprovalResponsePayload` value with the audit correlation — over the runtime's `PendingWait` carrier; the broadcast `approval/resolved` frame settles the wait and restores the composer. The sidebar mirrors the blocked state through the manager-tracked `waitingApproval` list bit (lit for uninstantiated sessions too), which outranks the running ring until the question resolves. Pending waits leave the message flow entirely: questions (ui-question) and approvals (ApprovalPanel) both answer through the composer takeover, so no display-only placeholder card remains. The composer's bottom-row Access seat mounts `PermissionSelect`, fed by the host-computed `permissions` projection through the standard-kit `useProjection` (key absence hides the chip); a pick submits the `/permission <preset>` command line through the bar's injected `command` callback.
+Approvals take over the composer through the chain this package declares: `ApprovalPanel` registers as a selector-routed `'conversation.composer'` entry (the ui-question pattern) and occupies the composer in place of the InputBar while an approval wait is pending (amber strip, justification headline, paired command line from the running call's args, one-shot refuse/allow). The `PendingApproval` domain face in `contract/slots.ts` owns the wire encoding — the `ApprovalResponsePayload` value with the audit correlation — over the runtime's `PendingWait` carrier; the broadcast `approval/resolved` frame settles the wait and restores the composer. The sidebar mirrors the blocked state through the manager-tracked `waitingApproval` list bit (lit for uninstantiated sessions too), which outranks the running ring until the question resolves. Pending waits leave the message flow entirely: questions (ui-question) and approvals (ApprovalPanel) both answer through the composer takeover, so no display-only placeholder card remains. The composer's bottom-row Access seat mounts `PermissionSelect`, fed by the host-computed `permissions` projection through the standard-kit `useProjection` (key absence hides the chip); the chip opens a Menu-primitive dropdown whose kebab-case preset names render as title-case labels (the `/permission` popup's display transform twin), and a pick submits the `/permission <preset>` command line through the bar's injected `command` callback.
 
 Generic tool rows classify the built-in bash, read, search, write, edit, and run_code names into dedicated visual variants. The filesystem variants render the edit icon and a path summary; that path is a hover-underline link that opens the file with the host OS default application (`host.openPath`, relative paths resolve against the session cwd). Tool rows are not whole-row click targets and do not open the details panel. The code variant summarizes with the model-authored `description` and expands to the program itself; its logged sub-dispatches render as always-visible nested rows through the SAME keyed toolview hole (custom registrations and the GenericToolCard fallback apply to sub-rows unchanged). Cordis lifecycle tools reuse those generic variants while presenting `Inspect`, `Mount temporary Plugin`, and `Unmount temporary Plugin` with a shared Cordis accent; mount keeps the code variant's expandable source rendering.
 
+A tool call declaring the `terminal` render intent renders its command output inline, at both conversation render sites, through ui-primitives' `TerminalBlock`. `contract/terminal-card-model.ts` is the single derivation from the snapshot's `callView`/`resultView` pair, so the sites cannot disagree about a command, its cwd, or its exit status; it yields null — the generic path — for any other card tag, including one this client version does not know. Both sites therefore also show the card's run-state dot, which is the same `StateDot` semantic a tool row's leading icon carries, so a row and its own card always agree about one command's state. A multi-line command gets one prompt row per line, with the dot marking the call once on the first row — the exit status is the whole call's, so a dot per line would claim a per-line outcome bash does not report. The keyed `BashRow` carries the card resident below its summary row; since tool rows are no longer details-panel click targets, the card's copy and expand controls are the row's only interactions. The render-site fallback row keeps the card behind its existing expand control. Rows cap at `CHAT_TERMINAL_MAX_LINES` (8) against the panel's 16, which is what keeps a summary surface bounded — the panel stays the single-call reading surface. Inline output is licensed for this intent alone; a generic tool's content remains panel-only ([decision](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)).
+
 Tool rows are slots too — the standalone tool ring (`ToolViewRegistry`/`ctx.toolviews`/outlet) is retired. The chat entry declares the keyed `'conversation.chat.toolview'` hole (session scope; the key space is runtime-open); its render site dispatches per row via `entryKey: toolName` with `GenericToolCard` as the call-site `fallback`. The owner payload is the uniform `ToolRowOwnerProps` (`callId`/`toolName`/`block`/`openFile`) and `ToolRowProps` pre-composes it with the session standard kit. A registrant is a plain plugin: `ctx.slots.register({ name: 'conversation.chat.toolview', key: '<tool>', inject? }, Row)` with `inject: ['slots', 'conversation']` as the load-order seam (apply mounts ConversationService after the chat registration, so the service being present guarantees the slot is declared); session differentiation happens inside the component (`useSessions` reading `parentId` — the bash sample is the third-party-posture exemplar). Trajectory/waterfall toolview slots share this shape and land with their own render sites (RendersCheck rejects a declaration nobody renders).
 
 The todo surfaces are two registrations over that shape, both plain registrant plugins with `inject: ['slots', 'conversation']`. `TodoRow` takes the `'conversation.chat.toolview'` key `todo_write` and summarizes what the call attempted (`<done>/<total> 已完成 · <active item>` parsed from its args, falling back to the generic summary on malformed or wrongly-shaped model JSON, and keeping the generic dot for non-ok execution states so a cancelled call never reads as a completed update). `TodoDock` takes the `'conversation.input.dock'` list slot at `order: -1` — above the queue rows — and is the plan strip: it reads the host-computed `todos` projection via `useProjection` (standing plan: latest `todo/write` with no later `turn/start`) and renders `TodoPanel`, which takes the plain list, hides itself while the list is empty, and starts collapsed as a header of title plus `"<done>/<total> tasks · <n> in progress"` (status glyphs are the figma check / progress / dashed-pending set). The dock adapter owns the selection so the panel stays a pure function of its props; the standing list lives here rather than in the row so the row stays one line. Anything the input-zone composer chain hides (a `conversation.composer` takeover such as ui-question's) hides the whole dock, this strip included.
 
 Per-session UI state for selection and the active view lives in the declared chat store (`stores.ts` `createChatStore`); the InputHub owns the composer state machine and mirrors its draft into that store for persistence. Apply passes one store handle to the strict session subtree, chat view, and details registrations, so each session shares one instance and the framework owns its lifecycle. Components are pure: the framework standard kit supplies `useSession`/`sessionId`, global `useSessions`/`useWorkspaces`, and the input machine's `useInput`/`inputActions`; store faces and inject factories supply the remaining state and callbacks.
 
-The composer bar declares session-scoped single seats for `'conversation.input.plan'` (right of the local access-mode control) and `'conversation.input.model'` (immediately before the pending indicator and send/stop button), plus list slots for overlay, dock, left, and right input extensions. Feature packages own each control and its state; ui-conversation supplies placement, the `locked` owner prop, and the standard slot shares. While the `plan` projection's effective target is plan mode, InputBar swaps its textarea placeholder to the plan-task wording (a host-folded value read through the standard-kit `useProjection`; owner-supplied placeholders win). The resident no-session shell uses `DisabledInputBar` and therefore dispatches no session-scoped control seats.
+The composer bar declares session-scoped single seats for `'conversation.input.plan'` (right of the local access-mode control) and `'conversation.input.model'` (immediately before the pending indicator and send/stop button), plus list slots for overlay, dock, left, and right input extensions. Feature packages own each control and its state; ui-conversation supplies placement, the `locked` owner prop, and the standard slot shares. While the `plan` projection's effective target is plan mode, InputBar swaps its textarea placeholder to the plan-task wording, localized through the `command.hint` locale namespace this package registers and shared verbatim with the claimed `/plan` command hint (a host-folded value read through the standard-kit `useProjection`; owner-supplied placeholders win). A pending composer takeover remains mounted when another conversation view is active so the blocked agent can still receive its answer; without a pending interaction, the active-session composer belongs to Chat. The resident no-session shell uses `DisabledInputBar` and therefore dispatches no session-scoped control seats.
 
 `src/client/` is organized for the future package split: `contract/` is the sole inter-domain shared face (`slots.ts` slot declarations + composed slot props including the tool-row contract, `views.ts` shared primitives, `tool-call-model.ts`); the `skeleton/`, `chat/`, and `toolviews/` (sample registrants) domain directories import contract files and never each other; `apply.ts` is the only assembly point allowed to import all three domains. The `/client` export surface is the contract only — `apply`/`inject`, the two service classes, and the `contract/` type families; implementation components (skeleton, chat rows) and the store factory stay internal and reach the page exclusively through apply's slot registrations (tests take them via the `./src/*` subpath).
 
@@ -33,8 +35,8 @@ None; this package neither assembles nor sends a provider request.
 ## Known Limitations and Deferred Work
 
 - **The stats line has no duration segment** — assistant `usage` carries token accounting only; elapsed-time needs a host data source.
-- **Details panel is the minimal form** — selected call args/result raw display; the Input/Output/Metadata switch, Prev/Next stepping, and See-in-trajectory deep link are deferred.
-- **Assistant footer extensions (IconActions row, per-message paging) are reserved slots** — drawn in the design, not implemented.
+- **Details panel is the minimal form and currently has no entry point** — selected call args/result raw display; the Input/Output/Metadata switch, Prev/Next stepping, and See-in-trajectory deep link are deferred. Tool rows stopped being details-panel click targets and nothing replaced that gesture, so `ChatViewInjected.openDetails` is implemented but uncalled and the panel (including its terminal card) is unreachable in the assembled application; its rendering stays covered by mounting it with a selection directly.
+- **Assistant per-message paging is a reserved slot** — drawn in the design, not implemented. The finalized IconActions row (copy / branch / clock) ships; branch remains a chrome stub.
 - **The sparkle icon for the others tool row is a hand-drawn approximation** — the design glyph's vector geometry is not exportable locally; promotion into ui-primitives waits on an exact export.
 - **The approval panel's "Always allow this type" is deferred** — durable grants need a grant-storage design; only allow-once/reject answer today.
 - **TodoPanel truncates long item text to one ellipsized line** — the figma strip has no wrap or expand affordance; full text is not readable inline.
diff --git a/packages/client/ui-conversation/README.zh.md b/packages/client/ui-conversation/README.zh.md
index 9336d3d360..89a34041e1 100644
--- a/packages/client/ui-conversation/README.zh.md
+++ b/packages/client/ui-conversation/README.zh.md
@@ -2,23 +2,25 @@
 
 [English](README.md) | 中文
 
-会话领域：骨架（标题栏／标签页／编辑器／空状态）、聊天视图（分组步骤摘要流、流式尾部隔离、统计行、逐工具行 slot 及一个 bash 示例注册方与 todo 行）、输入区 dock（队列行加 todo 计划条）、最小详情面板、按 scope 寻址的 ConversationService。契约：api-contracts v3 §7 加 slot 终端设计（store seat／props share）。
+会话领域：骨架（标题栏／标签页／编辑器／空状态）、聊天视图（分组步骤摘要流、流式尾部隔离、逐工具行 slot 及一个 bash 示例注册方与 todo 行）、编辑器 dock（与输入区一同 sticky 的会话统计行）、输入区 dock（队列行加 todo 计划条）、最小详情面板、按 scope 寻址的 ConversationService。契约：api-contracts v3 §7 加 slot 终端设计（store seat／props share）。
 
-常驻会话壳会跨无会话与会话状态切换而保留。没有当前会话时，它会渲染禁用输入栏；其根作用域的 `conversation.hero.workspace` slot 承载 Workspace 选择器。选择 Workspace 会连接或复用由 Host 拥有的空白会话，并在不替换会话壳的情况下打开该会话。空白会话与活跃会话渲染相同的输入区主体；InputHub 则在 Workspace 切换间携带草稿，并将草稿镜像到会话 store。
+常驻会话壳会跨无会话与会话状态切换而保留。没有当前会话时，它会渲染禁用输入栏；其根作用域的 `conversation.hero.workspace` slot 承载 Workspace 选择器。选择 Workspace 会连接或复用由 Host 拥有的空白会话，并在不替换会话壳的情况下打开该会话。空白会话与活跃会话渲染相同的输入区主体；InputHub 则在 Workspace 切换间携带草稿，并将草稿镜像到会话 store。活跃阶段会话标题栏以普通列 chrome 占据顶部；其下滚动容器（`data-conversation-scroll`）承载流动排版的各视图与 sticky 编辑器栈（统计 dock＋输入区 dock＋输入栏）。textarea 上的滚轮会链式处理：限高草稿先在本地滚动，到达边缘后再转交给该宿主。
 
 视图环本身就是 slot：会话注册声明 `'conversation.view'` 列表 slot（Session scope），并将其列在 `children` 表中；ConversationRoot 通过 renderSlot share 渲染活跃配置项（`only: <active id>`）；视图标签页从环账本的注册选项（`id`／`order`／`label`）投影而来。聊天视图是该包自身的环配置项；其他插件（ui-trajectory）通过普通的 `ctx.slots.register` 贡献标签页。先前包内的视图注册表（`registerView`／`ViewEntry`／`ConversationViewMap` 及 chrome 附加表）已退役，逐视图 chrome 则被拆入视图组件自身。
 
 通用工具行把内置的 bash、read、search、write、edit 和 run_code 名称归入专用视觉变体。文件系统变体会渲染 edit 图标和路径摘要；该路径是悬停下划线链接，点击后通过宿主操作系统的默认应用打开文件（`host.openPath`，相对路径相对会话 cwd 解析）。工具行不再是整行点击目标，也不会打开 details 面板。code 变体以模型撰写的 `description` 作摘要，展开后显示程序本身；其已记录的子调用经由同一个键控 toolview 空位渲染为始终可见的嵌套行（自定义注册和 GenericToolCard fallback 原样适用于子行）。Cordis 生命周期工具复用这些通用变体，同时以统一的 Cordis 强调色呈现 `Inspect`、`Mount temporary Plugin` 和 `Unmount temporary Plugin`；mount 行保留 code 变体的可展开源码渲染。
 
+声明 `terminal` 渲染意图的工具调用，会在两个对话渲染点上都通过 ui-primitives 的 `TerminalBlock` 内联渲染其命令输出。`contract/terminal-card-model.ts` 是从快照的 `callView`／`resultView` 对推导的唯一位置，因此两个渲染点不可能在命令、cwd 或退出状态上产生分歧；对任何其他 card 标签——包括当前客户端版本不认识的标签——它返回 null，落回通用路径。因此两个渲染点也都显示卡片的运行状态点，它与工具行行首图标承载同一套 `StateDot` 语义，所以一行与其自身的卡片对同一条命令的状态总是一致。多行命令的每一行各占一个提示行，状态点只在第一行为整次调用标记一次——退出状态属于整次调用，因此每行一枚就会声称一个 bash 并不报告的逐行结果。键控的 `BashRow` 把卡片常驻在摘要行下方；由于工具行已不再是详情面板的点击目标，卡片的复制与展开控件就是该行唯一的交互。渲染点兜底行则保持其既有的展开控件。行的上限是 `CHAT_TERMINAL_MAX_LINES`（8），面板为 16，正是这一点让摘要面保持有界——面板仍是单次调用的阅读面。内联输出只对该意图开放；通用工具的内容仍然只在面板中呈现（[决策](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)）。
+
 工具行同样是 slot：独立工具环（`ToolViewRegistry`／`ctx.toolviews`／outlet）已经退役。聊天配置项声明键控的 `'conversation.chat.toolview'` 空位（Session scope；key 空间在运行时开放）；其渲染点逐行通过 `entryKey: toolName` 分发，并以 `GenericToolCard` 作为调用点 `fallback`。owner 载荷是统一的 `ToolRowOwnerProps`（`callId`／`toolName`／`block`／`openFile`），`ToolRowProps` 则预先将其与 Session 标准工具包组合。注册方只是普通插件：`ctx.slots.register({ name: 'conversation.chat.toolview', key: '<tool>', inject? }, Row)`，以 `inject: ['slots', 'conversation']` 作为加载顺序 seam（apply 在聊天注册后挂载 ConversationService，因此服务存在即可保证 slot 已声明）；Session 区分在组件内部完成（`useSessions` 读取 `parentId`，bash 示例是第三方姿态的范例）。Trajectory/waterfall 工具视图 slot 共享此形状，并随各自的渲染点落地（RendersCheck 会拒绝没有任何渲染方的声明）。
 
-审批经由本包声明的链接管编辑器：`ApprovalPanel` 注册为按选择器路由的 `'conversation.composer'` 配置项（ui-question 模式），在审批等待未决期间取代 InputBar 占据编辑器（琥珀色条、理由标题、来自运行中调用参数的配对命令行、一次性的拒绝／允许）。`contract/slots.ts` 中的 `PendingApproval` 领域面在运行时 `PendingWait` 载体之上拥有 wire 编码——带审计关联的 `ApprovalResponsePayload` 值；广播的 `approval/resolved` 帧使等待落定并恢复编辑器。侧边栏通过 manager 跟踪的 `waitingApproval` 列表位（未实例化会话同样点亮）镜像该阻塞状态，其优先级高于运行中圆环，直至问题解决。未决等待完全离开消息流：问题（ui-question）与审批（ApprovalPanel）都经编辑器接管作答，不再保留只读占位卡。编辑器底行的 Access 席位挂载 `PermissionSelect`，由 host 计算的 `permissions` 投影经标准工具包 `useProjection` 供数（key 缺席即隐藏 chip）；选中会经由输入栏注入的 `command` 回调提交 `/permission <preset>` 命令行。
+审批经由本包声明的链接管编辑器：`ApprovalPanel` 注册为按选择器路由的 `'conversation.composer'` 配置项（ui-question 模式），在审批等待未决期间取代 InputBar 占据编辑器（琥珀色条、理由标题、来自运行中调用参数的配对命令行、一次性的拒绝／允许）。`contract/slots.ts` 中的 `PendingApproval` 领域面在运行时 `PendingWait` 载体之上拥有 wire 编码——带审计关联的 `ApprovalResponsePayload` 值；广播的 `approval/resolved` 帧使等待落定并恢复编辑器。侧边栏通过 manager 跟踪的 `waitingApproval` 列表位（未实例化会话同样点亮）镜像该阻塞状态，其优先级高于运行中圆环，直至问题解决。未决等待完全离开消息流：问题（ui-question）与审批（ApprovalPanel）都经编辑器接管作答，不再保留只读占位卡。编辑器底行的 Access 席位挂载 `PermissionSelect`，由 host 计算的 `permissions` 投影经标准工具包 `useProjection` 供数（key 缺席即隐藏 chip）；chip 打开 Menu 原语下拉，kebab-case 预设名渲染为 Title Case 标签（与 `/permission` popup 的显示变换孪生），选中会经由输入栏注入的 `command` 回调提交 `/permission <preset>` 命令行。
 
 todo 两个面就是在该形状上的两个注册项，都是普通注册方插件，`inject: ['slots', 'conversation']`。`TodoRow` 占用 `'conversation.chat.toolview'` 的 `todo_write` key，摘要该次调用「试图写入」的内容（从其 args 解析出 `<已完成>/<总数> 已完成 · <进行中条目>`；模型 JSON 残缺或形状不对时回落到通用摘要；非 ok 执行状态保留通用状态点，使被取消的调用绝不读成一次已完成的更新）。`TodoDock` 以 `order: -1` 占用 `'conversation.input.dock'` 列表 slot（位于队列行之上），是计划条：它经 `useProjection` 读取 host 计算的 `todos` 投影（站立计划：其后没有更晚 `turn/start` 的最近一次 `todo/write`）并渲染 `TodoPanel`，后者接收纯列表，在列表为空时自我隐藏；列表非空时面板初始折叠，表头显示标题加 `"<已完成>/<总数> tasks · <n> in progress"`（状态图标为 figma 的勾选／进行中／虚线未开始一组）。选取由 dock 适配器负责，因此面板保持为其 props 的纯函数；站立列表放在此处而非行内，行才能保持单行。输入区 composer 链隐藏的一切（例如 ui-question 对 `conversation.composer` 的接管）也会隐藏整个 dock，包括这条计划条。
 
 逐 Session UI 状态中的选择与活跃视图位于已声明的聊天 store（`stores.ts` `createChatStore`）中；InputHub 拥有输入区状态机，并将草稿镜像到该 store 以便持久化。apply 将同一个 store handle 传给严格限定于会话的子树、聊天视图和详情注册，因此每个会话内共享一个实例，框架拥有其生命周期。组件保持纯粹：框架标准工具包提供 `useSession`／`sessionId`、全局 `useSessions`／`useWorkspaces`，以及输入状态机的 `useInput`／`inputActions`；store 表层与 inject factory 提供其余状态和回调。
 
-输入栏为 `'conversation.input.plan'`（位于本地 access 模式控件右侧）和 `'conversation.input.model'`（渲染在 pending 指示器与发送／停止按钮之前）声明会话作用域的单实例 seat，并为 overlay、dock、left 和 right 输入扩展声明列表 slot。各功能包拥有相应控件及其状态；ui-conversation 提供放置位置、`locked` owner prop 和标准 slot share。当 `plan` 投影的有效目标为 plan mode 时，InputBar 将文本框 placeholder 切换为 plan 任务措辞（经标准套件 `useProjection` 读取的 host 折叠值；owner 提供的 placeholder 优先）。常驻无会话壳使用 `DisabledInputBar`，因此不会分发任何会话作用域的控件 seat。
+输入栏为 `'conversation.input.plan'`（位于本地 access 模式控件右侧）和 `'conversation.input.model'`（渲染在 pending 指示器与发送／停止按钮之前）声明会话作用域的单实例 seat，并为 overlay、dock、left 和 right 输入扩展声明列表 slot。各功能包拥有相应控件及其状态；ui-conversation 提供放置位置、`locked` owner prop 和标准 slot share。当 `plan` 投影的有效目标为 plan mode 时，InputBar 将文本框 placeholder 切换为 plan 任务措辞，经本包注册的 `command.hint` locale 命名空间本地化，并与已认领 `/plan` 命令的提示逐字共用同一份文案（经标准套件 `useProjection` 读取的 host 折叠值；owner 提供的 placeholder 优先）。另一个会话视图活跃时，待处理的 composer 接管仍保持挂载，使被阻塞的 agent（智能体）仍能收到回答；没有待处理交互时，活跃会话的 composer 归 Chat 所有。常驻无会话壳使用 `DisabledInputBar`，因此不会分发任何会话作用域的控件 seat。
 
 `src/client/` 按未来的包拆分组织：`contract/` 是唯一的跨领域共享表层（`slots.ts` slot 声明 + 组合后的 slot props，包括工具行契约、`views.ts` 共享原语、`tool-call-model.ts`）；`skeleton/`、`chat/` 和 `toolviews/`（示例注册方）领域目录只导入 contract 文件，彼此绝不导入；`apply.ts` 是唯一允许导入全部三个领域的组装点。`/client` 导出表层只包含契约：`apply`／`inject`、两个服务类和 `contract/` 类型家族；实现组件（骨架、聊天行）与 store factory 保持内部状态，只能通过 apply 的 slot 注册到达页面（测试通过 `./src/*` 子路径获取它们）。
 
@@ -33,8 +35,8 @@ todo 两个面就是在该形状上的两个注册项，都是普通注册方插
 ## 已知限制与暂缓事项
 
 - **统计行没有耗时区段**：assistant `usage` 只携带 token 计数；耗时需要主机数据源。
-- **详情面板是最小形态**：以原始形式显示已选择调用的参数／结果；Input/Output/Metadata 切换、Prev/Next 步进与 See-in-trajectory 深链接暂缓实现。
-- **assistant footer 扩展（IconActions 行、逐消息分页）是预留 slot**：设计中已有图稿，尚未实现。
+- **详情面板是最小形态，且当前没有入口**：以原始形式显示已选择调用的参数／结果；Input/Output/Metadata 切换、Prev/Next 步进与 See-in-trajectory 深链接暂缓实现。工具行已不再是详情面板的点击目标，且没有任何手势接替它，因此 `ChatViewInjected.openDetails` 虽已实现却无人调用，该面板（含其终端卡片）在组装后的应用中不可达；其渲染仍由直接以选中态挂载它来覆盖。
+- **assistant 逐消息分页是预留 slot**：设计中已有图稿，尚未实现。已定稿的 IconActions 行（复制／分支／时钟）已落地；分支仍是 chrome stub。
 - **others 工具行的闪光图标是手绘近似版本**：无法在本地导出设计字形的矢量几何；等到存在精确导出后再将其提升到 ui-primitives。
 - **审批面板的「始终允许此类」暂缓**：持久授权需要授权存储设计；今天只能回答允许一次／拒绝。
 - **TodoPanel 将过长条目截成单行省略号**：figma 条没有换行或展开入口，完整文本无法在行内读完。
diff --git a/packages/client/ui-conversation/package.json b/packages/client/ui-conversation/package.json
index 42812dce24..8a9173982b 100644
--- a/packages/client/ui-conversation/package.json
+++ b/packages/client/ui-conversation/package.json
@@ -39,6 +39,7 @@
     "clsx": "^2.0.0"
   },
   "peerDependencies": {
+    "@deepseek-ai/dsh-client-locale": "^0.0.1",
     "@deepseek-ai/dsh-client-runtime": "^0.0.1",
     "@deepseek-ai/dsh-client-ui-primitives": "^0.0.1",
     "@deepseek-ai/dsh-client-ui-slash": "^0.0.1",
@@ -48,10 +49,10 @@
     "react": "^18.2.0"
   },
   "devDependencies": {
+    "@deepseek-ai/dsh-client-locale": "workspace:^",
     "@deepseek-ai/dsh-client-runtime": "workspace:^",
+    "@deepseek-ai/dsh-goal": "workspace:^",
     "@deepseek-ai/dsh-plan-mode": "workspace:^",
-    "@deepseek-ai/dsh-session-projection": "workspace:^",
-    "@deepseek-ai/dsh-tool-todo": "workspace:^",
     "@deepseek-ai/dsh-client-ui-layout": "workspace:^",
     "@deepseek-ai/dsh-client-ui-primitives": "workspace:^",
     "@deepseek-ai/dsh-client-ui-slash": "workspace:^",
diff --git a/packages/client/ui-conversation/src/client/apply.ts b/packages/client/ui-conversation/src/client/apply.ts
index 2b845e6c83..7f3aeb38cc 100644
--- a/packages/client/ui-conversation/src/client/apply.ts
+++ b/packages/client/ui-conversation/src/client/apply.ts
@@ -3,6 +3,8 @@ import type { Context } from 'cordis'
 import type { BoundActions } from '@deepseek-ai/dsh-client-ui-slots'
 import type { ISessions, SessionId } from '@deepseek-ai/dsh-client-runtime/client'
 import type {} from '@deepseek-ai/dsh-client-ui-layout/client'
+// Type-only: pulls the locale plugin's Context merge (ctx.locale).
+import type {} from '@deepseek-ai/dsh-client-locale/client'
 import type { ViewTab } from './contract/views.ts'
 import type {
   ApprovalWait, ChatViewInjected, ComposerBarInjected, ComposerChainProps, ConversationInjected,
@@ -15,6 +17,7 @@ import type { IConversation } from './service.ts'
 import { InputHub } from './input/hub.ts'
 import { InputBar } from './skeleton/InputBar.tsx'
 import { ChatView } from './chat/ChatView.tsx'
+import { StatsLine } from './chat/StatsLine.tsx'
 import { bashToolviewSample } from './toolviews/bash-sample.tsx'
 import { ApprovalPanel } from './skeleton/ApprovalPanel.tsx'
 import { todoToolview } from './toolviews/todo-row.tsx'
@@ -25,7 +28,7 @@ import { ConversationSession } from './skeleton/ConversationSession.tsx'
 import { DetailsPanel } from './skeleton/DetailsPanel.tsx'
 
 /** Services required by the conversation plugin. */
-export const inject = ['slots', 'layout', 'sessions', 'workspaces']
+export const inject = ['slots', 'layout', 'sessions', 'workspaces', 'locale']
 
 /** Resolve the session-scoped conversation face (scope-addressed send/cancel), failing loud. */
 function scopedConversation(sessions: ISessions, id: SessionId): IConversation {
@@ -50,6 +53,33 @@ export function apply(ctx: Context): void {
   const layout = ctx.layout
   const slots = ctx.slots
 
+  // Command hint locale: friendly placeholder text for claimed commands. The
+  // claimed /plan hint and the plan-mode textarea placeholder share one
+  // string: both describe the same next action.
+  const HINT_NS = 'command.hint'
+  const PLAN_HINT_ZH = '描述你的任务以生成计划'
+  const PLAN_HINT_EN = 'describe your task to generate plan'
+  ctx.effect(() => {
+    const disposers = [
+      ctx.locale.register(HINT_NS, 'zh', {
+        plan: PLAN_HINT_ZH,
+        goal: '输入目标，智能体将持续执行',
+        'goal.active': '当前目标进行中。可输入 edit 修改 / pause 暂停 / resume 继续 / clear 清除',
+        'placeholder.plan': PLAN_HINT_ZH,
+        'placeholder.default': '给智能体发消息',
+      }),
+      ctx.locale.register(HINT_NS, 'en', {
+        plan: PLAN_HINT_EN,
+        goal: 'describe the objective for a long-running task',
+        'goal.active': 'goal active — edit / pause / resume / clear',
+        'placeholder.plan': PLAN_HINT_EN,
+        'placeholder.default': 'Message the agent',
+      }),
+    ]
+    return () => { for (const dispose of disposers) dispose() }
+  }, 'ui-conversation: command hint dictionaries')
+  const translateHint = ctx.locale.bind(HINT_NS)
+
   // Apply-time construction keeps store identity bound to this fiber.
   const chatStore = createChatStore()
 
@@ -159,6 +189,7 @@ export function apply(ctx: Context): void {
           const result = await session.command(line)
           return result.ok && result.value.matched
         },
+        translateHint,
         hooks: { notices: shell.notices, lexicon: shell.lexicon },
       }
     },
@@ -208,6 +239,9 @@ export function apply(ctx: Context): void {
     },
   }, ChatView)
 
+  // Session stats stick with the composer (composer.dock = stats-line family).
+  slots.register({ name: 'conversation.composer.dock', id: 'stats', order: 0 }, StatsLine)
+
   // Class-plugin mount (packages/AGENTS.md service form): the service
   // registers itself as `conversation` and lives on its own child fiber.
   // Mounted AFTER the chat entry register above — construction guarantee for
diff --git a/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.module.css b/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.module.css
index 9177d721c3..d988cf52f8 100644
--- a/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.module.css
+++ b/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.module.css
@@ -1,14 +1,22 @@
-/* Assistant flow body: full-width narration (figma 16/28), block gap 16. */
+/* Assistant flow body: full-width narration (figma 16/28), block gap 16.
+   IconActions sit below the body with an explicit 16px top margin (figma
+   43:32997) — separate from the body's internal gap so the footer spacing
+   stays fixed when the body is a single block. */
 
 .root {
   display: flex;
   flex-direction: column;
-  gap: 16px;
   font-size: 16px;
   line-height: 28px;
   color: var(--dsw-alias-label-primary);
 }
 
+.body {
+  display: flex;
+  flex-direction: column;
+  gap: 16px;
+}
+
 /* Interrupted-turn terminal marker: quiet inline tag, no animation. */
 .stopped {
   align-self: flex-start;
@@ -19,3 +27,18 @@
   font-size: 11px;
   line-height: 18px;
 }
+
+/* Finalized footer offset (figma 43:32997); chrome lives in MessageIconActions. */
+.actions {
+  margin-top: 16px;
+  /* Optical align with 28px icon hit targets that pad 6px past the glyph. */
+  margin-left: -6px;
+}
+
+/* Hover-capable pointers: reveal shared actions on root hover/focus. */
+@media (hover: hover) {
+  .root:hover .actions,
+  .root:focus-within .actions {
+    opacity: 1;
+  }
+}
diff --git a/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx b/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx
index e3573b14f5..e5a89a9e88 100644
--- a/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx
+++ b/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx
@@ -4,10 +4,14 @@
 // view groups them into tool rows through its keyed toolview slot (figma
 // step-summary flow). Shared by finalized nodes and the streaming partial;
 // the turn-level loading dots live in the chat view's tail, not here.
+// Finalized nodes append IconActions (copy / branch / clock) once streaming ends.
 
 import { memo } from 'react'
 import type { AssistantBlock } from '@deepseek-ai/dsh-client-runtime/client'
-import { IconThinkOutline14, JsonBlock, MarkdownText } from '@deepseek-ai/dsh-client-ui-primitives'
+import {
+  IconThinkOutline14, JsonBlock, MarkdownText,
+} from '@deepseek-ai/dsh-client-ui-primitives'
+import { MessageIconActions } from './MessageIconActions.tsx'
 import { ToolRow } from './ToolRow.tsx'
 import css from './AssistantMarkdown.module.css'
 
@@ -16,6 +20,8 @@ export interface AssistantMarkdownProps {
   streaming: boolean
   /** Frozen partial of an aborted turn: rendered with a 已停止 marker. */
   interrupted?: boolean | undefined
+  /** Unix epoch ms for the finalized IconActions clock; omitted while streaming. */
+  time?: number | undefined
 }
 
 function firstLine(text: string): string {
@@ -23,6 +29,15 @@ function firstLine(text: string): string {
   return nl === -1 ? text : text.slice(0, nl)
 }
 
+/** Joined text blocks for the copy action (reasoning / tool heads stay out). */
+function copyText(blocks: readonly AssistantBlock[]): string {
+  const parts: string[] = []
+  for (const block of blocks) {
+    if (block.kind === 'text') parts.push(block.text)
+  }
+  return parts.join('')
+}
+
 /** Reasoning block as the Think variant summary row (figma 39:28304). */
 function ThinkRow({ text, running }: { text: string; running: boolean }) {
   return (
@@ -38,7 +53,9 @@ function ThinkRow({ text, running }: { text: string; running: boolean }) {
   )
 }
 
-export const AssistantMarkdown = memo(function AssistantMarkdown({ blocks, streaming, interrupted }: AssistantMarkdownProps) {
+export const AssistantMarkdown = memo(function AssistantMarkdown({
+  blocks, streaming, interrupted, time,
+}: AssistantMarkdownProps) {
   const last = blocks.length - 1
   // Tool-call heads render as tool rows in the chat view's grouping pass, so
   // a node that is only those heads (or empty) would paint an empty root
@@ -47,18 +64,30 @@ export const AssistantMarkdown = memo(function AssistantMarkdown({ blocks, strea
     || interrupted === true
     || blocks.some(block => block.kind !== 'tool-call')
   if (!hasVisible) return null
+  // Footer only after the turn settles with a known event time; streaming omits it.
+  const showActions = !streaming && time !== undefined
   return (
     <div className={css.root} data-streaming={streaming || undefined}>
-      {blocks.map((block, i) => {
-        switch (block.kind) {
-          case 'text': return <MarkdownText key={i} text={block.text} streaming={streaming} />
-          case 'reasoning': return <ThinkRow key={i} text={block.text} running={streaming && i === last} />
-          // Grouped into tool rows by ChatView; hasVisible above skips an empty shell.
-          case 'tool-call': return null
-          default: return <JsonBlock key={i} label="未知内容块" payload={block.block} />
-        }
-      })}
-      {interrupted && <span className={css.stopped}>已停止</span>}
+      <div className={css.body}>
+        {blocks.map((block, i) => {
+          switch (block.kind) {
+            case 'text': return <MarkdownText key={i} text={block.text} streaming={streaming} />
+            case 'reasoning': return <ThinkRow key={i} text={block.text} running={streaming && i === last} />
+            // Grouped into tool rows by ChatView; hasVisible above skips an empty shell.
+            case 'tool-call': return null
+            default: return <JsonBlock key={i} label="未知内容块" payload={block.block} />
+          }
+        })}
+        {interrupted && <span className={css.stopped}>已停止</span>}
+      </div>
+      {showActions && (
+        <MessageIconActions
+          text={copyText(blocks)}
+          time={time}
+          clock="end"
+          className={css.actions}
+        />
+      )}
     </div>
   )
 })
diff --git a/packages/client/ui-conversation/src/client/chat/ChatView.module.css b/packages/client/ui-conversation/src/client/chat/ChatView.module.css
index fae96e5f6c..80e461b518 100644
--- a/packages/client/ui-conversation/src/client/chat/ChatView.module.css
+++ b/packages/client/ui-conversation/src/client/chat/ChatView.module.css
@@ -1,6 +1,8 @@
 /* Chat flow: one 16px rhythm everywhere — between blocks (prose <-> tool
    runs) via the column gap and between consecutive tool rows via the group
-   gap. Input padding cap rides the skeleton. */
+   gap. Input padding cap rides the skeleton. Under
+   `[data-conversation-scroll]` the column host owns overflow and this view
+   is ordinary flow (see ConversationRoot active-phase rules). */
 
 .root {
   position: relative;
@@ -17,6 +19,18 @@
   padding: 16px 24px;
 }
 
+:global([data-conversation-scroll]) .root {
+  flex: 0 0 auto;
+  min-height: auto;
+  height: auto;
+}
+
+:global([data-conversation-scroll]) .scroll {
+  overflow: visible;
+  flex: 0 0 auto;
+  min-height: auto;
+}
+
 /* Message column: 736px fixed width, centered on the same axis as the
    input box; the scroller itself stays full-bleed. */
 .column {
@@ -113,16 +127,34 @@
   opacity: 0.6;
 }
 
-/* Back-to-bottom: 34px circular icon button at the column's right edge. */
-.toBottom {
-  position: absolute;
-  right: max(24px, calc((100% - 736px) / 2));
+/* Back-to-bottom: zero-height sticky slot so the control does not extend
+   scrollHeight; the button translates up into the viewport. Under the
+   conversation host, clearance sits above the sticky composer stack. */
+.toBottomSlot {
+  position: sticky;
   bottom: 16px;
-  width: 34px;
-  height: 34px;
+  /* Above the sticky composer (z-index 7) so the control stays clickable and
+     visible over the input card. */
+  z-index: 8;
+  height: 0;
+  display: flex;
+  justify-content: flex-end;
+  padding-right: max(0px, calc((100% - 736px) / 2));
+  pointer-events: none;
+}
+
+:global([data-conversation-scroll]) .toBottomSlot {
+  /* Clears the sticky composer stack (stats + docks + input card). */
+  bottom: 168px;
+}
+
+.toBottom {
   display: flex;
   align-items: center;
   justify-content: center;
+  width: 34px;
+  height: 34px;
+  margin-top: -34px;
   padding: 0;
   border: 1px solid var(--dsw-alias-border-l2);
   border-radius: 100px;
@@ -130,6 +162,7 @@
   background: var(--dsw-alias-button-floating-fill);
   box-shadow: var(--dsw-shadow-lv2);
   cursor: pointer;
+  pointer-events: auto;
 }
 
 .toBottom:hover {
diff --git a/packages/client/ui-conversation/src/client/chat/ChatView.tsx b/packages/client/ui-conversation/src/client/chat/ChatView.tsx
index 1cbb00a4ce..b9e1e3351f 100644
--- a/packages/client/ui-conversation/src/client/chat/ChatView.tsx
+++ b/packages/client/ui-conversation/src/client/chat/ChatView.tsx
@@ -1,11 +1,16 @@
 // ChatView: the default conversation view — message flow with user bubbles,
 // assistant narration, tool summary rows grouped into step runs, pending
-// cards, paging, bottom-follow, and the session stats line under the flow
-// (chrome dissolved into the view: the footer is part of what a chat view
-// IS, not registration metadata). Pure component registered directly; its
-// registration declares the keyed 'conversation.chat.toolview' hole, so tool
-// rows render through the props renderSlot share (entryKey = tool name,
-// GenericToolCard as the render-site fallback).
+// cards, paging, and bottom-follow. Session stats live on
+// 'conversation.composer.dock' (sticky with the composer). Pure component
+// registered directly; its registration declares the keyed
+// 'conversation.chat.toolview' hole, so tool rows render through the props
+// renderSlot share (entryKey = tool name, GenericToolCard as the render-site
+// fallback).
+//
+// Scroll: when nested under `[data-conversation-scroll]` (active conversation
+// column), that host is the scrollport and this view is flow content; when
+// mounted alone (unit tests), `.scroll` owns overflow. Bottom-follow and
+// prepend anchoring always target the resolved scrollport.
 //
 // Render economics (architecture RFC performance model): the list parent
 // subscribes to snapshot segments that do NOT change per streaming chunk
@@ -17,7 +22,7 @@
 // memoized rows never churns them.
 
 import {
-  memo, useLayoutEffect, useMemo, useRef, useState, type ReactNode,
+  memo, useEffect, useLayoutEffect, useMemo, useRef, useState, type ReactNode,
 } from 'react'
 import type {
   CodeSubCall, CommandNode, ConversationNode, ConversationSnapshot, RunningToolCall, ToolResultNode,
@@ -30,11 +35,15 @@ import { AssistantMarkdown } from './AssistantMarkdown.tsx'
 import { GenericCommandCard } from './GenericCommandCard.tsx'
 import { GenericToolCard } from './GenericToolCard.tsx'
 import { MessageItem } from './MessageItem.tsx'
-import { StatsLine } from './StatsLine.tsx'
 import css from './ChatView.module.css'
 
 const FOLLOW_THRESHOLD = 24
 
+/** Active column host when present; otherwise the view-local scroller. */
+function scrollerOf(from: HTMLElement): HTMLElement {
+  return (from.closest('[data-conversation-scroll]')) ?? from
+}
+
 type OpenFile = (path: string) => void
 
 /** The declared toolview hole's render share (stable framework binding, passed through memoized rows). */
@@ -244,26 +253,34 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessio
   const firstSeqRef = useRef<number | null>(null)
   const openedRef = useRef(false)
   const lastKeyRef = useRef<string | null>(null)
+  /** Flow tip signature — follow-scroll only when this moves, never on a
+   *  scroll-driven at-bottom chrome re-render (that was snapping inertial
+   *  scrolls the rest of the way to the floor). */
+  const followSigRef = useRef<string | null>(null)
 
   const firstSeq = nodes[0]?.seq ?? null
   const lastItem = items[items.length - 1]
+  const lastKey = lastItem?.key ?? null
+  const followSig = `${openState}:${firstSeq}:${lastKey}:${nodes.length}:${running ? 1 : 0}:${runningCalls.length}`
 
-  const toBottom = (el: HTMLDivElement): void => {
+  const toBottom = (el: HTMLElement): void => {
     el.scrollTop = el.scrollHeight
     atBottomRef.current = true
     setAtBottom(true)
   }
 
   useLayoutEffect(() => {
-    const el = listRef.current
+    const local = listRef.current
     /* v8 ignore next -- ref-null guard: React attaches the ref before layout effects run. */
-    if (el === null) return
+    if (local === null) return
+    const el = scrollerOf(local)
     // Open completed: jump to the bottom once.
     if (openState === 'open' && !openedRef.current) {
       openedRef.current = true
       toBottom(el)
       firstSeqRef.current = firstSeq
-      lastKeyRef.current = lastItem?.key ?? null
+      lastKeyRef.current = lastKey
+      followSigRef.current = followSig
       return
     }
     // Prepend (head seq decreased): compensate by the height delta.
@@ -272,42 +289,65 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessio
       anchorRef.current = null
       firstSeqRef.current = firstSeq
       /* v8 ignore next -- ?? arm: a prepend adds nodes, so the flow list here is never empty. */
-      lastKeyRef.current = lastItem?.key ?? null
+      lastKeyRef.current = lastKey
+      followSigRef.current = followSig
       return
     }
     firstSeqRef.current = firstSeq
     // Own words must be visible: a new trailing user node force-scrolls
     // (send lives in the composer, so arrival is detected here, not armed there).
-    const lastKey = lastItem?.key ?? null
     const appendedUser = lastKey !== lastKeyRef.current
       && lastItem !== undefined && lastItem.kind === 'node' && lastItem.node.kind === 'user'
+    const tipMoved = followSigRef.current !== followSig
     lastKeyRef.current = lastKey
-    if (appendedUser || atBottomRef.current) toBottom(el)
+    followSigRef.current = followSig
+    // Follow new flow content while pinned; do NOT re-pin on every render
+    // merely because atBottomRef is true (scroll threshold → setState → snap).
+    if (appendedUser || (tipMoved && atBottomRef.current)) toBottom(el)
   })
 
-  const onScroll = (): void => {
-    const el = listRef.current
-    /* v8 ignore next -- ref-null guard: the handler only fires on the mounted element. */
-    if (el === null) return
+  const onScrollRef = useRef(() => {})
+  onScrollRef.current = () => {
+    const local = listRef.current
+    /* v8 ignore next -- ref-null guard: the handler only fires while mounted. */
+    if (local === null) return
+    const el = scrollerOf(local)
     const isAtBottom = el.scrollHeight - el.scrollTop - el.clientHeight <= FOLLOW_THRESHOLD + 1
     atBottomRef.current = isAtBottom
     setAtBottom(isAtBottom)
   }
 
+  // Bind scroll to the resolved scrollport (host or local) once per mount.
+  useEffect(() => {
+    const local = listRef.current
+    /* v8 ignore next -- ref-null guard: effect runs after the list node commits. */
+    if (local === null) return
+    const el = scrollerOf(local)
+    const onScroll = (): void => { onScrollRef.current() }
+    el.addEventListener('scroll', onScroll, { passive: true })
+    return () => { el.removeEventListener('scroll', onScroll) }
+  }, [])
+
   // Follow streaming growth the parent never re-renders for (stable ref).
   // The ref starts null and is assigned every render, so the placeholder
   // initializer a function initial value would need never exists.
   const followRef = useRef<(() => void) | null>(null)
   followRef.current = () => {
-    const el = listRef.current
-    if (el !== null && atBottomRef.current) el.scrollTop = el.scrollHeight
+    const local = listRef.current
+    if (local !== null && atBottomRef.current) {
+      const el = scrollerOf(local)
+      el.scrollTop = el.scrollHeight
+    }
   }
   const onGrow = useRef(() => followRef.current?.()).current
 
   const loadOlderAnchored = (): void => {
-    const el = listRef.current
+    const local = listRef.current
     /* v8 ignore next -- ref-null guard: the paging button renders inside the list tree. */
-    if (el !== null) anchorRef.current = { h: el.scrollHeight, t: el.scrollTop }
+    if (local !== null) {
+      const el = scrollerOf(local)
+      anchorRef.current = { h: el.scrollHeight, t: el.scrollTop }
+    }
     loadOlder()
   }
 
@@ -330,7 +370,15 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessio
     }
     const node: ConversationNode = item.node
     if (node.kind === 'assistant') {
-      return <AssistantMarkdown key={item.key} blocks={node.blocks} streaming={false} interrupted={node.interrupted} />
+      return (
+        <AssistantMarkdown
+          key={item.key}
+          blocks={node.blocks}
+          streaming={false}
+          interrupted={node.interrupted}
+          time={node.time}
+        />
+      )
     }
     if (node.kind === 'command') {
       return <CommandRow key={item.key} renderSlot={renderSlot} node={node} />
@@ -342,7 +390,7 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessio
 
   return (
     <div className={css.root}>
-      <div ref={listRef} className={css.scroll} onScroll={onScroll}>
+      <div ref={listRef} className={css.scroll}>
         <div className={css.column}>
           {openState === 'loading' && <div className={css.hint}>载入历史…</div>}
           {openState === 'error' && <div className={css.openError}>历史加载失败：{openErrorMessage}</div>}
@@ -380,22 +428,23 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessio
               wait, tool execution, streaming) so it never flickers per step. */}
           {running && <TurnDots />}
         </div>
+        {!atBottom && (
+          <div className={css.toBottomSlot}>
+            <button
+              type="button"
+              className={css.toBottom}
+              aria-label="回到底部"
+              onClick={() => {
+                const local = listRef.current
+                /* v8 ignore next -- ref-null guard: the button only renders alongside the mounted list. */
+                if (local !== null) toBottom(scrollerOf(local))
+              }}
+            >
+              <IconChevronDownOutline14 />
+            </button>
+          </div>
+        )}
       </div>
-      <StatsLine useSession={useSession} />
-      {!atBottom && (
-        <button
-          type="button"
-          className={css.toBottom}
-          aria-label="回到底部"
-          onClick={() => {
-            const el = listRef.current
-            /* v8 ignore next -- ref-null guard: the button only renders alongside the mounted list. */
-            if (el !== null) toBottom(el)
-          }}
-        >
-          <IconChevronDownOutline14 />
-        </button>
-      )}
     </div>
   )
 }
diff --git a/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx b/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx
index 412ff9fe59..ce55d84f57 100644
--- a/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx
+++ b/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx
@@ -10,6 +10,7 @@ import {
   IconThinkOutline14,
 } from '@deepseek-ai/dsh-client-ui-primitives'
 import type { ToolRowOwnerProps } from '../contract/slots.ts'
+import { terminalCardModel } from '../contract/terminal-card-model.ts'
 import { toolRowModel, type ToolRowVariant } from '../contract/tool-call-model.ts'
 import { ToolRow } from './ToolRow.tsx'
 
@@ -27,6 +28,7 @@ const VARIANT_ICONS: Record<ToolRowVariant, ReactNode> = {
 
 export function GenericToolCard({ toolName, block, cwd, openFile }: ToolRowOwnerProps) {
   const model = toolRowModel(toolName, block, cwd)
+  const terminal = terminalCardModel(block, cwd)
   const singleFile = model.filePath !== undefined
   return (
     <ToolRow
@@ -34,9 +36,12 @@ export function GenericToolCard({ toolName, block, cwd, openFile }: ToolRowOwner
       toolName={toolName}
       icon={VARIANT_ICONS[model.variant]}
       title={model.title}
-      summary={model.summary}
+      // A terminal presenter's description is the contract's above-card text, so
+      // it outranks the args-derived summary here exactly as it does in BashRow.
+      summary={terminal?.description ?? model.summary}
       // Single-file tools never expose an args body — the path link is the only action.
       body={singleFile ? null : model.body}
+      terminal={terminal}
       state={model.state}
       filePath={model.filePath}
       onOpenFile={singleFile ? openFile : undefined}
diff --git a/packages/client/ui-conversation/src/client/chat/MessageIconActions.module.css b/packages/client/ui-conversation/src/client/chat/MessageIconActions.module.css
new file mode 100644
index 0000000000..30d6920609
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/chat/MessageIconActions.module.css
@@ -0,0 +1,53 @@
+/* Shared message IconActions row (user + assistant). Parent modules own
+   hover-reveal selectors and layout offsets via the composed className. */
+
+.actions {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  height: 28px;
+}
+
+/* Clock before icons (user figma 388:20051) / after (assistant 43:32997). */
+.timeStart {
+  padding-right: 12px;
+  font-size: 14px;
+  line-height: 24px;
+  color: var(--dsw-alias-label-tertiary);
+  white-space: nowrap;
+}
+
+.timeEnd {
+  padding-left: 12px;
+  font-size: 14px;
+  line-height: 24px;
+  color: var(--dsw-alias-label-tertiary);
+  white-space: nowrap;
+}
+
+/* Hover-capable pointers: hide until a parent hover/focus rule reveals. */
+@media (hover: hover) {
+  .actions {
+    opacity: 0;
+    transition: opacity var(--ds-transition-duration) var(--ds-ease-in-out);
+  }
+}
+
+.action {
+  display: inline-flex;
+  align-items: center;
+  justify-content: center;
+  width: 28px;
+  height: 28px;
+  padding: 6px;
+  border: none;
+  border-radius: 28px;
+  background: transparent;
+  color: var(--dsw-alias-label-tertiary);
+  cursor: pointer;
+}
+
+.action:hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+  color: var(--dsw-alias-label-secondary);
+}
diff --git a/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx b/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx
new file mode 100644
index 0000000000..7579a4c249
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx
@@ -0,0 +1,65 @@
+// Shared IconActions chrome for user and assistant messages: copy / branch
+// live (branch still a stub), date-aware clock, optional edit stub.
+
+import { useCallback } from 'react'
+import {
+  IconBranchOutline16, IconCopyOutline16, IconEditOutline16, Tooltip,
+} from '@deepseek-ai/dsh-client-ui-primitives'
+import { formatMessageClock, writeClipboard } from './message-chrome.ts'
+import { useCalendarDay } from './use-calendar-day.ts'
+import css from './MessageIconActions.module.css'
+
+export interface MessageIconActionsProps {
+  /** Plain text the copy action writes. */
+  text: string
+  /** Unix epoch ms for the clock label. */
+  time: number
+  /** Clock before icons (user) or after (assistant). */
+  clock: 'start' | 'end'
+  /** When true, append the stub edit control (user bubble). */
+  edit?: boolean | undefined
+  /** Parent layout / hover-reveal class composed onto the actions row. */
+  className?: string | undefined
+}
+
+/**
+ * Copy / branch (/ clock) IconActions row shared by user and assistant chrome.
+ * @param props - Copy text, event time, clock side, optional edit, className.
+ * @returns The actions row element.
+ */
+export function MessageIconActions({
+  text, time, clock, edit, className,
+}: MessageIconActionsProps) {
+  const day = useCalendarDay()
+  const onCopy = useCallback(() => {
+    void writeClipboard(text)
+  }, [text])
+  const clockEl = (
+    <span className={clock === 'start' ? css.timeStart : css.timeEnd}>
+      {formatMessageClock(time, day)}
+    </span>
+  )
+  return (
+    <div className={className === undefined ? css.actions : `${css.actions} ${className}`}>
+      {clock === 'start' ? clockEl : null}
+      <Tooltip label="复制" side="bottom">
+        <button type="button" className={css.action} aria-label="复制" onClick={onCopy}>
+          <IconCopyOutline16 />
+        </button>
+      </Tooltip>
+      <Tooltip label="在新对话中分支" side="bottom">
+        <button type="button" className={css.action} aria-label="在新对话中分支">
+          <IconBranchOutline16 />
+        </button>
+      </Tooltip>
+      {edit === true && (
+        <Tooltip label="编辑" side="bottom">
+          <button type="button" className={css.action} aria-label="编辑">
+            <IconEditOutline16 />
+          </button>
+        </Tooltip>
+      )}
+      {clock === 'end' ? clockEl : null}
+    </div>
+  )
+}
diff --git a/packages/client/ui-conversation/src/client/chat/MessageItem.module.css b/packages/client/ui-conversation/src/client/chat/MessageItem.module.css
index 22537f9cfe..260382d530 100644
--- a/packages/client/ui-conversation/src/client/chat/MessageItem.module.css
+++ b/packages/client/ui-conversation/src/client/chat/MessageItem.module.css
@@ -20,46 +20,14 @@
   color: var(--dsw-alias-label-primary);
 }
 
-.actions {
-  display: flex;
-  align-items: center;
-  gap: 10px;
-  height: 28px;
-}
-
-/* Hover-capable pointers: hide until the row is hovered/focused. Touch /
-   hover:none keeps actions visible (opacity:0 still hit-tests). */
+/* Hover-capable pointers: reveal shared MessageIconActions on row hover/focus. */
 @media (hover: hover) {
-  .actions {
-    opacity: 0;
-    transition: opacity var(--ds-transition-duration) var(--ds-ease-in-out);
-  }
-
   .userRow:hover .actions,
   .userRow:focus-within .actions {
     opacity: 1;
   }
 }
 
-.action {
-  display: inline-flex;
-  align-items: center;
-  justify-content: center;
-  width: 28px;
-  height: 28px;
-  padding: 6px;
-  border: none;
-  border-radius: 28px;
-  background: transparent;
-  color: var(--dsw-alias-label-tertiary);
-  cursor: pointer;
-}
-
-.action:hover {
-  background: var(--dsw-alias-interactive-bg-hover);
-  color: var(--dsw-alias-label-secondary);
-}
-
 .badge {
   display: inline-block;
   margin-bottom: 4px;
diff --git a/packages/client/ui-conversation/src/client/chat/MessageItem.tsx b/packages/client/ui-conversation/src/client/chat/MessageItem.tsx
index 571104dac8..a149d37337 100644
--- a/packages/client/ui-conversation/src/client/chat/MessageItem.tsx
+++ b/packages/client/ui-conversation/src/client/chat/MessageItem.tsx
@@ -1,18 +1,16 @@
 // MessageItem: the four simple node kinds — user bubble (right-aligned, with
-// copy / branch / edit IconActions), steering (badged bubble), context
+// clock + copy / branch / edit IconActions), steering (badged bubble), context
 // injection and unknown-surface JSON rows. Props are frozen node slices off
 // the snapshot cache; memo holds across streaming because unchanged nodes
 // keep their references.
 
-import { memo, useCallback } from 'react'
+import { memo } from 'react'
 import type { ReactNode } from 'react'
 import type {
   ContextMessageNode, SteeringMessageNode, UnknownSurfaceNode, UserMessageNode,
 } from '@deepseek-ai/dsh-client-runtime/client'
-import {
-  IconBranchOutline16, IconCopyOutline16, IconEditOutline16,
-  JsonBlock, MessageText, Tooltip,
-} from '@deepseek-ai/dsh-client-ui-primitives'
+import { JsonBlock, MessageText } from '@deepseek-ai/dsh-client-ui-primitives'
+import { MessageIconActions } from './MessageIconActions.tsx'
 import css from './MessageItem.module.css'
 
 export interface MessageItemProps {
@@ -30,42 +28,6 @@ function contentText(content: readonly unknown[]): { text: string; rest: unknown
   return { text: texts.join(''), rest }
 }
 
-/** Best-effort clipboard write; rejections stay swallowed (no success chrome). */
-async function writeClipboard(text: string): Promise<void> {
-  // lib.dom types clipboard non-optional, but insecure contexts omit it —
-  // that runtime gap is exactly what this guard detects.
-  /* eslint-disable-next-line @typescript-eslint/no-unnecessary-condition */
-  if (navigator.clipboard?.writeText) {
-    try {
-      await navigator.clipboard.writeText(text)
-    } catch {
-      // Denied permissions / iframe policy.
-    }
-    return
-  }
-  // execCommand('copy') is the only clipboard fallback where the async API
-  // is missing (insecure contexts); deprecated but deliberately retained.
-  /* eslint-disable @typescript-eslint/no-deprecated */
-  const exec = typeof document.execCommand === 'function'
-    ? document.execCommand.bind(document)
-    : undefined
-  if (exec === undefined) return
-  const el = document.createElement('textarea')
-  el.value = text
-  el.setAttribute('readonly', '')
-  el.style.position = 'fixed'
-  el.style.left = '-9999px'
-  document.body.appendChild(el)
-  el.select()
-  try {
-    exec('copy')
-  } catch {
-    // Clipboard unavailable; the button stays idle.
-  }
-  /* eslint-enable @typescript-eslint/no-deprecated */
-  el.remove()
-}
-
 /**
  * Display projection of reference forms in a user bubble (free geometry — no
  * textarea alignment constraint here); everything else stays plain text. The
@@ -98,32 +60,6 @@ function projectUserText(text: string): ReactNode {
   return <>{parts}</>
 }
 
-/** User-bubble IconActions (figma 659:38820): copy is live; branch/edit are chrome stubs. */
-function UserActions({ text }: { text: string }) {
-  const onCopy = useCallback(() => {
-    void writeClipboard(text)
-  }, [text])
-  return (
-    <div className={css.actions}>
-      <Tooltip label="复制" side="bottom">
-        <button type="button" className={css.action} aria-label="复制" onClick={onCopy}>
-          <IconCopyOutline16 />
-        </button>
-      </Tooltip>
-      <Tooltip label="在新对话中分支" side="bottom">
-        <button type="button" className={css.action} aria-label="在新对话中分支">
-          <IconBranchOutline16 />
-        </button>
-      </Tooltip>
-      <Tooltip label="编辑" side="bottom">
-        <button type="button" className={css.action} aria-label="编辑">
-          <IconEditOutline16 />
-        </button>
-      </Tooltip>
-    </div>
-  )
-}
-
 export const MessageItem = memo(function MessageItem({ node }: MessageItemProps) {
   switch (node.kind) {
     case 'user': {
@@ -134,7 +70,13 @@ export const MessageItem = memo(function MessageItem({ node }: MessageItemProps)
             {projectUserText(text)}
             {rest.map((block, i) => <JsonBlock key={i} label="附加内容块" payload={block} />)}
           </div>
-          <UserActions text={text} />
+          <MessageIconActions
+            text={text}
+            time={node.time}
+            clock="start"
+            edit
+            className={css.actions}
+          />
         </div>
       )
     }
diff --git a/packages/client/ui-conversation/src/client/chat/StatsLine.tsx b/packages/client/ui-conversation/src/client/chat/StatsLine.tsx
index 45b783f19b..7db5695030 100644
--- a/packages/client/ui-conversation/src/client/chat/StatsLine.tsx
+++ b/packages/client/ui-conversation/src/client/chat/StatsLine.tsx
@@ -1,4 +1,6 @@
 // Settled-node identity prevents stream-delta updates from rerendering this row.
+// Mounted on 'conversation.composer.dock' so it sticks with the composer in the
+// active conversation scrollport (see ConversationRoot data-conversation-scroll).
 
 import { memo, useMemo } from 'react'
 import type { ConversationSnapshot } from '@deepseek-ai/dsh-client-runtime/client'
@@ -49,7 +51,7 @@ export function deriveStats(nodes: ConversationSnapshot['nodes']): UsageTotals {
   }
 }
 
-/** Props: the conversation-snapshot selector hook (handed down by ChatView). */
+/** Props: the conversation-snapshot selector (dock registration or unit mount). */
 export interface StatsLineProps { useSession: SnapshotSelectorHook<ConversationSnapshot> }
 
 export const StatsLine = memo(function StatsLine({ useSession }: StatsLineProps) {
diff --git a/packages/client/ui-conversation/src/client/chat/ToolRow.module.css b/packages/client/ui-conversation/src/client/chat/ToolRow.module.css
index 7e2ed4db21..4ff289388e 100644
--- a/packages/client/ui-conversation/src/client/chat/ToolRow.module.css
+++ b/packages/client/ui-conversation/src/client/chat/ToolRow.module.css
@@ -175,9 +175,23 @@ button.leading {
   color: var(--dsw-alias-label-tertiary);
 }
 
-/* The code variant's expanded body is the run_code program, rendered through
-   the shared CodeBlock (shiki-highlighted TypeScript); only indentation is
-   this row's concern. */
-.codeBody {
+/* The two block-shaped expanded bodies: the code variant's run_code program
+   through CodeBlock (shiki-highlighted TypeScript) and a terminal card's
+   command output through TerminalBlock. Both are drawn by the shared
+   primitive, so only the row's indentation is this file's concern — the margin
+   also replaces each primitive's own standalone vertical spacing with the
+   flow's row rhythm. */
+.codeBody,
+.terminalBody {
   margin: 4px 0 4px 22px;
 }
+
+/* Indented to the body's own column so the description reads as the card's
+   heading rather than as another summary row, and sits tight against the card
+   below it. Its own rule: grouping it with a body would put description
+   typography on a `CodeBlock` wrapper and change that body's spacing. */
+.terminalDescription {
+  margin: 4px 0 0 22px;
+  color: var(--dsw-alias-label-secondary);
+  font: var(--dsw-font-xs-13);
+}
diff --git a/packages/client/ui-conversation/src/client/chat/ToolRow.tsx b/packages/client/ui-conversation/src/client/chat/ToolRow.tsx
index 9395813910..d4fae4f6c5 100644
--- a/packages/client/ui-conversation/src/client/chat/ToolRow.tsx
+++ b/packages/client/ui-conversation/src/client/chat/ToolRow.tsx
@@ -1,14 +1,18 @@
 // ToolRow: the single-line tool summary row (figma component set 122:9479) —
 // 16px leading slot (state dot / tool icon, chevron on hover or expanded) + title +
-// separator dot + FILL-truncated summary. Expanded body is indented gray text;
-// no inline output (full results live in the details panel). Expand state is
+// separator dot + FILL-truncated summary. The collapsed row is always one
+// line; the expanded body is indented gray text, the run_code program through
+// CodeBlock, or — for a call whose render intent is a terminal card — the
+// command's own output through TerminalBlock, capped at
+// CHAT_TERMINAL_MAX_LINES so the message flow stays scannable. Expand state is
 // component-local view state. File-tool summaries are path links that open
 // through the host; the row itself is not a details-panel control.
 
 import { useState, type KeyboardEvent, type MouseEvent, type ReactNode } from 'react'
 import clsx from 'clsx'
-import { CodeBlock, StateDot } from '@deepseek-ai/dsh-client-ui-primitives'
+import { CodeBlock, StateDot, TerminalBlock } from '@deepseek-ai/dsh-client-ui-primitives'
 import { IconChevronDownOutline14 } from '@deepseek-ai/dsh-client-ui-primitives'
+import { CHAT_TERMINAL_MAX_LINES, type TerminalCardModel } from '../contract/terminal-card-model.ts'
 import type { ToolRowState, ToolRowVariant } from '../contract/tool-call-model.ts'
 import css from './ToolRow.module.css'
 
@@ -20,8 +24,15 @@ export interface ToolRowProps {
   icon: ReactNode
   title: string
   summary: string
-  /** Expanded-body text; null = not expandable (leading slot never toggles). */
+  /** Expanded-body text; null = no text body (`terminal` is the other body source). */
   body: string | null
+  /**
+   * Terminal-card material for a call whose render intent is a terminal card
+   * (derived by `terminalCardModel`); it replaces the text body when present.
+   * Null or absent leaves the text body, and a row with neither is not
+   * expandable (its leading slot never toggles).
+   */
+  terminal?: TerminalCardModel | null | undefined
   state: ToolRowState
   /** Makes the row itself the expand control instead of only its leading icon. */
   expandOnRowClick?: boolean | undefined
@@ -52,17 +63,25 @@ export function ToolRow({
   title,
   summary,
   body,
+  terminal,
   state,
   expandOnRowClick = false,
   filePath,
   onOpenFile,
 }: ToolRowProps) {
   const [expanded, setExpanded] = useState(false)
+  const terminalBody = terminal ?? null
   // A row that names a single file keeps one interaction (open that path);
-  // args expand is off whether or not the open callback is wired yet.
+  // args expand is off whether or not the open callback is wired yet. Terminal
+  // material still expands: only the file variants carry a path, so a terminal
+  // card and a file link never land on the same row.
   const singleFile = filePath !== undefined
   const fileLink = singleFile && onOpenFile !== undefined
-  const expandable = body !== null && !singleFile
+  const expandable = (body !== null && !singleFile) || terminalBody !== null
+  // The text arms take the empty string for a null body: a row expandable
+  // only through its terminal material renders the terminal body instead, so
+  // this substitution never shows.
+  const text = body ?? ''
   const open = expanded && expandable
   const rowExpands = expandable && expandOnRowClick
   const toggleExpand = () => {
@@ -137,9 +156,17 @@ export function ToolRow({
           </>
         )}
       </div>
-      {open && (variant === 'code'
-        ? <CodeBlock code={body} lang="typescript" className={css.codeBody} />
-        : <div className={css.body}>{body}</div>)}
+      {/* The terminal presenter's description belongs ABOVE the card per the
+          render-intent contract, so an expanded terminal row keeps showing it
+          even though the collapsed summary is hidden while open. */}
+      {open && terminalBody?.description !== undefined && (
+        <div className={css.terminalDescription}>{terminalBody.description}</div>
+      )}
+      {open && (terminalBody !== null
+        ? <TerminalBlock {...terminalBody.card} maxLines={CHAT_TERMINAL_MAX_LINES} className={css.terminalBody} />
+        : variant === 'code'
+          ? <CodeBlock code={text} lang="typescript" className={css.codeBody} />
+          : <div className={css.body}>{text}</div>)}
     </div>
   )
 }
diff --git a/packages/client/ui-conversation/src/client/chat/message-chrome.ts b/packages/client/ui-conversation/src/client/chat/message-chrome.ts
new file mode 100644
index 0000000000..cf376473d0
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/chat/message-chrome.ts
@@ -0,0 +1,91 @@
+// Shared chrome helpers for user/assistant IconActions rows: clipboard write
+// and the compact date+clock label from a session-event epoch.
+
+/**
+ * Best-effort clipboard write; rejections stay swallowed (no success chrome).
+ * @param text - Plain text to place on the clipboard.
+ */
+export async function writeClipboard(text: string): Promise<void> {
+  // lib.dom types clipboard non-optional, but insecure contexts omit it —
+  // that runtime gap is exactly what this guard detects.
+  /* eslint-disable-next-line @typescript-eslint/no-unnecessary-condition */
+  if (navigator.clipboard?.writeText) {
+    try {
+      await navigator.clipboard.writeText(text)
+    } catch {
+      // Denied permissions / iframe policy.
+    }
+    return
+  }
+  // execCommand('copy') is the only clipboard fallback where the async API
+  // is missing (insecure contexts); deprecated but deliberately retained.
+  /* eslint-disable @typescript-eslint/no-deprecated */
+  const exec = typeof document.execCommand === 'function'
+    ? document.execCommand.bind(document)
+    : undefined
+  if (exec === undefined) return
+  const el = document.createElement('textarea')
+  el.value = text
+  el.setAttribute('readonly', '')
+  el.style.position = 'fixed'
+  el.style.left = '-9999px'
+  document.body.appendChild(el)
+  el.select()
+  try {
+    exec('copy')
+  } catch {
+    // Clipboard unavailable; the button stays idle.
+  }
+  /* eslint-enable @typescript-eslint/no-deprecated */
+  el.remove()
+}
+
+function pad2(n: number): string {
+  return String(n).padStart(2, '0')
+}
+
+/**
+ * Local calendar-day epoch (ms at local midnight) for an instant.
+ * @param ms - Unix epoch ms.
+ * @returns Midnight of that local calendar day.
+ */
+export function startOfLocalDay(ms: number): number {
+  const d = new Date(ms)
+  d.setHours(0, 0, 0, 0)
+  return d.getTime()
+}
+
+/**
+ * Delay until the next local midnight after `ms` (at least 1ms).
+ * @param ms - Unix epoch ms.
+ * @returns Milliseconds until the following local midnight.
+ */
+export function msUntilNextLocalMidnight(ms: number): number {
+  const next = new Date(ms)
+  next.setHours(24, 0, 0, 0)
+  return Math.max(next.getTime() - ms, 1)
+}
+
+/**
+ * Compact local timestamp for message IconActions.
+ * Same calendar day → `HH:mm`; earlier this year → `M月D日 HH:mm`;
+ * other years → `YYYY年M月D日 HH:mm`.
+ * @param time - Unix epoch ms from the source session event.
+ * @param now - Reference instant for the day/year cut (defaults to wall clock).
+ * @returns Date-aware clock string (24-hour, zero-padded time).
+ */
+export function formatMessageClock(time: number, now: number = Date.now()): string {
+  const d = new Date(time)
+  const n = new Date(now)
+  const clock = `${pad2(d.getHours())}:${pad2(d.getMinutes())}`
+  if (
+    d.getFullYear() === n.getFullYear()
+    && d.getMonth() === n.getMonth()
+    && d.getDate() === n.getDate()
+  ) {
+    return clock
+  }
+  const md = `${d.getMonth() + 1}月${d.getDate()}日`
+  if (d.getFullYear() === n.getFullYear()) return `${md} ${clock}`
+  return `${d.getFullYear()}年${md} ${clock}`
+}
diff --git a/packages/client/ui-conversation/src/client/chat/use-calendar-day.ts b/packages/client/ui-conversation/src/client/chat/use-calendar-day.ts
new file mode 100644
index 0000000000..29e5311383
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/chat/use-calendar-day.ts
@@ -0,0 +1,25 @@
+// Component-local calendar-day tick: memoized message rows keep stable props
+// across midnight, so the IconActions clock needs a local day seat that
+// re-fires at the next local midnight without reaching for framework hooks.
+
+import { useEffect, useState } from 'react'
+import { msUntilNextLocalMidnight, startOfLocalDay } from './message-chrome.ts'
+
+/**
+ * Local calendar-day epoch that advances at each local midnight.
+ * @returns Midnight ms for the current local day; updates after the boundary.
+ */
+export function useCalendarDay(): number {
+  const [day, setDay] = useState(() => startOfLocalDay(Date.now()))
+  useEffect(() => {
+    let timer: ReturnType<typeof setTimeout>
+    const arm = (): void => {
+      const now = Date.now()
+      setDay(startOfLocalDay(now))
+      timer = setTimeout(arm, msUntilNextLocalMidnight(now))
+    }
+    timer = setTimeout(arm, msUntilNextLocalMidnight(Date.now()))
+    return () => { clearTimeout(timer) }
+  }, [])
+  return day
+}
diff --git a/packages/client/ui-conversation/src/client/contract/slots.ts b/packages/client/ui-conversation/src/client/contract/slots.ts
index 9f3f3edf02..c468c418a4 100644
--- a/packages/client/ui-conversation/src/client/contract/slots.ts
+++ b/packages/client/ui-conversation/src/client/contract/slots.ts
@@ -117,6 +117,18 @@ declare module '@deepseek-ai/dsh-client-ui-slots' {
 
 /** Owner share of the strict session content seat. */
 export interface ConversationSessionOwnerProps {
+  /**
+   * Wrap the view ring in the transcript scrollport that also hosts the
+   * sticky composer seat (whole `'conversation.composer'` chain output).
+   * Supplied for every real session (hero/settling/active) so the composer
+   * keeps one tree seat across the blank → active flip; the header stays
+   * outside that wrapper as ordinary column chrome (`flex: none`), while
+   * active CSS sticks the seat to the bottom of the same scrollport so wheel
+   * over the footer scrolls the flow.
+   * @param view - the session view-ring content (null while blank chrome is hidden).
+   * @returns the scrollport containing `view` and the sticky composer seat.
+   */
+  wrapActiveBody?: (view: ReactNode) => ReactNode
 }
 
 /**
@@ -257,6 +269,8 @@ export interface ComposerBarInjected {
    * Resolves admission: false = rejected/unmatched/transport failure.
    */
   command: (line: string) => Promise<boolean>
+  /** Locale-aware hint translator for claimed command placeholders. */
+  translateHint: (key: string) => string
   /** Registrant hooks compartment: the renderer binds these to useNotices/useLexicon. */
   hooks: {
     /** Latest surfaced notice (null after none; seq keys re-render of repeats). */
diff --git a/packages/client/ui-conversation/src/client/contract/terminal-card-model.ts b/packages/client/ui-conversation/src/client/contract/terminal-card-model.ts
new file mode 100644
index 0000000000..e25d68cbbb
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/contract/terminal-card-model.ts
@@ -0,0 +1,189 @@
+/**
+ * Pure derivation of the terminal-card props from a frozen call slice: the
+ * `card:'terminal'` render intent the bash tool declares arrives on the
+ * snapshot as `callView`/`resultView`, and this is the one place that turns
+ * that pair into what {@link TerminalBlock} draws. Both conversation render
+ * sites (the chat tool row's expanded body and the details panel's Output
+ * section) call this, so the command, cwd, output and exit status they show
+ * are derived once.
+ * @module
+ */
+import type { TerminalBlockProps } from '@deepseek-ai/dsh-client-ui-primitives'
+import { resolveToolPath, type ToolCallBlock } from './tool-call-model.ts'
+
+/**
+ * Output lines the chat row's expanded terminal body shows before collapsing
+ * the middle — half the primitive's own default, which the details panel
+ * keeps. A chat row is a summary surface inside the message flow: the flow
+ * must stay scannable across many calls, while the details panel is the
+ * single-call reading surface. A design constant of this UI's row geometry,
+ * not a deployment choice, so it is fixed here rather than a plugin Config
+ * field.
+ */
+export const CHAT_TERMINAL_MAX_LINES = 8
+
+/**
+ * The {@link TerminalBlock} props this derivation owns. Picked off the
+ * primitive's props so the two stay in step; `home` is absent because the web
+ * client has no home path for the session host (a cwd renders as its last
+ * path segment), and `maxLines`/`className` belong to each render site.
+ */
+export interface TerminalCardModel {
+  /**
+   * The props {@link TerminalBlock} draws. Held as a nested object so a render
+   * site spreads exactly the primitive's own surface and can never leak a
+   * neighbouring field into it.
+   */
+  card: Pick<TerminalBlockProps, 'command' | 'cwd' | 'output' | 'exitCode' | 'signal' | 'running'>
+  /**
+   * The call view's model-authored description, which the contract defines as
+   * rendering ABOVE the card (the card itself has no description slot). Absent
+   * when the presenter supplied none, or when the window dropped the call side;
+   * a row then keeps its args-derived summary.
+   */
+  description: string | undefined
+}
+
+/**
+ * Resolve a terminal view's working directory the way the render-intent
+ * contract assigns to the UI bridge: an absolute path is used as-is, a relative
+ * one joins under the session workspace, and an omitted one IS the session
+ * workspace. A pure presenter cannot see the session cwd, which is why this
+ * resolution belongs here rather than in the tool. Without a session cwd there
+ * is nothing to resolve against, so a relative path stays as authored and an
+ * omitted one stays absent (the prompt row then draws a bare `$`).
+ * @param viewCwd - the cwd the terminal call view carries, if any.
+ * @param sessionCwd - the session workspace root, if the caller knows it.
+ * @returns the working directory for the prompt label, or undefined.
+ */
+function resolveTerminalCwd(viewCwd: string | undefined, sessionCwd: string | undefined): string | undefined {
+  if (viewCwd === undefined || viewCwd === '') return sessionCwd
+  if (sessionCwd === undefined || sessionCwd === '') return normalizeSegments(viewCwd)
+  return normalizeSegments(resolveToolPath(sessionCwd, viewCwd))
+}
+
+/**
+ * Collapse `.` and `..` segments so the prompt label names the directory the
+ * command actually ran in. The bash executor resolves the workdir before
+ * running, so a joined `/w/app/..` must display as `w`, not as `..`. Separators
+ * are preserved as authored (a Windows path keeps its backslashes) because this
+ * value is only ever displayed; a `..` that would climb past the root is
+ * dropped, which is what a filesystem does with it. A UNC path's `server` and
+ * `share` are part of its root, not poppable segments: Windows cannot climb
+ * above a share, so `\\\\server\\share` with a `..` stays there.
+ * @param path - a joined or absolute path, possibly carrying `.`/`..` segments.
+ * @returns the same path with those segments resolved.
+ */
+function normalizeSegments(path: string): string {
+  if (!/(?:^|[/\\])\.\.?(?:[/\\]|$)/.test(path)) return path
+  // A UNC path is `\\\\server\\share\\...`: the server and share form the root,
+  // so they are split off here and neither is a segment `..` may pop. Its
+  // separator is fixed to a backslash, since a joined relative part may have
+  // introduced a forward slash that UNC syntax does not use.
+  const unc = /^[/\\]{2}([^/\\]+)[/\\]+([^/\\]+)/.exec(path)
+  if (unc !== null) {
+    // Both groups are mandatory in the pattern, so destructuring types them as
+    // strings without an assertion.
+    const [matched, server, share] = unc
+    const root = `\\\\${String(server)}\\${String(share)}`
+    // Rooted: what follows the share hangs off it, so a `..` at the top is
+    // dropped rather than kept — Windows cannot climb above a share.
+    const rest = collapse(path.slice(matched.length), true)
+    return rest === '' ? root : `${root}\\${rest}`
+  }
+  const backslashed = path.includes('\\') && !path.includes('/')
+  const separator = backslashed ? '\\' : '/'
+  const rooted = /^[/\\]/.test(path)
+  const drive = /^[A-Za-z]:/.exec(path)?.[0] ?? ''
+  const body = collapse(path.slice(drive.length), rooted || drive !== '', separator)
+  const leading = rooted ? separator : ''
+  return drive === '' ? `${leading}${body}` : `${drive}${rooted ? leading : separator}${body}`
+}
+
+/**
+ * Collapse the `.`/`..` segments of a path body against a known root state.
+ * @param body - the path after any drive letter or UNC root.
+ * @param rooted - the body hangs off a root, so a `..` at its top is dropped
+ *   the way a filesystem drops one; without a root the `..` is kept, since it
+ *   stays meaningful against a cwd this function cannot see.
+ * @param separator - separator to rejoin with (default `/`).
+ * @returns the collapsed body, without leading or trailing separators.
+ */
+function collapse(body: string, rooted: boolean, separator = '/'): string {
+  const kept: string[] = []
+  for (const segment of body.split(/[/\\]/)) {
+    if (segment === '' || segment === '.') continue
+    if (segment === '..') {
+      if (kept.length > 0 && kept[kept.length - 1] !== '..') kept.pop()
+      else if (!rooted) kept.push(segment)
+      continue
+    }
+    kept.push(segment)
+  }
+  return kept.join(separator)
+}
+
+/**
+ * Derive the terminal-card props for a tool call, or null when this call is
+ * not a terminal card and belongs on the generic path.
+ *
+ * The call side supplies the command and its working directory; the result
+ * side supplies the captured output and exit status. Three cases produce
+ * null, all of them the documented generic-card default:
+ *
+ * - Neither side declares `card:'terminal'` — including a `card` value this
+ *   UI version does not know, which arrives over the wire and therefore
+ *   cannot be trusted to be one of the compiled variants.
+ * - A settled call whose result view is not a terminal card: the result
+ *   presentation decides how the settled call renders, and the bash tool
+ *   returns a generic fenced card for an execution error or a background
+ *   start, whose text and error styling the generic path preserves.
+ *
+ * Window truncation can drop the call head from a settled result (see
+ * `ToolResultNode.call`/`callView` in dsh-client-runtime), leaving a terminal
+ * result with no call side. That still renders: the command falls back to the
+ * result view's replacement title, then to an empty command (the prompt line
+ * draws bare), and the prompt shows no cwd.
+ * @param block - RunningToolCall or ToolResultNode off the snapshot caches.
+ * @param sessionCwd - the session workspace root, which resolves an omitted or
+ *   relative view cwd (see {@link resolveTerminalCwd}); absent leaves both unresolved.
+ * @returns the terminal-card props, or null for the generic path.
+ */
+export function terminalCardModel(block: ToolCallBlock, sessionCwd?: string): TerminalCardModel | null {
+  const call = block.callView?.card === 'terminal' ? block.callView : null
+  if (!('kind' in block)) {
+    // Running: the call view exists, the result view does not yet.
+    return call === null ? null : {
+      description: call.description,
+      card: {
+        command: call.title,
+        cwd: resolveTerminalCwd(call.cwd, sessionCwd),
+        output: undefined,
+        exitCode: undefined,
+        signal: undefined,
+        running: true,
+      },
+    }
+  }
+  const result = block.resultView?.card === 'terminal' ? block.resultView : null
+  if (result === null) return null
+  return {
+    description: call?.description,
+    card: {
+      // The result's title REPLACES the pending one when the tool supplies it
+      // (the presentation contract's replacement-title rule); the call title is
+      // what a result without one keeps.
+      command: result.title ?? call?.title ?? '',
+      // Only a PRESENT call view can mean "omitted the cwd, so use the
+      // workspace". When the window dropped the call head there is no cwd
+      // anywhere — the result view carries none — and the original call may
+      // well have used an explicit workdir, so the prompt draws a bare `$`
+      // rather than naming a directory this card cannot know.
+      cwd: call === null ? undefined : resolveTerminalCwd(call.cwd, sessionCwd),
+      output: result.output,
+      exitCode: result.exitCode,
+      signal: result.signal,
+      running: false,
+    },
+  }
+}
diff --git a/packages/client/ui-conversation/src/client/contract/tool-call-model.ts b/packages/client/ui-conversation/src/client/contract/tool-call-model.ts
index 04a8c66d40..b53ef95c01 100644
--- a/packages/client/ui-conversation/src/client/contract/tool-call-model.ts
+++ b/packages/client/ui-conversation/src/client/contract/tool-call-model.ts
@@ -1,7 +1,9 @@
 /**
  * Pure row-model derivation for tool summary rows: variant classification,
- * one-line summary and expanded-body text from the frozen call slice. No
- * inline output ever — full results live in the details panel.
+ * one-line summary and expanded-body text from the frozen call slice. This
+ * derivation reads the call ARGUMENTS only; a call whose render intent is a
+ * terminal card gets its expanded body from the views instead, through
+ * `terminalCardModel` in terminal-card-model.ts.
  */
 // The block union's defining home is runtime (fold-product types); this
 // contract only forwards it (type-definition authority stays with the layer
diff --git a/packages/client/ui-conversation/src/client/input/machine.ts b/packages/client/ui-conversation/src/client/input/machine.ts
index f9c5a479a4..48f6ddd872 100644
--- a/packages/client/ui-conversation/src/client/input/machine.ts
+++ b/packages/client/ui-conversation/src/client/input/machine.ts
@@ -297,14 +297,23 @@ export class InputMachine {
     return []
   }
 
-  /** Shared chip-insertion transaction: replace [span) with one placeholder occurrence (insert-ref and paste-upgrade both land here). */
-  private replaceSpanWithChip(reference: ReferenceInsert, span: TokenSpan): void {
+  /**
+   * Shared chip-insertion transaction: replace [span) with one placeholder
+   * occurrence (insert-ref and paste-upgrade both land here). A separating
+   * space follows the chip unless one is already next.
+   * @returns the inserted length (placeholder plus optional gap).
+   */
+  private replaceSpanWithChip(reference: ReferenceInsert, span: TokenSpan): number {
     this.pushTxn()
     this.typingRun = undefined
-    this.reconcile({ start: span.start, end: span.end, insertedLength: 1 })
+    const tail = this.draft.slice(span.end)
+    const gap = tail.length === 0 || tail[0] !== ' ' ? ' ' : ''
+    const inserted = PLACEHOLDER + gap
+    this.reconcile({ start: span.start, end: span.end, insertedLength: inserted.length })
     this.withMinted([this.mint(reference, span.start)])
-    this.adopt(this.draft.slice(0, span.start) + PLACEHOLDER + this.draft.slice(span.end))
+    this.adopt(this.draft.slice(0, span.start) + inserted + tail)
     this.watchClaim()
+    return inserted.length
   }
 
   /**
@@ -442,10 +451,10 @@ export class InputMachine {
     if (attempt === undefined || attempt.attemptId !== attemptId) return []
     if (this.phase !== 'plain' && this.phase !== 'claimed') return []
     if (!this.casOk(span) || span.start === span.end) return []
-    this.replaceSpanWithChip(reference, span)
+    const insertedLength = this.replaceSpanWithChip(reference, span)
     this.paste = {
       ...attempt,
-      insertedRange: { start: attempt.insertedRange.start, end: attempt.insertedRange.end + 1 - (span.end - span.start) },
+      insertedRange: { start: attempt.insertedRange.start, end: attempt.insertedRange.end + insertedLength - (span.end - span.start) },
     }
     return []
   }
diff --git a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css
index de974da86c..272bbae873 100644
--- a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css
+++ b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css
@@ -17,6 +17,12 @@
   border-bottom: 1px solid var(--dsw-alias-border-l2);
 }
 
+/* Blank hero/settling: keep the header node mounted (stable Session tree for
+   the wrapActiveBody composer) without taking column space. */
+.headerHidden {
+  display: none;
+}
+
 .crumbRow {
   display: flex;
   align-items: center;
@@ -127,6 +133,46 @@
   flex-direction: column;
 }
 
+/* Common seat for the composer chain (fallback + elected overlay siblings). */
+.composerSeat {
+  display: flex;
+  flex: none;
+  flex-direction: column;
+}
+
+/* Active phase: header is ordinary column chrome above the scrollport (not
+   sticky). The scroll body holds the transcript and the sticky composer seat
+   so wheel over the footer moves the flow. */
+.root[data-phase='active'] {
+  overflow: hidden;
+}
+
+.root[data-phase='active'] .header {
+  flex: none;
+}
+
+.scrollBody {
+  display: flex;
+  flex: 1;
+  flex-direction: column;
+  min-height: 0;
+  overflow-y: auto;
+}
+
+.root[data-phase='active'] .viewArea {
+  flex: 1 0 auto;
+  min-height: auto;
+}
+
+.root[data-phase='active'] .composerSeat {
+  position: sticky;
+  bottom: 0;
+  /* Above markdown CodeBlock sticky banners (z-index 6) so the footer never
+     paints under a sticking code header while scrolling. */
+  z-index: 7;
+  background: var(--dsw-alias-bg-base);
+}
+
 /* Hero phase: the composer stack (hero chrome + workspace row + card) is
    flex-centered in the column; composer phase docks it at the bottom. Flex,
    NOT absolute+transform: a transform would make this box the containing
@@ -165,12 +211,15 @@
   padding-left: 8px;
 }
 
-.root[data-phase='hero'] {
+/* Hero: the composer sits inside the session scroll body; center there so
+   the tree seat matches active (sticky footer) without a Root remount. */
+.root[data-phase='hero'] .scrollBody {
   justify-content: center;
+  overflow-y: auto;
 }
 
-/* Settling (session replaying, hero/docked unknown): keep the composer
+/* Settling (session replaying, hero/docked unknown): keep the composer seat
    mounted but invisible so no wrong layout flashes before the phase lands. */
-.root[data-phase='settling'] .composerStack {
+.root[data-phase='settling'] .composerSeat {
   visibility: hidden;
 }
diff --git a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx
index b110860685..3afae25de9 100644
--- a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx
@@ -2,7 +2,7 @@
 // chain stay mounted across no-session/session transitions. Only the inert
 // input body swaps for the strict session InputBar.
 
-import { useEffect, useRef, useState } from 'react'
+import { useEffect, useRef, useState, type ReactNode } from 'react'
 import clsx from 'clsx'
 import type { WorkspaceId } from '@deepseek-ai/dsh-client-runtime/client'
 import type { ConversationSlotProps, InputZone } from '../contract/slots.ts'
@@ -113,24 +113,53 @@ export function ConversationRoot({
       {hero && <HeroGlow className={css.heroGlow} />}
       {hero && <HeroShell />}
       {hero && heroWorkspaceRow}
-      {!hero && zone !== undefined && renderSlot('conversation.input.dock', zone)}
+      {/* Stats band above the input-dock strips so the prior ChatView footer
+          order (stats → todo/queue → card) is preserved under the sticky stack. */}
       {!hero && zone !== undefined && renderSlot('conversation.composer.dock', zone)}
+      {!hero && zone !== undefined && renderSlot('conversation.input.dock', zone)}
       {inputBar}
     </div>
   )
 
+  const phase = settling ? 'settling' : hero ? 'hero' : 'active'
+  const composer = renderSlotChain(
+    'conversation.composer',
+    { interactions: pending },
+    { fallback: composerBar, overlay: true },
+  )
+
+  // Sticky wraps the whole chain output (fallback + elected overlay), not
+  // only `.composerStack`: overlay:true renders those as siblings, and sticky
+  // on the fallback alone would leave Question/Approval panels at the content
+  // end off-screen when the user is not pinned to the floor.
+  const composerSeat = (
+    <div className={css.composerSeat} data-composer-seat="">
+      {composer}
+    </div>
+  )
+
+  // Header stays column chrome above this scrollport; the sticky composer
+  // seat lives inside it with the transcript. Always wrap while a session
+  // exists (hero/settling/active) so the composer keeps one tree seat across
+  // the blank → active flip — relocating it only in active remounted the textarea.
+  const wrapActiveBody = (view: ReactNode): ReactNode => (
+    <div className={css.scrollBody} data-conversation-scroll="">
+      {view}
+      {composerSeat}
+    </div>
+  )
+
   return (
-    <div className={css.root} data-phase={settling ? 'settling' : hero ? 'hero' : 'active'}>
+    <div className={css.root} data-phase={phase}>
       {/* Mounted for every real session, hero included: ConversationSession
-          renders no chrome while blank but owns the draft-persistence mirror
-          bind — unmounting it in the hero would lose pre-first-send text on
-          a refresh or scope rebuild. */}
-      {sessionId !== undefined && renderSlot('conversation.session', {})}
-      {renderSlotChain(
-        'conversation.composer',
-        { interactions: pending },
-        { fallback: composerBar, overlay: true },
+          keeps a chrome-hidden shell while blank and owns the draft-
+          persistence mirror bind — unmounting it in the hero would lose
+          pre-first-send text on a refresh or scope rebuild. */}
+      {sessionId !== undefined && renderSlot(
+        'conversation.session',
+        { wrapActiveBody },
       )}
+      {sessionId === undefined ? composerSeat : null}
     </div>
   )
 }
diff --git a/packages/client/ui-conversation/src/client/skeleton/ConversationSession.tsx b/packages/client/ui-conversation/src/client/skeleton/ConversationSession.tsx
index 0c1addcba7..33b24f5245 100644
--- a/packages/client/ui-conversation/src/client/skeleton/ConversationSession.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/ConversationSession.tsx
@@ -1,6 +1,6 @@
 /** Strict per-session conversation content: header, view ring, and chat store bindings. */
 
-import { useEffect, useSyncExternalStore } from 'react'
+import { useEffect, useSyncExternalStore, type ReactNode } from 'react'
 import clsx from 'clsx'
 import { shallowEqual } from '@deepseek-ai/dsh-client-runtime/client'
 import type { SessionId, SessionListState, SessionSummary } from '@deepseek-ai/dsh-client-runtime/client'
@@ -24,7 +24,7 @@ function deriveAncestry(list: SessionListState, id: SessionId): readonly Session
 
 export function ConversationSession({
   sessionId, useSession, useSessions, useInput, inputActions, useStore, actions,
-  renderSlot, views, bindDraftMirror, open,
+  renderSlot, views, bindDraftMirror, open, wrapActiveBody,
 }: ConversationSessionProps) {
   useSyncExternalStore(views.subscribe, views.version)
   const tabs = views.list()
@@ -44,52 +44,67 @@ export function ConversationSession({
     // the machine mirror, not this seed effect.
   }, [inputActions])
 
-  if (blank && composerPhase === 'blank') return null
+  // Blank hero/settling: keep the same header + body tree shape so a
+  // wrapActiveBody-hosted composer keeps its DOM identity across the first
+  // send (hero → active). Chrome is hidden; the draft-persistence mirror
+  // still runs because this component stays mounted.
+  const hideChrome = blank && composerPhase === 'blank'
+
+  const view: ReactNode = hideChrome ? null : (
+    <div className={css.viewArea}>
+      {active !== undefined && renderSlot('conversation.view', {}, { only: active.id })}
+    </div>
+  )
 
   return (
     <>
-      <header className={css.header}>
-        <div className={css.crumbRow}>
-          <nav className={css.crumbs} aria-label="Session hierarchy">
-            {ancestry.map((summary, index) => {
-              const last = index === ancestry.length - 1
-              return (
-                <span key={summary.id} className={css.crumbSeg}>
-                  {index > 0 && <span className={css.crumbSep}>/</span>}
+      <header
+        className={clsx(css.header, hideChrome && css.headerHidden)}
+        aria-hidden={hideChrome || undefined}
+      >
+        {!hideChrome && (
+          <>
+            <div className={css.crumbRow}>
+              <nav className={css.crumbs} aria-label="Session hierarchy">
+                {ancestry.map((summary, index) => {
+                  const last = index === ancestry.length - 1
+                  return (
+                    <span key={summary.id} className={css.crumbSeg}>
+                      {index > 0 && <span className={css.crumbSep}>/</span>}
+                      <button
+                        type="button"
+                        className={clsx(css.crumb, last && css.crumbCurrent)}
+                        disabled={last}
+                        onClick={() => { open(summary.id) }}
+                      >
+                        {summary.displayTitle}
+                      </button>
+                    </span>
+                  )
+                })}
+                {ancestry.length === 0 && <span className={css.crumbCurrent}>{sessionId}</span>}
+              </nav>
+            </div>
+            {tabs.length > 1 && (
+              <div className={css.tabs} role="tablist">
+                {tabs.map(viewTab => (
                   <button
+                    key={viewTab.id}
                     type="button"
-                    className={clsx(css.crumb, last && css.crumbCurrent)}
-                    disabled={last}
-                    onClick={() => { open(summary.id) }}
+                    role="tab"
+                    aria-selected={viewTab.id === active?.id}
+                    className={clsx(css.tab, viewTab.id === active?.id && css.tabActive)}
+                    onClick={() => { actions.setView(viewTab.id) }}
                   >
-                    {summary.displayTitle}
+                    {viewTab.label}
                   </button>
-                </span>
-              )
-            })}
-            {ancestry.length === 0 && <span className={css.crumbCurrent}>{sessionId}</span>}
-          </nav>
-        </div>
-        {tabs.length > 1 && (
-          <div className={css.tabs} role="tablist">
-            {tabs.map(view => (
-              <button
-                key={view.id}
-                type="button"
-                role="tab"
-                aria-selected={view.id === active?.id}
-                className={clsx(css.tab, view.id === active?.id && css.tabActive)}
-                onClick={() => { actions.setView(view.id) }}
-              >
-                {view.label}
-              </button>
-            ))}
-          </div>
+                ))}
+              </div>
+            )}
+          </>
         )}
       </header>
-      <div className={css.viewArea}>
-        {active !== undefined && renderSlot('conversation.view', {}, { only: active.id })}
-      </div>
+      {wrapActiveBody !== undefined ? wrapActiveBody(view) : view}
     </>
   )
 }
diff --git a/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.module.css b/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.module.css
index abdece3e25..143174fe42 100644
--- a/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.module.css
+++ b/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.module.css
@@ -92,3 +92,17 @@
 .code[data-error] {
   color: var(--dsw-alias-state-error-primary);
 }
+
+/* Above the card, which is where the render-intent contract puts a terminal
+   call's description; the panel has no summary row to carry it. */
+.terminalDescription {
+  margin: 0 0 6px;
+  color: var(--dsw-alias-label-secondary);
+  font: var(--dsw-font-xs-13);
+}
+
+/* The terminal card sits directly under its section label, so it drops the
+   primitive's standalone vertical margin; the section owns the spacing. */
+.terminal {
+  margin: 0;
+}
diff --git a/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.tsx b/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.tsx
index 650a95f833..9fc5a04ff6 100644
--- a/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/DetailsPanel.tsx
@@ -1,47 +1,59 @@
 // DetailsPanel, P-I minimal form: close button + the selected call's args and
-// result rendered raw. The three-段 Switch / Prev-Next stepping / See-in-
-// trajectory are deferred (ledger). Reads the selection from the shared chat
+// result — args as JSON, the result raw except for a terminal-card call, whose
+// Output section is the command's terminal card. The three-段 Switch /
+// Prev-Next stepping / See-in-trajectory are deferred (ledger). Reads the
+// selection from the shared chat
 // store (conversation writes, this panel reads — the cross-registration
 // share the store seat exists for) and derives the call material from the
 // session snapshot — no data of its own.
 
-import { CodeBlock } from '@deepseek-ai/dsh-client-ui-primitives'
+import { CodeBlock, TerminalBlock } from '@deepseek-ai/dsh-client-ui-primitives'
 import { shallowEqual } from '@deepseek-ai/dsh-client-runtime/client'
-import type { ConversationSnapshot, ToolResultNode } from '@deepseek-ai/dsh-client-runtime/client'
+import type { ConversationSnapshot, RunningToolCall, ToolResultNode } from '@deepseek-ai/dsh-client-runtime/client'
 import type { DetailsSlotProps } from '../contract/slots.ts'
+import { terminalCardModel } from '../contract/terminal-card-model.ts'
+import type { ToolCallBlock } from '../contract/tool-call-model.ts'
 import css from './DetailsPanel.module.css'
 
 /** Full props composed by reference from the contract (automatic shares & injected share). */
 export type DetailsPanelProps = DetailsSlotProps
 
-/** Selected call material: resolved result node, or the in-flight running call's args. */
+/**
+ * Selected call material: the call's display name and args plus the frozen
+ * block slice it came from. `block` is a snapshot-cached reference, so the
+ * wrapper stays shallow-equal across unrelated snapshot frames; the settled /
+ * running split is read off it with the `'kind' in block` discrimination
+ * instead of duplicated as flags.
+ */
 interface CallMaterial {
   name: string
   argsRaw: string | null
-  result: ToolResultNode | null
-  running: boolean
+  block: ToolCallBlock
+}
+
+/** Material of a settled result node (native call or run_code sub-dispatch). */
+function settledMaterial(node: ToolResultNode, callId: string): CallMaterial {
+  return { name: node.call?.name ?? callId, argsRaw: node.call?.argsRaw ?? null, block: node }
+}
+
+/** Material of an in-flight call (native call or run_code sub-dispatch). */
+function runningMaterial(call: RunningToolCall): CallMaterial {
+  return { name: call.name, argsRaw: call.argsRaw, block: call }
 }
 
 function materialFor(s: ConversationSnapshot, callId: string): CallMaterial | null {
   for (const node of s.nodes) {
-    if (node.kind === 'tool-result' && node.callId === callId) {
-      return { name: node.call?.name ?? callId, argsRaw: node.call?.argsRaw ?? null, result: node, running: false }
-    }
+    if (node.kind === 'tool-result' && node.callId === callId) return settledMaterial(node, callId)
   }
   const open = s.runningCalls.find(c => c.callId === callId)
-  if (open !== undefined) {
-    return { name: open.name, argsRaw: open.argsRaw, result: null, running: true }
-  }
+  if (open !== undefined) return runningMaterial(open)
   // run_code sub-dispatches: the native call-block shapes, so a selected
   // sub-row resolves through the same material as a native call — the
   // settled ToolResultNode form, or the RunningToolCall form mid-flight.
   for (const subs of s.codeDispatches.values()) {
     for (const sub of subs) {
       if (sub.callId !== callId) continue
-      if ('kind' in sub) {
-        return { name: sub.call?.name ?? callId, argsRaw: sub.call?.argsRaw ?? null, result: sub, running: false }
-      }
-      return { name: sub.name, argsRaw: sub.argsRaw, result: null, running: true }
+      return 'kind' in sub ? settledMaterial(sub, callId) : runningMaterial(sub)
     }
   }
   return null
@@ -56,8 +68,11 @@ function pretty(raw: string): string {
   }
 }
 
-export function DetailsPanel({ useSession, useStore, closeDetails }: DetailsPanelProps) {
+export function DetailsPanel({ useSession, useSessions, sessionId, useStore, closeDetails }: DetailsPanelProps) {
   const selection = useStore(s => s.selection)
+  // Session workspace root: an omitted or relative terminal cwd resolves
+  // against it, which the pure presenter cannot see.
+  const sessionCwd = useSessions(list => list.byId[sessionId]?.cwd)
   const callId = selection?.callId
   // materialFor builds a fresh wrapper; shallowEqual short-circuits on its
   // stable members (result node reference rides the snapshot's structural sharing).
@@ -95,15 +110,11 @@ export function DetailsPanel({ useSession, useStore, closeDetails }: DetailsPane
                 )}
                 <section className={css.section}>
                   <div className={css.sectionLabel}>Output</div>
-                  {/* materialFor invariant: result===null ⇔ running (a settled
-                        material always carries its result node). */}
-                  {material.result === null
-                    ? <div className={css.empty}>运行中…</div>
-                    : (
-                      <pre className={css.code} data-error={material.result.isError || undefined}>
-                        {renderResult(material.result)}
-                      </pre>
-                    )}
+                  {/* Keyed by the selected call: the body owns per-call view
+                      state (the terminal card's expand and copy), which React
+                      would otherwise carry into the next selection because the
+                      panel does not unmount between calls. */}
+                  <OutputBody key={callId} material={material} cwd={sessionCwd} />
                 </section>
               </>
             )}
@@ -112,6 +123,41 @@ export function DetailsPanel({ useSession, useStore, closeDetails }: DetailsPane
   )
 }
 
+/**
+ * The Output section's body for the selected call. A terminal-card call — a
+ * shell command's call/result views — renders through the shared TerminalBlock
+ * at the primitive's own full height allowance, so column-aligned output keeps
+ * its alignment and scrolls sideways instead of folding. Every other call, and
+ * a running call with no terminal card yet, keeps the flattened text form.
+ * @param props.material - the selected call's material from {@link materialFor}.
+ * @param props.cwd - the session workspace root, resolving the terminal view's cwd.
+ * @returns the Output section's body element.
+ */
+function OutputBody({ material, cwd }: { material: CallMaterial; cwd: string | undefined }) {
+  const terminal = terminalCardModel(material.block, cwd)
+  if (terminal !== null) {
+    // The contract renders the presenter's description above the card, and the
+    // panel has no summary row to carry it, so it is drawn here.
+    return (
+      <>
+        {terminal.description !== undefined && (
+          <div className={css.terminalDescription}>{terminal.description}</div>
+        )}
+        <TerminalBlock {...terminal.card} className={css.terminal} />
+      </>
+    )
+  }
+  // A settled call always carries the result node the flattened form needs;
+  // the running shape has no result to flatten.
+  if (!('kind' in material.block)) return <div className={css.empty}>运行中…</div>
+  const result = material.block
+  return (
+    <pre className={css.code} data-error={result.isError || undefined}>
+      {renderResult(result)}
+    </pre>
+  )
+}
+
 /** Flatten result content blocks to display text (text blocks verbatim, others as JSON). */
 function renderResult(node: ToolResultNode): string {
   const parts: string[] = []
diff --git a/packages/client/ui-conversation/src/client/skeleton/EmptyHero.tsx b/packages/client/ui-conversation/src/client/skeleton/EmptyHero.tsx
index 7fbed0ee25..3b2d60417a 100644
--- a/packages/client/ui-conversation/src/client/skeleton/EmptyHero.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/EmptyHero.tsx
@@ -115,9 +115,9 @@ export function HeroShell({ children }: HeroShellProps) {
           Let&apos;s start building
         </div>
         <div className={css.body}>
-          {/* The resident composer (rendered by ConversationRoot at its stable
-              tree position; the workspace row rides its accessory hole) is
-              CSS-positioned into this gap during the hero phase — see
+          {/* The resident composer (ConversationRoot wrapActiveBody seat; the
+              workspace row rides the stack above the card) is CSS-centered in
+              the session scroll body during hero — see
               ConversationRoot.module.css [data-phase='hero']. */}
         </div>
       </div>
diff --git a/packages/client/ui-conversation/src/client/skeleton/InputBar.module.css b/packages/client/ui-conversation/src/client/skeleton/InputBar.module.css
index 1bada4391d..8837752830 100644
--- a/packages/client/ui-conversation/src/client/skeleton/InputBar.module.css
+++ b/packages/client/ui-conversation/src/client/skeleton/InputBar.module.css
@@ -125,20 +125,18 @@
   position: absolute;
   inset: 0;
   overflow: hidden;
-  color: transparent;
+  color: var(--dsw-alias-label-primary);
   pointer-events: none;
 }
 
 .hlToken {
-  border-radius: 4px;
-  /* GOAL 稿 amber token emphasis (state warn pair; glyphs stay the textarea's). */
-  background: var(--dsw-alias-state-warn-tertiary);
-  color: transparent;
+  background-color: transparent;
+  color: var(--dsw-alias-state-warn-label);
 }
 
 .hlSegment {
   border-radius: 4px;
-  background: var(--dsw-alias-interactive-bg-hover);
+  background-color: transparent;
   color: transparent;
 }
 
@@ -170,7 +168,7 @@
   border: none;
   outline: none;
   background: transparent;
-  color: var(--dsw-alias-label-primary);
+  color: transparent;
   /* Business blue, not brand-primary: that token resolves to ink in this sheet. */
   caret-color: var(--dsw-alias-state-business-primary);
 }
@@ -348,25 +346,13 @@
    draft's own glyphs — advance untouched, so the two layers cannot drift.
    Chip family colors; clone keeps rounded ends on soft-wrap fragments. */
 .textRef {
-  color: transparent;
   background-color: transparent;
+  color: var(--dsw-alias-state-business-primary);
   box-decoration-break: clone;
   -webkit-box-decoration-break: clone;
-  position: relative;
 }
 .textRef:after {
-  content: "";
-  position: absolute;
-  left: 0;
-  top: 0;
-
-  width: 100%;
-  height: 100%;
-
-  border-radius: 6px;
-  background: rgba(97, 135, 216, 0.22);
-  transform: translate(-2px, -1px);
-  padding: 2px 4px;
+  display: none;
 }
 
 /* Reference chip: rendered in the backdrop at the placeholder offset. Hard
diff --git a/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx b/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx
index a98ee09114..24359965f5 100644
--- a/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx
@@ -13,6 +13,8 @@ import { IconPlusOutline16 } from '@deepseek-ai/dsh-client-ui-primitives'
 // Type-only: the `plan` projection key merge (the TodoDock posture — the
 // composer reads a host-computed value; the domain owns the key).
 import type {} from '@deepseek-ai/dsh-plan-mode/client'
+// Type-only: the `goal` projection key merge (hint disambiguation).
+import type {} from '@deepseek-ai/dsh-goal/client'
 import type { ComposerBarProps } from '../contract/slots.ts'
 import { deriveDecorations } from '../input/decorations.ts'
 import { PermissionSelect } from './PermissionSelect.tsx'
@@ -27,7 +29,7 @@ export interface InputBarError {
 export type InputBarProps = ComposerBarProps
 
 export function InputBar({
-  useSession, useInput, inputActions, keyboard, stop, command, renderSlot, useNotices, useLexicon, useProjection,
+  useSession, useInput, inputActions, keyboard, stop, command, translateHint, renderSlot, useNotices, useLexicon, useProjection,
   variant, placeholder, accessory, overlay, leftItems, rightItems, onAdd, addLabel = 'Add attachment',
 }: InputBarProps) {
   const input = useInput(s => s)
@@ -39,6 +41,8 @@ export function InputBar({
   // Plan mode swaps the textarea placeholder (the projection is the folded
   // host value; owner-prop placeholders — hero, session-unavailable — win).
   const planActive = useProjection('plan', plan => plan !== undefined && (plan.pending ? !plan.active : plan.active))
+  // Absent (undefined: no frame yet) and cleared (null) both mean no goal.
+  const hasGoal = useProjection('goal', goal => goal != null)
   // Prompt failures are ordinary failures (no create/attach transaction
   // exists anymore): the strip renders promptError, the draft stays in the
   // machine, and the user resubmits.
@@ -75,6 +79,27 @@ export function InputBar({
     if (!locked) inputRef.current?.focus()
   }, [locked])
 
+  // Active conversation scrollport: chain the wheel. While the textarea (capped
+  // at 14 lines with overflow-y:auto) can still move in this direction, keep
+  // the native scroll; only at its own edge forward delta to the host so a
+  // short draft never traps the gesture and a long draft stays scrollable.
+  // Hero mounts have no host and keep native wheel scrolling.
+  useEffect(() => {
+    const el = inputRef.current
+    if (el === null) return
+    const onWheel = (e: WheelEvent): void => {
+      const host = el.closest('[data-conversation-scroll]')
+      if (!(host instanceof HTMLElement) || e.deltaY === 0) return
+      const atTop = el.scrollTop <= 0
+      const atEnd = el.scrollTop + el.clientHeight >= el.scrollHeight - 1
+      if ((e.deltaY < 0 && !atTop) || (e.deltaY > 0 && !atEnd)) return
+      e.preventDefault()
+      host.scrollTop += e.deltaY
+    }
+    el.addEventListener('wheel', onWheel, { passive: false })
+    return () => { el.removeEventListener('wheel', onWheel) }
+  }, [])
+
   const onKeyDown = (e: KeyboardEvent<HTMLTextAreaElement>): void => {
     // Shift+Enter is the native newline UNCONDITIONALLY — decided before the
     // IME guard so a composition-closing Shift+Enter still breaks the line.
@@ -296,7 +321,12 @@ export function InputBar({
     }
     pushPlain(draft.length)
     if (deco.hint !== null) {
-      backdrop.push(<span key="hint" className={css.hint} data-decoration="hint">{deco.hint}</span>)
+      // Claim tokens are shaped `/name ` (trailing space); trim to the bare name.
+      const commandName = input.claim?.token.slice(1).trim() ?? ''
+      const hintKey = commandName === 'goal' && hasGoal ? 'goal.active' : commandName
+      const translated = translateHint(hintKey)
+      const displayHint = translated !== hintKey ? translated : deco.hint
+      backdrop.push(<span key="hint" className={css.hint} data-decoration="hint">{displayHint}</span>)
     }
   }
 
@@ -312,7 +342,7 @@ export function InputBar({
           {notice.text}
         </div>
       )}
-      <div className={css.card}>
+      <div className={css.card} data-composer-card>
         {overlay !== undefined && <div className={css.overlayAnchor}>{overlay}</div>}
         {accessory !== undefined && <div className={css.accessory}>{accessory}</div>}
         {/* Mirror-div auto-grow: the hidden mirror renders draft+'\n' and stretches the wrapper
@@ -329,7 +359,7 @@ export function InputBar({
             data-phase={input.phase}
             placeholder={placeholder ?? (disabled
               ? 'Session unavailable'
-              : planActive ? 'describe your task to generate plan' : 'Message the agent')}
+              : planActive ? translateHint('placeholder.plan') : translateHint('placeholder.default'))}
             rows={2}
             onChange={onChange}
             onKeyDown={onKeyDown}
diff --git a/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.module.css b/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.module.css
index dd5986992c..50dce3913f 100644
--- a/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.module.css
+++ b/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.module.css
@@ -1,49 +1,43 @@
-/* Composer bottom-row permission chip (draft start.jpeg `Read-only ∨`): a
-   quiet text chip with a chevron; hover paints the standard interactive pill.
-   The native select is stretched invisibly over the chip so the platform
-   dropdown does the menu work — keyboard/AT semantics come free. */
-
-.root {
-  position: relative;
-  display: inline-flex;
-  align-items: center;
-}
-
-.chip {
+.trigger {
   display: inline-flex;
   align-items: center;
   gap: 4px;
-  padding: 6px 8px;
-  border-radius: 8px;
-  color: var(--dsw-alias-label-secondary);
-  font-size: 14px;
-  line-height: 20px;
-  pointer-events: none; /* the overlaid select owns the interaction */
-}
-
-.root:hover .chip {
-  background: var(--dsw-alias-interactive-bg-hover);
-}
-
-.chevron {
-  color: var(--dsw-alias-label-caption);
-}
-
-/* Invisible native select stretched over the chip: real menu, zero drawing. */
-.select {
-  position: absolute;
-  inset: 0;
-  width: 100%;
-  height: 100%;
-  opacity: 0;
+  min-width: 0;
+  max-width: 220px;
+  height: 28px;
+  padding: 0 4px 0 8px;
   border: none;
+  border-radius: 8px;
+  outline: none;
+  background: transparent;
+  color: var(--dsw-alias-label-secondary);
+  font-size: 13px;
+  line-height: 20px;
+  font-weight: 500;
   cursor: pointer;
 }
 
-.select:disabled {
+.trigger:hover:not(:disabled) {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.trigger:focus-visible {
+  box-shadow: 0 0 0 2px var(--dsw-alias-border-l3);
+}
+
+.trigger:disabled {
+  color: var(--dsw-alias-label-dimmed);
   cursor: default;
 }
 
-.root:has(.select:disabled) .chip {
-  opacity: 0.5;
+.triggerLabel {
+  min-width: 0;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.chevron {
+  flex: 0 0 auto;
+  color: var(--dsw-alias-label-caption);
 }
diff --git a/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.tsx b/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.tsx
index 0622e64500..873c8c11c4 100644
--- a/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/PermissionSelect.tsx
@@ -1,27 +1,14 @@
-// PermissionSelect: the composer bottom-row permission chip (draft
-// start.jpeg's `Read-only ∨` control), the Access seat's wired occupant.
-// Options and the current value read from the host-computed `permissions`
-// projection (baseline block + push frames — no fetch, no mount timing);
-// key absence (a permission-less composition, or a Draft with no host
-// session yet) renders nothing. The visible chip is presentation only — an
-// invisible native select stretched over it owns the menu and interaction.
-// A switch submits the `/permission <preset>` command line (the one write
-// path); the control shows the picked value optimistically and disables
-// until the admission result, then re-follows the projection — the pushed
-// frame confirms the switch, and a failed/unmatched submit falls back to
-// the still-authoritative projection value (`custom` is shown as the
-// current value but never offered as a target — the host omits it from
-// switchable options).
-
 import { useState } from 'react'
 import type { PermissionSelect as PermissionSelectValue } from '@deepseek-ai/dsh-permission/client'
+import { Menu } from '@deepseek-ai/dsh-client-ui-primitives'
+import type { MenuEntry } from '@deepseek-ai/dsh-client-ui-primitives'
 import css from './PermissionSelect.module.css'
 
 /**
  * Display transform: kebab-case machine names render as title-case labels
- * (`workspace-write` → `Workspace Write`). Presentation-only — the wire
- * vocabulary and the host's advertised names are untouched; a host-configured
- * name that is not kebab-case (contains spaces or uppercase) passes through.
+ * (`workspace-write` → `Workspace Write`); non-kebab host-configured names
+ * pass through. Twin of the /permission popup's (client ui-permission) — the
+ * two permission surfaces must show the same text.
  */
 function displayName(name: string): string {
   if (!/^[a-z0-9]+(-[a-z0-9]+)*$/.test(name)) return name
@@ -29,52 +16,57 @@ function displayName(name: string): string {
 }
 
 export interface PermissionSelectProps {
-  /** The host-computed select, or undefined while the capability is absent. */
   value: PermissionSelectValue | undefined
-  /** Session-removed lock (the bar's chrome disable state). */
   locked: boolean
-  /** Submit one slash-command line; resolves admission (false = rejected/unmatched). */
   command: (line: string) => Promise<boolean>
 }
 
 export function PermissionSelect({ value, locked, command }: PermissionSelectProps) {
-  // Optimistic pick, shown while the admission round-trip runs; null follows
-  // the projection (the pushed frame lands the confirmed value there).
   const [pick, setPick] = useState<string | null>(null)
+  const [open, setOpen] = useState(false)
+
   if (value === undefined) return null
 
   const currentValue = pick ?? value.currentValue
   const current = value.options.find(option => option.value === currentValue)
+  const busy = pick !== null
 
-  const onChange = (next: string): void => {
-    if (next === value.currentValue) return
-    setPick(next)
-    void command(`/permission ${next}`)
+  const items: MenuEntry[] = value.options
+    .filter(o => o.value !== 'custom')
+    .map(option => ({ id: option.value, label: displayName(option.name) }))
+
+  const choose = (id: string): void => {
+    setOpen(false)
+    if (id === value.currentValue) return
+    setPick(id)
+    void command(`/permission ${id}`)
       .catch(() => false)
       .then(() => { setPick(null) })
   }
 
   return (
-    <label className={css.root} title={current?.description}>
-      <span className={css.chip}>
-        {displayName(current?.name ?? currentValue)}
-        <svg className={css.chevron} viewBox="0 0 12 12" width="12" height="12" aria-hidden>
-          <path d="M3 4.5L6 7.5L9 4.5" stroke="currentColor" strokeWidth="1.5" strokeLinecap="round" strokeLinejoin="round" fill="none" />
-        </svg>
-      </span>
-      <select
-        className={css.select}
-        aria-label="Access mode"
-        value={currentValue}
-        disabled={locked || pick !== null}
-        onChange={(e) => { onChange(e.target.value) }}
-      >
-        {value.options.map(option => (
-          <option key={option.value} value={option.value} disabled={option.value === 'custom'}>
-            {displayName(option.name)}
-          </option>
-        ))}
-      </select>
-    </label>
+    <Menu
+      open={open}
+      items={items}
+      selectedId={currentValue}
+      onSelect={choose}
+      onClose={() => { setOpen(false) }}
+      side="top"
+      anchor={
+        <button
+          type="button"
+          className={css.trigger}
+          aria-label={`Access mode, current: ${displayName(current?.name ?? currentValue)}`}
+          title={current?.description}
+          disabled={locked || busy}
+          onClick={() => { setOpen(!open) }}
+        >
+          <span className={css.triggerLabel}>{displayName(current?.name ?? currentValue)}</span>
+          <svg className={css.chevron} viewBox="0 0 12 12" width="12" height="12" aria-hidden>
+            <path d="M3 4.5L6 7.5L9 4.5" stroke="currentColor" strokeWidth="1.5" strokeLinecap="round" strokeLinejoin="round" fill="none" />
+          </svg>
+        </button>
+      }
+    />
   )
 }
diff --git a/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css b/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css
index 2e5d601e62..e9ee5286dc 100644
--- a/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css
+++ b/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css
@@ -1,4 +1,18 @@
-/* Bash toolview: same geometry/tokens as ToolRow (figma Bash · description). */
+/* Bash toolview: same geometry/tokens as ToolRow (figma Bash · description),
+   plus the terminal card the row stacks under its summary line. */
+
+/* Summary line over the terminal card; the summary row keeps its own 24px
+   height, so the card is a column around it rather than a change to it. */
+.card {
+  display: flex;
+  flex-direction: column;
+}
+
+/* Row indentation matches ToolRow's expanded bodies (16px leading + 6px gap),
+   and replaces the primitive's standalone vertical margin with the flow's. */
+.terminal {
+  margin: 4px 0 4px 22px;
+}
 
 .root {
   position: relative; /* sweep-glare overlay anchor */
diff --git a/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx b/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx
index d972a35274..c9385cab04 100644
--- a/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx
+++ b/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx
@@ -3,10 +3,20 @@
 // Product chrome matches ToolRow / Think (figma: Bash · {description}).
 // Child sessions keep a scoped badge so session-dimension differentiation stays
 // observable inside the component (no parallel registry).
+//
+// A bash call declares the terminal render intent, so this row also renders
+// the command's own output through TerminalBlock. This row has no expand
+// control and is not a details-panel target either (tool rows stopped being
+// one), so its terminal body is resident rather than expand-gated as in
+// ToolRow, and the card's own copy and expand controls are the row's only
+// interactions. CHAT_TERMINAL_MAX_LINES is passed as `maxLines` — the chat
+// flow's tighter cap over the block's own default of 16 — and the block's
+// internal expander keeps a long output from taking over the message flow.
 
 import type { Context } from 'cordis'
-import { IconApiOutline14, StateDot } from '@deepseek-ai/dsh-client-ui-primitives'
+import { IconApiOutline14, StateDot, TerminalBlock } from '@deepseek-ai/dsh-client-ui-primitives'
 import type { ToolRowProps } from '../contract/slots.ts'
+import { CHAT_TERMINAL_MAX_LINES, terminalCardModel } from '../contract/terminal-card-model.ts'
 import { toolRowModel, type ToolRowState } from '../contract/tool-call-model.ts'
 import css from './bash-sample.module.css'
 
@@ -29,24 +39,40 @@ function stateStatus(state: ToolRowState): string | null {
   }
 }
 
-/** Bash row: icon + Bash · {description}, matching the shared ToolRow chrome. */
+/**
+ * Bash row: icon + Bash · {description} in the shared ToolRow chrome, with the
+ * command's terminal card resident below it. The summary row is not a
+ * details-panel control (tool rows stopped being one), so the card's copy and
+ * expand controls are the row's only interactions.
+ */
 export function BashRow({ toolName, block, sessionId, useSessions }: ToolRowProps) {
   const model = toolRowModel(toolName, block)
+  // Session workspace root: the terminal view's cwd resolves against it (an
+  // omitted workdir IS the workspace), which the pure presenter cannot do.
+  const cwd = useSessions(list => list.byId[sessionId]?.cwd)
+  const terminal = terminalCardModel(block, cwd)
   const isChild = useSessions(list => list.byId[sessionId]?.parentId !== undefined)
   const status = stateStatus(model.state)
   return (
-    <div
-      className={css.root}
-      data-sample={isChild ? 'bash-scoped' : 'bash-global'}
-      data-variant="bash"
-      data-state={model.state}
-    >
-      <span className={css.leading}>{leadingFor(model.state)}</span>
-      {status !== null && <span className={css.visuallyHidden}>{status}</span>}
-      {isChild && <span className={css.scopeBadge}>scoped</span>}
-      <span className={css.title}>{model.title}</span>
-      <span className={css.sep} aria-hidden />
-      <span className={css.summary}>{model.summary}</span>
+    <div className={css.card}>
+      <div
+        className={css.root}
+        data-sample={isChild ? 'bash-scoped' : 'bash-global'}
+        data-variant="bash"
+        data-state={model.state}
+      >
+        <span className={css.leading}>{leadingFor(model.state)}</span>
+        {status !== null && <span className={css.visuallyHidden}>{status}</span>}
+        {isChild && <span className={css.scopeBadge}>scoped</span>}
+        <span className={css.title}>{model.title}</span>
+        <span className={css.sep} aria-hidden />
+        {/* The terminal presenter's description is the contractual
+            above-card summary; it outranks the args-derived one. */}
+        <span className={css.summary}>{terminal?.description ?? model.summary}</span>
+      </div>
+      {terminal !== null && (
+        <TerminalBlock {...terminal.card} maxLines={CHAT_TERMINAL_MAX_LINES} className={css.terminal} />
+      )}
     </div>
   )
 }
diff --git a/packages/client/ui-conversation/tests/apply-inject.spec.tsx b/packages/client/ui-conversation/tests/apply-inject.spec.tsx
index 416f3fa4ef..e73f959341 100644
--- a/packages/client/ui-conversation/tests/apply-inject.spec.tsx
+++ b/packages/client/ui-conversation/tests/apply-inject.spec.tsx
@@ -17,6 +17,7 @@
 import { describe, expect, it, vi } from 'vitest'
 import { SlotTestRuntime } from '@deepseek-ai/dsh-client-test-runtime'
 import type { SessionBehaviorOverrides } from '@deepseek-ai/dsh-client-test-runtime'
+import { LocaleService } from '@deepseek-ai/dsh-client-locale/client'
 import type { ISession, SessionId } from '@deepseek-ai/dsh-client-runtime/client'
 import { apply, inject } from '@deepseek-ai/dsh-client-ui-conversation/client'
 import type {
@@ -49,6 +50,7 @@ async function bench() {
   })
   const layoutFake = { openDetails: vi.fn(), closeDetails: vi.fn() }
   runtime.provide('layout', layoutFake)
+  runtime.provide('locale', new LocaleService(runtime.ctx))
 
   // The AppFrame role: the conversation-package slots must be declared by a
   // live entry before apply can contribute into them.
diff --git a/packages/client/ui-conversation/tests/chat-apply.spec.tsx b/packages/client/ui-conversation/tests/chat-apply.spec.tsx
index 4d6dc99f4a..d7b9125b34 100644
--- a/packages/client/ui-conversation/tests/chat-apply.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-apply.spec.tsx
@@ -10,6 +10,7 @@
 
 import { describe, expect, it, vi } from 'vitest'
 import { SlotTestRuntime } from '@deepseek-ai/dsh-client-test-runtime'
+import { LocaleService } from '@deepseek-ai/dsh-client-locale/client'
 import type { SessionId } from '@deepseek-ai/dsh-client-runtime/client'
 import { apply, inject } from '@deepseek-ai/dsh-client-ui-conversation/client'
 
@@ -22,6 +23,7 @@ async function bench() {
   await runtime.sessions.add(
     { id: CHILD, summary: { title: 'C', displayTitle: 'C', parentId: ROOT } }, { current: false })
   runtime.provide('layout', { openDetails: vi.fn(), closeDetails: vi.fn() })
+  runtime.provide('locale', new LocaleService(runtime.ctx))
 
   // Declared by ui-layout's root entry in production; the test root declares
   // them here so the contributions land.
@@ -84,6 +86,8 @@ describe('apply wiring', () => {
     // service being present implies the chat entry declared the hole first.
     const entries = b.slots.entries('conversation.chat.toolview')
     expect(entries.map(e => e.options.key)).toEqual(['bash', 'todo_write'])
+    // Stats stick with the composer (not inside ChatView).
+    expect(b.slots.entries('conversation.composer.dock').map(e => e.options.id)).toEqual(['stats'])
     await b.runtime.dispose()
   })
 
diff --git a/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx b/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx
index 9fb33d1d24..9bb6ba539a 100644
--- a/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx
@@ -5,9 +5,12 @@
 // with the keyed-slot machinery specs since the tool ring dissolved into
 // renderSlot.)
 
-import { afterEach, describe, expect, it, vi } from 'vitest'
-import { cleanup, fireEvent, render, screen } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
 import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react'
+import {
+  formatMessageClock, msUntilNextLocalMidnight, startOfLocalDay,
+} from '../src/client/chat/message-chrome.ts'
 import { MessageItem } from '../src/client/chat/MessageItem.tsx'
 import { AssistantMarkdown } from '../src/client/chat/AssistantMarkdown.tsx'
 import { StatsLine, type StatsLineProps } from '../src/client/chat/StatsLine.tsx'
@@ -15,19 +18,24 @@ import { StatsLine, type StatsLineProps } from '../src/client/chat/StatsLine.tsx
 afterEach(cleanup)
 
 describe('MessageItem arms', () => {
-  it('user bubbles expose copy / branch / edit actions; copy writes the text', () => {
+  it('user bubbles expose clock / copy / branch / edit; copy writes the text', () => {
     const writeText = vi.fn().mockResolvedValue(undefined)
     Object.defineProperty(navigator, 'clipboard', {
       configurable: true,
       value: { writeText },
     })
+    // Same-day clock: construct "today at 14:24" so the label stays `HH:mm`.
+    const now = new Date()
+    const time = new Date(now.getFullYear(), now.getMonth(), now.getDate(), 14, 24).getTime()
     render(
       <MessageItem node={{
-        kind: 'user', seq: 1,
+        kind: 'user', seq: 1, time,
         content: [{ type: 'text', text: 'hello bubble' }] as never,
-      } as never}
+        source: null,
+      }}
       />,
     )
+    expect(screen.getByText('14:24')).toBeTruthy()
     expect(screen.getByRole('button', { name: '复制' })).toBeTruthy()
     expect(screen.getByRole('button', { name: '在新对话中分支' })).toBeTruthy()
     expect(screen.getByRole('button', { name: '编辑' })).toBeTruthy()
@@ -47,9 +55,10 @@ describe('MessageItem arms', () => {
     })
     render(
       <MessageItem node={{
-        kind: 'user', seq: 1,
+        kind: 'user', seq: 1, time: 1_000,
         content: [{ type: 'text', text: 'fallback body' }] as never,
-      } as never}
+        source: null,
+      }}
       />,
     )
     fireEvent.click(screen.getByRole('button', { name: '复制' }))
@@ -69,9 +78,10 @@ describe('MessageItem arms', () => {
     })
     render(
       <MessageItem node={{
-        kind: 'user', seq: 1,
+        kind: 'user', seq: 1, time: 1_000,
         content: [{ type: 'text', text: 'quiet' }] as never,
-      } as never}
+        source: null,
+      }}
       />,
     )
     fireEvent.click(screen.getByRole('button', { name: '复制' }))
@@ -109,6 +119,56 @@ describe('MessageItem arms', () => {
   })
 })
 
+describe('formatMessageClock', () => {
+  const now = new Date(2026, 6, 29, 10, 0).getTime()
+
+  it('keeps HH:mm on the same calendar day', () => {
+    expect(formatMessageClock(new Date(2026, 6, 29, 14, 24).getTime(), now)).toBe('14:24')
+  })
+
+  it('prefixes month and day across days in the same year', () => {
+    expect(formatMessageClock(new Date(2026, 0, 1, 14, 24).getTime(), now)).toBe('1月1日 14:24')
+  })
+
+  it('prefixes year, month, and day across years', () => {
+    expect(formatMessageClock(new Date(2025, 11, 31, 9, 5).getTime(), now)).toBe('2025年12月31日 09:05')
+  })
+
+  it('arms the next local midnight from an in-day instant', () => {
+    const noon = new Date(2026, 6, 29, 12, 0).getTime()
+    expect(startOfLocalDay(noon)).toBe(new Date(2026, 6, 29).getTime())
+    expect(msUntilNextLocalMidnight(noon)).toBe(12 * 3_600_000)
+  })
+})
+
+describe('useCalendarDay boundary refresh', () => {
+  beforeEach(() => {
+    vi.useFakeTimers()
+  })
+  afterEach(() => {
+    vi.useRealTimers()
+  })
+
+  it('widens a same-day user clock after local midnight', () => {
+    const dayStart = new Date(2026, 6, 29, 23, 50).getTime()
+    vi.setSystemTime(dayStart)
+    const time = new Date(2026, 6, 29, 14, 24).getTime()
+    render(
+      <MessageItem node={{
+        kind: 'user', seq: 1, time,
+        content: [{ type: 'text', text: 'night bubble' }] as never,
+        source: null,
+      }}
+      />,
+    )
+    expect(screen.getByText('14:24')).toBeTruthy()
+    act(() => {
+      vi.advanceTimersByTime(msUntilNextLocalMidnight(dayStart) + 1)
+    })
+    expect(screen.getByText('7月29日 14:24')).toBeTruthy()
+  })
+})
+
 describe('small branch tails', () => {
   it('AssistantMarkdown single-line reasoning summary skips the newline cut', () => {
     const view = render(
@@ -117,6 +177,35 @@ describe('small branch tails', () => {
     expect(view.getByText('one-liner')).toBeTruthy()
   })
 
+  it('finalized assistant messages expose copy / branch / clock after the body; streaming omits them', () => {
+    const writeText = vi.fn().mockResolvedValue(undefined)
+    Object.defineProperty(navigator, 'clipboard', {
+      configurable: true,
+      value: { writeText },
+    })
+    const now = new Date()
+    const time = new Date(now.getFullYear(), now.getMonth(), now.getDate(), 14, 24).getTime()
+    const settled = render(
+      <AssistantMarkdown
+        blocks={[{ kind: 'text', text: 'answer body' }, { kind: 'reasoning', text: 'hidden' }]}
+        streaming={false}
+        time={time}
+      />,
+    )
+    expect(settled.getByText('14:24')).toBeTruthy()
+    expect(settled.getByRole('button', { name: '复制' })).toBeTruthy()
+    expect(settled.getByRole('button', { name: '在新对话中分支' })).toBeTruthy()
+    fireEvent.click(settled.getByRole('button', { name: '复制' }))
+    expect(writeText).toHaveBeenCalledWith('answer body')
+    settled.unmount()
+
+    const streaming = render(
+      <AssistantMarkdown blocks={[{ kind: 'text', text: 'partial' }]} streaming time={time} />,
+    )
+    expect(streaming.queryByRole('button', { name: '复制' })).toBeNull()
+    expect(streaming.queryByText('14:24')).toBeNull()
+  })
+
   it('StatsLine omits the cache-hit segment when no input accounting exists at all', () => {
     // cacheHitPct is null only when input+cacheRead are both zero (pure
     // output accounting) — any input makes it a real 0%.
diff --git a/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx b/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx
index 1b4d1ee158..6b75f940d4 100644
--- a/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx
@@ -17,6 +17,7 @@ import type {
   ToolResultNode, WorkspaceListState,
 } from '@deepseek-ai/dsh-client-runtime/client'
 import { createSlotRenderer } from '@deepseek-ai/dsh-client-web-react'
+import { LocaleService } from '@deepseek-ai/dsh-client-locale/client'
 import type { PropsRenderSlots } from '@deepseek-ai/dsh-client-ui-slots'
 import { apply, inject } from '@deepseek-ai/dsh-client-ui-conversation/client'
 
@@ -125,7 +126,7 @@ async function bench(snapshot: ConversationSnapshot) {
   }
   ctx.provide('workspaces', workspaces)
   ctx.provide('layout', layout)
-  ctx.provide('i18n', { bind: () => (key: string) => key })
+  ctx.provide('locale', new LocaleService(ctx))
 
   slots.install(createSlotRenderer())
   slots.register({
diff --git a/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx b/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx
index 6985991074..38edc48cca 100644
--- a/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx
@@ -1,5 +1,5 @@
 // @vitest-environment jsdom
-// StatsLine (rendered inside the chat view body): totals derivation + the RFC
+// StatsLine (composer.dock entry): totals derivation + the RFC
 // hard acceptance — zero renders during streaming. Bash sample row: the
 // canonical sub-agent differential decided INSIDE the component off the
 // standard useSessions kit (no registry predicates — tool ring dissolved).
diff --git a/packages/client/ui-conversation/tests/chat-tool-row.spec.tsx b/packages/client/ui-conversation/tests/chat-tool-row.spec.tsx
index 4b17c616cf..546b4a18e8 100644
--- a/packages/client/ui-conversation/tests/chat-tool-row.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-tool-row.spec.tsx
@@ -97,6 +97,11 @@ describe('tool-call-model', () => {
     expect(toolRowModel('bash', result({ call: null })).body).toBeNull()
   })
 
+  it('a code row with an empty program falls back to the args JSON envelope', () => {
+    expect(toolRowModel('run_code', running({ name: 'run_code', argsRaw: '{"code":""}' })).body)
+      .toBe('{\n  "code": ""\n}')
+  })
+
   it('gives Cordis lifecycle tools action titles over their generic variants', () => {
     expect(toolRowModel('cordis_inspect', running({
       name: 'cordis_inspect',
@@ -165,6 +170,22 @@ describe('ToolRow', () => {
     expect(view.queryByTestId('tool-icon')).not.toBeNull()
   })
 
+  it('an expandOnRowClick row toggles from Enter and Space, ignoring other keys', () => {
+    const view = render(<ToolRow {...rowProps} expandOnRowClick />)
+    const row = view.getByRole('button')
+    fireEvent.keyDown(row, { key: 'Tab' })
+    expect(row.getAttribute('aria-expanded')).toBe('false')
+    fireEvent.keyDown(row, { key: 'Enter' })
+    expect(row.getAttribute('aria-expanded')).toBe('true')
+    fireEvent.keyDown(row, { key: ' ' })
+    expect(row.getAttribute('aria-expanded')).toBe('false')
+  })
+
+  it('a non-expandable expandOnRowClick row exposes no row button', () => {
+    const view = render(<ToolRow {...rowProps} body={null} expandOnRowClick />)
+    expect(view.queryByRole('button')).toBeNull()
+  })
+
   it('file-path summary opens through onOpenFile; the leading slot is not an expand control', () => {
     const open = vi.fn()
     const view = render(
diff --git a/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx b/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx
index 3d2e9ea0e6..02bb6b92dc 100644
--- a/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx
@@ -15,6 +15,7 @@ import { cleanup, fireEvent } from '@testing-library/react'
 import type { ISession, SessionId, ToolResultNode } from '@deepseek-ai/dsh-client-runtime/client'
 import type { PropsRenderSlots } from '@deepseek-ai/dsh-client-ui-slots'
 import { SlotTestRuntime } from '@deepseek-ai/dsh-client-test-runtime'
+import { LocaleService } from '@deepseek-ai/dsh-client-locale/client'
 import { apply, inject } from '@deepseek-ai/dsh-client-ui-conversation/client'
 import type { ToolRowProps } from '@deepseek-ai/dsh-client-ui-conversation/client'
 
@@ -53,6 +54,7 @@ async function bench(nodes: ToolResultNode[]) {
   const runtime = await SlotTestRuntime.create()
   const layout = { openDetails: vi.fn(), closeDetails: vi.fn() }
   runtime.provide('layout', layout)
+  runtime.provide('locale', new LocaleService(runtime.ctx))
   await runtime.sessions.add({
     id: SID,
     summary: { title: 'S', displayTitle: 'S' },
@@ -180,6 +182,7 @@ describe('registrant load-order seam', () => {
   it("suspends a registrant on inject: ['slots', 'conversation'] until the service (and the hole) exists", async () => {
     const runtime = await SlotTestRuntime.create()
     runtime.provide('layout', { openDetails: vi.fn(), closeDetails: vi.fn() })
+    runtime.provide('locale', new LocaleService(runtime.ctx))
     await runtime.root.declare(LAYOUT_CHILDREN, AppRoot)
 
     // Third-party posture, mounted BEFORE ui-conversation: real fiber inject
diff --git a/packages/client/ui-conversation/tests/chat-view.spec.tsx b/packages/client/ui-conversation/tests/chat-view.spec.tsx
index 328ba38340..c0cb4dcb78 100644
--- a/packages/client/ui-conversation/tests/chat-view.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-view.spec.tsx
@@ -371,6 +371,42 @@ describe('ChatView', () => {
     expect(view.queryByLabelText('回到底部')).toBeNull()
   })
 
+  it('entering the at-bottom threshold does not snap the remaining scroll distance', () => {
+    const h = makeHarness({ nodes: [user(1, 'q'), assistant(2, 'a')] })
+    const view = render(<h.ChatView {...h.props} />)
+    const scroller = view.container.querySelector('[class*="scroll"]') as HTMLDivElement
+    Object.defineProperty(scroller, 'scrollHeight', { value: 1000, writable: true })
+    Object.defineProperty(scroller, 'clientHeight', { value: 300, writable: true })
+    // Inside FOLLOW_THRESHOLD (24) but not flush with the floor — the chrome
+    // re-render from setAtBottom must not force scrollTop to scrollHeight.
+    scroller.scrollTop = 690 // distance-to-bottom = 10
+    fireEvent.scroll(scroller)
+    expect(view.queryByLabelText('回到底部')).toBeNull()
+    expect(scroller.scrollTop).toBe(690)
+  })
+
+  it('under data-conversation-scroll, bottom-follow targets the host scrollport', () => {
+    const host = document.createElement('div')
+    host.setAttribute('data-conversation-scroll', '')
+    Object.defineProperty(host, 'scrollHeight', { value: 2000, writable: true, configurable: true })
+    Object.defineProperty(host, 'clientHeight', { value: 500, writable: true, configurable: true })
+    Object.defineProperty(host, 'scrollTop', { value: 0, writable: true, configurable: true })
+    document.body.appendChild(host)
+    try {
+      const h = makeHarness({ nodes: [user(1, 'q'), assistant(2, 'a')] })
+      const view = render(<h.ChatView {...h.props} />, { container: host })
+      // Open jump uses the host, not the local .scroll node.
+      expect(host.scrollTop).toBe(2000)
+      host.scrollTop = 100
+      fireEvent.scroll(host)
+      expect(view.getByLabelText('回到底部')).toBeTruthy()
+      fireEvent.click(view.getByLabelText('回到底部'))
+      expect(host.scrollTop).toBe(2000)
+    } finally {
+      host.remove()
+    }
+  })
+
   it('paging button loads older and shows its busy label', () => {
     const h = makeHarness({ nodes: [user(5, 'later')], hasMore: true })
     const view = render(<h.ChatView {...h.props} />)
diff --git a/packages/client/ui-conversation/tests/input-bar.spec.tsx b/packages/client/ui-conversation/tests/input-bar.spec.tsx
index 23d853dd1a..c0bfda7f4a 100644
--- a/packages/client/ui-conversation/tests/input-bar.spec.tsx
+++ b/packages/client/ui-conversation/tests/input-bar.spec.tsx
@@ -42,6 +42,7 @@ interface BenchOptions {
   promptError?: ConversationSnapshot['promptError']
   variant?: 'hero' | 'composer'
   placeholder?: string
+  translateHint?: (key: string) => string
   accessory?: React.ReactNode
   overlay?: React.ReactNode
   leftItems?: React.ReactNode
@@ -100,6 +101,11 @@ function bench(over?: BenchOptions) {
     useLexicon: bindSnapshotSelector(shell.lexicon),
     stop,
     command: () => Promise.resolve(true),
+    // Mirrors the en 'command.hint' locale entries the production apply wires in.
+    translateHint: over?.translateHint ?? ((key: string) => ({
+      'placeholder.default': 'Message the agent',
+      'placeholder.plan': 'describe your task to generate plan',
+    } as Record<string, string>)[key] ?? key),
     renderSlot,
     variant: over?.variant ?? 'composer',
     ...(over?.placeholder !== undefined ? { placeholder: over.placeholder } : {}),
@@ -227,6 +233,56 @@ describe('running and lock semantics (queue cut 1)', () => {
     expect((textarea).value).toBe('typed')
   })
 
+  it('wheel over a non-overflowing textarea forwards to the conversation host', () => {
+    const host = document.createElement('div')
+    host.setAttribute('data-conversation-scroll', '')
+    Object.defineProperty(host, 'scrollTop', { value: 40, writable: true, configurable: true })
+    const { view, textarea } = bench()
+    host.appendChild(view.container)
+    document.body.appendChild(host)
+    try {
+      const wheeled = fireEvent.wheel(textarea, { deltaY: 30 })
+      expect(wheeled).toBe(false) // preventDefault
+      expect(host.scrollTop).toBe(70)
+    } finally {
+      host.remove()
+    }
+  })
+
+  it('wheel chains: long drafts scroll inside the textarea until each edge, then the host', () => {
+    const host = document.createElement('div')
+    host.setAttribute('data-conversation-scroll', '')
+    Object.defineProperty(host, 'scrollTop', { value: 40, writable: true, configurable: true })
+    const { view, textarea } = bench()
+    host.appendChild(view.container)
+    document.body.appendChild(host)
+    Object.defineProperty(textarea, 'clientHeight', { value: 100, configurable: true })
+    Object.defineProperty(textarea, 'scrollHeight', { value: 400, configurable: true })
+    let scrollTop = 150
+    Object.defineProperty(textarea, 'scrollTop', {
+      configurable: true,
+      get: () => scrollTop,
+      set: (value: number) => { scrollTop = value },
+    })
+    try {
+      // Mid-draft: both directions stay local — host must not move.
+      expect(fireEvent.wheel(textarea, { deltaY: 30 })).toBe(true)
+      expect(fireEvent.wheel(textarea, { deltaY: -30 })).toBe(true)
+      expect(host.scrollTop).toBe(40)
+      // At the bottom edge, further down-scroll forwards to the host.
+      scrollTop = 300
+      expect(fireEvent.wheel(textarea, { deltaY: 30 })).toBe(false)
+      expect(host.scrollTop).toBe(70)
+      // At the top edge, further up-scroll forwards to the host.
+      scrollTop = 0
+      host.scrollTop = 70
+      expect(fireEvent.wheel(textarea, { deltaY: -20 })).toBe(false)
+      expect(host.scrollTop).toBe(50)
+    } finally {
+      host.remove()
+    }
+  })
+
   it('disabled state shows the unavailable placeholder; custom placeholder wins', () => {
     const { textarea } = bench({ disabled: true })
     expect(textarea.placeholder).toBe('Session unavailable')
@@ -292,6 +348,19 @@ describe('decorations', () => {
     expect(view.container.querySelector('[data-decoration="token"]')).not.toBeNull()
   })
 
+  it('a locale entry for the claimed command overrides the raw claim hint (trailing-space token)', () => {
+    const dict: Record<string, string> = { goal: '输入目标，智能体将持续执行' }
+    const { view, shell } = bench({ translateHint: key => dict[key] ?? key })
+    act(() => {
+      shell.setDraft('/goal ')
+      shell.beginCommand(
+        { token: '/goal ', hint: '[<objective>|clear|edit <objective>|pause|resume]', submit: () => Promise.resolve({ kind: 'success' as const }) },
+        { start: 0, end: 6, draftRev: shell.snapshot.draftRev },
+      )
+    })
+    expect(view.container.querySelector('[data-decoration="hint"]')?.textContent).toBe('输入目标，智能体将持续执行')
+  })
+
   it('an inserted reference renders as a chip at its placeholder offset', () => {
     const { view, shell } = bench()
     act(() => {
@@ -374,7 +443,7 @@ describe('placeholder chrome and control seats', () => {
     const { view, slotCalls } = bench()
     expect(view.getByLabelText('Add attachment')).toBeTruthy()
     // Capability absent (no projection value): the chip renders nothing.
-    expect(view.queryByLabelText('Access mode')).toBeNull()
+    expect(view.queryByLabelText(/^Access mode/)).toBeNull()
     // Both seats dispatched, nothing rendered.
     expect(slotCalls.map(c => c.key)).toEqual(['conversation.input.plan', 'conversation.input.model'])
     expect(view.queryByLabelText('Plan mode')).toBeNull()
@@ -390,15 +459,19 @@ describe('placeholder chrome and control seats', () => {
       currentValue: 'workspace-write',
     }
     const { view } = bench({ permissions })
-    const select = view.getByLabelText('Access mode') as HTMLSelectElement
-    expect(select.value).toBe('workspace-write')
-    // Title-case display is presentation only; the option values stay machine names.
-    expect([...select.options].map(o => o.textContent)).toEqual(['Workspace Write', 'Danger Full Access'])
-    fireEvent.change(select, { target: { value: 'danger-full-access' } })
+    const trigger = view.getByLabelText(/^Access mode/) as HTMLButtonElement
+    // Title-case display is presentation only; the menu ids stay machine names.
+    expect(trigger.textContent).toBe('Workspace Write')
+    fireEvent.click(trigger)
+    const items = view.getAllByRole('menuitem')
+    expect(items.map(o => o.textContent)).toEqual(['Workspace Write', 'Danger Full Access'])
+    fireEvent.click(items[1]!)
     // Optimistic pick + disable until admission resolves (command stub resolves true).
-    expect(select.disabled).toBe(true)
+    const busy = view.getByLabelText(/^Access mode/) as HTMLButtonElement
+    expect(busy.textContent).toBe('Danger Full Access')
+    expect(busy.disabled).toBe(true)
     await act(async () => {})
-    expect(select.disabled).toBe(false)
+    expect((view.getByLabelText(/^Access mode/) as HTMLButtonElement).disabled).toBe(false)
   })
 
   it('a registered entry fills its seat and receives the locked owner prop', () => {
@@ -420,9 +493,9 @@ describe('placeholder chrome and control seats', () => {
     const permissions = { options: [{ value: 'workspace-write', name: 'workspace-write' }], currentValue: 'workspace-write' }
     const { view } = bench({ disabled: true, permissions })
     expect((view.getByLabelText('Add attachment') as HTMLButtonElement).disabled).toBe(true)
-    expect((view.getByLabelText('Access mode') as HTMLSelectElement).disabled).toBe(true)
+    expect((view.getByLabelText(/^Access mode/) as HTMLButtonElement).disabled).toBe(true)
     cleanup()
     const live = bench({ running: true, permissions })
-    expect((live.view.getByLabelText('Access mode') as HTMLSelectElement).disabled).toBe(false)
+    expect((live.view.getByLabelText(/^Access mode/) as HTMLButtonElement).disabled).toBe(false)
   })
 })
diff --git a/packages/client/ui-conversation/tests/input-machine.spec.ts b/packages/client/ui-conversation/tests/input-machine.spec.ts
index 206a66e4c6..9ce23c4a10 100644
--- a/packages/client/ui-conversation/tests/input-machine.spec.ts
+++ b/packages/client/ui-conversation/tests/input-machine.spec.ts
@@ -260,10 +260,10 @@ describe('input-machine: insert-ref and the occurrence table', () => {
     m.dispatch({ type: 'insert-ref', reference: refOf('alpha'), span: spanOf(m, 0, 4) })
     m.dispatch({ type: 'draft-changed', draft: `${P} and /alp`, editRange: { start: 1, end: 1, insertedLength: 9 } })
     m.dispatch({ type: 'insert-ref', reference: refOf('alpha'), span: spanOf(m, 6, 10) })
-    expect(m.state.draft).toBe(`${P} and ${P}`)
+    expect(m.state.draft).toBe(`${P} and ${P} `)
     expect(m.state.occurrences.map(o => o.occurrenceId)).toEqual([1, 2])
     // Delete the first chip whole; the second survives with its own identity.
-    m.dispatch({ type: 'draft-changed', draft: ` and ${P}`, editRange: { start: 0, end: 1, insertedLength: 0 } })
+    m.dispatch({ type: 'draft-changed', draft: ` and ${P} `, editRange: { start: 0, end: 1, insertedLength: 0 } })
     expect(m.state.occurrences).toEqual([expect.objectContaining({ occurrenceId: 2, offset: 5 })])
   })
 
@@ -273,7 +273,7 @@ describe('input-machine: insert-ref and the occurrence table', () => {
     m.dispatch({ type: 'begin-command', claim: claimOf('goal'), span: spanOf(m, 0, 3) })
     m.dispatch({ type: 'draft-changed', draft: '/goal ask @wor' })
     m.dispatch({ type: 'insert-ref', reference: refOf('worker-1', 'subagent'), span: spanOf(m, 10, 14) })
-    expect(m.state.draft).toBe(`/goal ask ${P}`)
+    expect(m.state.draft).toBe(`/goal ask ${P} `)
     expect(m.state.phase).toBe('claimed')
     expect(m.state.occurrences).toHaveLength(1)
   })
@@ -350,10 +350,10 @@ describe('input-machine: newline transaction (F1)', () => {
     m.dispatch({ type: 'draft-changed', draft: 'ab @wor' })
     m.dispatch({ type: 'insert-ref', reference: refOf('w'), span: spanOf(m, 3, 7) })
     m.dispatch({ type: 'newline', selection: { start: 2, end: 2 } })
-    expect(m.state.draft).toBe(`ab\n ${P}`)
+    expect(m.state.draft).toBe(`ab\n ${P} `)
     expect(m.state.occurrences[0]?.offset).toBe(4)
     m.dispatch({ type: 'undo' })
-    expect(m.state.draft).toBe(`ab ${P}`)
+    expect(m.state.draft).toBe(`ab ${P} `)
   })
 
   it('replaces a selection, breaks the claim prefix when leading, and rejects out-of-bounds', () => {
@@ -410,7 +410,7 @@ describe('input-machine: consume-token guards', () => {
     m.dispatch({ type: 'draft-changed', draft: '/model @wor' })
     m.dispatch({ type: 'insert-ref', reference: refOf('w'), span: spanOf(m, 7, 11) })
     m.dispatch({ type: 'consume-token', guard: { kind: 'span', span: spanOf(m, 0, 7) } })
-    expect(m.state.draft).toBe(P)
+    expect(m.state.draft).toBe(`${P} `)
     expect(m.state.occurrences[0]?.offset).toBe(0)
   })
 })
@@ -490,7 +490,7 @@ describe('input-machine: undo / redo', () => {
     m.dispatch({ type: 'draft-changed', draft: '', editRange: { start: 0, end: 1, insertedLength: 0 } })
     expect(m.state.occurrences).toEqual([])
     m.dispatch({ type: 'undo' })
-    expect(m.state.draft).toBe(P)
+    expect(m.state.draft).toBe(`${P} `)
     expect(m.state.occurrences).toHaveLength(1)
   })
 
@@ -555,9 +555,9 @@ describe('input-machine: paste plane', () => {
     m.dispatch({ type: 'paste-upgrade', attemptId: 1, span: spanOf(m, 0, 6), reference: refOf('alpha') })
     expect(m.state.paste?.insertedRange).toEqual({ start: 0, end: 7 })
     m.dispatch({ type: 'paste-upgrade', attemptId: 1, span: spanOf(m, 2, 7), reference: refOf('beta') })
-    expect(m.state.draft).toBe(`${P} ${P}`)
+    expect(m.state.draft).toBe(`${P} ${P} `)
     expect(m.state.occurrences.map(o => o.ref)).toEqual(['alpha', 'beta'])
-    expect(m.state.paste?.insertedRange).toEqual({ start: 0, end: 3 })
+    expect(m.state.paste?.insertedRange).toEqual({ start: 0, end: 4 })
   })
 
   it('a stale span CAS drops one upgrade without ending the attempt', () => {
@@ -634,8 +634,8 @@ describe('input-machine: projectClipboard', () => {
     m.dispatch({ type: 'insert-ref', reference: refOf('alpha'), span: spanOf(m, 4, 8) })
     m.dispatch({ type: 'draft-changed', draft: `use ${P} then /bet`, editRange: { start: 5, end: 5, insertedLength: 10 } })
     m.dispatch({ type: 'insert-ref', reference: refOf('beta'), span: spanOf(m, 11, 15) })
-    expect(m.state.draft).toBe(`use ${P} then ${P}`)
-    expect(projectClipboard(m.state)).toBe('use /alpha then /beta')
+    expect(m.state.draft).toBe(`use ${P} then ${P} `)
+    expect(projectClipboard(m.state)).toBe('use /alpha then /beta ')
   })
 
   it('is the identity on a chip-free draft', () => {
diff --git a/packages/client/ui-conversation/tests/input-matrix.spec.tsx b/packages/client/ui-conversation/tests/input-matrix.spec.tsx
index f6694f8cb4..a9c00b0748 100644
--- a/packages/client/ui-conversation/tests/input-matrix.spec.tsx
+++ b/packages/client/ui-conversation/tests/input-matrix.spec.tsx
@@ -48,6 +48,7 @@ function mountBar(shell: SessionInputShell, over?: { running?: boolean; disabled
     renderSlot: (() => null) as InputBarProps['renderSlot'],
     stop: vi.fn(),
     command: () => Promise.resolve(true),
+    translateHint: (key: string) => key,
     variant: 'composer',
   }
   return render(<InputBar {...props} />)
diff --git a/packages/client/ui-conversation/tests/input-scenarios.spec.tsx b/packages/client/ui-conversation/tests/input-scenarios.spec.tsx
index 9d9ace032c..1c7bbe50ec 100644
--- a/packages/client/ui-conversation/tests/input-scenarios.spec.tsx
+++ b/packages/client/ui-conversation/tests/input-scenarios.spec.tsx
@@ -134,6 +134,7 @@ async function scopedBench(register?: (slash: SlashService) => void) {
     renderSlot: (() => null) as InputBarProps['renderSlot'],
     stop: vi.fn(),
     command: () => Promise.resolve(true),
+    translateHint: (key: string) => key,
     variant: 'composer',
   }
   const view = render(<InputBar {...barProps} />)
diff --git a/packages/client/ui-conversation/tests/skeleton.spec.tsx b/packages/client/ui-conversation/tests/skeleton.spec.tsx
index 0d32e2edea..b174bace53 100644
--- a/packages/client/ui-conversation/tests/skeleton.spec.tsx
+++ b/packages/client/ui-conversation/tests/skeleton.spec.tsx
@@ -17,7 +17,9 @@ import { ConversationRoot } from '../src/client/skeleton/ConversationRoot.tsx'
 import { ConversationSession } from '../src/client/skeleton/ConversationSession.tsx'
 import { InputBar } from '../src/client/skeleton/InputBar.tsx'
 import type { InputBarProps } from '../src/client/skeleton/InputBar.tsx'
-import type { ComposerBarOwnerProps } from '../src/client/contract/slots.ts'
+import type {
+  ComposerBarOwnerProps,
+} from '../src/client/contract/slots.ts'
 
 /** Machine-backed wiring over a sink spy. */
 function fakeWiring() {
@@ -59,6 +61,8 @@ function mount(
   snapshot: ConversationSnapshot,
   workspaceRows: WorkspaceView[] = [{ ...workspace('one'), sessionIds: [SID] }],
   retargetWorkspace = vi.fn(async (_workspaceId: WorkspaceId) => {}),
+  /** When true, mimic overlay:true chain siblings (hidden fallback + takeover). */
+  overlayTakeover = false,
 ) {
   const root = sid('root')
   const sessions = createSnapshotStore<SessionListState>({
@@ -99,9 +103,17 @@ function mount(
           useStore={bindSnapshotSelector(chat)}
           actions={chat.actions}
           renderSlot={renderSlot as never}
-          views={{ list: () => [{ id: 'chat', label: 'Chat' }], subscribe: () => () => {}, version: () => 1 }}
+          views={{
+            list: () => [
+              { id: 'chat', label: 'Chat' },
+              { id: 'trajectory', label: 'Trajectory' },
+            ],
+            subscribe: () => () => {},
+            version: () => 1,
+          }}
           bindDraftMirror={write => wiring.bindMirror(write)}
           open={open}
+          {...owner}
         />
       )
     }
@@ -124,6 +136,7 @@ function mount(
           useLexicon={bindSnapshotSelector(wiring.lexicon)}
           stop={stop}
           command={() => Promise.resolve(true)}
+          translateHint={(key: string) => key}
           renderSlot={(() => null) as InputBarProps['renderSlot']}
           {...bar}
         />
@@ -131,7 +144,18 @@ function mount(
     }
     return <div data-testid={`view-${opts?.only ?? key}`} />
   }) as ConversationRootProps['renderSlot']
-  const renderSlotChain = ((_key, _owner, opts) => opts?.fallback ?? null) as ConversationRootProps['renderSlotChain']
+  const renderSlotChain = ((_key, _owner, opts) => (
+    overlayTakeover
+      ? (
+        <>
+          <div data-chain-overlay-fallback="conversation.composer" style={{ display: 'none' }}>
+            {opts?.fallback ?? null}
+          </div>
+          <div data-testid="composer-takeover">TAKEOVER</div>
+        </>
+      )
+      : (opts?.fallback ?? null)
+  )) as ConversationRootProps['renderSlotChain']
   const props: ConversationRootProps = {
     sessionId: SID,
     SessionProvider: ({ children }) => children(SID),
@@ -166,6 +190,30 @@ describe('ConversationRoot resident composer', () => {
     expect(b.open).toHaveBeenCalledWith(sid('root'))
   })
 
+  it('active phase: fixed header outside the scrollport; sticky composer seat inside it', () => {
+    const b = mount(conversationSnapshot())
+    const host = b.view.container.querySelector('[data-conversation-scroll]')
+    const seat = b.view.container.querySelector('[data-composer-seat]')
+    const header = b.view.container.querySelector('header')
+    const textarea = b.view.container.querySelector('textarea')
+    expect(host).not.toBeNull()
+    expect(seat).not.toBeNull()
+    expect(header).not.toBeNull()
+    // Header is column chrome above the scrollport; the seat sticks inside it.
+    expect(host?.contains(header)).toBe(false)
+    expect(host?.contains(seat)).toBe(true)
+    expect(seat?.contains(textarea)).toBe(true)
+  })
+
+  it('sticky composer seat wraps the whole overlay chain, not only the fallback stack', () => {
+    const b = mount(conversationSnapshot(), undefined, undefined, true)
+    const seat = b.view.container.querySelector('[data-composer-seat]')
+    const takeover = b.view.getByTestId('composer-takeover')
+    const fallback = b.view.container.querySelector('[data-chain-overlay-fallback="conversation.composer"]')
+    expect(seat?.contains(takeover)).toBe(true)
+    expect(seat?.contains(fallback)).toBe(true)
+  })
+
   it('hero phase: same textarea, hero chrome, no header, picker switches the workspace', () => {
     const b = mount(
       conversationSnapshot({ composerPhase: 'blank', blank: true }),
@@ -174,13 +222,19 @@ describe('ConversationRoot resident composer', () => {
         { ...workspace('second'), title: 'Selected Folder' },
       ],
     )
-    // Hero chrome present, view ring absent.
+    // Hero chrome present, view ring absent; scroll host already wraps the
+    // resident composer so the blank → active flip does not remount it.
+    const host = b.view.container.querySelector('[data-conversation-scroll]')
+    const header = b.view.container.querySelector('header')
+    expect(host).not.toBeNull()
+    expect(header?.getAttribute('aria-hidden')).toBe('true')
     expect(b.view.getByText("Let's start building")).toBeTruthy()
     expect(b.view.queryByTestId('view-chat')).toBeNull()
     // The same machine-backed textarea is live in the hero, and the
-    // persistence mirror stays bound (ConversationSession mounts chrome-less
+    // persistence mirror stays bound (ConversationSession mounts chrome-hidden
     // for blank sessions): hero typing reaches the chat store.
     const box = b.view.getByRole('textbox')
+    expect(host?.contains(box)).toBe(true)
     fireEvent.change(box, { target: { value: 'draft in hero' } })
     expect(b.chat.store.getSnapshot().draft).toBe('draft in hero')
     // Picker: open through the chip; a pick switches to the other
@@ -193,20 +247,31 @@ describe('ConversationRoot resident composer', () => {
     expect(b.view.getByText('Selected Folder')).toBeTruthy()
   })
 
-  it('textarea DOM identity survives the hero → active flip', () => {
+  it('same textarea DOM node survives the hero → active flip into the sticky scrollport', () => {
     const b = mount(conversationSnapshot({ composerPhase: 'blank', blank: true }))
     const before = b.view.getByRole('textbox')
     fireEvent.change(before, { target: { value: 'kept across flip' } })
-    // First message landed: content exists, phase leaves blank.
+    // First message landed: content exists, phase leaves blank. Composer
+    // already sat in the Session scrollport during hero, so the textarea
+    // node and InputHub draft both survive.
     b.session.set(conversationSnapshot({ composerPhase: 'active', blank: false }))
     b.rerender()
-    const after = b.view.getByRole('textbox')
+    const after = b.view.getByRole('textbox') as HTMLTextAreaElement
     expect(after).toBe(before)
-    expect((after as HTMLTextAreaElement).value).toBe('kept across flip')
+    expect(after.value).toBe('kept across flip')
+    expect(b.chat.store.getSnapshot().draft).toBe('kept across flip')
+    expect(b.view.container.querySelector('[data-conversation-scroll]')?.contains(after)).toBe(true)
     expect(b.view.queryByText("Let's start building")).toBeNull()
     expect(b.view.getByTestId('view-chat')).toBeTruthy()
   })
 
+  it('keeps pending takeover interaction accessible outside the Chat view', () => {
+    const b = mount(conversationSnapshot({ pending: [{} as never] }))
+    act(() => { b.chat.actions.setView('trajectory') })
+    expect(b.view.getByTestId('view-trajectory')).toBeTruthy()
+    expect(b.view.getByRole('textbox')).toBeTruthy()
+  })
+
   it('rolls the pending workspace label back when switching fails', async () => {
     const selectWorkspace = vi.fn(async () => { throw new Error('connect failed') })
     const b = mount(
diff --git a/packages/client/ui-conversation/tests/terminal-card.spec.tsx b/packages/client/ui-conversation/tests/terminal-card.spec.tsx
new file mode 100644
index 0000000000..968c1e4461
--- /dev/null
+++ b/packages/client/ui-conversation/tests/terminal-card.spec.tsx
@@ -0,0 +1,600 @@
+// @vitest-environment jsdom
+// The terminal render intent on the web side: the pure terminalCardModel
+// derivation over callView/resultView, and both conversation render sites that
+// consume it — the chat tool row's expanded body (GenericToolCard / BashRow)
+// and the details panel's Output section.
+
+import { afterEach, describe, expect, it, vi } from 'vitest'
+import { cleanup, fireEvent, render } from '@testing-library/react'
+import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react'
+import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
+import type {
+  ConversationSnapshot, RunningToolCall, SessionId, SessionListState, ToolResultNode, WorkspaceListState,
+} from '@deepseek-ai/dsh-client-runtime/client'
+import type { ToolCallView, ToolResultView } from '@deepseek-ai/dsh-client-connection/client'
+import type { SelectionTarget, ToolRowOwnerProps, ToolRowProps } from '@deepseek-ai/dsh-client-ui-conversation/client'
+import { CHAT_TERMINAL_MAX_LINES, terminalCardModel } from '../src/client/contract/terminal-card-model.ts'
+import { createChatStore } from '../src/client/stores.ts'
+import { GenericToolCard } from '../src/client/chat/GenericToolCard.tsx'
+import { DetailsPanel } from '../src/client/skeleton/DetailsPanel.tsx'
+import { BashRow } from '../src/client/toolviews/bash-sample.tsx'
+
+afterEach(cleanup)
+
+/**
+ * Match an output line with its interior whitespace intact: the column
+ * alignment this card exists to preserve is exactly what the default
+ * whitespace-collapsing matcher would hide.
+ */
+const RAW = { normalizer: (text: string) => text }
+
+/** The rendered card's run-state dot state, so a render site cannot silently drop it. */
+function runStateOf(container: HTMLElement): string | null {
+  return container.querySelector('[data-terminal] [data-state]')?.getAttribute('data-state') ?? null
+}
+
+const SID = 's1' as SessionId
+
+const ARGS = '{"command":"ls -la","description":"List files"}'
+
+/** The bash tool's own call view for a foreground command. */
+const callTerminal = (over?: Partial<Extract<ToolCallView, { card: 'terminal' }>>): ToolCallView => ({
+  card: 'terminal', title: 'ls -la', description: 'List files', ...over,
+})
+
+/** The bash tool's own result view for a settled foreground command. */
+const resultTerminal = (over?: Partial<Extract<ToolResultView, { card: 'terminal' }>>): ToolResultView => ({
+  card: 'terminal', output: 'a.ts  b.ts\nc.ts  d.ts\n', exitCode: 0, ...over,
+})
+
+const running = (over?: Partial<RunningToolCall>): RunningToolCall => ({
+  callId: 'c1', name: 'bash', argsRaw: ARGS,
+  turn: 1, step: 1, time: 1_000, callView: callTerminal(), ...over,
+})
+
+const settled = (over?: Partial<ToolResultNode>): ToolResultNode => ({
+  kind: 'tool-result', seq: 10, time: 2_000, callId: 'c1',
+  call: { name: 'bash', argsRaw: ARGS },
+  callTime: 1_000,
+  content: [{ type: 'text', text: 'a.ts  b.ts\nc.ts  d.ts\n' }], isError: false,
+  callView: callTerminal(), resultView: resultTerminal(), ...over,
+})
+
+describe('terminalCardModel', () => {
+  it('derives a running card from the call view alone', () => {
+    expect(terminalCardModel(running({ callView: callTerminal({ cwd: '/projects/app' }) }))).toEqual({
+      description: 'List files',
+      card: {
+        command: 'ls -la', cwd: '/projects/app', output: undefined,
+        exitCode: undefined, signal: undefined, running: true,
+      },
+    })
+  })
+
+  it('derives a settled card from both sides, carrying the exit status', () => {
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '/projects/app' }),
+      resultView: resultTerminal({ output: 'boom\n', exitCode: 2 }),
+    }))).toEqual({
+      description: 'List files',
+      card: {
+        command: 'ls -la', cwd: '/projects/app', output: 'boom\n',
+        exitCode: 2, signal: undefined, running: false,
+      },
+    })
+    expect(terminalCardModel(settled({
+      resultView: { card: 'terminal', output: '', signal: 'SIGTERM' },
+    }))?.card.signal).toBe('SIGTERM')
+  })
+
+  it('takes the result view\'s replacement title over the pending one', () => {
+    // The presentation contract defines a result title as REPLACING the pending
+    // title, so a tool that rewrites it at settle time must win here.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ title: 'pnpm run check' }),
+      resultView: resultTerminal({ title: 'pnpm run check --filter web' }),
+    }))?.card.command).toBe('pnpm run check --filter web')
+    // Without one, the call's title is what the card keeps.
+    expect(terminalCardModel(settled())?.card.command).toBe('ls -la')
+  })
+
+  it('resolves the cwd against the session workspace the way the bridge must', () => {
+    // Omitted workdir — the common bash call — IS the session workspace.
+    expect(terminalCardModel(settled(), '/w/app')?.card.cwd).toBe('/w/app')
+    // A relative workdir joins under it.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: 'packages/ui' }),
+    }), '/w/app')?.card.cwd).toBe('/w/app/packages/ui')
+    // An absolute one is used as-is.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '/srv/other' }),
+    }), '/w/app')?.card.cwd).toBe('/srv/other')
+    // With no session cwd there is nothing to resolve against: a relative path
+    // stays as authored and an omitted one stays absent (a bare `$` prompt).
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: 'packages/ui' }),
+    }))?.card.cwd).toBe('packages/ui')
+    expect(terminalCardModel(settled())?.card.cwd).toBeUndefined()
+    // The running arm resolves identically.
+    expect(terminalCardModel(running(), '/w/app')?.card.cwd).toBe('/w/app')
+  })
+
+  it('normalizes a relative workdir so the label names the directory actually used', () => {
+    // The bash executor resolves the workdir before running, so `..` against
+    // /w/app runs in /w — the card must say `w`, not `..`.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '..' }),
+    }), '/w/app')?.card.cwd).toBe('/w')
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '.' }),
+    }), '/w/app')?.card.cwd).toBe('/w/app')
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '../sibling' }),
+    }), '/w/app')?.card.cwd).toBe('/w/sibling')
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: './nested/../other' }),
+    }), '/w/app')?.card.cwd).toBe('/w/app/other')
+    // A `..` that would climb past the root is dropped, as a filesystem does.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '../../..' }),
+    }), '/w')?.card.cwd).toBe('/')
+    // An absolute path carrying segments normalizes too.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '/srv/./app/../other' }),
+    }), '/w/app')?.card.cwd).toBe('/srv/other')
+    // A Windows path keeps its separators.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: 'C:\\ws\\app\\..' }),
+    }), '/w')?.card.cwd).toBe('C:\\ws')
+    // Without a session cwd a relative `..` has nothing to resolve against, so
+    // it survives as authored rather than being silently dropped.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '../elsewhere' }),
+    }))?.card.cwd).toBe('../elsewhere')
+  })
+
+  it('keeps a UNC server and share as an unpoppable root', () => {
+    // Windows cannot climb above a share, so `..` from the share root stays put.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '..' }),
+    }), '\\\\server\\share')?.card.cwd).toBe('\\\\server\\share')
+    // Below the share it pops normally, keeping the UNC separators.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '..' }),
+    }), '\\\\server\\share\\app')?.card.cwd).toBe('\\\\server\\share')
+    // Several `..` cannot escape the root either.
+    expect(terminalCardModel(settled({
+      callView: callTerminal({ cwd: '../../..' }),
+    }), '\\\\server\\share\\app')?.card.cwd).toBe('\\\\server\\share')
+  })
+
+  it('draws a bare $ when the window dropped the call head, rather than guessing', () => {
+    // A truncated call carries no cwd anywhere: the result view has none, and
+    // the original call may have used an explicit workdir. Falling back to the
+    // session workspace here would name a directory the card cannot know.
+    expect(terminalCardModel(settled({
+      call: null, callView: null, resultView: resultTerminal({ title: 'ls -la' }),
+    }), '/w/app')?.card.cwd).toBeUndefined()
+    // A present call view that omits its cwd still means the workspace.
+    expect(terminalCardModel(settled(), '/w/app')?.card.cwd).toBe('/w/app')
+  })
+
+  it('carries the call view\'s description, which the contract renders above the card', () => {
+    expect(terminalCardModel(settled())?.description).toBe('List files')
+    expect(terminalCardModel(running())?.description).toBe('List files')
+    // A presenter that supplies none, and a window-truncated call side, both
+    // leave it absent so the row keeps its args-derived summary.
+    expect(terminalCardModel(settled({
+      callView: { card: 'terminal', title: 'ls' },
+    }))?.description).toBeUndefined()
+    expect(terminalCardModel(settled({ call: null, callView: null }))?.description).toBeUndefined()
+  })
+
+  it('a window-truncated call side falls back to the result title, then to an empty command', () => {
+    // Truncation drops both the call head and its view (conversation.ts).
+    const truncated = { call: null, callView: null }
+    expect(terminalCardModel(settled({
+      ...truncated, resultView: resultTerminal({ title: 'ls -la' }),
+    }))?.card).toMatchObject({ command: 'ls -la', cwd: undefined, running: false })
+    expect(terminalCardModel(settled(truncated))?.card).toMatchObject({ command: '', cwd: undefined })
+  })
+
+  it('returns null for every non-terminal call: no views, generic views, unknown cards', () => {
+    expect(terminalCardModel(running({ callView: null }))).toBeNull()
+    expect(terminalCardModel(settled({ callView: null, resultView: null }))).toBeNull()
+    expect(terminalCardModel(running({ callView: { card: 'generic', title: 'read x' } }))).toBeNull()
+    // A generic result settles a terminal call as a generic card (the bash
+    // tool's own execution-error and background paths).
+    expect(terminalCardModel(settled({ resultView: { card: 'generic' } }))).toBeNull()
+    // A card tag this UI version does not know arrives over the wire; the
+    // documented generic-card default takes it, not a crash.
+    const future = { card: 'chart', title: 'plot' } as unknown as ToolCallView
+    expect(terminalCardModel(running({ callView: future }))).toBeNull()
+    expect(terminalCardModel(settled({
+      callView: future, resultView: { card: 'chart' } as unknown as ToolResultView,
+    }))).toBeNull()
+  })
+})
+
+describe('chat row terminal body', () => {
+  const ownerProps = (block: RunningToolCall | ToolResultNode): ToolRowOwnerProps => ({
+    callId: 'c1', toolName: 'bash', block, openFile: vi.fn(),
+  })
+
+  it('the expanded body is the command output, capped tighter than the panel', () => {
+    expect(CHAT_TERMINAL_MAX_LINES).toBeLessThan(16)
+    const view = render(<GenericToolCard {...ownerProps(settled())} />)
+    // Collapsed: the one-line summary row only, no output.
+    expect(view.getByText('List files')).toBeTruthy()
+    expect(view.queryByText(/a\.ts/)).toBeNull()
+    fireEvent.click(view.container.querySelector('button')!)
+    expect(view.getByText('a.ts  b.ts', RAW)).toBeTruthy()
+    expect(view.getByText('ls -la')).toBeTruthy()
+    // The args JSON body the generic path would have shown is gone.
+    expect(view.queryByText(/"command"/)).toBeNull()
+  })
+
+  it('the cap collapses a long output inside the row, expandable in place', () => {
+    const lines = Array.from({ length: CHAT_TERMINAL_MAX_LINES + 3 }, (_, i) => `line-${i}`)
+    const view = render(<GenericToolCard {...ownerProps(settled({
+      resultView: resultTerminal({ output: `${lines.join('\n')}\n` }),
+    }))} />)
+    fireEvent.click(view.container.querySelector('button')!)
+    expect(view.getByText('… 其余 3 行')).toBeTruthy()
+    expect(view.queryByText('line-5')).toBeNull()
+    fireEvent.click(view.getByRole('button', { name: '展开其余 3 行输出' }))
+    expect(view.getByText('line-5')).toBeTruthy()
+  })
+
+  it('renders a multi-line command as one prompt row per line', () => {
+    const view = render(<GenericToolCard {...ownerProps(settled({
+      callView: callTerminal({ title: 'ls -la\necho done' }),
+    }))} />)
+    fireEvent.click(view.container.querySelector('button')!)
+    const rows = view.container.querySelectorAll('[class^="_promptLine_"]')
+    expect([...rows].map(row => row.textContent)).toEqual(['$ls -la', '$echo done'])
+    // Still one dot for the call, on the first row.
+    expect(view.container.querySelectorAll('[data-terminal] [data-state]')).toHaveLength(1)
+  })
+
+  it('the fallback row shows the presenter description, not the args summary', () => {
+    // Any terminal-declaring tool without its own keyed row lands here, so the
+    // contract's above-card description has to win at this render site as well.
+    const view = render(<GenericToolCard {...ownerProps(settled({
+      callView: callTerminal({ description: 'Terminal 3' }),
+    }))} />)
+    expect(view.getByText('Terminal 3')).toBeTruthy()
+    expect(view.queryByText('List files')).toBeNull()
+  })
+
+  it('keeps the presenter description visible once the terminal card is expanded', () => {
+    // The contract puts the description ABOVE the card. The collapsed summary is
+    // hidden while a row is open, so an expanded terminal row has to draw it
+    // itself or the description would only ever be visible collapsed.
+    const view = render(<GenericToolCard {...ownerProps(settled({
+      callView: callTerminal({ description: 'Terminal 3' }),
+    }))} />)
+    expect(view.getByText('Terminal 3')).toBeTruthy()
+    fireEvent.click(view.container.querySelector('button')!)
+    expect(view.container.querySelector('[data-terminal]')).not.toBeNull()
+    expect(view.getByText('Terminal 3')).toBeTruthy()
+  })
+
+  it('a running terminal call expands to the prompt line with no output yet', () => {
+    const view = render(<GenericToolCard {...ownerProps(running())} />)
+    fireEvent.click(view.container.querySelector('button')!)
+    expect(view.getByText('ls -la')).toBeTruthy()
+    expect(view.queryByText('复制')).toBeNull()
+    // The card states its own run state: a running command reads as running
+    // even though it has no output yet to distinguish it from an empty settle.
+    expect(runStateOf(view.container)).toBe('ongoing')
+  })
+
+  it('a non-terminal call keeps the args-JSON text body', () => {
+    const view = render(<GenericToolCard {...ownerProps(settled({
+      callView: null, resultView: null,
+    }))} />)
+    fireEvent.click(view.container.querySelector('button')!)
+    expect(view.getByText(/"command"/)).toBeTruthy()
+  })
+
+  it('a terminal call with no args still expands, through its terminal body alone', () => {
+    // Empty args make the text body null; the terminal material carries the row.
+    const view = render(<GenericToolCard {...ownerProps(settled({
+      call: { name: 'bash', argsRaw: '' },
+    }))} />)
+    fireEvent.click(view.container.querySelector('button')!)
+    expect(view.getByText('a.ts  b.ts', RAW)).toBeTruthy()
+  })
+})
+
+describe('BashRow terminal card', () => {
+  const list = () => createSnapshotStore<SessionListState>({
+    ids: [SID],
+    byId: { [SID]: { id: SID, displayTitle: 'r', running: false, blank: false, waitingApproval: false, updatedAt: 0 } },
+    current: undefined,
+    phase: 'ready',
+  })
+
+  const rowProps = (block: RunningToolCall | ToolResultNode): ToolRowProps => ({
+    callId: 'c1', toolName: 'bash', block, openFile: vi.fn(),
+    sessionId: SID, useSessions: bindSnapshotSelector(list()),
+  } as unknown as ToolRowProps)
+
+  it('renders the command output under the summary row, without an expand gesture', () => {
+    const view = render(<BashRow {...rowProps(settled())} />)
+    expect(view.getByText('List files')).toBeTruthy()
+    expect(view.getByText('a.ts  b.ts', RAW)).toBeTruthy()
+    // The card's controls are the row's only interactions: a bash row is not a
+    // path link and no longer a details-panel target, so nothing here navigates.
+    expect(view.container.querySelector('[data-clickable]')).toBeNull()
+    expect(view.getByText('复制')).toBeTruthy()
+  })
+
+  // The row's leading StateDot and the card's run-state dot describe the same
+  // command, so a running row whose card claimed 'done' would be a contradiction
+  // the reader sees on one line.
+  it('agrees with the summary row about the run state', () => {
+    const runningView = render(<BashRow {...rowProps(running())} />)
+    expect(runningView.container.querySelector('[data-variant="bash"]')?.getAttribute('data-state')).toBe('running')
+    expect(runStateOf(runningView.container)).toBe('ongoing')
+    cleanup()
+    const settledView = render(<BashRow {...rowProps(settled())} />)
+    expect(settledView.container.querySelector('[data-variant="bash"]')?.getAttribute('data-state')).toBe('ok')
+    expect(runStateOf(settledView.container)).toBe('done')
+  })
+
+  it('shows the terminal presenter\'s description instead of the args summary', () => {
+    // `terminal_send`-style presenters author a description the args do not
+    // repeat; the contract puts it above the card, which is this row's summary.
+    const view = render(<BashRow {...rowProps(settled({
+      callView: callTerminal({ description: 'Terminal 3' }),
+    }))} />)
+    expect(view.getByText('Terminal 3')).toBeTruthy()
+    expect(view.queryByText('List files')).toBeNull()
+  })
+
+  it('keeps the args-derived summary when the presenter authored no description', () => {
+    const view = render(<BashRow {...rowProps(settled({
+      callView: { card: 'terminal', title: 'ls -la' },
+    }))} />)
+    expect(view.getByText('List files')).toBeTruthy()
+  })
+
+  it('a non-terminal bash call (background start) renders the summary row alone', () => {
+    const view = render(<BashRow {...rowProps(settled({
+      callView: { card: 'generic', title: 'sleep 30', kind: 'execute' },
+      resultView: { card: 'generic' },
+    }))} />)
+    expect(view.getByText('List files')).toBeTruthy()
+    expect(view.queryByText(/a\.ts/)).toBeNull()
+  })
+})
+
+describe('DetailsPanel Output section', () => {
+  function mount(snapshot: ConversationSnapshot, selection: SelectionTarget | null, cwd?: string) {
+    localStorage.clear()
+    const chat = createChatStore().create()
+    if (selection !== null) chat.actions.select(selection)
+    const sessions = createSnapshotStore<SessionListState>(cwd === undefined
+      ? { ids: [], byId: {}, current: undefined, phase: 'ready' }
+      : {
+        ids: [SID],
+        byId: { [SID]: { id: SID, displayTitle: 'r', running: false, blank: false, waitingApproval: false, updatedAt: 0, cwd } },
+        current: SID,
+        phase: 'ready',
+      })
+    const workspaces = createSnapshotStore<WorkspaceListState>({
+      items: [], state: 'idle', phase: 'ready', error: null,
+      baselinesReady: true, recentWorkspaceId: undefined,
+    })
+    return render(
+      <DetailsPanel
+        sessionId={SID}
+        useSession={bindSnapshotSelector({ getSnapshot: () => snapshot, subscribe: () => () => {} })}
+        useSessions={bindSnapshotSelector(sessions)}
+        useWorkspaces={bindSnapshotSelector(workspaces)}
+        useInput={(() => { throw new Error('unused') })}
+        inputActions={{ setDraft: () => {}, submit: () => {} }}
+        useProjection={(() => undefined)}
+        useStore={bindSnapshotSelector(chat)}
+        actions={chat.actions}
+        closeDetails={vi.fn()}
+      />,
+    )
+  }
+
+  function snapshot(over: Partial<ConversationSnapshot> = {}): ConversationSnapshot {
+    return {
+      sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [], codeDispatches: new Map(),
+      pending: [], queue: [], running: false, composerPhase: 'active', removed: false,
+      openState: 'open', openError: null, hasMore: false, loadingOlder: false,
+      promptError: null, blank: false, lastAgentError: null, ...over,
+    }
+  }
+
+  const target: SelectionTarget = { turnSeq: 10, callId: 'c1', toolName: 'bash' }
+
+  // The panel never unmounts between selections, so per-call view state has to
+  // be keyed off the selected call or it leaks into the next one.
+  it('resets the card\'s expand state when the selected call changes', () => {
+    const long = Array.from({ length: 20 }, (_, i) => `row-${i}`)
+    const view = mount(snapshot({
+      nodes: [settled({ resultView: resultTerminal({ output: `${long.join('\n')}\n` }) })],
+    }), target)
+    fireEvent.click(view.getByRole('button', { name: '展开其余 4 行输出' }))
+    expect(view.getByRole('button', { name: '收起输出' })).toBeTruthy()
+    // A second call, selected without unmounting the panel, starts collapsed.
+    cleanup()
+    const second = mount(snapshot({
+      nodes: [settled({
+        callId: 'c2', resultView: resultTerminal({ output: `${long.join('\n')}\n` }),
+      })],
+    }), { turnSeq: 10, callId: 'c2', toolName: 'bash' })
+    expect(second.getByRole('button', { name: '展开其余 4 行输出' })).toBeTruthy()
+  })
+
+  it('renders the presenter description above the card', () => {
+    const view = mount(snapshot({
+      nodes: [settled({ callView: callTerminal({ description: 'Terminal 3' }) })],
+    }), target)
+    const description = view.getByText('Terminal 3')
+    const card = view.container.querySelector('[data-terminal]')
+    expect(card).not.toBeNull()
+    // Above, not below: document order is what places it as the card's heading.
+    expect(description.compareDocumentPosition(card!) & Node.DOCUMENT_POSITION_FOLLOWING).toBeTruthy()
+  })
+
+  it('resolves the prompt cwd against the session workspace', () => {
+    const view = mount(snapshot({ nodes: [settled()] }), target, '/w/app')
+    // No workdir in the call view: the prompt label is the workspace basename.
+    expect(view.getByText('app')).toBeTruthy()
+  })
+
+  it('renders the terminal card at full height, keeping the JSON Input section', () => {
+    const long = Array.from({ length: 20 }, (_, i) => `row-${i}`)
+    const view = mount(snapshot({
+      nodes: [settled({ resultView: resultTerminal({ output: `${long.join('\n')}\n` }) })],
+    }), target)
+    expect(view.getByText(/"command"/)).toBeTruthy()
+    expect(view.getByText('ls -la')).toBeTruthy()
+    // The panel takes the primitive's own default cap (16), not the row's.
+    expect(view.getByText(`… 其余 ${20 - 16} 行`)).toBeTruthy()
+    expect(view.getByText('row-0')).toBeTruthy()
+  })
+
+  it('a running terminal call shows the prompt line, not the 运行中… placeholder', () => {
+    const view = mount(snapshot({ runningCalls: [running()] }), target)
+    expect(view.getByText('ls -la')).toBeTruthy()
+    expect(view.queryByText('运行中…')).toBeNull()
+    expect(runStateOf(view.container)).toBe('ongoing')
+  })
+
+  it('a running non-terminal call keeps the 运行中… placeholder', () => {
+    const view = mount(snapshot({ runningCalls: [running({ callView: null })] }), target)
+    expect(view.getByText('运行中…')).toBeTruthy()
+  })
+
+  it('a non-terminal result keeps the flattened pre with its error styling', () => {
+    const view = mount(snapshot({
+      nodes: [settled({
+        callView: null, resultView: null, isError: true,
+        content: [{ type: 'text', text: 'permission denied' }],
+      })],
+    }), target)
+    const pre = view.container.querySelector('pre[data-error]')
+    expect(pre?.textContent).toBe('permission denied')
+  })
+
+  // The panel resolves a sub-dispatch through the same material as a native
+  // call, so a sub-call that DID carry terminal views would render the card.
+  // The shipped wire cannot produce that yet: `session.ts` folds
+  // `tool/code-dispatch(-start)` with `callView: null`/`resultView: null`, and
+  // the host's `viewFor` only presents top-level `tool/call`/`tool/result`. This
+  // pins the resolution path with views injected directly, and the arm below
+  // pins what the shipped path actually shows today.
+  it('a run_code sub-dispatch resolves to its own terminal card once views reach it', () => {
+    const view = mount(snapshot({
+      codeDispatches: new Map([['p1', [settled({ callId: 'c1' })]]]),
+    }), target)
+    expect(view.getByText('a.ts  b.ts', RAW)).toBeTruthy()
+  })
+
+  it('a sub-dispatch as the wire actually delivers it (no views) keeps the flattened form', () => {
+    const view = mount(snapshot({
+      codeDispatches: new Map([['p1', [settled({ callId: 'c1', callView: null, resultView: null })]]]),
+    }), target)
+    // No terminal card: the generic path renders the result text in the Output
+    // section's <pre> (the Input section has its own, hence the scoping).
+    expect(view.container.querySelector('[data-terminal]')).toBeNull()
+    const output = view.getByText('Output').closest('section')
+    expect(output?.querySelector('pre')?.textContent).toContain('a.ts  b.ts')
+  })
+
+  it('a running run_code sub-dispatch resolves through the running material', () => {
+    const view = mount(snapshot({
+      // The leading non-matching sub-call exercises the scan's skip.
+      codeDispatches: new Map([['p1', [running({ callId: 'other' }), running()]]]),
+    }), target)
+    expect(view.getByText('ls -la')).toBeTruthy()
+  })
+
+  it('a window-truncated call head titles the panel by callId and drops the Input section', () => {
+    const view = mount(snapshot({
+      nodes: [settled({ call: null, callView: null, resultView: resultTerminal({ title: 'ls -la' }) })],
+    }), target)
+    expect(view.getByText('c1')).toBeTruthy()
+    expect(view.queryByText('Input')).toBeNull()
+    expect(view.getByText('Output')).toBeTruthy()
+  })
+
+  it('scans past other nodes and other calls before reporting the call out of window', () => {
+    const view = mount(snapshot({
+      nodes: [
+        { kind: 'assistant', seq: 1, time: 1_000, turn: 1, step: 1, blocks: [] },
+        settled({ callId: 'elsewhere' }),
+      ],
+      runningCalls: [running({ callId: 'also-elsewhere' })],
+    }), target)
+    expect(view.getByText('该调用不在当前窗口内')).toBeTruthy()
+  })
+
+  it('no selection at all renders the guidance line and the default title', () => {
+    const view = mount(snapshot(), null)
+    expect(view.getByText('详情')).toBeTruthy()
+    expect(view.getByText('点击消息流中的工具行查看详情')).toBeTruthy()
+  })
+
+  it('a step selection without a callId renders the guidance line too', () => {
+    const view = mount(snapshot(), { turnSeq: 3, stepSeq: 1 })
+    expect(view.getByText('点击消息流中的工具行查看详情')).toBeTruthy()
+  })
+
+  it('the close button reaches closeDetails', () => {
+    localStorage.clear()
+    const chat = createChatStore().create()
+    const closeDetails = vi.fn()
+    const snap = snapshot()
+    const view = render(
+      <DetailsPanel
+        sessionId={SID}
+        useSession={bindSnapshotSelector({ getSnapshot: () => snap, subscribe: () => () => {} })}
+        useSessions={bindSnapshotSelector(createSnapshotStore<SessionListState>(
+          { ids: [], byId: {}, current: undefined, phase: 'ready' }))}
+        useWorkspaces={bindSnapshotSelector(createSnapshotStore<WorkspaceListState>({
+          items: [], state: 'idle', phase: 'ready', error: null,
+          baselinesReady: true, recentWorkspaceId: undefined,
+        }))}
+        useInput={(() => { throw new Error('unused') })}
+        inputActions={{ setDraft: () => {}, submit: () => {} }}
+        useProjection={(() => undefined)}
+        useStore={bindSnapshotSelector(chat)}
+        actions={chat.actions}
+        closeDetails={closeDetails}
+      />,
+    )
+    fireEvent.click(view.getByRole('button', { name: '关闭详情' }))
+    expect(closeDetails).toHaveBeenCalledTimes(1)
+  })
+
+  it('a non-text result block renders as JSON, and an empty result falls back to its error', () => {
+    const nonText = mount(snapshot({
+      nodes: [settled({
+        callView: null, resultView: null,
+        content: [{ type: 'reasoning', text: 'why' }],
+      })],
+    }), target)
+    // Scope to the Output section: the Input section's CodeBlock renders a
+    // <pre> of its own, and it comes first in document order.
+    expect(nonText.getByText('Output').closest('section')?.querySelector('pre')?.textContent)
+      .toBe('{\n  "type": "reasoning",\n  "text": "why"\n}')
+    cleanup()
+    const empty = mount(snapshot({
+      nodes: [settled({
+        callView: null, resultView: null, content: [], isError: true,
+        error: { name: 'ToolError', code: 'interrupted' },
+      })],
+    }), target)
+    expect(empty.getByText('ToolError: interrupted')).toBeTruthy()
+  })
+})
diff --git a/packages/client/ui-conversation/tsconfig.json b/packages/client/ui-conversation/tsconfig.json
index 14ae91598a..1aa28c9c62 100644
--- a/packages/client/ui-conversation/tsconfig.json
+++ b/packages/client/ui-conversation/tsconfig.json
@@ -29,6 +29,9 @@
     {
       "path": "../../plan/plan-mode"
     },
+    {
+      "path": "../../goal/goal"
+    },
     {
       "path": "../../todo/tool-todo"
     },
diff --git a/packages/client/ui-goal/README.i18n.yaml b/packages/client/ui-goal/README.i18n.yaml
index 9cda25e2e2..666a6e472e 100644
--- a/packages/client/ui-goal/README.i18n.yaml
+++ b/packages/client/ui-goal/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-goal/README.md
-README.md: 476096a43532a0bf514cd191585872ef17f65c50
-README.zh.md: 27bd9a2e735cb4895d30eaf3b08dd939a00436fc
+README.md: fed4870f73277b22760417297d668853b8afb2db
+README.zh.md: cc607edc856e04c6ee42cc8f596aa658679a02ca
diff --git a/packages/client/ui-goal/README.md b/packages/client/ui-goal/README.md
index 476096a435..fed4870f73 100644
--- a/packages/client/ui-goal/README.md
+++ b/packages/client/ui-goal/README.md
@@ -2,13 +2,13 @@
 
 English | [中文](README.zh.md)
 
-Goal surface plugin, browser half: the `GoalBar` strip in the `conversation.input.dock` list (order 1, tucked against the composer). The live goal arrives through `useProjection('goal')` — the host-computed whole value seeded by the history tail page and updated by `session/projection` frames — so the plugin owns no store, no refresh chain, and no event listener. The slot inject face carries only the three mutation verbs (edit / resume / clear over the `goal.*` wire domain); each reads the CAS ref from the session's current projected value at call time and surfaces the settled RPC error inline (the RPC's compare-and-set is the staleness guard — there is no client fence). Goal creation stays on the `/goal` host command; loading, absent, and completed goals render nothing.
+Goal surface plugin, browser half: the `GoalBar` strip in the `conversation.input.dock` list (order 1, tucked against the composer). The live goal arrives through `useProjection('goal')` — the host-computed whole value seeded by the history tail page and updated by `session/projection` frames — so the plugin owns no store, no refresh chain, and no event listener. The slot inject face carries only the four mutation verbs (edit / pause / resume / clear over the `goal.*` wire domain — an active goal offers the pause action, a paused one resume); each reads the CAS ref from the session's current projected value at call time and surfaces the settled RPC error inline (the RPC's compare-and-set is the staleness guard — there is no client fence). Goal creation stays on the `/goal` host command; loading, absent, and completed goals render nothing.
 
 The `/client` export surface is the plugin body (`apply`/`inject`), the `GoalBar`/`GoalDock` components, and the injected verb face types.
 
 ## Model Experience
 
-Indirectly, through the `goal.edit`/`goal.resume`/`goal.clear` RPCs the strip's verbs submit: each accepted mutation appends a model-visible `goal/change` context message to the session (the same durable event the projection folds), so the model sees the updated goal state on its next turn. The strip itself adds no prompt content.
+Indirectly, through the `goal.edit`/`goal.pause`/`goal.resume`/`goal.clear` RPCs the strip's verbs submit: each accepted mutation appends a model-visible `goal/change` context message to the session (the same durable event the projection folds), so the model sees the updated goal state on its next turn. The strip itself adds no prompt content.
 
 #### KV Cache effect
 
diff --git a/packages/client/ui-goal/README.zh.md b/packages/client/ui-goal/README.zh.md
index 27bd9a2e73..cc607edc85 100644
--- a/packages/client/ui-goal/README.zh.md
+++ b/packages/client/ui-goal/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-Goal 表面插件（浏览器半件）：`conversation.input.dock` 列表中的 `GoalBar` 条带（order 1，紧贴 composer）。活值经 `useProjection('goal')` 到达——host 计算的全量值由历史尾页播种、由 `session/projection` 帧更新——因此本插件不持有 store、不设刷新链、不挂事件监听。slot 注入面只携带三个变更动词（edit / resume / clear，走 `goal.*` 协议域）；每个动词在调用时从会话当前投影值读取 CAS ref，并把结算后的 RPC 错误内联呈现（RPC 的 compare-and-set 即陈旧性防护——客户端没有任何栅栏）。goal 的创建仍归 `/goal` host 命令；加载中、无 goal、已完成三种状态一律不渲染。
+Goal 表面插件（浏览器半件）：`conversation.input.dock` 列表中的 `GoalBar` 条带（order 1，紧贴 composer）。活值经 `useProjection('goal')` 到达——host 计算的全量值由历史尾页播种、由 `session/projection` 帧更新——因此本插件不持有 store、不设刷新链、不挂事件监听。slot 注入面只携带四个变更动词（edit / pause / resume / clear，走 `goal.*` 协议域——active 的 goal 提供暂停动作，paused 的提供恢复）；每个动词在调用时从会话当前投影值读取 CAS ref，并把结算后的 RPC 错误内联呈现（RPC 的 compare-and-set 即陈旧性防护——客户端没有任何栅栏）。goal 的创建仍归 `/goal` host 命令；加载中、无 goal、已完成三种状态一律不渲染。
 
 `/client` 出口面为插件本体（`apply`/`inject`）、`GoalBar`/`GoalDock` 组件与注入动词面类型。
 
 ## Model Experience
 
-间接影响：条带动词提交的 `goal.edit`/`goal.resume`/`goal.clear` RPC 每次被接受后，会向会话追加一条模型可见的 `goal/change` 上下文消息（与投影折叠的正是同一条持久事件），模型在下一轮即可看到更新后的 goal 状态。条带自身不添加任何提示词内容。
+间接影响：条带动词提交的 `goal.edit`/`goal.pause`/`goal.resume`/`goal.clear` RPC 每次被接受后，会向会话追加一条模型可见的 `goal/change` 上下文消息（与投影折叠的正是同一条持久事件），模型在下一轮即可看到更新后的 goal 状态。条带自身不添加任何提示词内容。
 
 #### KV Cache effect
 
diff --git a/packages/client/ui-goal/src/client/GoalBar.module.css b/packages/client/ui-goal/src/client/GoalBar.module.css
index fe07bace1b..80c87be57b 100644
--- a/packages/client/ui-goal/src/client/GoalBar.module.css
+++ b/packages/client/ui-goal/src/client/GoalBar.module.css
@@ -91,7 +91,7 @@
 .actions {
   display: flex;
   align-items: center;
-  gap: 2px;
+  gap: 8px;
   flex: none;
 }
 
diff --git a/packages/client/ui-goal/src/client/GoalBar.tsx b/packages/client/ui-goal/src/client/GoalBar.tsx
index 76308734fc..b1b0fd7398 100644
--- a/packages/client/ui-goal/src/client/GoalBar.tsx
+++ b/packages/client/ui-goal/src/client/GoalBar.tsx
@@ -11,7 +11,7 @@
 import { useCallback, useEffect, useState } from 'react'
 import type { GoalSnapshot } from '@deepseek-ai/dsh-goal/client'
 import {
-  IconCheckOutline16, IconCloseOutline16, IconEditOutline16, IconPlayOutline16, IconSparkle16, IconTrashOutline16,
+  IconCheckOutline16, IconCloseOutline16, IconEditOutline16, IconPauseOutline16, IconPlayOutline16, IconSparkle16, IconTrashOutline16,
 } from '@deepseek-ai/dsh-client-ui-primitives'
 import type { GoalActionResult, GoalBarActions } from './slots.ts'
 import css from './GoalBar.module.css'
@@ -28,7 +28,7 @@ const PHASE_LABELS = {
   blocked: 'Blocked Goal',
 } as const
 
-export function GoalBar({ goal, onEdit, onResume, onClear }: GoalBarProps) {
+export function GoalBar({ goal, onEdit, onPause, onResume, onClear }: GoalBarProps) {
   const [editing, setEditing] = useState(false)
   const [draft, setDraft] = useState('')
   const [pending, setPending] = useState(false)
@@ -120,6 +120,11 @@ export function GoalBar({ goal, onEdit, onResume, onClear }: GoalBarProps) {
         <span className={css.objective}>{goal.objective}</span>
         {actionError !== null && <span className={css.error} role="alert">{actionError}</span>}
         <div className={css.actions}>
+          {goal.phase === 'active' && (
+            <button type="button" className={css.iconBtn} disabled={pending} onClick={() => { void runAction(onPause) }} title="Pause goal" aria-label="Pause goal">
+              <IconPauseOutline16 />
+            </button>
+          )}
           {goal.phase === 'paused' && (
             <button type="button" className={css.iconBtn} disabled={pending} onClick={() => { void runAction(onResume) }} title="Resume goal" aria-label="Resume goal">
               <IconPlayOutline16 />
@@ -148,12 +153,13 @@ export function GoalBar({ goal, onEdit, onResume, onClear }: GoalBarProps) {
 export type GoalDockProps = import('@deepseek-ai/dsh-client-ui-slots').PropsRuntime<'conversation.input.dock'> & GoalBarActions
 
 /** Dock adapter: reads the host-computed 'goal' projection (whole value; absent or null renders nothing). */
-export function GoalDock({ useProjection, onEdit, onResume, onClear }: GoalDockProps) {
+export function GoalDock({ useProjection, onEdit, onPause, onResume, onClear }: GoalDockProps) {
   const projection = useProjection('goal')
   return (
     <GoalBar
       goal={projection === undefined ? undefined : projection === null ? null : projection.goal}
       onEdit={onEdit}
+      onPause={onPause}
       onResume={onResume}
       onClear={onClear}
     />
diff --git a/packages/client/ui-goal/src/client/index.ts b/packages/client/ui-goal/src/client/index.ts
index 9c9d3f298d..fc1cf1e035 100644
--- a/packages/client/ui-goal/src/client/index.ts
+++ b/packages/client/ui-goal/src/client/index.ts
@@ -66,6 +66,11 @@ export function apply(ctx: ClientContext): void {
           if (ref === undefined) return noCurrentGoal
           return settle((await goals.edit({ sessionId, ref, objective })).result)
         },
+        onPause: async () => {
+          const ref = refOf(sessionId)
+          if (ref === undefined) return noCurrentGoal
+          return settle((await goals.pause({ sessionId, ref })).result)
+        },
         onResume: async () => {
           const ref = refOf(sessionId)
           if (ref === undefined) return noCurrentGoal
diff --git a/packages/client/ui-goal/src/client/slots.ts b/packages/client/ui-goal/src/client/slots.ts
index 5791f26373..7a1e869afc 100644
--- a/packages/client/ui-goal/src/client/slots.ts
+++ b/packages/client/ui-goal/src/client/slots.ts
@@ -19,6 +19,8 @@ export interface GoalBarActions {
    * @param objective - replacement objective text.
    */
   onEdit: (objective: string) => Promise<GoalActionResult>
+  /** Pause an active goal. */
+  onPause: () => Promise<GoalActionResult>
   /** Resume a paused goal. */
   onResume: () => Promise<GoalActionResult>
   /** Clear the current goal (tombstone). */
diff --git a/packages/client/ui-goal/tests/browser-plugin.spec.tsx b/packages/client/ui-goal/tests/browser-plugin.spec.tsx
index e1eae47529..348d06d1a6 100644
--- a/packages/client/ui-goal/tests/browser-plugin.spec.tsx
+++ b/packages/client/ui-goal/tests/browser-plugin.spec.tsx
@@ -57,6 +57,7 @@ function bench(options: { projection?: GoalProjection | null | undefined; failWi
   const ref = { id: 'g-1', revision: 3 }
   ctx.provide('connection', { api: { goals: {
     edit: answer('goal.edit', { ref }),
+    pause: answer('goal.pause', { ref }),
     resume: answer('goal.resume', { ref }),
     clear: answer('goal.clear', { cleared: true as const }),
   } } })
@@ -100,13 +101,15 @@ describe('ui-goal browser plugin', () => {
     await b.fiber.await()
     const verbs = b.entry()!.inject!(sid('s1'))
     expect(await verbs.onEdit('New objective')).toEqual({ ok: true })
+    expect(await verbs.onPause()).toEqual({ ok: true })
     expect(await verbs.onResume()).toEqual({ ok: true })
     expect(await verbs.onClear()).toEqual({ ok: true })
-    expect(b.calls.map(c => c.method)).toEqual(['goal.edit', 'goal.resume', 'goal.clear'])
+    expect(b.calls.map(c => c.method)).toEqual(['goal.edit', 'goal.pause', 'goal.resume', 'goal.clear'])
     const ref = { id: 'g-1', revision: 5 }
     expect(b.calls[0]?.payload).toEqual({ sessionId: 's1', ref, objective: 'New objective' })
     expect(b.calls[1]?.payload).toEqual({ sessionId: 's1', ref })
     expect(b.calls[2]?.payload).toEqual({ sessionId: 's1', ref })
+    expect(b.calls[3]?.payload).toEqual({ sessionId: 's1', ref })
   })
 
   it('a null or absent projection short-circuits every verb without touching the wire', async () => {
@@ -114,7 +117,7 @@ describe('ui-goal browser plugin', () => {
       const b = bench({ projection })
       await b.fiber.await()
       const verbs = b.entry()!.inject!(sid('s1'))
-      for (const result of [await verbs.onEdit('x'), await verbs.onResume(), await verbs.onClear()]) {
+      for (const result of [await verbs.onEdit('x'), await verbs.onPause(), await verbs.onResume(), await verbs.onClear()]) {
         expect(result).toEqual({ ok: false, error: { code: 'no-current-goal', message: 'no current goal to mutate' } })
       }
       expect(b.calls).toHaveLength(0)
@@ -143,6 +146,7 @@ describe('GoalDock adapter', () => {
     const useProjection = vi.fn(() => projection)
     const actions: GoalBarActions = {
       onEdit: () => Promise.resolve({ ok: true }),
+      onPause: () => Promise.resolve({ ok: true }),
       onResume: () => Promise.resolve({ ok: true }),
       onClear: () => Promise.resolve({ ok: true }),
     }
diff --git a/packages/client/ui-goal/tests/goalbar.spec.tsx b/packages/client/ui-goal/tests/goalbar.spec.tsx
index ece447f8b0..38a943d457 100644
--- a/packages/client/ui-goal/tests/goalbar.spec.tsx
+++ b/packages/client/ui-goal/tests/goalbar.spec.tsx
@@ -25,6 +25,7 @@ function makeGoal(over: Partial<GoalSnapshot> = {}): GoalSnapshot {
 function makeActions() {
   return {
     onEdit: vi.fn<GoalBarActions['onEdit']>(() => Promise.resolve({ ok: true })),
+    onPause: vi.fn<GoalBarActions['onPause']>(() => Promise.resolve({ ok: true })),
     onResume: vi.fn<GoalBarActions['onResume']>(() => Promise.resolve({ ok: true })),
     onClear: vi.fn<GoalBarActions['onClear']>(() => Promise.resolve({ ok: true })),
   } satisfies GoalBarActions
@@ -103,6 +104,13 @@ describe('GoalBar', () => {
     expect(screen.getByRole('textbox', { name: 'Goal objective' })).toBeTruthy()
   })
 
+  it('active goal: the pause action pauses', () => {
+    const actions = makeActions()
+    render(<GoalBar goal={makeGoal()} {...actions} />)
+    fireEvent.click(screen.getByRole('button', { name: 'Pause goal' }))
+    expect(actions.onPause).toHaveBeenCalledTimes(1)
+  })
+
   it('paused goal: "Paused Goal" with a resume action before edit', () => {
     const actions = makeActions()
     render(<GoalBar goal={makeGoal({ phase: 'paused' })} {...actions} />)
diff --git a/packages/client/ui-layout/README.i18n.yaml b/packages/client/ui-layout/README.i18n.yaml
index 589f78a23c..eff1fbe925 100644
--- a/packages/client/ui-layout/README.i18n.yaml
+++ b/packages/client/ui-layout/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-layout/README.md
-README.md: 26e909b96412985792eeae72d51a2ab2a315c943
-README.zh.md: 2e5799fd32c41328f8ca8b9e1a439fbccb3cdba2
+README.md: 9354f4b79f7b1af7d8a20a295e77913ff443c2e4
+README.zh.md: c949236557e7eb3eed0c698566fb5aa9e9cdd18a
diff --git a/packages/client/ui-layout/README.md b/packages/client/ui-layout/README.md
index 26e909b964..9354f4b79f 100644
--- a/packages/client/ui-layout/README.md
+++ b/packages/client/ui-layout/README.md
@@ -4,7 +4,7 @@ English | [中文](README.zh.md)
 
 Shell plugin: three-column AppFrame (drag handles and concession chain) plus the `ctx.layout` panel-geometry service; it registers into the runtime-owned `root` slot and declares `sidebar`, `conversation`, `details`, and `conversation.empty`. The sidebar is fixed-width (only details shrinks, then auto-closes); a closed sidebar retains a 56px control rail while details closes to zero width. The package also seats the theme presenter: it consumes resolved `ctx.theme` snapshots and projects them onto the document (`html { color-scheme }` for native UA chrome, `body[data-ds-dark-theme]` from the active color scheme, plus the theme's alias tokens as inline variables on body).
 
-AppFrame reads the runtime Session projection: `baselinesReady` selects loading, a page-local `SessionListState.intent` selects the empty composer, and a connected Session renders through `SessionProvider`. The conversation and empty-state owner shares are empty; each registrant obtains business data from standard hooks and actions from its own inject face. The sidebar owner share contains only `collapsed` and `width`; navigation actions belong to sidebar's own injected service face.
+AppFrame always mounts the conversation and details columns; a connected Session renders through `SessionProvider`. The transient layout store starts both panels at their default widths and never reads or writes `localStorage`. Hero and other unselected states derive a zero rendered details width without changing that stored preference. AppFrame retains the last non-blank Session id across those states: the first Session opens at the default width, returning to the same Session restores its unchanged width, and selecting a different Session closes details before paint. The conversation owner share is empty, while the sidebar owner share contains only `collapsed` and `width`; registrants obtain business data from standard hooks and actions from their own inject faces.
 
 The `/client` export surface is the plugin body (`apply`/`inject`), `LayoutService`, and the four owner-share interfaces. AppFrame, the panel store, and the concession solver remain package-internal; tests import internals through `/src`.
 
@@ -18,6 +18,6 @@ None; this package neither assembles nor sends a provider request.
 
 ## Known Limitations and Deferred Work
 
-- **Details open/width state is global** — it does not follow the session (arbitrated for P-I); the per-session keyed upgrade slot is reserved.
-- **Concession-chain auto-close derives a zero width without touching the persisted open flag** — the panel restores itself when the window widens; consumers must not read `details.open` as the rendered truth.
+- **Panel geometry is transient** — reload restores both panels to their defaults; switching between distinct Session ids closes details and forgets its dragged width, while unselected surfaces render details at zero width without modifying geometry.
+- **Concession-chain auto-close derives a zero width without touching the preferred width** — the panel restores itself when the window widens; consumers must not read the stored details width as the rendered truth.
 - **Scroll anchoring during squeeze reflow is not implemented** — deferred with the virtualized-list project.
diff --git a/packages/client/ui-layout/README.zh.md b/packages/client/ui-layout/README.zh.md
index 2e5799fd32..c949236557 100644
--- a/packages/client/ui-layout/README.zh.md
+++ b/packages/client/ui-layout/README.zh.md
@@ -4,7 +4,7 @@
 
 外壳插件：三栏 AppFrame（拖动手柄与让步链）加 `ctx.layout` 面板几何服务；它注册到运行时拥有的 `root` slot，并声明 `sidebar`、`conversation`、`details` 和 `conversation.empty`。侧边栏宽度固定（只会收缩详情栏，然后将其自动关闭）；关闭的侧边栏仍保留 56px 控制轨道，详情栏则关闭到零宽度。该包还提供主题呈现器：它消费解析后的 `ctx.theme` 快照，并将其投影到 document（用 `html { color-scheme }` 驱动原生 UA 控件，依据当前配色方案设置 `body[data-ds-dark-theme]`，并将主题的别名 token 设为 body 上的内联变量）。
 
-AppFrame 读取运行时 Session 投影：`baselinesReady` 选择加载状态，页面局部的 `SessionListState.intent` 选择空白编辑器，已连接 Session 则通过 `SessionProvider` 渲染。会话及空状态的 owner share 为空；每个注册方通过标准 hook 获取业务数据，并从自身的 inject 表层获取操作。侧边栏 owner share 只包含 `collapsed` 和 `width`；导航操作属于侧边栏自身注入的服务表层。
+AppFrame 始终挂载会话栏和详情栏；已连接 Session 通过 `SessionProvider` 渲染。布局 store 是瞬时状态，两个面板均以默认宽度启动，且从不读写 `localStorage`。hero 和其他未选中状态会将详情栏的渲染宽度派生为零，但不会改变存储的首选宽度。AppFrame 会跨越这些状态保留最后一个非 blank 会话 id：首个会话以默认宽度打开；返回同一会话时恢复其未改变的宽度；选择不同会话时，详情栏会在绘制前关闭。会话 owner share 为空，侧边栏 owner share 只包含 `collapsed` 和 `width`；注册方通过标准钩子获取业务数据，并从各自的 inject 表层获取操作。
 
 `/client` 导出表层包含插件主体（`apply`／`inject`）、`LayoutService` 和四个 owner-share 接口。AppFrame、面板 store 与让步求解器仍属于包内部；测试通过 `/src` 导入内部实现。
 
@@ -18,6 +18,6 @@ AppFrame 读取运行时 Session 投影：`baselinesReady` 选择加载状态，
 
 ## 已知限制与暂缓事项
 
-- **详情栏打开／宽度状态是全局状态**：它不会随会话变化（P-I 已裁定）；为逐会话键控升级预留了 slot。
-- **让步链自动关闭通过推导零宽度实现，不会改动持久化的打开标志**：窗口变宽时面板会自行恢复；消费方禁止把 `details.open` 当作实际渲染状态。
+- **面板几何信息是瞬时状态**：重新加载会将两个面板恢复为默认值；在不同会话 id 之间切换会关闭详情栏，并忘记拖动后的宽度，而未选中表面会以零宽度渲染详情栏，但不会修改几何信息。
+- **让步链自动关闭通过推导零宽度实现，不会改动首选宽度**：窗口变宽时面板会自行恢复；消费方禁止把 store 中的详情宽度当作实际渲染状态。
 - **挤压重排期间尚未实现滚动锚定**：与虚拟化列表项目一并暂缓。
diff --git a/packages/client/ui-layout/src/client/AppFrame.tsx b/packages/client/ui-layout/src/client/AppFrame.tsx
index a7c7696222..8aa16d8675 100644
--- a/packages/client/ui-layout/src/client/AppFrame.tsx
+++ b/packages/client/ui-layout/src/client/AppFrame.tsx
@@ -10,7 +10,7 @@
  * through the three framework shares — zero cordis or framework imports,
  * zero self-made hooks.
  */
-import { useCallback, useEffect, useRef, useState } from 'react'
+import { useCallback, useEffect, useLayoutEffect, useRef, useState } from 'react'
 import type { ReactNode } from 'react'
 import type { PropsRenderSlots, PropsRuntime, PropsStore } from '@deepseek-ai/dsh-client-ui-slots'
 import { computeColumns } from './columns.ts'
@@ -86,13 +86,27 @@ function DragHandle(props: { side: 'sidebar' | 'details'; left: number; onStart:
 /** The three-column frame (see module doc). */
 export function AppFrame({
   useStore,
+  useSessions,
   actions,
   renderSlot,
 }: AppFrameProps) {
   const panels = useStore(s => s)
+  const detailsSession = useSessions((s) => {
+    const current = s.current
+    return current !== undefined && s.byId[current]?.blank === false ? current : undefined
+  })
   const frameRef = useRef<HTMLDivElement | null>(null)
   const [viewport, setViewport] = useState(() => window.innerWidth)
 
+  const lastSession = useRef(detailsSession)
+  useLayoutEffect(() => {
+    if (detailsSession === undefined) return
+    if (lastSession.current !== undefined && lastSession.current !== detailsSession) {
+      actions.closeDetails()
+    }
+    lastSession.current = detailsSession
+  }, [actions, detailsSession])
+
   // Track the frame's own box (not the window): rAF-throttled ResizeObserver.
   useEffect(() => {
     const el = frameRef.current
@@ -113,12 +127,12 @@ export function AppFrame({
     }
   }, [])
 
-  const cols = computeColumns(viewport, panels.sidebar, panels.details)
+  const cols = computeColumns(viewport, panels.sidebar, detailsSession === undefined ? 0 : panels.details)
   const colsRef = useRef(cols)
   colsRef.current = cols
 
   // The drag base is the rendered width captured at drag start (grabbing a
-  // concession-clamped panel must not jump back to the persisted preference);
+  // concession-clamped panel must not jump back to the stored preference);
   // it stays frozen for the whole gesture so dx deltas do not compound.
   const sidebarBase = useRef(0)
   const detailsBase = useRef(0)
diff --git a/packages/client/ui-layout/src/client/columns.ts b/packages/client/ui-layout/src/client/columns.ts
index 7cd5f8c2d8..125bb92a70 100644
--- a/packages/client/ui-layout/src/client/columns.ts
+++ b/packages/client/ui-layout/src/client/columns.ts
@@ -1,7 +1,7 @@
 /**
  * Pure concession-chain column solver for the three-column AppFrame.
  * Chain order is fixed by contract: keep center >= CENTER_MIN by shrinking
- * details, then auto-closing it (derived zero width — persisted width
+ * details, then auto-closing it (derived zero width — preferred width
  * preferences are never rewritten, so widening the window restores them).
  * The sidebar never concedes: its rendered width is always the drag
  * preference (or the collapsed rail), and center absorbs any remaining
@@ -45,8 +45,8 @@ export function clampWidth(px: number, min: number, max: number): number {
 /**
  * Solve the three column widths for one viewport frame. Pure: no hysteresis —
  * the output is a function of (viewport, preferences) only, so recovery on
- * re-widening is automatic. Preferences re-clamp here because they cross a
- * durable boundary (localStorage rehydration may carry stale ranges).
+ * re-widening is automatic. Preferences re-clamp here because they cross the
+ * store boundary and callers may still supply stale ranges.
  * @param viewport - available frame width in px.
  * @param sidebar - sidebar width preference in px (0 = closed).
  * @param details - details width preference in px (0 = closed).
diff --git a/packages/client/ui-layout/src/client/stores.ts b/packages/client/ui-layout/src/client/stores.ts
index 06bcbe5ae3..01115c12a5 100644
--- a/packages/client/ui-layout/src/client/stores.ts
+++ b/packages/client/ui-layout/src/client/stores.ts
@@ -1,7 +1,7 @@
 /**
- * The root entry's layout store: panel geometry as plain widths in px
- * (0 = closed), persisted across reloads. Module level exports the factory
- * only — a module-level handle would pin the store's identity in the module
+ * The root entry's transient layout store: panel geometry as plain widths in
+ * px (0 = closed). Module level exports the factory only — a module-level
+ * handle would pin the store's identity in the module
  * cache (a de-facto singleton surviving plugin reloads). register() receives
  * the factory (exclusive use: the framework instantiates per entry), AppFrame
  * derives its PropsStore share from the return type, and the service face
@@ -29,17 +29,16 @@ type LayoutActions = {
 }
 
 /**
- * Create the layout panel store handle. The persisted preference IS the
- * width, so closing a panel forgets its drag width — reopening restores the
- * contract default. Actions are the complete write set: drag writes clamp
+ * Create the layout panel store handle. The preference IS the width, so
+ * closing a panel forgets its drag width — reopening restores the contract
+ * default. Actions are the complete write set: drag writes clamp
  * into the panel's contract range and never cross the open/closed line;
  * open/close transitions write 0 / the default explicitly.
  * @returns the store handle (spec + type + identity + factory in one).
  */
 export function createLayoutStore(): EngineStoreHandle<LayoutState, LayoutActions>  {
   const handle = defineStore({
-    init: (): LayoutState => ({ sidebar: SIDEBAR_DEFAULT, details: 0 }),
-    persist: 'dsh.layout.panels',
+    init: (): LayoutState => ({ sidebar: SIDEBAR_DEFAULT, details: DETAILS_DEFAULT }),
     actions: {
       setSidebar: (d, px: number) => { d.sidebar = clampWidth(px, SIDEBAR_MIN, SIDEBAR_MAX) },
       setDetails: (d, px: number) => { d.details = clampWidth(px, DETAILS_MIN, DETAILS_MAX) },
diff --git a/packages/client/ui-layout/src/invariant.ts b/packages/client/ui-layout/src/invariant.ts
index fa46392b5d..dd572e679d 100644
--- a/packages/client/ui-layout/src/invariant.ts
+++ b/packages/client/ui-layout/src/invariant.ts
@@ -15,8 +15,8 @@ export const name = 'client-ui-layout-invariant'
 export const inject = ['invariants']
 
 /**
- * No runtime invariant: shell viewing-state stores (zustand+persist) behind
- * ctx.layout — it emits no cordis events; clamp/prune/concession-chain
+ * No runtime invariant: the shell viewing-state store behind ctx.layout emits
+ * no cordis events; clamp/prune/concession-chain
  * sequencing is asserted directly by this package's columns and service specs.
  */
 const install: InvariantInstaller = () => {}
diff --git a/packages/client/ui-layout/tests/app-frame.spec.tsx b/packages/client/ui-layout/tests/app-frame.spec.tsx
index 4d5f6de30d..95933b2783 100644
--- a/packages/client/ui-layout/tests/app-frame.spec.tsx
+++ b/packages/client/ui-layout/tests/app-frame.spec.tsx
@@ -21,8 +21,9 @@ import type {
   SessionId, SessionListState, WorkspaceListState,
 } from '@deepseek-ai/dsh-client-runtime/client'
 
-// Session-mode switch for the SessionProvider stub prop.
-const sessionMode = { current: true }
+// Session selection controls for the SessionProvider and useSessions stubs.
+const selectedSession = { current: 's-test' as SessionId | undefined }
+const selectedSessionBlank = { current: false }
 const baselinesReady = { current: true }
 
 // Render-prop contract stub fed through the standard seat prop (the renderer
@@ -31,7 +32,7 @@ const baselinesReady = { current: true }
 // shape. Typed as the seat's own component type so the branded sessionId
 // parameter stays contract-checked.
 const SessionProviderStub: AppFrameProps['SessionProvider'] = ({ children, empty }) =>
-  sessionMode.current ? <>{children('s-test' as Parameters<typeof children>[0])}</> : <>{empty?.() ?? null}</>
+  selectedSession.current === undefined ? <>{empty?.() ?? null}</> : <>{children(selectedSession.current)}</>
 
 
 /** Observer stub: captures the callback so tests can fire resizes manually. */
@@ -54,7 +55,6 @@ function hookOf<T>(inst: { subscribe: (fn: () => void) => () => void; getSnapsho
 function mountFrame() {
   window.innerWidth = frameWidth // first-render viewport source before the observer fires
   const instance = createLayoutStore().create()
-  instance.actions.openDetails() // seed: sidebar at default 280, details open at default 360
   const slotCalls: { key: string; props: unknown }[] = []
   const renderSlot = ((key: string, owner: object) => {
     slotCalls.push({ key, props: owner })
@@ -64,32 +64,35 @@ function mountFrame() {
     if (key === 'conversation.empty') return <div data-testid="empty-content" />
     return <div data-testid="other-content" />
   }) as AppFrameProps['renderSlot']
-  const sessionId = 's-test' as SessionId
-  const sessionState = {
-    ids: sessionMode.current ? [sessionId] : [],
-    byId: sessionMode.current
-      ? { [sessionId]: { id: sessionId, displayTitle: 'Test', running: false, blank: false, updatedAt: 1 } }
-      : {},
-    current: sessionMode.current ? sessionId : undefined,
-    phase: 'ready',
-  } as SessionListState
-  const useSessions = ((sel: (s: SessionListState) => unknown) => sel(sessionState)) as never
+  const useSessions = ((sel: (s: SessionListState) => unknown) => {
+    const current = selectedSession.current
+    const sessionState = {
+      ids: current === undefined ? [] : [current],
+      byId: current === undefined
+        ? {}
+        : { [current]: { id: current, displayTitle: 'Test', running: false, blank: selectedSessionBlank.current, updatedAt: 1 } },
+      current,
+      phase: 'ready',
+    } as SessionListState
+    return sel(sessionState)
+  }) as never
   const workspaceState: WorkspaceListState = {
     items: [], state: 'idle', phase: 'ready', error: null,
     baselinesReady: baselinesReady.current, recentWorkspaceId: undefined,
   }
-  const utils = render(
+  const element = () => (
     <AppFrame
-      useStore={hookOf(instance) as never}
+      useStore={hookOf(instance)}
       actions={instance.actions}
       renderSlot={renderSlot}
       useSessions={useSessions}
       useWorkspaces={((sel: (s: WorkspaceListState) => unknown) => sel(workspaceState)) as never}
       SessionProvider={SessionProviderStub}
-    />,
+    />
   )
+  const utils = render(element())
   const frame = utils.container.firstElementChild as HTMLElement
-  return { instance, frame, slotCalls, ...utils }
+  return { instance, frame, slotCalls, rerenderFrame: () => { utils.rerender(element()) }, ...utils }
 }
 
 function tracks(frame: HTMLElement): number[] {
@@ -109,9 +112,9 @@ function drag(handle: Element, fromX: number, toX: number): void {
 
 beforeEach(() => {
   frameWidth = 1920
-  sessionMode.current = true
+  selectedSession.current = 's-test' as SessionId
+  selectedSessionBlank.current = false
   baselinesReady.current = true
-  localStorage.clear() // the layout store persists; instances must not bleed across tests
   vi.useFakeTimers()
   vi.stubGlobal('ResizeObserver', ResizeObserverStub)
   vi.stubGlobal('requestAnimationFrame', (cb: FrameRequestCallback) => setTimeout(() => { cb(0) }, 16) as unknown as number)
@@ -154,7 +157,7 @@ describe('AppFrame', () => {
   it('keeps the conversation slot mounted while no session is current', () => {
     // No current session: the session-maybe conversation shell owns the New
     // Session view itself — the center column renders it unconditionally.
-    sessionMode.current = false
+    selectedSession.current = undefined
     const { slotCalls, getByTestId } = mountFrame()
     expect(getByTestId('center-content')).toBeTruthy()
     expect(slotCalls.map(c => c.key)).toContain('conversation')
@@ -169,6 +172,45 @@ describe('AppFrame', () => {
     expect(slotCalls.map(c => c.key)).toContain('details')
   })
 
+  it('ignores unselected states and closes only when the Session id changes', () => {
+    const { frame, instance, rerenderFrame } = mountFrame()
+    expect(tracks(frame)).toEqual([280, 360])
+
+    selectedSession.current = 's-next' as SessionId
+    act(() => { rerenderFrame() })
+    expect(tracks(frame)).toEqual([280, 0])
+
+    act(() => { instance.actions.openDetails() })
+    selectedSession.current = 's-blank' as SessionId
+    selectedSessionBlank.current = true
+    act(() => { rerenderFrame() })
+    expect(tracks(frame)).toEqual([280, 0])
+    expect(instance.getSnapshot().details).toBe(360)
+
+    selectedSession.current = 's-next' as SessionId
+    selectedSessionBlank.current = false
+    act(() => { rerenderFrame() })
+    expect(tracks(frame)).toEqual([280, 360])
+
+    selectedSession.current = undefined
+    act(() => { rerenderFrame() })
+    expect(tracks(frame)).toEqual([280, 0])
+    selectedSession.current = 's-test' as SessionId
+    act(() => { rerenderFrame() })
+    expect(tracks(frame)).toEqual([280, 0])
+  })
+
+  it('keeps the default details width when the first Session materializes', () => {
+    selectedSession.current = undefined
+    const { frame, instance, rerenderFrame } = mountFrame()
+    expect(tracks(frame)).toEqual([280, 0])
+    expect(instance.getSnapshot().details).toBe(360)
+
+    selectedSession.current = 's-first' as SessionId
+    act(() => { rerenderFrame() })
+    expect(tracks(frame)).toEqual([280, 360])
+  })
+
   it('sidebar slot receives live concession output as owner props', () => {
     const { slotCalls } = mountFrame()
     expect(slotCalls.find(c => c.key === 'sidebar')!.props).toEqual({ collapsed: false, width: 280 })
diff --git a/packages/client/ui-layout/tests/layout-store.spec.ts b/packages/client/ui-layout/tests/layout-store.spec.ts
index e5938d0a1b..3ec3cb2c7e 100644
--- a/packages/client/ui-layout/tests/layout-store.spec.ts
+++ b/packages/client/ui-layout/tests/layout-store.spec.ts
@@ -1,8 +1,8 @@
 // @vitest-environment jsdom
 /**
  * createLayoutStore unit account: init shape, the action write set (clamp
- * inside actions), and the persist key round-trip over jsdom localStorage.
- * Uses the test-sanctioned path: factory self-call + .create() gives the
+ * inside actions), and the absence of browser persistence. Uses the
+ * test-sanctioned path: factory self-call + .create() gives the
  * real engine instance (same create path as production).
  */
 import { beforeEach, describe, expect, it } from 'vitest'
@@ -17,9 +17,9 @@ const PERSIST_KEY = 'dsh.layout.panels'
 beforeEach(() => { localStorage.clear() })
 
 describe('createLayoutStore', () => {
-  it('initializes with sidebar open at default and details closed', () => {
+  it('initializes both panels at their default widths', () => {
     const { store } = createLayoutStore().create()
-    expect(store.getSnapshot()).toEqual({ sidebar: SIDEBAR_DEFAULT, details: 0 })
+    expect(store.getSnapshot()).toEqual({ sidebar: SIDEBAR_DEFAULT, details: DETAILS_DEFAULT })
   })
 
   it('each create() is an independent instance (factory is not a singleton)', () => {
@@ -52,6 +52,7 @@ describe('createLayoutStore', () => {
 
   it('openDetails is a no-op when already open; closeDetails zeroes', () => {
     const { store, actions } = createLayoutStore().create()
+    actions.closeDetails()
     actions.openDetails()
     expect(store.getSnapshot().details).toBe(DETAILS_DEFAULT)
     actions.setDetails(500)
@@ -61,13 +62,16 @@ describe('createLayoutStore', () => {
     expect(store.getSnapshot().details).toBe(0)
   })
 
-  it('persists under dsh.layout.panels and rehydrates on the next create', () => {
+  it('does not persist panel geometry', () => {
     const first = createLayoutStore().create()
-    first.actions.setSidebar(320)
-    first.actions.openDetails()
-    expect(JSON.parse(localStorage.getItem(PERSIST_KEY) ?? '{}')).toEqual({ sidebar: 320, details: DETAILS_DEFAULT })
+    first.actions.setSidebar(400)
+    first.actions.closeDetails()
+    expect(localStorage.getItem(PERSIST_KEY)).toBeNull()
 
     const second = createLayoutStore().create()
-    expect(second.store.getSnapshot()).toEqual({ sidebar: 320, details: DETAILS_DEFAULT })
+    expect(second.store.getSnapshot()).toEqual({
+      sidebar: SIDEBAR_DEFAULT,
+      details: DETAILS_DEFAULT,
+    })
   })
 })
diff --git a/packages/client/ui-model/README.i18n.yaml b/packages/client/ui-model/README.i18n.yaml
index c45190753a..9620b15513 100644
--- a/packages/client/ui-model/README.i18n.yaml
+++ b/packages/client/ui-model/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-model/README.md
 README.md: 267717c78434f7a73b1c1eebca0cc0f9d65c3642
-README.zh.md: 325b1d93d99ed22e0945c26f5a3a9e5b3b209c85
+README.zh.md: 6d6f433315336812a51b5110ceeac3eecbd9bbd4
diff --git a/packages/client/ui-model/README.zh.md b/packages/client/ui-model/README.zh.md
index 325b1d93d9..6d6f433315 100644
--- a/packages/client/ui-model/README.zh.md
+++ b/packages/client/ui-model/README.zh.md
@@ -2,20 +2,20 @@
 
 [English](README.md) | 中文
 
-模型选择插件（浏览器半侧）：**两个入口共用一份 per-session 目录**，由 `ModelService`（`ctx.models`）持有。`/model` popupSelect contribution（经 `ctx.command` 注册）与 composer 的具名 `conversation.input.model` 坑位都通过同一个 `ModelDirectory` 实例，经 `session.models` 加载会话的建议目录，并经 `session.selectModel` 提交。紧凑型 composer 触发器会打开两级 Model/Effort 菜单：模型仍按提供方分组，所选确切模型则提供由其适配器持有的推理强度名称、说明和默认值。Host 报告的提供方／模型／推理（reasoning）目标是两个入口共同回显的唯一事实；`/model` 应用所选模型的默认推理强度，composer 随后可以选择任一已公布的推理强度。目录加载与选择共享一个代次计数器，旧响应不会覆盖新结果；连接重置会丢弃所有常驻目录投影，并在显示前重新拉取 Host 恢复的目标。逐提供方元数据失败会内联列出，同时可用分组仍可选择；选择失败会保留先前的目标和目录。目录按会话惰性解析（`ctx.models.directoryFor(sessionId)`），随会话 scope 一并释放。
+模型选择插件（浏览器侧）：**两个入口共用一份会话级目录**，由 `ModelService`（`ctx.models`）持有。`/model` popupSelect 贡献项（经 `ctx.command` 注册）与 composer 的具名 `conversation.input.model` 坑位都通过同一个 `ModelDirectory` 实例，经 `session.models` 加载会话的建议目录，并经 `session.selectModel` 提交。紧凑型 composer 触发器会打开两级 Model/Effort 菜单：模型仍按提供方分组，所选具体模型则提供由其适配器持有的推理强度名称、说明和默认值。Host 报告的提供方／模型／推理（reasoning）目标是两个入口共同回显的唯一事实；`/model` 应用所选模型的默认推理强度，composer 随后可以选择任一已公布的推理强度。目录加载与选择共享一个代次计数器，旧响应不会覆盖新结果；连接重置会丢弃所有常驻目录投影，并在显示前重新拉取 Host 恢复的目标。各提供方的元数据获取失败会内联列出，同时可用分组仍可选择；选择失败会保留先前的目标和目录。目录按会话惰性解析（`ctx.models.directoryFor(sessionId)`），随会话作用域一并释放。
 
 `/client` 导出面为插件本体（`apply`/`inject`）、`ModelService`、`ModelDirectory` 及其状态形状、坑位注入面类型。
 
-## Model Experience
+## 模型体验
 
-间接影响，经两个入口共同提交的 `session.selectModel` RPC：Host 在下一次提示词组装边界快照所选提供方／模型／推理强度目标，因此后续请求采用所选路由和推理强度，而运行中的步骤保留已组装目标。只有当现有请求头记录一次实际采用该选择的请求后，选择才会持久化；菜单交互不会添加提示词内容。
+间接影响，经两个入口共同提交的 `session.selectModel` RPC：Host 在下一次提示词组装边界快照所选提供方／模型／推理强度目标，因此下一次请求采用所选路由和推理强度，而运行中的步骤保留已组装目标。只有当现有请求头记录一次实际采用该选择的请求后，选择才会持久化；菜单交互不会添加提示词内容。
 
-#### KV Cache effect
+#### KV Cache 影响
 
-切换路由可能降低或作废提供方侧后续请求的缓存复用；提示词前缀本身不受影响。
+切换路由可能减少提供方侧后续请求的缓存复用，或使其失效；提示词前缀本身不受影响。
 
-## Known Limitations and Deferred Work
+## 已知限制与暂缓事项
 
-- **无创建期选择**——两个入口都寻址既有会话的 agent；没有 Draft 期模型选择折入会话创建的通道（host `targetFor` 处的种子序注释记录了该层未来的落点）。
-- **目录名仅供呈现**——选择与持久化使用提供方／模型／推理强度 id；目录查询或确切模型元数据查询失败的提供方以不可选失败行列出，重新加载前保持原样。
-- **不能任意输入推理强度**——composer 仅提供确切模型由适配器公布的推理强度；适配器没有推理元数据时不显示 Effort 行。
+- **无创建期选择**——两个入口都面向既有会话的 agent（智能体）；没有将草稿阶段的模型选择纳入会话创建的通道（host 的 `targetFor` 中的种子顺序说明了该层未来的落点）。
+- **目录名仅供呈现**——选择与持久化使用提供方／模型／推理强度 id；目录查询或具体模型元数据查询失败的提供方以不可选失败行列出，重新加载前保持原样。
+- **不能任意输入推理强度**——composer 仅提供具体模型由适配器公布的推理强度；适配器没有推理元数据时不显示 Effort 行。
diff --git a/packages/client/ui-models/README.i18n.yaml b/packages/client/ui-models/README.i18n.yaml
index 4ecb711730..fd750959f4 100644
--- a/packages/client/ui-models/README.i18n.yaml
+++ b/packages/client/ui-models/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/ui-models/README.md
 README.md: 13f51d5338affd65d0705cec6a3b4ef78a534f0f
-README.zh.md: 466505beb27c729246afe04e6378235b91d072cf
+README.zh.md: 90f4eb5959e105178844c7bd5b07596aff81e706
diff --git a/packages/client/ui-models/README.zh.md b/packages/client/ui-models/README.zh.md
index 466505beb2..90f4eb5959 100644
--- a/packages/client/ui-models/README.zh.md
+++ b/packages/client/ui-models/README.zh.md
@@ -10,7 +10,7 @@
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
diff --git a/packages/client/ui-permission/README.i18n.yaml b/packages/client/ui-permission/README.i18n.yaml
index f963bb9dda..12fef93f39 100644
--- a/packages/client/ui-permission/README.i18n.yaml
+++ b/packages/client/ui-permission/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-permission/README.md
-README.md: 0cd8e7f878a151ffacd749eb625afcb20d44ad93
-README.zh.md: 6bc299529c9795ef44cbe5429e78d6355c02a6ca
+README.md: 3377a1c5907b67b065879b012923427685c106d6
+README.zh.md: 34cf6f72394632968ded1671a5ac0377e5c78cc6
diff --git a/packages/client/ui-permission/README.md b/packages/client/ui-permission/README.md
index 0cd8e7f878..3377a1c590 100644
--- a/packages/client/ui-permission/README.md
+++ b/packages/client/ui-permission/README.md
@@ -2,7 +2,7 @@
 
 English | [中文](README.zh.md)
 
-Permission preset selection plugin, browser half: a popupSelect DECORATION hung on the host `/permission` command (`ctx.command.decorate`). A decoration is not a second command — the host command keeps its slash-menu row, the argued path (`/permission <preset>` switches directly), and the durable lifecycle logging; the decoration replaces only the bare invocation with the picker: one flat preset list with the current value marked active, where a pick submits the `/permission <preset>` command line. Options and the active mark read the session's `permissions` projection (the same host-computed select the composer chip renders), so both surfaces share one read source and one write path, and the pushed projection frame is the single confirmation both follow. The decoration is available exactly while the projection key is present; a permission-less composition shows no picker (a decoration never manufactures a catalog row).
+Permission preset selection plugin, browser half: a popupSelect DECORATION hung on the host `/permission` command (`ctx.command.decorate`). A decoration is not a second command — the host command keeps its slash-menu row, the argued path (`/permission <preset>` switches directly), and the durable lifecycle logging; the decoration replaces only the bare invocation with the picker: one flat preset list with the current value marked active and kebab-case preset names rendered as title-case labels (`workspace-write` → `Workspace Write`, the composer chip's display transform twin), where a pick submits the `/permission <preset>` command line. Options and the active mark read the session's `permissions` projection (the same host-computed select the composer chip renders), so both surfaces share one read source and one write path, and the pushed projection frame is the single confirmation both follow. The decoration is available exactly while the projection key is present; a permission-less composition shows no picker (a decoration never manufactures a catalog row).
 
 The `/client` export surface is the plugin body (`apply`/`inject`).
 
diff --git a/packages/client/ui-permission/README.zh.md b/packages/client/ui-permission/README.zh.md
index 6bc299529c..34cf6f7239 100644
--- a/packages/client/ui-permission/README.zh.md
+++ b/packages/client/ui-permission/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-权限预设选择插件（浏览器半侧）：挂在 host `/permission` 命令上的 popupSelect **装饰**（`ctx.command.decorate`）。装饰不是第二条命令——host 命令保留斜杠菜单行、带参路径（`/permission <preset>` 直接切换）与持久生命周期记账；装饰只把裸调用替换为选择框：一张扁平预设列表，当前值标记为 active，选中即提交 `/permission <preset>` 命令行。选项与 active 标记读取会话的 `permissions` 投影（与 composer chip 渲染的同一份 host 计算 select），因此两个界面共享同一读源与同一写路径，推送的投影帧是两者共同跟随的唯一确认。装饰恰在投影 key 存在时可用；无权限组合不显示选择框（装饰绝不无中生有目录行）。
+权限预设选择插件（浏览器半侧）：挂在 host `/permission` 命令上的 popupSelect **装饰**（`ctx.command.decorate`）。装饰不是第二条命令——host 命令保留斜杠菜单行、带参路径（`/permission <preset>` 直接切换）与持久生命周期记账；装饰只把裸调用替换为选择框：一张扁平预设列表，当前值标记为 active，kebab-case 预设名渲染为 Title Case 标签（`workspace-write` → `Workspace Write`，与 composer chip 的显示变换孪生），选中即提交 `/permission <preset>` 命令行。选项与 active 标记读取会话的 `permissions` 投影（与 composer chip 渲染的同一份 host 计算 select），因此两个界面共享同一读源与同一写路径，推送的投影帧是两者共同跟随的唯一确认。装饰恰在投影 key 存在时可用；无权限组合不显示选择框（装饰绝不无中生有目录行）。
 
 `/client` 导出面为插件本体（`apply`/`inject`）。
 
diff --git a/packages/client/ui-permission/src/client/index.ts b/packages/client/ui-permission/src/client/index.ts
index 42043ee008..30fc6d2dd5 100644
--- a/packages/client/ui-permission/src/client/index.ts
+++ b/packages/client/ui-permission/src/client/index.ts
@@ -23,13 +23,24 @@ function selectOf(session: SessionFace | undefined): PermissionSelect | undefine
   return session?.projections.faceOf('permissions').getSnapshot() as PermissionSelect | undefined
 }
 
+/**
+ * Display transform twin of the composer chip's (ui-conversation
+ * PermissionSelect): kebab-case machine names render as title-case labels
+ * (`workspace-write` → `Workspace Write`) so both permission surfaces show
+ * the same text; non-kebab host-configured names pass through.
+ */
+function displayName(name: string): string {
+  if (!/^[a-z0-9]+(-[a-z0-9]+)*$/.test(name)) return name
+  return name.split('-').map(word => word.charAt(0).toUpperCase() + word.slice(1)).join(' ')
+}
+
 /** Flatten the projection select into popup rows; `custom` is display state, never a target. */
 function optionsOf(value: PermissionSelect): SelectOption[] {
   return value.options
     .filter(option => option.value !== 'custom')
     .map(option => ({
       id: option.value,
-      label: option.name,
+      label: displayName(option.name),
       ...(option.description !== undefined ? { detail: option.description } : {}),
       ...(option.value === value.currentValue ? { active: true } : {}),
     }))
diff --git a/packages/client/ui-permission/tests/browser-plugin.spec.ts b/packages/client/ui-permission/tests/browser-plugin.spec.ts
index 167cf17362..5f9125db53 100644
--- a/packages/client/ui-permission/tests/browser-plugin.spec.ts
+++ b/packages/client/ui-permission/tests/browser-plugin.spec.ts
@@ -85,6 +85,11 @@ describe('ui-permission browser plugin', () => {
     const again = await c.ui.options(proj, new AbortController().signal)
     expect(again.find(option => option.id === 'workspace-write')?.active).toBe(true)
     expect(again.find(option => option.id === 'read-only')?.detail).toBe('Reads only.')
+    // Kebab-case names title-case; non-kebab host-configured names pass through.
+    expect(again.map(option => option.label)).toEqual(['Read Only', 'Workspace Write', 'Danger Full Access'])
+    b.values.set(sid('s1'), { ...SELECT, options: [{ value: 'plain', name: 'Ask Every Time' }] })
+    const passthrough = await c.ui.options(proj, new AbortController().signal)
+    expect(passthrough[0]?.label).toBe('Ask Every Time')
     // A projection that vanished between availability and open throws.
     expect(() => c.ui.options({ sessionId: sid('ghost') }, new AbortController().signal))
       .toThrow(/not available on this host/)
diff --git a/packages/client/ui-plan/README.i18n.yaml b/packages/client/ui-plan/README.i18n.yaml
index f7572c9bfa..199210a863 100644
--- a/packages/client/ui-plan/README.i18n.yaml
+++ b/packages/client/ui-plan/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-plan/README.md
-README.md: de43ce66d17498d31e05f8c64092ea0843103054
-README.zh.md: b4d2f4fd1a6d45f814d4a20195434f34d207e9c8
+README.md: 1d22c057b439ff337bf9daadcdba96dd4cca4540
+README.zh.md: 183b8ef7776b60c1f0afa630e04627a474d40391
diff --git a/packages/client/ui-plan/README.md b/packages/client/ui-plan/README.md
index de43ce66d1..1d22c057b4 100644
--- a/packages/client/ui-plan/README.md
+++ b/packages/client/ui-plan/README.md
@@ -4,7 +4,7 @@ English | [中文](README.zh.md)
 
 Plan-mode status chip, a pure browser surface plugin. The browser half occupies the conversation-declared `conversation.input.plan` single seat (to the right of the access-mode control); the node half is an empty apply (the roster row). Plan behavior itself — the `/plan` command, the boundary-or-idle-committed `plan/mode` state, the `plan` projection unit, and the policy section — is owned by [`@deepseek-ai/dsh-plan-mode`](../../plan/plan-mode/README.md), composed independently on the host roster.
 
-Plan mode is entered through the `/plan` command only; there is no UI control that turns it on. While the host-computed `plan` projection's effective target is plan mode (`pending ? !active : active` — a folded host value, not client optimism, so an arriving frame corrects the chip either way), the seat renders a read-only "Plan" chip whose hover × executes `/plan off` through `command.execute`; otherwise the seat stays empty — a host without plan-mode (or a Draft with no session) shows nothing. While plan mode is the effective target, the composer textarea's placeholder switches to "describe your task to generate plan" (rendered by the composer from the same projection; owner-supplied placeholders win).
+Plan mode is entered through the `/plan` command only; there is no UI control that turns it on. While the host-computed `plan` projection's effective target is plan mode (`pending ? !active : active` — a folded host value, not client optimism, so an arriving frame corrects the chip either way), the seat renders a read-only "Plan" chip whose hover × executes `/plan off` through `command.execute`; otherwise the seat stays empty — a host without plan-mode (or a Draft with no session) shows nothing. While plan mode is the effective target, the composer textarea's placeholder switches to the plan-task hint — "describe your task to generate plan", localized through ui-conversation's `command.hint` locale namespace and shared verbatim with the claimed `/plan` command hint (rendered by the composer from the same projection; owner-supplied placeholders win).
 
 The chip carries the accessible description "Plan mode on, press to turn off". Admission failures (`matched: false`, business errors, transport faults) surface as an inline error and the chip stays until the projection confirms the exit.
 
diff --git a/packages/client/ui-plan/README.zh.md b/packages/client/ui-plan/README.zh.md
index b4d2f4fd1a..183b8ef777 100644
--- a/packages/client/ui-plan/README.zh.md
+++ b/packages/client/ui-plan/README.zh.md
@@ -4,7 +4,7 @@
 
 Plan mode 状态徽章，纯浏览器 surface 插件。浏览器侧占据会话声明的 `conversation.input.plan` 单座（位于 access 模式控件右侧）；node 侧是空 apply（roster 行）。plan 行为本身——`/plan` 命令、边界或空闲即时提交的 `plan/mode` 状态、`plan` 投影单元与 policy 段——归 [`@deepseek-ai/dsh-plan-mode`](../../plan/plan-mode/README.md) 所有，由 host roster 独立组合。
 
-plan mode 只经 `/plan` 命令进入；UI 上没有打开它的控件。当 host 计算的 `plan` 投影有效目标为 plan mode 时（`pending ? !active : active`——折叠的 host 值而非客户端乐观态，帧到达即自动纠正），座位渲染一个只读 "Plan" chip，hover 出现的 × 经 `command.execute` 执行 `/plan off`；否则座位保持为空——未组合 plan-mode 的 host（或尚无会话的 Draft）不显示任何内容。plan mode 为有效目标期间，composer 文本框的 placeholder 切换为 "describe your task to generate plan"（由 composer 从同一投影渲染；owner 提供的 placeholder 优先）。
+plan mode 只经 `/plan` 命令进入；UI 上没有打开它的控件。当 host 计算的 `plan` 投影有效目标为 plan mode 时（`pending ? !active : active`——折叠的 host 值而非客户端乐观态，帧到达即自动纠正），座位渲染一个只读 "Plan" chip，hover 出现的 × 经 `command.execute` 执行 `/plan off`；否则座位保持为空——未组合 plan-mode 的 host（或尚无会话的 Draft）不显示任何内容。plan mode 为有效目标期间，composer 文本框的 placeholder 切换为 plan 任务提示——"describe your task to generate plan"（中文「描述你的任务以生成计划」），经 ui-conversation 的 `command.hint` locale 命名空间本地化，并与已认领 `/plan` 命令的提示逐字共用同一份文案（由 composer 从同一投影渲染；owner 提供的 placeholder 优先）。
 
 chip 携带无障碍描述 "Plan mode on, press to turn off"。准入失败（`matched: false`、业务错误、传输故障）以内联错误呈现，chip 保持显示直至投影确认退出。
 
diff --git a/packages/client/ui-plan/src/client/PlanModeControl.module.css b/packages/client/ui-plan/src/client/PlanModeControl.module.css
index 4bf42bd2a1..9e7e11d828 100644
--- a/packages/client/ui-plan/src/client/PlanModeControl.module.css
+++ b/packages/client/ui-plan/src/client/PlanModeControl.module.css
@@ -39,13 +39,10 @@
   display: inline-flex;
   align-items: center;
   color: var(--dsw-alias-label-caption);
-  opacity: 0;
-  transition: opacity 0.12s ease;
 }
 
 .chip:hover .close,
 .chip:focus-visible .close {
-  opacity: 1;
   color: var(--dsw-alias-label-secondary);
 }
 
diff --git a/packages/client/ui-primitives/README.i18n.yaml b/packages/client/ui-primitives/README.i18n.yaml
index 6162494def..b5e4b5c078 100644
--- a/packages/client/ui-primitives/README.i18n.yaml
+++ b/packages/client/ui-primitives/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 58e450451ab64f69762817dfb277b8a888e2177f
-README.zh.md: 6824f3efe4981adf9549941afa7e2f5db2ac005d
+#   pnpm run verify-translation-pairing --write packages/client/ui-primitives/README.md
+README.md: 0ef3c20f848b3d331c007911d0837f11cd72c024
+README.zh.md: af94551bfb9e12dbadcef6a96a54f9bf7ea71299
diff --git a/packages/client/ui-primitives/README.md b/packages/client/ui-primitives/README.md
index 58e450451a..0ef3c20f84 100644
--- a/packages/client/ui-primitives/README.md
+++ b/packages/client/ui-primitives/README.md
@@ -2,11 +2,15 @@
 
 English | [中文](README.zh.md)
 
-Pure React atoms (zero cordis): StateDot, ic_ds_* icons, Button/Pill/Menu/Modal/Input, markdown family (MessageText/MarkdownText/JsonBlock). Contract: api-contracts v3 §8.
+Pure React atoms (zero cordis): StateDot, ic_ds_* icons, Button/Pill/Menu/Modal/Input, the markdown family (MessageText/MarkdownText/JsonBlock), the read-only JsonTree inspector, the `useAnchoredMaxHeight` hook that clamps a bottom-anchored overlay to the viewport space above its anchor (re-measured on resize, scroll, and a caller-supplied dependency), and TerminalBlock. Contract: api-contracts v3 §8.
 
 ## Markdown rendering
 
-`MarkdownText` renders GFM from untrusted assistant output through React elements. It omits raw HTML, neutralizes relative and non-HTTP(S)/mailto links, opens HTTP(S) links with safe external-link attributes, and renders image alt text without loading remote resources; `MessageText` remains the literal-text primitive for user-authored content. Element spacing, tables, links, and inline code use the same `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` tokens as deepsuite `@deepseek/md`. Fenced blocks render through `CodeBlock` (language banner, copy control, shiki for the registered grammars).
+`MarkdownText` renders GFM from untrusted assistant output through React elements. It omits raw HTML, neutralizes relative and non-HTTP(S)/mailto links, opens HTTP(S) links with safe external-link attributes, and renders image alt text without loading remote resources; `MessageText` remains the literal-text primitive for user-authored content. `extractMarkdownPlainText` removes Markdown presentation markup for compact labels while preserving raw HTML as literal text. Element spacing, tables, links, and inline code use the same `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` tokens as deepsuite `@deepseek/md`. Fenced blocks render through `CodeBlock` (language banner, copy control, shiki for the registered grammars).
+
+## Terminal output
+
+`TerminalBlock` renders a shell command as a terminal surface: one prompt row per line of the command (the shortened `cwd` label on the first row only, since the view knows one working directory and a `cd` moves later lines elsewhere, then that line), the command's output, a status pill for a non-zero exit code or a terminating signal, and a copy control that writes the raw `output` prop. A run-state `StateDot` marks the call once, on the first row, out of flow in a gutter the card reserves as its own left padding, so the dot sits inside the card box yet left of the prompt text. It reaches three of `StateDot`'s states — the chase while `running`, red for the same exit status that renders the pill, green otherwise — so a card states whether its command is still running rather than leaving that to be inferred from the presence of output; it carries one visually hidden text label because `StateDot` is `aria-hidden`. One dot regardless of line count is deliberate: the exit status is the whole call's, so a dot per line would claim a per-line outcome the view does not carry. Command text is `white-space: pre`, so repeated spaces, tabs, and an indented continuation render verbatim while the row stays single-line and ellipsizes. ANSI escape sequences are parsed with the `anser` runtime dependency into React spans; cursor movements replay into a per-line column buffer before inert controls are stripped, since carriage return and backspace only MOVE the cursor: `100%` + CR + `OK` alone shows `OK0%`, while the `\x1b[K` a spinner writes with its redraw erases the tail so `100%\r\x1b[KOK` shows `OK`. Erase-in-line is honored in all three parameter forms, the cursor advances by terminal columns (8-column tab stops, two for emoji and CJK, none for a combining mark), and SGR state is normalized per cell as a terminal stores it, threading across lines and closing at the state the line ended in; basic-16 foreground colors map onto `--dsw-*` tokens, while 256-palette and truecolor values pass through as literal rgb. Output keeps `white-space: pre` with horizontal scrolling, so column-aligned output holds its alignment instead of soft-wrapping, and collapses to a head slice plus a tail slice past `maxLines` (default 16, the TUI transcript's split arithmetic) behind an expand button. Rationale: [the web terminal card note](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md).
 
 ## Model Experience
 
@@ -21,3 +25,5 @@ None; this package neither assembles nor sends a provider request.
 - **Glyph-level icons are redrawn approximations** — the fish logo (and the sparkle held by ui-conversation) come from font glyphs whose vector geometry is not exportable from the local design data; hand-authored recreations stand in until an exact export path exists.
 - **Pill and Input have no design source** — both atoms are self-defined; the sidebar search field and view-tab strip that resemble them are consumer-owned compositions, not these atoms.
 - **StateDot `Active` variant is a hidden placeholder in the design** — not implemented; the four shipped states (done/warning/ongoing/error) are the complete P-I surface.
+- **This package's user-facing copy is inline Chinese, not localized** — the atoms are zero-cordis and so cannot reach `ctx.locale`; `TerminalBlock`'s exit-code and signal pills, its copy and expand controls, and `CodeBlock`'s copy control are all hardcoded. This matches the repo-wide state the locale package records (only the Settings surface is translated); extracting these into the `zh`/`en` dictionaries needs a localization channel for zero-cordis atoms and belongs to that repo-wide extraction.
+- **`TerminalBlock` is not a terminal emulator** — it renders settled or still-running command output, not an interactive session: SGR color and attributes are honored, and so are the in-line cursor movements a progress line uses — carriage return, backspace, erase-in-line, tab stops and character width. Absolute cursor positioning, screen clearing, and alternate-screen sequences are stripped. Basic-16 magenta and cyan have no token equivalent and stay literal rgb.
diff --git a/packages/client/ui-primitives/README.zh.md b/packages/client/ui-primitives/README.zh.md
index 6824f3efe4..af94551bfb 100644
--- a/packages/client/ui-primitives/README.zh.md
+++ b/packages/client/ui-primitives/README.zh.md
@@ -2,15 +2,18 @@
 
 [English](README.md) | 中文
 
-纯 React 原子组件（零 cordis）：StateDot、ic_ds_* 图标、Button/Pill/Menu/Modal/Input，以及 markdown 家族（MessageText/MarkdownText/JsonBlock）。契约：api-contracts v3 §8。
+纯 React 原子组件（零 cordis）：StateDot、ic_ds_* 图标、Button/Pill/Menu/Modal/Input、markdown 家族（MessageText/MarkdownText/JsonBlock）、只读 JsonTree 检查器、`useAnchoredMaxHeight` hook（把底部锚定的浮层高度收敛到锚点上方的视口空间，并在 resize、scroll 与调用方提供的依赖变化时重新测量），以及 TerminalBlock。契约：api-contracts v3 §8。
 
 ## Markdown 渲染
 
-`MarkdownText` 通过 React 元素渲染来自不受信任 assistant 输出的 GFM。它会省略原始 HTML，使相对链接及非 HTTP(S)/mailto 链接失效，以安全的外部链接属性打开 HTTP(S) 链接，并只渲染图片 alt 文本而不加载远程资源；`MessageText` 仍是用户创作内容使用的字面文本原语。元素间距、表格、链接与行内代码使用与 deepsuite `@deepseek/md` 相同的 `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` token。围栏代码块通过 `CodeBlock` 渲染（语言横幅、复制控件，以及对已注册语法使用 shiki）。
+`MarkdownText` 通过 React 元素渲染来自不受信任 assistant 输出的 GFM。它会省略原始 HTML，使相对链接及非 HTTP(S)/mailto 链接失效，以安全的外部链接属性打开 HTTP(S) 链接，并只渲染图片 alt 文本而不加载远程资源；`MessageText` 仍是用户创作内容使用的字面文本原语。`extractMarkdownPlainText` 会移除 Markdown 呈现标记以用于紧凑标签，同时将原始 HTML 保留为字面文本。元素间距、表格、链接与行内代码使用与 deepsuite `@deepseek/md` 相同的 `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` token。围栏代码块通过 `CodeBlock` 渲染（语言横幅、复制控件，以及对已注册语法使用 shiki）。
+## 终端输出
+
+`TerminalBlock` 将一条 shell 命令渲染为终端表层：命令的每一行各占一个提示行（缩短后的 `cwd` 标签只出现在第一行，因为视图只知道一个工作目录，而一个 `cd` 就会让后面的行去到别处，标签之后是该行）、命令输出、非零退出码或终止信号对应的状态胶囊，以及写入原始 `output` prop 的复制控件。一枚运行状态 `StateDot` 为整次调用标记一次，位于第一行，以脱离文档流的方式落在卡片以自身左内边距预留的落区中，因此它位于卡片盒之内、提示文字之左。它用到 `StateDot` 的三种状态——`running` 期间为追逐动画，与渲染状态胶囊相同的退出状态为红色，其余为绿色——因此卡片直接陈述其命令是否仍在运行，而不是让人从有无输出中推断；由于 `StateDot` 是 `aria-hidden`，它携带一处视觉隐藏的文本标签。无论多少行都只有一枚状态点是有意为之：退出状态属于整次调用，因此每行一枚就会声称一个视图并不携带的逐行结果。命令文本使用 `white-space: pre`，因此重复空格、制表符与缩进续行都原样呈现，同时该行仍保持单行并以省略号截断。ANSI 转义序列通过运行时依赖 `anser` 解析为 React span；光标移动在剥除无显示意义控制符之前先重放进逐行的列缓冲，因为回车与退格**只移动**光标：单是 `100%` 加回车再加 `OK` 显示为 `OK0%`，而 spinner 随重绘写出的 `\x1b[K` 会擦掉尾巴，因此 `100%\r\x1b[KOK` 显示为 `OK`。行内擦除的三种参数形式都被遵循，光标按终端列推进（8 列制表位；emoji 与 CJK 占两列；组合标记不占列），SGR 状态按单元格归一化存储，与终端一致，并跨行延续、在行结束时的状态处收束；基础 16 色前景色映射到 `--dsw-*` token，而 256 色板与真彩色值按字面 rgb 透传。输出保持 `white-space: pre` 并支持横向滚动，因此按列对齐的输出保留其对齐而不会软换行；超过 `maxLines`（默认 16，与 TUI 转录相同的切分算法）时折叠为头部切片加尾部切片，由展开按钮控制。原理：[Web 终端卡片笔记](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)。
 
 ## 模型体验
 
-无。该包在浏览器中渲染纯 React 原子组件；这里没有任何内容进入模型请求。
+无。该包（package）在浏览器中渲染纯 React 原子组件；这里没有任何内容进入模型请求。
 
 #### KV Cache 影响
 
@@ -21,3 +24,5 @@
 - **字形级图标是重新绘制的近似版本**：鱼形标志（以及 ui-conversation 持有的闪光图标）来自字体字形，而本地设计数据无法导出其矢量几何；在获得精确导出路径前，使用手工重建版本代替。
 - **Pill 与 Input 没有设计来源**：两个原子组件均自行定义；与其相似的侧边栏搜索字段和视图标签条由消费方组合，不是这些原子组件。
 - **StateDot 的 `Active` 变体是设计中的隐藏占位符**：尚未实现；已交付的四种状态（done/warning/ongoing/error）构成完整的 P-I 表层。
+- **本包面向用户的文案是内联中文，未做本地化**：这些原子组件是 zero-cordis 的，因此拿不到 `ctx.locale`；`TerminalBlock` 的退出码与信号胶囊、它的复制与展开控件，以及 `CodeBlock` 的复制控件全部硬编码。这与 locale 包记录的全仓现状一致（只有 Settings 表面做了翻译）；把它们抽取进 `zh`/`en` 字典需要为 zero-cordis 原子组件提供一条本地化通道，属于那次全仓抽取的范围。
+- **`TerminalBlock` 不是终端模拟器**：它渲染已结束或仍在运行的命令输出，而不是交互式会话：SGR 颜色与属性会被遵循，进度行所用的行内光标移动同样被遵循——回车、退格、行内擦除、制表位与字符宽度。绝对光标定位、清屏与备用屏幕序列会被剥离。基础 16 色中的洋红与青色没有对应 token，保持字面 rgb。
diff --git a/packages/client/ui-primitives/package.json b/packages/client/ui-primitives/package.json
index 9ce2bc8676..2ac25a6708 100644
--- a/packages/client/ui-primitives/package.json
+++ b/packages/client/ui-primitives/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@deepseek-ai/dsh-client-ui-primitives",
-  "description": "Pure React atoms for the dsh web UI: StateDot, ic_ds_* icon set, Button/Pill/Menu/Modal/Input, markdown family (zero cordis)",
+  "description": "Pure React atoms for the dsh web UI: controls, icons, markdown, and JSON inspectors (zero cordis)",
   "version": "0.0.1",
   "private": true,
   "type": "module",
@@ -21,7 +21,11 @@
   "license": "BSD-3-Clause",
   "dependencies": {
     "@shikijs/langs": "^4.3.1",
+    "anser": "^2.3.5",
     "clsx": "^2.0.0",
+    "mdast-util-from-markdown": "^2.0.3",
+    "mdast-util-gfm": "^3.1.0",
+    "micromark-extension-gfm": "^3.0.0",
     "react": "^18.2.0",
     "react-dom": "^18.2.0",
     "react-markdown": "^10.1.0",
diff --git a/packages/client/ui-primitives/src/JsonTree.module.css b/packages/client/ui-primitives/src/JsonTree.module.css
new file mode 100644
index 0000000000..33ec3f30d3
--- /dev/null
+++ b/packages/client/ui-primitives/src/JsonTree.module.css
@@ -0,0 +1,222 @@
+.root {
+  --json-tree-property: #881391;
+  --json-tree-string: #c41a16;
+  --json-tree-number: #1c00cf;
+  --json-tree-keyword: #1c00cf;
+  --json-tree-punctuation: #202124;
+  --json-tree-icon: #5f6368;
+  --json-tree-hover: rgb(60 64 67 / 4%);
+
+  min-width: 0;
+  overflow: auto;
+  position: relative;
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-bg-layer-1);
+  font: 12px/16px var(--ds-font-family-code);
+  overscroll-behavior-x: contain;
+  overscroll-behavior-y: auto;
+}
+
+:global(body[data-ds-dark-theme]) .root {
+  --json-tree-property: #5db0d7;
+  --json-tree-string: #f28b82;
+  --json-tree-number: #99c8ff;
+  --json-tree-keyword: #99c8ff;
+  --json-tree-punctuation: #e8eaed;
+  --json-tree-icon: #9aa0a6;
+  --json-tree-hover: rgb(232 234 237 / 5%);
+}
+
+.container {
+  box-sizing: border-box;
+  width: max-content;
+  min-width: 100%;
+  margin: 0;
+  padding: 6px 8px 8px;
+  white-space: pre;
+}
+
+.expandedTopLevel {
+  box-sizing: border-box;
+  width: max-content;
+  min-width: 100%;
+  padding: 6px 8px 8px 14px;
+}
+
+.expandedTopLevel:has(> .topLevelBracket[data-json-root-row]:hover),
+.expandedTopLevel:has(> .topLevelBracket[data-json-root-row][data-json-copy-active]) {
+  background: var(--json-tree-hover);
+}
+
+.expandedTopLevelContainer {
+  padding: 0;
+}
+
+.row.topLevelBracket {
+  margin-left: 0;
+  padding-left: 0;
+}
+
+.children {
+  margin: 0;
+  padding: 0;
+  list-style: none;
+}
+
+.row {
+  position: relative;
+  box-sizing: border-box;
+  min-width: 100%;
+  min-height: 16px;
+  margin: 0;
+  padding: 0 0 0 10px;
+  list-style: none;
+}
+
+.row:not(.topLevelBracket):hover:not(:has(.row:hover))::after,
+.row:not(.topLevelBracket)[data-json-copy-active]::after,
+.row:has(> .expander:focus-visible)::after {
+  position: absolute;
+  z-index: 0;
+  top: 0;
+  right: 0;
+  left: 0;
+  height: 16px;
+  background: var(--json-tree-hover);
+  content: '';
+  pointer-events: none;
+}
+
+.row > span:not(.expander) {
+  position: relative;
+  z-index: 1;
+}
+
+.label {
+  margin-right: 3px;
+  color: var(--json-tree-property);
+  font-weight: 400;
+}
+
+.clickableLabel {
+  cursor: pointer;
+}
+
+.stringValue {
+  color: var(--json-tree-string);
+}
+
+.numberValue {
+  color: var(--json-tree-number);
+}
+
+.keywordValue {
+  color: var(--json-tree-keyword);
+}
+
+.otherValue {
+  color: var(--dsw-alias-label-secondary);
+}
+
+.punctuation {
+  color: var(--json-tree-punctuation);
+}
+
+.preview {
+  color: var(--json-tree-punctuation);
+}
+
+.previewProperty {
+  color: var(--json-tree-punctuation);
+}
+
+.previewEllipsis {
+  color: var(--dsw-alias-label-tertiary);
+}
+
+.copyAnchor {
+  position: fixed;
+  z-index: 3;
+  display: inline-flex;
+}
+
+.copyButton {
+  box-sizing: border-box;
+  display: inline-flex;
+  align-items: center;
+  justify-content: center;
+  width: 20px;
+  height: 16px;
+  margin: 0;
+  padding: 0;
+  border: 0;
+  border-radius: 3px;
+  color: var(--dsw-alias-label-secondary);
+  background: var(--dsw-alias-bg-layer-1);
+  box-shadow: -5px 0 5px var(--dsw-alias-bg-layer-1);
+  cursor: pointer;
+}
+
+.copyButton:hover {
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.copyButton:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: -1px;
+}
+
+.copyButton[data-state='failed'] {
+  color: var(--dsw-alias-state-error-primary);
+}
+
+.expander {
+  position: absolute;
+  z-index: 2;
+  top: 0;
+  left: 0;
+  display: inline-flex;
+  align-items: center;
+  justify-content: flex-start;
+  box-sizing: border-box;
+  width: 8px;
+  height: 16px;
+  margin: 0;
+  color: var(--json-tree-icon);
+  cursor: pointer;
+  user-select: none;
+}
+
+.expander::before {
+  width: 0;
+  height: 0;
+  border-top: 4px solid transparent;
+  border-bottom: 4px solid transparent;
+  border-left: 6px solid currentColor;
+  content: '';
+  transform: scale(0.75);
+  transform-origin: 33.333% center;
+}
+
+.collapseIcon::before {
+  transform: rotate(90deg) scale(0.75);
+}
+
+.expander:hover {
+  color: var(--dsw-alias-label-primary);
+}
+
+.expander:focus-visible {
+  outline: none;
+}
+
+.collapsedContent {
+  margin: 0 1px;
+  color: var(--json-tree-punctuation);
+  cursor: pointer;
+}
+
+.collapsedContent::after {
+  content: '…';
+}
diff --git a/packages/client/ui-primitives/src/JsonTree.tsx b/packages/client/ui-primitives/src/JsonTree.tsx
new file mode 100644
index 0000000000..1bb8fa62cc
--- /dev/null
+++ b/packages/client/ui-primitives/src/JsonTree.tsx
@@ -0,0 +1,602 @@
+import clsx from 'clsx'
+import { useEffect, useId, useRef, useState } from 'react'
+import type {
+  KeyboardEvent as ReactKeyboardEvent,
+  MouseEvent as ReactMouseEvent,
+  ReactNode,
+  UIEvent as ReactUIEvent,
+} from 'react'
+import { IconCheckOutline16, IconCopyOutline16 } from './icons/index.tsx'
+import { Menu } from './Menu.tsx'
+import type { MenuEntry } from './Menu.tsx'
+import css from './JsonTree.module.css'
+
+const OBJECT_PREVIEW_LIMIT = 4
+const ARRAY_PREVIEW_LIMIT = 5
+const PREVIEW_DEPTH_LIMIT = 2
+const VALUE_COPY_MENU_ITEMS: readonly MenuEntry[] = [
+  { id: 'value', label: 'Copy value' },
+  { id: 'json', label: 'Copy JSON' },
+  { id: 'path', label: 'Copy property path' },
+]
+const OBJECT_COPY_MENU_ITEMS: readonly MenuEntry[] = [
+  { id: 'prettyJson', label: 'Copy pretty JSON' },
+  { id: 'json', label: 'Copy compact JSON' },
+  { id: 'path', label: 'Copy property path' },
+]
+
+type JsonPath = readonly (number | string)[]
+
+interface RowTarget {
+  path: JsonPath
+  value: unknown
+}
+
+interface CopyTarget extends RowTarget {
+  left: number
+  side: 'bottom' | 'top'
+  top: number
+}
+
+function isExpandableValue(value: unknown): value is object | unknown[] {
+  return typeof value === 'object' && value !== null && !(value instanceof Date)
+}
+
+function entriesOf(value: object | unknown[]): readonly (readonly [string, unknown])[] {
+  if (Array.isArray(value)) {
+    return value.map((item, index) => [String(index), item] as const)
+  }
+  return Object.keys(value).map(key => [
+    key,
+    (value as Record<string, unknown>)[key],
+  ] as const)
+}
+
+function bracketOf(value: object | unknown[]): readonly [string, string] {
+  return Array.isArray(value) ? ['[', ']'] : ['{', '}']
+}
+
+function previewPrimitive(value: unknown): ReactNode {
+  if (value === null) return <span className={css.keywordValue}>null</span>
+  if (typeof value === 'string') {
+    return <span className={css.stringValue}>{JSON.stringify(value)}</span>
+  }
+  if (typeof value === 'number') {
+    return <span className={css.numberValue}>{String(value)}</span>
+  }
+  if (typeof value === 'boolean') {
+    return <span className={css.keywordValue}>{String(value)}</span>
+  }
+  if (typeof value === 'bigint') {
+    return <span className={css.otherValue}>{value.toString()}</span>
+  }
+  if (typeof value === 'undefined') {
+    return <span className={css.otherValue}>undefined</span>
+  }
+  if (typeof value === 'symbol') {
+    return <span className={css.otherValue}>{value.description ?? 'Symbol'}</span>
+  }
+  if (typeof value === 'function') {
+    return <span className={css.otherValue}>{value.name || 'Function'}</span>
+  }
+  return null
+}
+
+function previewValue(value: unknown, depth: number): ReactNode {
+  if (!isExpandableValue(value)) return previewPrimitive(value)
+
+  const array = Array.isArray(value)
+  const entries = entriesOf(value)
+  const limit = array ? ARRAY_PREVIEW_LIMIT : OBJECT_PREVIEW_LIMIT
+  const visible = entries.slice(0, limit)
+  const [open, close] = bracketOf(value)
+
+  return (
+    <>
+      <span className={css.punctuation}>{open}</span>
+      {depth >= PREVIEW_DEPTH_LIMIT
+        ? <span className={css.previewEllipsis}>…</span>
+        : visible.map(([key, item], index) => (
+          <span key={key}>
+            {index > 0 && <span className={css.punctuation}>, </span>}
+            {!array && (
+              <>
+                <span className={css.previewProperty}>{key}</span>
+                <span className={css.punctuation}>: </span>
+              </>
+            )}
+            {previewValue(item, depth + 1)}
+          </span>
+        ))}
+      {depth < PREVIEW_DEPTH_LIMIT && entries.length > limit && (
+        <span className={css.previewEllipsis}>, …</span>
+      )}
+      <span className={css.punctuation}>{close}</span>
+    </>
+  )
+}
+
+function primitiveValue(value: unknown): ReactNode {
+  if (value === null) return <span className={css.keywordValue}>null</span>
+  if (typeof value === 'string') {
+    return <span className={css.stringValue}>{JSON.stringify(value)}</span>
+  }
+  if (typeof value === 'boolean') {
+    return <span className={css.keywordValue}>{String(value)}</span>
+  }
+  if (typeof value === 'number') {
+    return <span className={css.numberValue}>{String(value)}</span>
+  }
+  if (typeof value === 'bigint') {
+    return <span className={css.numberValue}>{`${value.toString()}n`}</span>
+  }
+  if (value instanceof Date) {
+    return <span className={css.otherValue}>{value.toISOString()}</span>
+  }
+  if (typeof value === 'function') {
+    return <span className={css.otherValue}>function() {'{ }'}</span>
+  }
+  if (typeof value === 'undefined') {
+    return <span className={css.otherValue}>undefined</span>
+  }
+  return <span className={css.otherValue}>{(value as symbol).toString()}</span>
+}
+
+function fieldText(field: string): string {
+  return field === '' ? '""' : field
+}
+
+function pathId(path: JsonPath): string {
+  return path.map(part => (
+    typeof part === 'number' ? `n${String(part)}` : `s${String(part.length)}:${part}`
+  )).join('/')
+}
+
+function claimFocus(button: HTMLElement): void {
+  button.focus()
+}
+
+function moveFocus(button: HTMLElement, direction: -1 | 1): void {
+  const tree = button.closest<HTMLElement>('[role="tree"]')
+  /* v8 ignore next -- JsonTree attaches expander handlers only beneath its owning role=tree. */
+  if (tree === null) return
+  const expanders = Array.from(tree.querySelectorAll<HTMLElement>('[data-json-expander]'))
+  const current = expanders.indexOf(button)
+  /* v8 ignore next -- the current expander is a member of the queried non-empty set. */
+  if (current < 0 || expanders.length === 0) return
+  const next = (current + direction + expanders.length) % expanders.length
+  const nextExpander = expanders[next]
+  /* v8 ignore next -- modulo over the non-empty expander set always resolves a member. */
+  if (nextExpander !== undefined) claimFocus(nextExpander)
+}
+
+function NodeField({
+  field,
+  expandable,
+  onToggle,
+}: {
+  field: string | undefined
+  expandable: boolean
+  onToggle: () => void
+}) {
+  if (field === undefined) return null
+  return (
+    <span
+      className={clsx(css.label, expandable && css.clickableLabel)}
+      onClick={expandable ? onToggle : undefined}
+    >
+      {fieldText(field)}:
+    </span>
+  )
+}
+
+interface JsonTreeNodeProps {
+  field?: string
+  initialExpanded: boolean
+  lastElement: boolean
+  onClaimTabStop: (id: string) => void
+  onRowHover: (row: HTMLElement, target: RowTarget) => void
+  path: JsonPath
+  tabStopId: string | null
+  value: unknown
+}
+
+function JsonTreeNode({
+  field,
+  initialExpanded,
+  lastElement,
+  onClaimTabStop,
+  onRowHover,
+  path,
+  tabStopId,
+  value,
+}: JsonTreeNodeProps) {
+  const contentsId = useId()
+  const expanderRef = useRef<HTMLSpanElement>(null)
+  const [expanded, setExpanded] = useState(initialExpanded)
+  const nodeId = pathId(path)
+  const container = isExpandableValue(value)
+  const entries = container ? entriesOf(value) : []
+  const expandable = entries.length > 0
+
+  const toggle = () => {
+    setExpanded(current => !current)
+    claimFocus(expanderRef.current as HTMLSpanElement)
+  }
+
+  const onExpanderKeyDown = (event: ReactKeyboardEvent<HTMLSpanElement>) => {
+    if (event.key === 'ArrowRight' || event.key === 'ArrowLeft') {
+      event.preventDefault()
+      setExpanded(event.key === 'ArrowRight')
+      return
+    }
+    if (event.key === 'ArrowUp' || event.key === 'ArrowDown') {
+      event.preventDefault()
+      moveFocus(event.currentTarget, event.key === 'ArrowUp' ? -1 : 1)
+    }
+  }
+
+  const row = (children: ReactNode, ariaExpanded?: boolean) => (
+    <div
+      className={css.row}
+      role="treeitem"
+      aria-expanded={ariaExpanded}
+      onMouseOver={(event) => {
+        event.stopPropagation()
+        onRowHover(event.currentTarget, { path, value })
+      }}
+    >
+      {children}
+    </div>
+  )
+
+  if (!container) {
+    return row((
+      <>
+        <NodeField field={field} expandable={false} onToggle={toggle} />
+        {primitiveValue(value)}
+        {!lastElement && <span className={css.punctuation}>,</span>}
+      </>
+    ))
+  }
+
+  const [open, close] = bracketOf(value)
+  if (!expandable) {
+    return row((
+      <>
+        <NodeField field={field} expandable={false} onToggle={toggle} />
+        <span className={css.punctuation}>{open}</span>
+        <span className={css.punctuation}>{close}</span>
+        {!lastElement && <span className={css.punctuation}>,</span>}
+      </>
+    ))
+  }
+
+  return row((
+    <>
+      <span
+        ref={expanderRef}
+        className={clsx(css.expander, expanded ? css.collapseIcon : css.expandIcon)}
+        data-json-expander
+        role="button"
+        aria-label={expanded ? 'Collapse JSON node' : 'Expand JSON node'}
+        aria-expanded={expanded}
+        aria-controls={expanded ? contentsId : undefined}
+        tabIndex={tabStopId === nodeId ? 0 : -1}
+        onFocus={() => { onClaimTabStop(nodeId) }}
+        onClick={toggle}
+        onKeyDown={onExpanderKeyDown}
+      />
+      <NodeField field={field} expandable onToggle={toggle} />
+      <span className={css.preview}>{previewValue(value, 0)}</span>
+      {!lastElement && <span className={css.punctuation}>,</span>}
+      {expanded && (
+        <ul id={contentsId} role="group" className={css.children}>
+          {entries.map(([key, item], index) => (
+            <JsonTreeNode
+              key={key}
+              field={key}
+              value={item}
+              path={[...path, Array.isArray(value) ? index : key]}
+              lastElement={index === entries.length - 1}
+              initialExpanded={false}
+              tabStopId={tabStopId}
+              onClaimTabStop={onClaimTabStop}
+              onRowHover={onRowHover}
+            />
+          ))}
+        </ul>
+      )}
+    </>
+  ), expanded)
+}
+
+function formattedPath(path: JsonPath): string {
+  return path.reduce<string>((result, part) => {
+    if (typeof part === 'number') return `${result}[${String(part)}]`
+    return /^[A-Za-z_$][\w$]*$/.test(part)
+      ? `${result}.${part}`
+      : `${result}[${JSON.stringify(part)}]`
+  }, '$')
+}
+
+function copyText(target: CopyTarget, mode: 'json' | 'path' | 'prettyJson' | 'value'): string {
+  if (mode === 'path') return formattedPath(target.path)
+  if (mode === 'prettyJson') return JSON.stringify(target.value, null, 2)
+  if (mode === 'json') return JSON.stringify(target.value)
+  if (typeof target.value === 'string') return target.value
+  if (typeof target.value === 'undefined') return 'undefined'
+  if (typeof target.value === 'bigint') return target.value.toString()
+  if (typeof target.value === 'symbol') return target.value.description ?? 'Symbol'
+  if (typeof target.value === 'function') return target.value.name || 'Function'
+  return JSON.stringify(target.value)
+}
+
+/** Props for the read-only, token-themed JSON tree. */
+export interface JsonTreeProps {
+  /** Parsed JSON object or array. */
+  data: object | unknown[]
+  /** Accessible label for the tree. */
+  label?: string
+  /** Optional positioning class owned by the caller. */
+  className?: string | undefined
+  /** Whether JSON rows expose copy actions. */
+  copyable?: boolean
+  /** Whether the top-level object or array is always expanded. */
+  expandTopLevel?: boolean
+}
+
+/**
+ * Render parsed JSON as a compact, keyboard-accessible inspector tree.
+ * @param props - Parsed data, accessible label, and display options.
+ * @returns A read-only JSON tree with an optionally fixed-open top level.
+ */
+export function JsonTree({
+  data,
+  label = 'JSON',
+  className,
+  copyable = true,
+  expandTopLevel = true,
+}: JsonTreeProps) {
+  const rootEntries = entriesOf(data)
+  const firstExpandableIndex = rootEntries.findIndex(([, value]) => (
+    isExpandableValue(value) && entriesOf(value).length > 0
+  ))
+  const firstExpandableEntry = rootEntries[firstExpandableIndex]
+  const initialTabStopId = expandTopLevel
+    ? firstExpandableEntry === undefined
+      ? null
+      : pathId([Array.isArray(data) ? firstExpandableIndex : firstExpandableEntry[0]])
+    : isExpandableValue(data) && rootEntries.length > 0 ? pathId([]) : null
+  const rootRef = useRef<HTMLDivElement>(null)
+  const activeRowRef = useRef<HTMLElement>()
+  const copyButtonRef = useRef<HTMLButtonElement>(null)
+  const copyMenuOpenRef = useRef(false)
+  const resetTimer = useRef<ReturnType<typeof setTimeout>>()
+  const [copyTarget, setCopyTarget] = useState<CopyTarget>()
+  const [copyState, setCopyState] = useState<'idle' | 'copied' | 'failed'>('idle')
+  const [copyMenuOpen, setCopyMenuOpen] = useState(false)
+  const [tabStopId, setTabStopId] = useState<string | null>(initialTabStopId)
+
+  const setActiveRow = (row: HTMLElement | undefined) => {
+    activeRowRef.current?.removeAttribute('data-json-copy-active')
+    activeRowRef.current = row
+    row?.setAttribute('data-json-copy-active', '')
+  }
+
+  const clearCopyTarget = () => {
+    setActiveRow(undefined)
+    setCopyTarget(undefined)
+    setCopyState('idle')
+    copyMenuOpenRef.current = false
+    setCopyMenuOpen(false)
+  }
+
+  const copyPosition = (row: HTMLElement): Pick<CopyTarget, 'left' | 'side' | 'top'> => {
+    const root = rootRef.current
+    /* v8 ignore next -- row events and viewport listeners run only after the root ref mounts. */
+    if (root === null) throw new Error('JsonTree root is not mounted')
+    const rootRect = root.getBoundingClientRect()
+    const rowRect = row.getBoundingClientRect()
+    return {
+      left: rootRect.left + root.clientWidth - 26,
+      side: rowRect.top - rootRect.top > root.clientHeight / 2 ? 'top' : 'bottom',
+      top: rowRect.top,
+    }
+  }
+
+  const positionCopyButton = (row: HTMLElement, target: RowTarget) => {
+    const position = copyPosition(row)
+    setCopyTarget({ ...target, ...position })
+  }
+
+  const repositionCopyButton = (row: HTMLElement) => {
+    const position = copyPosition(row)
+    setCopyTarget((current) => {
+      /* v8 ignore next -- an active row and its copy target are installed together. */
+      if (current === undefined) return current
+      return { ...current, ...position }
+    })
+  }
+
+  useEffect(() => () => {
+    if (resetTimer.current !== undefined) clearTimeout(resetTimer.current)
+    activeRowRef.current?.removeAttribute('data-json-copy-active')
+  }, [])
+
+  useEffect(() => {
+    activeRowRef.current?.removeAttribute('data-json-copy-active')
+    activeRowRef.current = undefined
+    copyMenuOpenRef.current = false
+    setCopyTarget(undefined)
+    setCopyState('idle')
+    setCopyMenuOpen(false)
+    setTabStopId(initialTabStopId)
+  }, [data, expandTopLevel, initialTabStopId])
+
+  useEffect(() => {
+    const reposition = () => {
+      const row = activeRowRef.current
+      if (row !== undefined) repositionCopyButton(row)
+    }
+    window.addEventListener('scroll', reposition, true)
+    window.addEventListener('resize', reposition)
+    return () => {
+      window.removeEventListener('scroll', reposition, true)
+      window.removeEventListener('resize', reposition)
+    }
+  }, [])
+
+  const handleRowHover = (row: HTMLElement, target: RowTarget) => {
+    if (!copyable || copyMenuOpenRef.current) return
+    if (activeRowRef.current === row) return
+    setActiveRow(row)
+    setCopyState('idle')
+    copyMenuOpenRef.current = false
+    setCopyMenuOpen(false)
+    positionCopyButton(row, target)
+  }
+
+  const handleRootMouseOver = (event: ReactMouseEvent<HTMLDivElement>) => {
+    if (!copyable || copyMenuOpenRef.current) return
+    /* v8 ignore next -- browser mouse events delivered through React target an Element. */
+    if (!(event.target instanceof Element)) return
+    if (event.target.closest('[data-json-copy-button]') === null) clearCopyTarget()
+  }
+
+  const handleScroll = (_event: ReactUIEvent<HTMLDivElement>) => {
+    const row = activeRowRef.current
+    if (row !== undefined) repositionCopyButton(row)
+  }
+
+  const copy = async (mode: 'json' | 'path' | 'prettyJson' | 'value') => {
+    /* v8 ignore next -- copy controls only render while their target exists. */
+    if (copyTarget === undefined) return
+    try {
+      await navigator.clipboard.writeText(copyText(copyTarget, mode))
+      setCopyState('copied')
+    } catch {
+      setCopyState('failed')
+    }
+    if (resetTimer.current !== undefined) clearTimeout(resetTimer.current)
+    resetTimer.current = setTimeout(() => { setCopyState('idle') }, 1_500)
+  }
+
+  const [rootOpen, rootClose] = bracketOf(data)
+  const copyTargetIsObject = typeof copyTarget?.value === 'object' && copyTarget.value !== null
+  const defaultCopyMode = copyTargetIsObject ? 'prettyJson' : 'value'
+  const copyTitle = copyState === 'copied'
+    ? 'Copied'
+    : copyState === 'failed'
+      ? 'Copy failed'
+      : copyTargetIsObject ? 'Copy pretty JSON' : 'Copy value'
+
+  return (
+    <div
+      ref={rootRef}
+      className={clsx(css.root, className)}
+      onMouseOver={handleRootMouseOver}
+      onMouseLeave={() => {
+        if (!copyMenuOpenRef.current) clearCopyTarget()
+      }}
+      onScroll={handleScroll}
+    >
+      {expandTopLevel
+        ? (
+          <div className={css.expandedTopLevel}>
+            <div
+              className={clsx(css.row, css.topLevelBracket)}
+              data-json-root-row
+              onMouseOver={(event) => {
+                event.stopPropagation()
+                handleRowHover(event.currentTarget, { path: [], value: data })
+              }}
+            >
+              <span className={css.punctuation}>{rootOpen}</span>
+            </div>
+            <div
+              aria-label={label}
+              className={clsx(css.container, css.expandedTopLevelContainer)}
+              role="tree"
+            >
+              {rootEntries.map(([key, value], index) => (
+                <JsonTreeNode
+                  key={key}
+                  field={key}
+                  value={value}
+                  path={[Array.isArray(data) ? index : key]}
+                  lastElement={index === rootEntries.length - 1}
+                  initialExpanded={false}
+                  tabStopId={tabStopId}
+                  onClaimTabStop={setTabStopId}
+                  onRowHover={handleRowHover}
+                />
+              ))}
+            </div>
+            <div className={clsx(css.row, css.topLevelBracket)}>
+              <span className={css.punctuation}>{rootClose}</span>
+            </div>
+          </div>
+        )
+        : (
+          <div aria-label={label} className={css.container} role="tree">
+            <JsonTreeNode
+              value={data}
+              path={[]}
+              lastElement
+              initialExpanded
+              tabStopId={tabStopId}
+              onClaimTabStop={setTabStopId}
+              onRowHover={handleRowHover}
+            />
+          </div>
+        )}
+      {copyTarget !== undefined && (
+        <span
+          className={css.copyAnchor}
+          style={{ left: copyTarget.left, top: copyTarget.top }}
+        >
+          <Menu
+            open={copyMenuOpen}
+            compact
+            portal
+            align="end"
+            side={copyTarget.side}
+            anchor={(
+              <button
+                ref={copyButtonRef}
+                type="button"
+                className={css.copyButton}
+                data-json-copy-button
+                data-state={copyState}
+                aria-label={copyTitle}
+                title={`${copyTitle}; right-click for copy options`}
+                onClick={() => void copy(defaultCopyMode)}
+                onContextMenu={(event) => {
+                  event.preventDefault()
+                  event.stopPropagation()
+                  copyMenuOpenRef.current = true
+                  setCopyMenuOpen(true)
+                }}
+              >
+                {copyState === 'copied'
+                  ? <IconCheckOutline16 size={12} />
+                  : <IconCopyOutline16 size={12} />}
+              </button>
+            )}
+            items={copyTargetIsObject ? OBJECT_COPY_MENU_ITEMS : VALUE_COPY_MENU_ITEMS}
+            onSelect={(id) => {
+              void copy(id as 'json' | 'path' | 'prettyJson' | 'value')
+              copyMenuOpenRef.current = false
+              setCopyMenuOpen(false)
+            }}
+            onClose={clearCopyTarget}
+            getAnchorRect={() => (
+              copyButtonRef.current as HTMLButtonElement
+            ).getBoundingClientRect()}
+          />
+        </span>
+      )}
+    </div>
+  )
+}
diff --git a/packages/client/ui-primitives/src/Menu.module.css b/packages/client/ui-primitives/src/Menu.module.css
index 873c54e179..20c1584c1a 100644
--- a/packages/client/ui-primitives/src/Menu.module.css
+++ b/packages/client/ui-primitives/src/Menu.module.css
@@ -115,6 +115,37 @@
   background: var(--dsw-alias-interactive-bg-hover);
 }
 
+.list.compactList,
+.submenu.compactList {
+  min-width: 164px;
+  padding: 2px;
+  border-radius: 7px;
+}
+
+.compactList .item {
+  min-height: 26px;
+  gap: 6px;
+  padding: 3px 7px;
+  border-radius: 5px;
+  font-size: 12px;
+  line-height: 18px;
+}
+
+.compactList .itemIcon {
+  width: 14px;
+  height: 14px;
+}
+
+.compactList .separator {
+  margin: 2px;
+}
+
+.compactList .label {
+  padding: 4px 7px;
+  font-size: 11px;
+  line-height: 16px;
+}
+
 .item:disabled {
   opacity: 0.4;
   cursor: not-allowed;
diff --git a/packages/client/ui-primitives/src/Menu.tsx b/packages/client/ui-primitives/src/Menu.tsx
index 1a6409f42b..747750363d 100644
--- a/packages/client/ui-primitives/src/Menu.tsx
+++ b/packages/client/ui-primitives/src/Menu.tsx
@@ -71,6 +71,7 @@ const MEASURE_STYLE: CSSProperties = { visibility: 'hidden', left: 0, top: 0 }
  * keeps the pure-CSS in-place behavior.
  * @param props.closeOnPointerLeave - close the list when the pointer leaves
  * it (default false keeps it open until outside click/Escape/selection).
+ * @param props.compact - use reduced menu typography and spacing.
  * @param props.getAnchorRect - portal mode only: supply the anchor rect
  * directly (e.g. from a host-owned trigger button) instead of measuring the
  * Menu's own wrapper span. Required when the wrapper isn't itself laid out at
@@ -81,7 +82,7 @@ const MEASURE_STYLE: CSSProperties = { visibility: 'hidden', left: 0, top: 0 }
  * by a hairline; they stay visible while the items above scroll.
  * @returns anchor wrapper with the conditional list.
  */
-export function Menu({ open, anchor, items, selectedId, onSelect, onClose, align = 'start', side = 'bottom', portal = false, closeOnPointerLeave = false, getAnchorRect, footer, className }: {
+export function Menu({ open, anchor, items, selectedId, onSelect, onClose, align = 'start', side = 'bottom', portal = false, closeOnPointerLeave = false, compact = false, getAnchorRect, footer, className }: {
   open: boolean
   anchor: ReactNode
   items: readonly MenuEntry[]
@@ -93,6 +94,7 @@ export function Menu({ open, anchor, items, selectedId, onSelect, onClose, align
   side?: 'bottom' | 'top' | 'right'
   portal?: boolean
   closeOnPointerLeave?: boolean
+  compact?: boolean
   getAnchorRect?: () => DOMRect | null
   className?: string
 }) {
@@ -219,7 +221,7 @@ export function Menu({ open, anchor, items, selectedId, onSelect, onClose, align
           {entry.id === selectedId && <IconCheckOutline16 className={css.check} />}
         </button>
         {subOpen && entry.submenu !== undefined && (
-          <div className={css.submenu} role="menu">
+          <div className={clsx(css.submenu, compact && css.compactList)} role="menu">
             {entry.submenu.map(sub => (
               <button
                 key={sub.id}
@@ -246,7 +248,7 @@ export function Menu({ open, anchor, items, selectedId, onSelect, onClose, align
   const list = open && (
     <div
       ref={listRef}
-      className={clsx(css.list, scrollable && css.scrollable, portal && css.portal, side === 'top' && !portal && css.sideTop, align === 'end' && !portal && css.alignEnd)}
+      className={clsx(css.list, compact && css.compactList, scrollable && css.scrollable, portal && css.portal, side === 'top' && !portal && css.sideTop, align === 'end' && !portal && css.alignEnd)}
       style={portal ? fixedPos ?? MEASURE_STYLE : undefined}
       role="menu"
       onPointerLeave={closeOnPointerLeave ? () => { onClose() } : undefined}
diff --git a/packages/client/ui-primitives/src/Pill.tsx b/packages/client/ui-primitives/src/Pill.tsx
index 76b9301972..2c3c24c1ad 100644
--- a/packages/client/ui-primitives/src/Pill.tsx
+++ b/packages/client/ui-primitives/src/Pill.tsx
@@ -12,7 +12,9 @@ import css from './Pill.module.css'
  */
 export function Pill({ active = false, className, children, onClick, ...rest }: {
   active?: boolean
-  className?: string
+  // `| undefined` so a caller can forward an optional class straight through
+  // under exactOptionalPropertyTypes (a CSS-module lookup is string|undefined).
+  className?: string | undefined
   children?: ReactNode
 } & ButtonHTMLAttributes<HTMLButtonElement>) {
   if (!onClick) {
diff --git a/packages/client/ui-primitives/src/StateDot.tsx b/packages/client/ui-primitives/src/StateDot.tsx
index 643f9283f4..77f9c0a794 100644
--- a/packages/client/ui-primitives/src/StateDot.tsx
+++ b/packages/client/ui-primitives/src/StateDot.tsx
@@ -23,8 +23,8 @@ const MATRIX_CELLS: readonly (readonly [number, number])[] = [
  */
 export function StateDot({ state, size = 10, className }: {
   state: StateDotState
-  size?: number
-  className?: string
+  size?: number | undefined
+  className?: string | undefined
 }) {
   if (state === 'ongoing') {
     return (
diff --git a/packages/client/ui-primitives/src/TerminalBlock.module.css b/packages/client/ui-primitives/src/TerminalBlock.module.css
new file mode 100644
index 0000000000..704ea0e808
--- /dev/null
+++ b/packages/client/ui-primitives/src/TerminalBlock.module.css
@@ -0,0 +1,152 @@
+/* Geometry mirrors CodeBlock (12px radius, code-block surface + banner rows,
+   markdown code-block font) so a terminal card and a fenced code block read as
+   one family. The one deliberate divergence: output keeps `white-space: pre`
+   and scrolls horizontally, because folding a column-aligned command's output
+   destroys its alignment. */
+
+.block {
+  --dsl-terminal-radius: 12px;
+  --dsl-terminal-line-height: 22px;
+  /* The card's own left inset, holding the run-state dot in a column of its own
+     so it never competes with the commands for horizontal space. */
+  --dsl-terminal-gutter: 30px;
+
+  position: relative;
+  margin: 16px 0;
+  /* The gutter is the card's OWN padding, not a margin: every consumer rewrites
+     `margin` wholesale (each render site sets its own indent), which silently
+     cancelled the reservation and let the dot fall outside the card into a
+     container that clips it. Owning the reservation here keeps the invariant
+     with the component that depends on it. */
+  padding-left: var(--dsl-terminal-gutter);
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-markdown-code-block);
+  border-radius: var(--dsl-terminal-radius);
+}
+
+/* Top-aligned: the status pill and copy control stay on the first prompt row
+   however many command lines the card carries. */
+.header {
+  display: flex;
+  align-items: flex-start;
+  gap: 12px;
+  /* Pulled back across the card's gutter padding so the banner background and
+     its top-left radius span the FULL surface, then re-inset by the same amount
+     so the prompt text and the dot keep their positions. A plain block child
+     only reaches the content box, which left the gutter column painted in the
+     body color and drew the card's top-left corner in it — invisible in the
+     light theme, where banner and body share a token, and visible in the dark
+     one, where they do not. */
+  margin-left: calc(-1 * var(--dsl-terminal-gutter));
+  padding: 9px 14px 9px var(--dsl-terminal-gutter);
+  background: var(--dsw-alias-markdown-code-block-banner);
+  border-top-left-radius: var(--dsl-terminal-radius);
+  border-top-right-radius: var(--dsl-terminal-radius);
+}
+
+/* One row per command line. The prompt column is the only element allowed to
+   shrink; the status pill and the copy control keep their intrinsic width. */
+.prompt {
+  display: flex;
+  flex-direction: column;
+  min-width: 0;
+  flex: 1;
+  font: var(--dsw-font-markdown-code-block);
+}
+
+.promptLine {
+  position: relative;
+  display: flex;
+  align-items: baseline;
+  gap: 8px;
+  min-width: 0;
+  line-height: var(--dsl-terminal-line-height);
+}
+
+/* Out of flow inside the card's own gutter padding, so the reservation and the
+   dot move together and no consumer margin can pull them apart; the dot neither
+   indents its command nor depends on the command's text metrics to line up.
+   Centered against the row's line box, not the code font's baseline. */
+.runState {
+  position: absolute;
+  left: calc(-1 * var(--dsl-terminal-gutter) + 8px);
+  top: 50%;
+  transform: translateY(-50%);
+}
+
+/* The dot is aria-hidden; this is its text label for assistive technology. */
+.runStateLabel {
+  position: absolute;
+  width: 1px;
+  height: 1px;
+  overflow: hidden;
+  clip-path: inset(50%);
+  white-space: nowrap;
+}
+
+.cwd {
+  flex: none;
+  color: var(--dsw-alias-label-tertiary);
+}
+
+/* `pre`, not `nowrap`: the prompt row renders the command verbatim, and
+   `nowrap` collapses the repeated spaces, tabs, and alignment of an indented
+   continuation. Both hold the single row and the ellipsis. */
+.command {
+  min-width: 0;
+  color: var(--dsw-alias-label-primary);
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: pre;
+}
+
+.status {
+  flex: none;
+  color: var(--dsw-alias-state-error-primary);
+}
+
+.copyButton {
+  flex: none;
+  background-color: transparent;
+  border: none;
+  padding: 0;
+  margin: 0;
+  color: var(--dsw-alias-label-secondary);
+  cursor: pointer;
+  font: var(--dsw-font-xs-13);
+}
+
+.output {
+  padding: 12px 14px 12px 0;
+  font: var(--dsw-font-markdown-code-block);
+  overflow-x: auto;
+  overflow-y: hidden;
+}
+
+/* No wrapping, no word-break: alignment is the payload of terminal output. */
+.line {
+  min-height: var(--dsl-terminal-line-height);
+  white-space: pre;
+}
+
+.expand {
+  display: block;
+  width: 100%;
+  padding: 0;
+  border: none;
+  background-color: transparent;
+  color: var(--dsw-alias-label-tertiary);
+  cursor: pointer;
+  font: inherit;
+  text-align: left;
+}
+
+.expand:hover {
+  color: var(--dsw-alias-label-secondary);
+}
+
+.empty {
+  padding: 12px 14px 12px 0;
+  font: var(--dsw-font-markdown-code-block);
+  color: var(--dsw-alias-label-tertiary);
+}
diff --git a/packages/client/ui-primitives/src/TerminalBlock.tsx b/packages/client/ui-primitives/src/TerminalBlock.tsx
new file mode 100644
index 0000000000..c707711f69
--- /dev/null
+++ b/packages/client/ui-primitives/src/TerminalBlock.tsx
@@ -0,0 +1,237 @@
+// TerminalBlock: the terminal surface for a shell command and its output —
+// prompt line (run-state dot + shortened cwd + command), ANSI-colored output,
+// settled exit status, and a copy control for the raw output. Output never soft-wraps:
+// column-aligned output (ls, tables, box drawing) keeps its alignment and
+// scrolls horizontally instead of folding. Colors resolve through --dsw-*
+// tokens; ANSI parsing lives in ansi.ts.
+
+import { useCallback, useMemo, useState } from 'react'
+import clsx from 'clsx'
+import { parseAnsiLines, type AnsiLine } from './ansi.ts'
+import { writeClipboard } from './clipboard.ts'
+import { Pill } from './Pill.tsx'
+import { StateDot, type StateDotState } from './StateDot.tsx'
+import css from './TerminalBlock.module.css'
+
+/**
+ * Output lines shown before the height cap collapses the middle. Matches the
+ * TUI transcript's default tool-output budget so both front ends cut a long
+ * command's output at the same place.
+ */
+export const DEFAULT_TERMINAL_MAX_LINES = 16
+
+export interface TerminalBlockProps {
+  /** The command line, rendered verbatim after the prompt label. */
+  command: string
+  /** Working directory for the prompt label; absent renders a plain `$`. */
+  cwd?: string | undefined
+  /** Absolute home directory, so a cwd equal to it collapses to `~`; absent disables that collapse. */
+  home?: string | undefined
+  /** The command's output text; may contain ANSI escape sequences. */
+  output?: string | undefined
+  /** Settled exit code; a non-zero value renders the status pill. */
+  exitCode?: number | undefined
+  /** Settled terminating signal name; any value renders the status pill, taking precedence over the exit code. */
+  signal?: string | undefined
+  /** The command is still running: the block shows the prompt line alone. */
+  running?: boolean | undefined
+  /** Height cap in output lines before the middle collapses (default {@link DEFAULT_TERMINAL_MAX_LINES}). */
+  maxLines?: number | undefined
+  /** Extra class merged onto the wrapper (callers position; this component draws). */
+  className?: string | undefined
+}
+
+/**
+ * Prompt label for a working directory: `~` for the home directory itself,
+ * otherwise the path's last segment (both separators accepted, trailing
+ * separators ignored), falling back to the path itself when it has no
+ * segment.
+ * @param cwd - the working directory path.
+ * @param home - absolute home directory, when the caller knows it.
+ * @returns the prompt label.
+ */
+function promptLabel(cwd: string, home: string | undefined): string {
+  const trimmed = cwd.replace(/[/\\]+$/, '')
+  if (home !== undefined && trimmed === home.replace(/[/\\]+$/, '')) return '~'
+  const segment = trimmed.split(/[/\\]/).pop()
+  return segment === undefined || segment === '' ? cwd : segment
+}
+
+/**
+ * Status pill text for a settled command, or undefined when the command
+ * settled cleanly (exit 0, no signal) and needs no pill — the same
+ * distinction the bash tool's own exit-status markers draw.
+ * @param exitCode - settled exit code, when known.
+ * @param signal - settled terminating signal name, when known.
+ * @returns the pill text, or undefined for a clean exit.
+ */
+function statusText(exitCode: number | undefined, signal: string | undefined): string | undefined {
+  if (signal !== undefined) return `信号 ${signal}`
+  if (exitCode !== undefined && exitCode !== 0) return `退出码 ${exitCode}`
+  return undefined
+}
+
+/**
+ * Run-state indicator for the command, shown at the head of the prompt line so
+ * the card states whether the command is still running without the reader
+ * having to infer it from the presence of output. Three of {@link StateDotState}'s
+ * four states are reachable: the running chase (the same
+ * indicator a running tool row's leading icon uses, so the row and its card
+ * never disagree), green for a clean settle, red for a signal or a non-zero
+ * exit — the same status distinction {@link statusText} draws for the pill. A
+ * settled command whose exit status never reached the view counts as a clean
+ * settle: the view says it finished and says nothing went wrong.
+ * @param running - the command has not settled.
+ * @param exitCode - settled exit code, when known.
+ * @param signal - settled terminating signal name, when known.
+ * @returns the dot's state and its text label, since the dot is aria-hidden.
+ */
+function runState(
+  running: boolean,
+  exitCode: number | undefined,
+  signal: string | undefined,
+): { state: StateDotState; label: string } {
+  if (running) return { state: 'ongoing', label: '运行中' }
+  if (statusText(exitCode, signal) !== undefined) return { state: 'error', label: '失败' }
+  return { state: 'done', label: '已完成' }
+}
+
+/**
+ * Render one parsed output line. Runs without SGR state render as bare text,
+ * so uncolored output carries no span wrappers.
+ * @param line - the line's styled runs.
+ * @returns the line's children.
+ */
+function renderLine(line: AnsiLine) {
+  return line.map((span, index) => span.style === undefined
+    ? span.text
+    : <span key={index} style={span.style}>{span.text}</span>)
+}
+
+/**
+ * Render a shell command as a terminal surface.
+ * @param props - see {@link TerminalBlockProps}.
+ * @returns the terminal block element.
+ */
+export function TerminalBlock({
+  command,
+  cwd,
+  home,
+  output,
+  exitCode,
+  signal,
+  running = false,
+  maxLines = DEFAULT_TERMINAL_MAX_LINES,
+  className,
+}: TerminalBlockProps) {
+  const text = output ?? ''
+  // A command's output ends with a newline; that terminator is not an extra
+  // blank line to draw or to count against the height cap. The check runs on the
+  // PARSED lines rather than on the raw text, because a reset after the final
+  // newline (`line\n\x1b[0m`) leaves the string not ending in one while still
+  // producing a last line with nothing visible in it. A genuinely blank final
+  // line — the double newline — survives, since it has a real empty line before
+  // the terminator. The copy control still copies `text` untouched.
+  const lines = useMemo(() => {
+    const parsed = parseAnsiLines(text)
+    const last = parsed[parsed.length - 1]
+    const terminated = parsed.length > 1 && last !== undefined
+      && last.every(span => span.text === '')
+    return terminated ? parsed.slice(0, -1) : parsed
+  }, [text])
+  const [expanded, setExpanded] = useState(false)
+  const [copied, setCopied] = useState(false)
+
+  const onCopy = useCallback(() => {
+    if (copied) return
+    // The raw output, never the rendered tree: the prompt line and the status
+    // pill are chrome the user did not run.
+    void writeClipboard(text).then((ok) => {
+      if (!ok) return
+      setCopied(true)
+      window.setTimeout(() => { setCopied(false) }, 1000)
+    })
+  }, [copied, text])
+
+  const onToggle = useCallback(() => { setExpanded(value => !value) }, [])
+
+  const status = statusText(exitCode, signal)
+  const state = runState(running, exitCode, signal)
+  // A multi-line command gets one prompt row per line, so a two-command shell
+  // snippet reads as the two commands it is instead of collapsing into one
+  // ellipsized row. A trailing newline is a terminator, not an empty command.
+  const commandLines = useMemo(() => {
+    const body = command.endsWith('\n') ? command.slice(0, -1) : command
+    return body.split('\n')
+  }, [command])
+  // Read from the parsed lines the card actually renders, not from the raw text:
+  // output that is only escapes or control bytes (a lone reset, an OSC title, an
+  // erase) survives `text.trim()` yet parses to nothing visible. Judging it on
+  // the raw text drew an output box of blank rows plus a copy control for
+  // invisible bytes, and hid the placeholder that belongs there.
+  const empty = lines.every(line => line.every(span => span.text.trim() === ''))
+  const hidden = lines.length - maxLines
+  const capped = hidden > 0 && !expanded
+  // Same split arithmetic as the TUI transcript's collapsed tool card, so a
+  // command's head and tail slices agree between the two front ends.
+  const headLines = Math.ceil(maxLines / 2)
+  const tailLines = maxLines - headLines
+
+  return (
+    <div className={clsx(css.block, className)} data-terminal="" data-running={running ? '' : undefined}>
+      <div className={css.header}>
+        <div className={css.prompt}>
+          <span className={css.runStateLabel}>{state.label}</span>
+          {commandLines.map((line, index) => (
+            <div key={index} className={css.promptLine}>
+              {/* One dot for the card, on the first row: the exit status the
+                  view carries is the whole call's, and bash reports no
+                  per-command status, so a dot per row would assert a
+                  per-line outcome nothing here knows. */}
+              {index === 0 && <StateDot state={state.state} className={css.runState} />}
+              {/* The cwd labels the CALL, so only its first row carries it. The
+                  view knows one working directory — where the call started —
+                  and a later line may well run somewhere else (a `cd` in the
+                  command is enough), so repeating the label down the rows would
+                  assert a directory per line that nothing here knows. Later
+                  rows keep a bare `$` to stay aligned as prompts. */}
+              <span className={css.cwd}>
+                {index > 0 || cwd === undefined ? '$' : promptLabel(cwd, home)}
+              </span>
+              <span className={css.command}>{line}</span>
+            </div>
+          ))}
+        </div>
+        {status !== undefined && <Pill className={css.status}>{status}</Pill>}
+        {!running && !empty && (
+          <button type="button" className={css.copyButton} onClick={onCopy}>
+            {copied ? '复制成功' : '复制'}
+          </button>
+        )}
+      </div>
+      {!running && (empty
+        ? <div className={css.empty}>无输出</div>
+        : (
+          <div className={css.output}>
+            {(capped ? lines.slice(0, headLines) : lines).map((line, index) => (
+              <div key={index} className={css.line}>{renderLine(line)}</div>
+            ))}
+            {hidden > 0 && (
+              <button
+                type="button"
+                className={css.expand}
+                aria-expanded={expanded}
+                aria-label={expanded ? '收起输出' : `展开其余 ${hidden} 行输出`}
+                onClick={onToggle}
+              >
+                {expanded ? '收起' : `… 其余 ${hidden} 行`}
+              </button>
+            )}
+            {capped && lines.slice(lines.length - tailLines).map((line, index) => (
+              <div key={index} className={css.line}>{renderLine(line)}</div>
+            ))}
+          </div>
+        ))}
+    </div>
+  )
+}
diff --git a/packages/client/ui-primitives/src/ansi.ts b/packages/client/ui-primitives/src/ansi.ts
new file mode 100644
index 0000000000..9eeba29e12
--- /dev/null
+++ b/packages/client/ui-primitives/src/ansi.ts
@@ -0,0 +1,447 @@
+// ANSI model behind TerminalBlock: anser splits the SGR runs, this module
+// resolves each run's colors and decorations into a plain style record and
+// folds the runs into per-line span arrays so a height cap can slice whole
+// lines. Sequences anser does not turn into color (OSC, cursor movement,
+// other C0 controls) are removed before parsing so they never reach the DOM
+// as literal characters.
+
+import Anser from 'anser'
+import type { CSSProperties } from 'react'
+
+/**
+ * The subset of one anser JSON chunk this module reads. anser's own types
+ * declare `fg`/`bg` as `string`, but its parser leaves them `null` for a run
+ * that sets no color, so the null is spelled out here.
+ */
+interface AnsiChunk {
+  /** Run text with its SGR codes already removed. */
+  content: string
+  /** Foreground as an `r, g, b` triple, or null when the run sets none. */
+  fg: string | null
+  /** Background as an `r, g, b` triple, or null when the run sets none. */
+  bg: string | null
+  /** SGR attributes in effect for the run, in the order they were declared. */
+  decorations: readonly string[]
+}
+
+/** One run of terminal text; `style` is undefined for text that carries no SGR state. */
+export interface AnsiSpan {
+  /** The run's plain text, free of escape sequences and newlines. */
+  text: string
+  /** Resolved inline style, or undefined when the run needs no wrapper. */
+  style: CSSProperties | undefined
+}
+
+/** The spans of one output line, in order. */
+export type AnsiLine = readonly AnsiSpan[]
+
+/**
+ * The 8/16 basic ANSI colors, keyed by the whitespace-free `r,g,b` triple
+ * anser emits for them, mapped onto the theme tokens that carry the same
+ * semantic. Black and white both resolve to the primary label color so text
+ * stays legible under either theme instead of matching the surface it sits
+ * on; bright black takes the tertiary label color (the muted-gray role).
+ * Magenta and cyan have no token equivalent in this design system and fall
+ * through to anser's literal rgb, as do all 256-palette and truecolor values.
+ */
+const TOKEN_BY_BASIC_RGB: Record<string, string> = {
+  '0,0,0': 'var(--dsw-alias-label-primary)',
+  '255,255,255': 'var(--dsw-alias-label-primary)',
+  '85,85,85': 'var(--dsw-alias-label-tertiary)',
+  '187,0,0': 'var(--dsw-alias-state-error-primary)',
+  '255,85,85': 'var(--dsw-alias-state-error-secondary)',
+  '0,187,0': 'var(--dsw-alias-state-success-primary)',
+  '0,255,0': 'var(--dsw-alias-state-success-secondary)',
+  '187,187,0': 'var(--dsw-alias-state-warn-primary)',
+  '255,255,85': 'var(--dsw-alias-state-warn-secondary)',
+  '0,0,187': 'var(--dsw-alias-state-business-primary)',
+  '85,85,255': 'var(--dsw-static-blue-400)',
+}
+
+/**
+ * CSS for each SGR attribute anser reports. `blink` is deliberately absent —
+ * animated text is not reproduced. `reverse` never arrives here: anser
+ * consumes it by swapping the run's foreground and background. Underline and
+ * strikethrough share `textDecoration`, so in a run declaring both, the
+ * later declaration wins.
+ */
+const STYLE_BY_DECORATION: Record<string, CSSProperties | undefined> = {
+  bold: { fontWeight: 700 },
+  dim: { opacity: 0.7 },
+  italic: { fontStyle: 'italic' },
+  underline: { textDecoration: 'underline' },
+  strikethrough: { textDecoration: 'line-through' },
+  hidden: { visibility: 'hidden' },
+}
+
+/** OSC strings (window title, hyperlinks), with or without their terminator. */
+const OSC_SEQUENCE = /\u001b\][^\u0007\u001b]*(?:\u0007|\u001b\\)?/g
+
+/** Escape sequences other than CSI: charset selection, single-shift, reset. */
+const NON_CSI_ESCAPE = /\u001b(?!\[)[\u0020-\u002f]*[\u0030-\u007e]?/g
+
+/**
+ * C0 controls with no display meaning here. Tab, newline, backspace and ESC
+ * survive: the first two for layout, backspace for the cursor replay, ESC
+ * for anser's CSI split.
+ */
+const INERT_CONTROL = /[\u0000-\u0007\u000b-\u001a\u001c-\u001f\u007f]/g
+
+/**
+ * Lines whose cursor movements have to be replayed: a carriage return, a
+ * backspace, or an erase-in-line. The erase pattern matches the SAME CSI shape
+ * `replayLine` parses (parameters may carry `;` and intermediate bytes), so a
+ * form like `\x1b[1;2K` cannot slip past this guard and skip its own erase.
+ */
+const NEEDS_REPLAY = /\r|\u0008|\u001b\[[\u0030-\u003f]*[\u0020-\u002f]*K/
+
+/** SGR sequences alone, for folding state through a line that needs no replay. */
+const SGR_SEQUENCE = /\u001b\[([\u0030-\u003f]*)[\u0020-\u002f]*m/g
+
+/** Terminal tab stop width; a tab advances to the next multiple of this. */
+const TAB_WIDTH = 8
+
+/**
+ * Combining marks and other zero-width code points: a terminal advances no
+ * column for them, so `e` + U+0301 occupies one cell and a two-column redraw
+ * covers both code points.
+ */
+const ZERO_WIDTH = /^[\p{Mn}\p{Me}\p{Cf}\u200b-\u200f\u2060]$/u
+
+/**
+ * Characters a terminal advances two columns for: CJK scripts, fullwidth forms,
+ * CJK punctuation, and characters with emoji presentation. Text-presentation
+ * symbols (`\u2713`, `\u26a0` and the rest of U+2600-U+27BF) are ONE column and
+ * must stay out of this set.
+ */
+const WIDE_CHAR = new RegExp(
+  '\\p{Script=Han}|\\p{Script=Hiragana}|\\p{Script=Katakana}|\\p{Script=Hangul}'
+  // Emoji presentation only: the U+2600-U+27BF symbol block is mostly SINGLE
+  // width — `\u2713` (the check every progress line writes, this fixture
+  // included) advances one column, verified against a real terminal, so taking
+  // the whole block as wide misaligned exactly the output this card exists for.
+  + '|\\p{Emoji_Presentation}'
+  + '|[\\uff01-\\uff60\\u3000-\\u303e]',
+  'u',
+)
+
+/**
+ * Whether a character occupies two terminal columns (CJK, fullwidth forms,
+ * emoji). Covers the ranges a command's output realistically carries; a
+ * narrower guess would misalign the columns this card exists to preserve.
+ * @param char - one character from the output.
+ * @returns true when the terminal advances two columns for it.
+ */
+function isWide(char: string): boolean {
+  const code = char.codePointAt(0)
+  if (code === undefined || code < 0x1100) return false
+  return WIDE_CHAR.test(char)
+}
+
+/**
+ * A cell's graphic state, normalized. Held as fields rather than as the raw
+ * sequence history because a terminal tracks CURRENT state, not a transcript:
+ * accumulating sequences made each state boundary re-emit the whole chain, so
+ * output that switches color without a full reset emitted O(n^2) characters
+ * (3200 such cells produced 25 MB and eventually a `RangeError`). It also makes
+ * the attribute closers every chalk-based tool writes — `39`, `49`, `22`, `23`,
+ * `24`, `27`, `29` — actually close their attribute instead of appending to it.
+ */
+interface SgrState {
+  fg: string
+  bg: string
+  /** Attribute parameters in force, e.g. `1` (bold) or `4` (underline). */
+  attrs: readonly string[]
+}
+
+/** The default state: no color, no attributes. */
+const SGR_NONE: SgrState = { fg: '', bg: '', attrs: [] }
+
+/** Attribute closers, mapped to the opener parameters each one turns off. */
+const ATTR_CLOSERS: Record<string, readonly string[]> = {
+  22: ['1', '2'], 23: ['3'], 24: ['4'], 25: ['5', '6'], 27: ['7'], 28: ['8'], 29: ['9'],
+}
+
+/**
+ * Fold one SGR sequence's parameters into the state it produces.
+ * @param state - state in force before the sequence.
+ * @param params - the sequence's raw parameter string (`31`, `1;4`, `38;5;208`).
+ * @returns the state the sequence leaves in force.
+ */
+function foldSgr(state: SgrState, params: string): SgrState {
+  const codes = params === '' ? ['0'] : params.split(';')
+  let next = state
+  for (let index = 0; index < codes.length; index++) {
+    const code = String(codes[index])
+    if (code === '' || code === '0') { next = SGR_NONE; continue }
+    // Extended color: `38;5;N` / `38;2;R;G;B` and the `48` background pair
+    // consume their own arguments, so they are taken whole.
+    if (code === '38' || code === '48') {
+      const kind = codes[index + 1] ?? ''
+      const span = kind === '2' ? 4 : kind === '5' ? 2 : 0
+      const value = codes.slice(index, index + span + 1).join(';')
+      next = code === '38' ? { ...next, fg: value } : { ...next, bg: value }
+      index += span
+      continue
+    }
+    const closes = ATTR_CLOSERS[code]
+    if (closes !== undefined) {
+      next = { ...next, attrs: next.attrs.filter(attr => !closes.includes(attr)) }
+      continue
+    }
+    const numeric = Number(code)
+    if (code === '39') { next = { ...next, fg: '' }; continue }
+    if (code === '49') { next = { ...next, bg: '' }; continue }
+    if ((numeric >= 30 && numeric <= 37) || (numeric >= 90 && numeric <= 97)) { next = { ...next, fg: code }; continue }
+    if ((numeric >= 40 && numeric <= 47) || (numeric >= 100 && numeric <= 107)) { next = { ...next, bg: code }; continue }
+    if (!next.attrs.includes(code)) next = { ...next, attrs: [...next.attrs, code] }
+  }
+  return next
+}
+
+/**
+ * Render a state as the one canonical sequence that establishes it from the
+ * default, so a boundary emits a bounded string no matter how the state was
+ * reached.
+ * @param state - the state to open.
+ * @returns the SGR sequence, or the empty string for the default state.
+ */
+function openSgr(state: SgrState): string {
+  const codes = [...state.attrs]
+  if (state.fg !== '') codes.push(state.fg)
+  if (state.bg !== '') codes.push(state.bg)
+  return codes.length === 0 ? '' : `\u001b[${codes.join(';')}m`
+}
+
+/** Whether two states are the same, so a boundary is only emitted on a change. */
+function sameSgr(a: SgrState, b: SgrState): boolean {
+  return a.fg === b.fg && a.bg === b.bg && a.attrs.length === b.attrs.length
+    && a.attrs.every((attr, index) => attr === b.attrs[index])
+}
+
+/**
+ * Replay one line's cursor movements the way a terminal paints it, into a
+ * column buffer. Carriage return and backspace only MOVE the cursor — neither
+ * erases anything — so what a reader sees is whatever each column last had
+ * written to it. That distinction is the whole point of doing this as a buffer
+ * rather than as string surgery: `100%\rOK` shows `OK0%` because the redraw is
+ * shorter than the frame beneath it, and a trailing `abc\b` still shows `abc`
+ * because nothing ever overwrote the `c`.
+ *
+ * A CSI sequence occupies no column; it changes the state that the NEXT writes
+ * are stamped with, which is how a terminal stores color per cell. `red bad`
+ * then three backspaces then `ok` therefore shows `okd` with the `d` still red:
+ * `ok` overwrote two cells and the third kept the state it was written with.
+ * The columns are re-emitted as runs, so anser sees that same styling.
+ * @param line - one output line, still carrying its CSI sequences.
+ * @param entrySgr - SGR state in force when the line begins, since a newline
+ *   does not reset it.
+ * @returns the line as the terminal would have it after every movement, plus the
+ *   SGR state at its end for the next line to enter with.
+ */
+function replayLine(line: string, entrySgr: SgrState): { text: string; sgr: SgrState } {
+  // Same shape anser splits on, so a sequence is one unit here as well.
+  const csi = /\u001b\[([\u0030-\u003f]*)[\u0020-\u002f]*([\u0040-\u007e])/g
+  /** Per column: the state in force when it was written, and its character. */
+  const columns: (Cell | undefined)[] = []
+  let cursor = 0
+  // State is tracked exactly as a terminal tracks it: each cell is stamped with
+  // whatever was in force at the moment of the write, so a later redraw cannot
+  // restyle the cells it does not reach. It enters carrying the previous line's
+  // state, since a newline does not reset it.
+  let sgr = entrySgr
+  let at = 0
+
+  /** Clear a cell and, for a wide pair, its partner: a terminal erases both. */
+  const clear = (index: number, fill: string): void => {
+    const cell = columns[index]
+    if (cell?.spacer === true && index > 0) columns[index - 1] = { sgr, char: fill }
+    else if (cell !== undefined && isWide(cell.char) && columns[index + 1]?.spacer === true) {
+      columns[index + 1] = { sgr, char: fill }
+    }
+    columns[index] = { sgr, char: fill }
+  }
+
+  const consume = (chunk: string): void => {
+    for (const char of chunk) {
+      if (char === '\r') { cursor = 0; continue }
+      if (char === '\u0008') { cursor = Math.max(0, cursor - 1); continue }
+      if (char === '\t') {
+        // A tab advances to the next 8-column stop, leaving the cells it skips
+        // as they were — which is how a redraw can leave a tabbed column
+        // standing. Column alignment is the whole point of this card.
+        const stop = cursor + TAB_WIDTH - (cursor % TAB_WIDTH)
+        for (; cursor < stop; cursor++) columns[cursor] ??= { sgr, char: ' ' }
+        continue
+      }
+      if (ZERO_WIDTH.test(char)) {
+        // No column of its own: it attaches to the cell already written, so a
+        // redraw that covers that cell covers the mark with it. With no cell to
+        // attach to (line start, or straight after a redraw to column 0) a
+        // terminal shows nothing rather than a lone accent.
+        const base = cursor > 0 ? columns[cursor - 1] : undefined
+        if (base !== undefined) columns[cursor - 1] = { sgr: base.sgr, char: base.char + char }
+        continue
+      }
+      // Writing over either half of a wide pair blanks the other half, since a
+      // terminal cannot leave one cell of a two-cell glyph standing.
+      clear(cursor, ' ')
+      columns[cursor] = { sgr, char }
+      cursor++
+      // A wide character occupies two columns; the trailing one is a spacer,
+      // marked so that overwriting the lead cell leaves a blank behind instead
+      // of closing the gap and shifting everything after it left.
+      if (isWide(char)) { columns[cursor] = { sgr, char: '', spacer: true }; cursor++ }
+    }
+  }
+
+  for (const match of line.matchAll(csi)) {
+    consume(line.slice(at, match.index))
+    at = match.index + match[0].length
+    // Both groups are mandatory in the pattern, so destructuring types them as
+    // strings without a fallback that could never run.
+    const params = String(match[1])
+    const final = String(match[2])
+    if (final === 'K') {
+      // Erase in line: the fixed companion of `\r` in every spinner and progress
+      // bar. Without it a shorter redraw leaves the previous frame's tail
+      // standing, which is text the terminal never showed. `1` blanks from the
+      // line start THROUGH the cursor column (inclusive, per the CSI spec)
+      // rather than dropping those cells, since the cursor does not move and a
+      // later write can still land past them. Only the FIRST parameter selects
+      // the mode; a terminal ignores the rest (`1;2K` erases exactly as `1K`).
+      const mode = String(params.split(';')[0])
+      if (mode === '1') for (let index = 0; index <= cursor; index++) clear(index, ' ')
+      else columns.length = mode === '2' ? 0 : cursor
+      continue
+    }
+    // Only SGR carries graphic state; every other final byte is a cursor or
+    // erase action that must not affect a cell's style.
+    if (final !== 'm') continue
+    sgr = foldSgr(sgr, params)
+  }
+  consume(line.slice(at))
+
+  // Re-emit the columns, opening a run only where its state changes, so anser
+  // sees the same styling a terminal shows. Each boundary emits ONE canonical
+  // sequence for the state it opens, which is what keeps the output linear in
+  // the number of cells however the state was reached.
+  let out = ''
+  let active = entrySgr
+  for (let index = 0; index < columns.length; index++) {
+    const column = columns[index] ?? { sgr: SGR_NONE, char: ' ' }
+    if (!sameSgr(column.sgr, active)) {
+      if (!sameSgr(active, SGR_NONE)) out += '\u001b[0m'
+      out += openSgr(column.sgr)
+      active = column.sgr
+    }
+    // A spacer still holds its column. While its lead cell survives, the wide
+    // glyph spans both and the spacer emits nothing; once a later write replaced
+    // that lead, the terminal blanks the spacer instead of closing the gap, so
+    // emitting nothing would shift everything after it one column left.
+    const leadIntact = index > 0 && isWide(columns[index - 1]?.char ?? '')
+    out += column.spacer === true && !leadIntact ? ' ' : column.char
+  }
+  // Converge to the state the SCAN ended in, not the last written cell's: a
+  // sequence after the final write (the `\x1b[0m` closing a colored line) changes
+  // no cell yet still ends the run, and it has to reach both the DOM and the
+  // next line. Without this a line ending in a reset leaked its color onward.
+  if (!sameSgr(active, sgr)) {
+    if (!sameSgr(active, SGR_NONE)) out += '\u001b[0m'
+    out += openSgr(sgr)
+  }
+  return { text: out, sgr }
+}
+
+/** One replayed column: the state it was written with, and its character. */
+interface Cell {
+  sgr: SgrState
+  char: string
+  /** The trailing half of a wide character's two-column pair. */
+  spacer?: boolean
+}
+
+/**
+ * Replay every line's cursor movements. A `\r` that only terminates a CRLF line
+ * is dropped first, so those lines keep their text instead of being redrawn onto
+ * themselves. SGR state threads across lines: a newline does not reset it, so a
+ * run opened before a redraw still colors the lines after it.
+ * @param text - output text, already free of OSC and non-CSI escapes.
+ * @returns the text with each line painted as the terminal would.
+ */
+function applyCursorMovements(text: string): string {
+  const replayed: string[] = []
+  let sgr = SGR_NONE
+  for (const raw of text.split('\n')) {
+    const line = raw.replace(/\r+$/, '')
+    if (NEEDS_REPLAY.test(line)) {
+      const result = replayLine(line, sgr)
+      replayed.push(result.text)
+      sgr = result.sgr
+      continue
+    }
+    // No cursor movement: the line needs no column buffer, and painting one
+    // would allocate a cell per character of output this card never redraws —
+    // an `ls -R` or a 5k-line log. Only its own SGR has to be folded, so a later
+    // line that DOES replay enters with the right state.
+    replayed.push(line)
+    for (const match of line.matchAll(SGR_SEQUENCE)) sgr = foldSgr(sgr, String(match[1]))
+  }
+  return replayed.join('\n')
+}
+
+/**
+ * Remove every escape sequence and control character that carries no color,
+ * leaving CSI sequences for anser and `\n`/`\t` for layout. Cursor movements
+ * (carriage return, backspace) replay first, since their effect on the visible
+ * text must land before the characters that expressed them are dropped.
+ * @param text - raw command output.
+ * @returns text whose only remaining escapes are CSI sequences.
+ */
+function sanitize(text: string): string {
+  const escaped = text.replace(OSC_SEQUENCE, '').replace(NON_CSI_ESCAPE, '')
+  return applyCursorMovements(escaped).replace(INERT_CONTROL, '')
+}
+
+/**
+ * Resolve one run's colors and decorations.
+ * @param chunk - the anser chunk to style.
+ * @returns the run's inline style, or undefined when it carries no SGR state.
+ */
+function resolveStyle(chunk: AnsiChunk): CSSProperties | undefined {
+  const style: CSSProperties = {}
+  const background = chunk.bg === null ? undefined : `rgb(${chunk.bg})`
+  if (background !== undefined) style.backgroundColor = background
+  if (chunk.fg !== null) {
+    const literal = `rgb(${chunk.fg})`
+    // A run that paints its own background keeps anser's literal pair so the
+    // authored foreground/background contrast survives; a foreground-only run
+    // maps onto a theme token, which adapts to light and dark surfaces.
+    style.color = background === undefined
+      ? TOKEN_BY_BASIC_RGB[chunk.fg.replace(/\s+/g, '')] ?? literal
+      : literal
+  }
+  for (const decoration of chunk.decorations) Object.assign(style, STYLE_BY_DECORATION[decoration])
+  return Object.keys(style).length === 0 ? undefined : style
+}
+
+/**
+ * Parse command output into styled spans grouped by line.
+ * @param text - raw output text, which may contain ANSI escape sequences.
+ * @returns one entry per output line (always at least one, possibly empty).
+ */
+export function parseAnsiLines(text: string): AnsiLine[] {
+  let current: AnsiSpan[] = []
+  const lines: AnsiSpan[][] = [current]
+  for (const chunk of Anser.ansiToJson(sanitize(text), { json: true, remove_empty: true })) {
+    const style = resolveStyle(chunk)
+    for (const [index, part] of chunk.content.split('\n').entries()) {
+      if (index > 0) {
+        current = []
+        lines.push(current)
+      }
+      if (part !== '') current.push({ text: part, style })
+    }
+  }
+  return lines
+}
diff --git a/packages/client/ui-primitives/src/clipboard.ts b/packages/client/ui-primitives/src/clipboard.ts
new file mode 100644
index 0000000000..c01e802f31
--- /dev/null
+++ b/packages/client/ui-primitives/src/clipboard.ts
@@ -0,0 +1,48 @@
+// Package-internal clipboard write, shared by every copy control in this
+// package (CodeBlock's code copy, TerminalBlock's output copy). Not part of the
+// public surface: consumers get the components, not the host detection.
+
+/**
+ * Write text to the host clipboard, preferring the async Clipboard API and
+ * falling back to `execCommand('copy')` on hosts (jsdom, insecure contexts)
+ * that omit it.
+ * @param text - the exact text to place on the clipboard.
+ * @returns true only when the host accepted the write.
+ */
+export async function writeClipboard(text: string): Promise<boolean> {
+  // lib.dom types clipboard non-optional, but insecure contexts omit it —
+  // that runtime gap is exactly what this guard detects.
+  /* eslint-disable-next-line @typescript-eslint/no-unnecessary-condition */
+  if (navigator.clipboard?.writeText) {
+    try {
+      await navigator.clipboard.writeText(text)
+      return true
+    } catch {
+      // Denied permissions / iframe policy — do not claim success.
+      return false
+    }
+  }
+  // jsdom and older hosts: best-effort execCommand path when present.
+  // execCommand('copy') is the only clipboard fallback where the async API
+  // is missing; deprecated but deliberately retained.
+  /* eslint-disable @typescript-eslint/no-deprecated */
+  const exec = typeof document.execCommand === 'function'
+    ? document.execCommand.bind(document)
+    : undefined
+  if (exec === undefined) return false
+  const el = document.createElement('textarea')
+  el.value = text
+  el.setAttribute('readonly', '')
+  el.style.position = 'fixed'
+  el.style.left = '-9999px'
+  document.body.appendChild(el)
+  el.select()
+  try {
+    return exec('copy')
+  } catch {
+    return false
+  } finally {
+    el.remove()
+  }
+  /* eslint-enable @typescript-eslint/no-deprecated */
+}
diff --git a/packages/client/ui-primitives/src/icons/index.tsx b/packages/client/ui-primitives/src/icons/index.tsx
index 8f5c333778..207f3afefa 100644
--- a/packages/client/ui-primitives/src/icons/index.tsx
+++ b/packages/client/ui-primitives/src/icons/index.tsx
@@ -512,6 +512,18 @@ export const IconPlayOutline16 = ({ size = 16, className }: IconProps) => (
   </svg>
 )
 
+/** ic_ds_pause_outline_16 */
+export const IconPauseOutline16 = ({ size = 16, className }: IconProps) => (
+  <svg width={size} height={size} className={className} viewBox="0 0 16 16" fill="none" xmlns="http://www.w3.org/2000/svg">
+    <path
+      d="M14.1448 8.00024C14.1448 4.60644 11.394 1.85563 8.00024 1.85563C4.60644 1.85563 1.85563 4.60644 1.85563 8.00024C1.85563 11.394 4.60644 14.1448 8.00024 14.1448C11.394 14.1448 14.1448 11.394 14.1448 8.00024ZM15.5112 8.00024C15.5112 12.1482 12.1482 15.5112 8.00024 15.5112C3.85226 15.5112 0.489258 12.1482 0.489258 8.00024C0.489258 3.85226 3.85226 0.489258 8.00024 0.489258C12.1482 0.489258 15.5112 3.85226 15.5112 8.00024Z"
+      fill="currentColor"
+    />
+    <path d="M7.14244 5.14258V10.8569H5.71387V5.14258H7.14244Z" fill="currentColor" />
+    <path d="M10.286 5.14258V10.8569H8.85742V5.14258H10.286Z" fill="currentColor" />
+  </svg>
+)
+
 /** ic_ds_fullscreen_outline_16 */
 export const IconFullscreenOutline16 = ({ size = 16, className }: IconProps) => (
   <svg width={size} height={size} className={className} viewBox="0 0 16 16" fill="none" xmlns="http://www.w3.org/2000/svg">
diff --git a/packages/client/ui-primitives/src/index.ts b/packages/client/ui-primitives/src/index.ts
index 330a3145bf..aa674f7a1a 100644
--- a/packages/client/ui-primitives/src/index.ts
+++ b/packages/client/ui-primitives/src/index.ts
@@ -10,6 +10,7 @@ export { Pill } from './Pill.tsx'
 export { Input } from './Input.tsx'
 export { Menu } from './Menu.tsx'
 export type { MenuEntry, MenuItem, MenuSeparator, MenuLabel } from './Menu.tsx'
+export { useAnchoredMaxHeight } from './useAnchoredMaxHeight.ts'
 export { HoverCard } from './HoverCard.tsx'
 export { Modal } from './Modal.tsx'
 export { ConnectionBanner } from './ConnectionBanner.tsx'
@@ -17,8 +18,14 @@ export { FishLogo } from './FishLogo.tsx'
 export { BrandWordmark } from './BrandWordmark.tsx'
 export { Tooltip } from './Tooltip.tsx'
 export type { TooltipSide } from './Tooltip.tsx'
+export { JsonTree } from './JsonTree.tsx'
+export type { JsonTreeProps } from './JsonTree.tsx'
+export { TerminalBlock, DEFAULT_TERMINAL_MAX_LINES } from './TerminalBlock.tsx'
+export type { TerminalBlockProps } from './TerminalBlock.tsx'
 export { CodeBlock } from './markdown/CodeBlock.tsx'
 export { JsonBlock } from './markdown/JsonBlock.tsx'
 export { MarkdownText } from './markdown/MarkdownText.tsx'
 export { MessageText } from './markdown/MessageText.tsx'
+export { extractMarkdownPlainText } from './markdown/plain-text.ts'
+export type { MarkdownPlainTextMode, MarkdownPlainTextOptions } from './markdown/plain-text.ts'
 export * from './icons/index.tsx'
diff --git a/packages/client/ui-primitives/src/markdown/CodeBlock.module.css b/packages/client/ui-primitives/src/markdown/CodeBlock.module.css
index 7222c3df44..4f78a49935 100644
--- a/packages/client/ui-primitives/src/markdown/CodeBlock.module.css
+++ b/packages/client/ui-primitives/src/markdown/CodeBlock.module.css
@@ -74,6 +74,10 @@
   white-space: pre-wrap;
   word-break: break-all;
   background: var(--dsw-alias-markdown-code-block);
+  /* Bottom radii live on <pre>: overflow:hidden on .block would kill the
+     sticky banner, and this opaque fill otherwise squares off the wrapper. */
+  border-bottom-left-radius: var(--dsl-code-block-border-radius);
+  border-bottom-right-radius: var(--dsl-code-block-border-radius);
 }
 
 /* Shiki inlines its theme background var; route it to the repo token. */
diff --git a/packages/client/ui-primitives/src/markdown/CodeBlock.tsx b/packages/client/ui-primitives/src/markdown/CodeBlock.tsx
index de6a478af4..9c7e968053 100644
--- a/packages/client/ui-primitives/src/markdown/CodeBlock.tsx
+++ b/packages/client/ui-primitives/src/markdown/CodeBlock.tsx
@@ -6,6 +6,7 @@
 
 import { useCallback, useMemo, useRef, useState } from 'react'
 import clsx from 'clsx'
+import { writeClipboard } from '../clipboard.ts'
 import { highlightToHtml } from './highlight.ts'
 import css from './CodeBlock.module.css'
 
@@ -18,45 +19,6 @@ export interface CodeBlockProps {
   className?: string | undefined
 }
 
-/** @returns true only when the host accepted the write. */
-async function writeClipboard(text: string): Promise<boolean> {
-  // lib.dom types clipboard non-optional, but insecure contexts omit it —
-  // that runtime gap is exactly what this guard detects.
-  /* eslint-disable-next-line @typescript-eslint/no-unnecessary-condition */
-  if (navigator.clipboard?.writeText) {
-    try {
-      await navigator.clipboard.writeText(text)
-      return true
-    } catch {
-      // Denied permissions / iframe policy — do not claim success.
-      return false
-    }
-  }
-  // jsdom and older hosts: best-effort execCommand path when present.
-  // execCommand('copy') is the only clipboard fallback where the async API
-  // is missing; deprecated but deliberately retained.
-  /* eslint-disable @typescript-eslint/no-deprecated */
-  const exec = typeof document.execCommand === 'function'
-    ? document.execCommand.bind(document)
-    : undefined
-  if (exec === undefined) return false
-  const el = document.createElement('textarea')
-  el.value = text
-  el.setAttribute('readonly', '')
-  el.style.position = 'fixed'
-  el.style.left = '-9999px'
-  document.body.appendChild(el)
-  el.select()
-  try {
-    return exec('copy')
-  } catch {
-    return false
-  } finally {
-    el.remove()
-  }
-  /* eslint-enable @typescript-eslint/no-deprecated */
-}
-
 export function CodeBlock({ code, lang, className }: CodeBlockProps) {
   const trimmed = code.endsWith('\n') ? code.slice(0, -1) : code
   const html = useMemo(() => highlightToHtml(trimmed, lang), [trimmed, lang])
diff --git a/packages/client/ui-primitives/src/markdown/plain-text.ts b/packages/client/ui-primitives/src/markdown/plain-text.ts
new file mode 100644
index 0000000000..af449dc774
--- /dev/null
+++ b/packages/client/ui-primitives/src/markdown/plain-text.ts
@@ -0,0 +1,124 @@
+/**
+ * Markdown-to-plain-text projection for compact summaries and labels.
+ * Parsing shares the renderer's GFM grammar; raw HTML stays literal, links
+ * keep their labels, images keep alt text, and code keeps its source text.
+ */
+
+import { fromMarkdown } from 'mdast-util-from-markdown'
+import { gfmFromMarkdown } from 'mdast-util-gfm'
+import { gfm } from 'micromark-extension-gfm'
+
+/** Amount of parsed Markdown content returned by the extractor. */
+export type MarkdownPlainTextMode = 'all' | 'first-line' | 'first-paragraph'
+
+/** Options for {@link extractMarkdownPlainText}. */
+export interface MarkdownPlainTextOptions {
+  /** Projection boundary; defaults to the complete document. */
+  mode?: MarkdownPlainTextMode
+}
+
+interface MarkdownNode {
+  type: string
+  value?: string
+  alt?: string
+  children?: MarkdownNode[]
+}
+
+function inlineText(node: MarkdownNode): string {
+  switch (node.type) {
+    case 'text':
+    case 'inlineCode':
+    case 'code':
+      return node.value ?? ''
+    case 'image':
+    case 'imageReference':
+      return node.alt ?? ''
+    case 'break':
+      return '\n'
+    case 'html':
+      return node.value ?? ''
+    default:
+      return node.children?.map(inlineText).join('') ?? ''
+  }
+}
+
+function compactInline(text: string): string {
+  return text.replace(/\s+/g, ' ').trim()
+}
+
+function blockText(node: MarkdownNode): string {
+  switch (node.type) {
+    case 'root':
+    case 'blockquote':
+      return node.children?.map(blockText).filter(Boolean).join('\n\n') ?? ''
+    case 'paragraph':
+    case 'heading':
+      return compactInline(inlineText(node))
+    case 'code':
+      return node.value?.trim() ?? ''
+    case 'list':
+      return node.children?.map(blockText).filter(Boolean).join('\n') ?? ''
+    case 'listItem':
+      return node.children?.map(blockText).filter(Boolean).join(' ') ?? ''
+    case 'table':
+      return node.children?.map(blockText).filter(Boolean).join('\n') ?? ''
+    case 'tableRow':
+      return node.children?.map(blockText).join('\t') ?? ''
+    case 'tableCell':
+      return compactInline(inlineText(node))
+    case 'html':
+      return node.value ?? ''
+    case 'thematicBreak':
+    case 'definition':
+      return ''
+    default:
+      return compactInline(inlineText(node))
+  }
+}
+
+function findFirstParagraph(node: MarkdownNode): string | undefined {
+  if (node.type === 'paragraph') {
+    const text = compactInline(inlineText(node))
+    if (text !== '') return text
+  }
+  for (const child of node.children ?? []) {
+    const text = findFirstParagraph(child)
+    if (text !== undefined) return text
+  }
+  return undefined
+}
+
+function fullText(root: MarkdownNode): string {
+  return blockText(root)
+    .split('\n')
+    .map(line => line.trim())
+    .join('\n')
+    .replace(/\n{3,}/g, '\n\n')
+    .trim()
+}
+
+/**
+ * Parse GFM Markdown, remove its presentation markup, and preserve raw HTML literally.
+ * @param markdown - Markdown source.
+ * @param options - Optional extraction boundary.
+ * @returns Plain text for the whole document, first visible line, or first semantic paragraph.
+ */
+export function extractMarkdownPlainText(
+  markdown: string,
+  options: MarkdownPlainTextOptions = {},
+): string {
+  const { mode = 'all' } = options
+  const root = fromMarkdown(markdown, {
+    extensions: [gfm()],
+    mdastExtensions: [gfmFromMarkdown()],
+  }) as MarkdownNode
+  const all = fullText(root)
+  switch (mode) {
+    case 'all':
+      return all
+    case 'first-line':
+      return all.split('\n').find(line => line !== '') ?? ''
+    case 'first-paragraph':
+      return findFirstParagraph(root) ?? all.split('\n').find(line => line !== '') ?? ''
+  }
+}
diff --git a/packages/client/ui-primitives/src/useAnchoredMaxHeight.ts b/packages/client/ui-primitives/src/useAnchoredMaxHeight.ts
new file mode 100644
index 0000000000..5f3bc52bab
--- /dev/null
+++ b/packages/client/ui-primitives/src/useAnchoredMaxHeight.ts
@@ -0,0 +1,38 @@
+/**
+ * Viewport-fit hook for bottom-anchored overlays (slash menu, popupSelect):
+ * the element's bottom edge is laid out independent of its height, so it
+ * grows upward and only the top edge can collide with the viewport — clamp
+ * the design cap to the space between that edge and the viewport top.
+ */
+import { useLayoutEffect, useState } from 'react'
+import type { RefObject } from 'react'
+
+/** Safe distance kept between the overlay and the viewport top edge (mirrors the Menu portal margin). */
+const MARGIN = 12
+
+/**
+ * Clamp a bottom-anchored overlay's max-height to the viewport.
+ * @param ref - the overlay element; a null current (overlay closed) skips measuring.
+ * @param cap - design max-height in px (the clamp never exceeds it).
+ * @param signal - re-measure trigger: pass the overlay's render state so anchor
+ *   moves (composer growth) re-fit; resize/scroll re-fit while mounted.
+ * @returns the max-height to apply inline, in px.
+ */
+export function useAnchoredMaxHeight(ref: RefObject<HTMLElement>, cap: number, signal: unknown): number {
+  const [maxHeight, setMaxHeight] = useState(cap)
+  useLayoutEffect(() => {
+    const el = ref.current
+    if (el === null) return
+    const fit = () => {
+      setMaxHeight(Math.min(cap, Math.max(0, el.getBoundingClientRect().bottom - MARGIN)))
+    }
+    fit()
+    window.addEventListener('resize', fit)
+    window.addEventListener('scroll', fit, true)
+    return () => {
+      window.removeEventListener('resize', fit)
+      window.removeEventListener('scroll', fit, true)
+    }
+  }, [ref, cap, signal])
+  return maxHeight
+}
diff --git a/packages/client/ui-primitives/tests/ansi.spec.ts b/packages/client/ui-primitives/tests/ansi.spec.ts
new file mode 100644
index 0000000000..008165f5ab
--- /dev/null
+++ b/packages/client/ui-primitives/tests/ansi.spec.ts
@@ -0,0 +1,513 @@
+// parseAnsiLines, the ANSI model behind TerminalBlock: anser's SGR runs
+// resolved into inline styles and folded into per-line span arrays, with every
+// escape and control character that carries no color removed first. The DOM
+// side of the same model (which runs get a span wrapper) is in
+// terminal-block.spec.tsx.
+
+import { describe, expect, it } from 'vitest'
+import { parseAnsiLines } from '../src/ansi.ts'
+
+const ESC = '\u001b'
+const BS = '\u0008'
+/** A combining acute accent: zero-width, so it takes no terminal column. */
+const ACCENT = '\u0301'
+
+/** Paint `text` with the SGR `codes`, then reset. */
+function sgr(codes: string, text: string): string {
+  return `${ESC}[${codes}m${text}${ESC}[0m`
+}
+
+/** The single span of a single-line, single-run parse. */
+function onlySpan(text: string) {
+  const lines = parseAnsiLines(text)
+  expect(lines).toHaveLength(1)
+  expect(lines[0]).toHaveLength(1)
+  return lines[0]![0]!
+}
+
+describe('parseAnsiLines: text without SGR state', () => {
+  it('leaves plain text as one unstyled span', () => {
+    expect(parseAnsiLines('hello')).toEqual([[{ text: 'hello', style: undefined }]])
+  })
+
+  it('returns exactly one empty line for empty input', () => {
+    expect(parseAnsiLines('')).toEqual([[]])
+  })
+
+  it('splits a multi-line run and drops the empty line between two blocks', () => {
+    expect(parseAnsiLines('a\n\nb')).toEqual([
+      [{ text: 'a', style: undefined }],
+      [],
+      [{ text: 'b', style: undefined }],
+    ])
+  })
+
+  it('keeps tabs, which the terminal surface needs for column layout', () => {
+    expect(onlySpan('a\tb')).toEqual({ text: 'a\tb', style: undefined })
+  })
+})
+
+describe('parseAnsiLines: basic colors mapped onto theme tokens', () => {
+  it.each<[string, string, string]>([
+    ['30', 'black', 'var(--dsw-alias-label-primary)'],
+    ['37', 'white', 'var(--dsw-alias-label-primary)'],
+    ['90', 'bright black', 'var(--dsw-alias-label-tertiary)'],
+    ['31', 'red', 'var(--dsw-alias-state-error-primary)'],
+    ['91', 'bright red', 'var(--dsw-alias-state-error-secondary)'],
+    ['32', 'green', 'var(--dsw-alias-state-success-primary)'],
+    ['92', 'bright green', 'var(--dsw-alias-state-success-secondary)'],
+    ['33', 'yellow', 'var(--dsw-alias-state-warn-primary)'],
+    ['93', 'bright yellow', 'var(--dsw-alias-state-warn-secondary)'],
+    ['34', 'blue', 'var(--dsw-alias-state-business-primary)'],
+    ['94', 'bright blue', 'var(--dsw-static-blue-400)'],
+  ])('SGR %s (%s) resolves to %s', (code, _name, token) => {
+    expect(onlySpan(sgr(code, 'x'))).toEqual({ text: 'x', style: { color: token } })
+  })
+})
+
+describe('parseAnsiLines: colors with no token equivalent', () => {
+  it.each<[string, string, string]>([
+    ['35', 'magenta', 'rgb(187, 0, 187)'],
+    ['36', 'cyan', 'rgb(0, 187, 187)'],
+    ['38;5;208', '256-palette orange', 'rgb(255, 135, 0)'],
+    ['38;2;10;20;30', 'truecolor', 'rgb(10, 20, 30)'],
+  ])('SGR %s (%s) falls through to %s', (code, _name, literal) => {
+    expect(onlySpan(sgr(code, 'x')).style).toEqual({ color: literal })
+  })
+})
+
+describe('parseAnsiLines: backgrounds', () => {
+  it('sets backgroundColor for a background-only run', () => {
+    expect(onlySpan(sgr('44', 'x')).style).toEqual({ backgroundColor: 'rgb(0, 0, 187)' })
+  })
+
+  it('keeps the literal foreground when the run paints its own background', () => {
+    expect(onlySpan(sgr('41;37', 'x')).style).toEqual({
+      backgroundColor: 'rgb(187, 0, 0)',
+      color: 'rgb(255,255,255)',
+    })
+  })
+
+  it('renders reverse video as the swapped pair anser reports', () => {
+    expect(onlySpan(sgr('31;7', 'x')).style).toEqual({
+      backgroundColor: 'rgb(187, 0, 0)',
+      color: 'rgb(0, 0, 0)',
+    })
+  })
+})
+
+describe('parseAnsiLines: decorations', () => {
+  it.each<[string, string, Record<string, unknown>]>([
+    ['1', 'bold', { fontWeight: 700 }],
+    ['2', 'dim', { opacity: 0.7 }],
+    ['3', 'italic', { fontStyle: 'italic' }],
+    ['4', 'underline', { textDecoration: 'underline' }],
+    ['9', 'strikethrough', { textDecoration: 'line-through' }],
+    ['8', 'hidden', { visibility: 'hidden' }],
+  ])('SGR %s (%s) resolves to %o', (code, _name, style) => {
+    expect(onlySpan(sgr(code, 'x')).style).toEqual(style)
+  })
+
+  it('lets the later textDecoration win when a run declares underline and strikethrough', () => {
+    expect(onlySpan(sgr('4;9', 'x')).style).toEqual({ textDecoration: 'line-through' })
+    expect(onlySpan(sgr('9;4', 'x')).style).toEqual({ textDecoration: 'underline' })
+  })
+
+  it('combines a color with several decorations in one style', () => {
+    expect(onlySpan(sgr('1;3;31', 'x')).style).toEqual({
+      color: 'var(--dsw-alias-state-error-primary)',
+      fontWeight: 700,
+      fontStyle: 'italic',
+    })
+  })
+
+  it('reproduces no animation for blink, leaving the run unstyled', () => {
+    expect(onlySpan(sgr('5', 'x'))).toEqual({ text: 'x', style: undefined })
+  })
+})
+
+describe('parseAnsiLines: sequences that carry no color', () => {
+  it('removes an OSC string with its BEL terminator', () => {
+    expect(onlySpan(`a${ESC}]0;window title\u0007b`)).toEqual({ text: 'ab', style: undefined })
+  })
+
+  it('removes an OSC string terminated by ST', () => {
+    expect(onlySpan(`a${ESC}]8;;https://example.com${ESC}\\b`)).toEqual({ text: 'ab', style: undefined })
+  })
+
+  it('removes non-CSI escapes such as charset selection and reset', () => {
+    expect(onlySpan(`x${ESC}(By${ESC}cz`)).toEqual({ text: 'xyz', style: undefined })
+  })
+
+  it('removes inert C0 controls', () => {
+    expect(onlySpan('\u0000ab\u001fc\u007f')).toEqual({ text: 'abc', style: undefined })
+  })
+
+  it('keeps CSI sequences that only move the cursor out of the text', () => {
+    expect(onlySpan(`${ESC}[2K${ESC}[1Adone`)).toEqual({ text: 'done', style: undefined })
+  })
+})
+
+describe('parseAnsiLines: carriage returns', () => {
+  it('keeps only the last redraw of a line', () => {
+    expect(onlySpan('10%\r55%\r100%')).toEqual({ text: '100%', style: undefined })
+  })
+
+  it('leaves the tail of a longer frame standing under a shorter redraw', () => {
+    // Verified against a real terminal: `100%\rOK` paints `OK0%`. A carriage
+    // return only moves the cursor, so the two columns the redraw never reaches
+    // still hold the frame beneath — truncating to the last `\r` would lose them.
+    expect(onlySpan('100%\rOK')).toEqual({ text: 'OK0%', style: undefined })
+    expect(onlySpan('abcdef\rXY')).toEqual({ text: 'XYcdef', style: undefined })
+  })
+
+  it('clamps a backspace run at the line start rather than going negative', () => {
+    // More backspaces than characters: the cursor stops at column 0, so the
+    // following write simply overwrites from there.
+    expect(onlySpan(`ab${BS}${BS}${BS}${BS}xyz`)).toEqual({ text: 'xyz', style: undefined })
+  })
+
+  it('keeps SGR state in force across a redraw, as a terminal does', () => {
+    // Verified against a real terminal: `\x1b[31mgone\rkept` paints `kept` RED.
+    // A carriage return moves the cursor; it does not reset the graphic state,
+    // so the redraw inherits the color the discarded frame was written with.
+    expect(onlySpan(`${ESC}[31mgone\rkept`))
+      .toEqual({ text: 'kept', style: { color: 'var(--dsw-alias-state-error-primary)' } })
+  })
+
+  it('preserves both lines of a CRLF pair instead of treating it as a redraw', () => {
+    expect(parseAnsiLines('a\r\r\nb\r\n')).toEqual([
+      [{ text: 'a', style: undefined }],
+      [{ text: 'b', style: undefined }],
+      [],
+    ])
+  })
+
+  it('applies the redraw per line, not across the whole text', () => {
+    expect(parseAnsiLines('one\rtwo\nthree')).toEqual([
+      [{ text: 'two', style: undefined }],
+      [{ text: 'three', style: undefined }],
+    ])
+  })
+})
+
+describe('parseAnsiLines: backspaces', () => {
+  it('applies a backspace as the overwrite a terminal draws', () => {
+    // `abc` then two backspaces then `XY` shows as `aXY`, not `abcXY`.
+    expect(onlySpan(`abc${BS}${BS}XY`)).toEqual({ text: 'aXY', style: undefined })
+  })
+
+  it('stops at the line start instead of eating the newline before it', () => {
+    expect(parseAnsiLines(`ab\n${BS}${BS}${BS}cd`)).toEqual([
+      [{ text: 'ab', style: undefined }],
+      [{ text: 'cd', style: undefined }],
+    ])
+  })
+
+  it('treats a trailing backspace as a cursor move, not a delete', () => {
+    // Verified against a real terminal: `abc\b` still shows `abc`. Only a later
+    // write overwrites; a backspace with nothing after it erases nothing.
+    expect(onlySpan(`abc${BS}`)).toEqual({ text: 'abc', style: undefined })
+    // Same at a line boundary: the newline ends the line before any overwrite.
+    expect(parseAnsiLines(`abc${BS}\ndef`)).toEqual([
+      [{ text: 'abc', style: undefined }],
+      [{ text: 'def', style: undefined }],
+    ])
+  })
+
+  it('steps over an SGR sequence instead of erasing its bytes', () => {
+    // `abc` reset then two backspaces then `XY`: erasing the reset's bytes would
+    // corrupt it and repaint the rest of the line with whatever the remainder
+    // parses as. The visible result is `aXY`, still red, with the reset intact.
+    expect(parseAnsiLines(`${sgr('31', 'abc')}${BS}${BS}XY`)).toEqual([[
+      { text: 'a', style: { color: 'var(--dsw-alias-state-error-primary)' } },
+      { text: 'XY', style: undefined },
+    ]])
+  })
+
+  it('erases across a style boundary without dropping the styles between', () => {
+    // The backspace reaches back past the reset to the last printed character.
+    expect(parseAnsiLines(`${sgr('32', 'ok')}${ESC}[31m${BS}bad`)).toEqual([[
+      { text: 'o', style: { color: 'var(--dsw-alias-state-success-primary)' } },
+      { text: 'bad', style: { color: 'var(--dsw-alias-state-error-primary)' } },
+    ]])
+  })
+
+  it('replays a redraw and a trailing backspace as pure cursor moves', () => {
+    // Verified against a real terminal: `old\rnew\b` shows `new`. The redraw
+    // repaints all three columns and the trailing backspace only moves the
+    // cursor left — nothing overwrites the `w`, so nothing is lost.
+    expect(onlySpan(`old\rnew${BS}`)).toEqual({ text: 'new', style: undefined })
+  })
+
+  it('overwrites only the columns the later write reaches, keeping the rest styled', () => {
+    // Verified against a real terminal: red `bad`, three backspaces, then `ok`
+    // shows `okd` — the cursor returned to column 0 and `ok` overwrote two of
+    // the three columns, so the untouched `d` keeps the run's red.
+    expect(parseAnsiLines(`${sgr('31', 'bad')}${BS}${BS}${BS}ok`)).toEqual([[
+      { text: 'ok', style: undefined },
+      { text: 'd', style: { color: 'var(--dsw-alias-state-error-primary)' } },
+    ]])
+  })
+})
+
+describe('parseAnsiLines: erase and column arithmetic', () => {
+  it('erases the rest of the line, the fixed companion of a redraw', () => {
+    // Verified in a real terminal: `100%\r\x1b[KOK` shows `OK`. Every spinner and
+    // progress bar writes `\r\x1b[K`; without the erase the previous frame's tail
+    // stands and the card shows text the terminal never displayed.
+    expect(onlySpan(`100%\r${ESC}[KOK`)).toEqual({ text: 'OK', style: undefined })
+    // The parameterless form and `0` are the same erase.
+    expect(onlySpan(`100%\r${ESC}[0KOK`)).toEqual({ text: 'OK', style: undefined })
+  })
+
+  it('erases the whole line for the 2K form and to the cursor for 1K', () => {
+    expect(onlySpan(`ab\r${ESC}[2Kxy`)).toEqual({ text: 'xy', style: undefined })
+    // 1K clears left of the cursor without moving it, so those columns read as
+    // blanks — verified in a real terminal, which shows `    |` for this input.
+    expect(onlySpan(`abcd${ESC}[1K|`)).toEqual({ text: '    |', style: undefined })
+  })
+
+  it('paints columns a 2K dropped as blanks when a later write lands past them', () => {
+    // 2K clears the line but leaves the cursor where it was, so writing there
+    // leaves the columns before it unwritten — blanks, as a terminal shows.
+    expect(onlySpan(`abcd${ESC}[2Kx`)).toEqual({ text: '    x', style: undefined })
+  })
+
+  it('advances a redraw cursor by tab stops, leaving a tabbed column standing', () => {
+    // Verified in a real terminal: `a\tb\rXY` shows `XY      b` — the `b` sits at
+    // column 8, which a two-character redraw cannot reach. Counting the tab as
+    // one column would have produced `XYb` and destroyed the alignment.
+    expect(onlySpan('a\tb\rXY')).toEqual({ text: 'XY      b', style: undefined })
+  })
+
+  it('counts a wide character as the two columns a terminal advances', () => {
+    // `中` occupies two cells, so a two-character redraw covers exactly it.
+    expect(onlySpan('中x\rab')).toEqual({ text: 'abx', style: undefined })
+  })
+
+  it('does not accumulate a cursor or erase sequence into a cell style', () => {
+    // Only SGR carries graphic state. An erase folded into the style string
+    // would grow it per redraw and emit boundaries anser has to discard.
+    expect(parseAnsiLines(`${ESC}[31ma\r${ESC}[Kb`)).toEqual([[
+      { text: 'b', style: { color: 'var(--dsw-alias-state-error-primary)' } },
+    ]])
+  })
+})
+
+describe('parseAnsiLines: line-end state and column widths', () => {
+  it('closes a run whose reset lands after the last written cell', () => {
+    // Verified in a real terminal: `\x1b[32mdone\rok\x1b[0m` then `plain` shows
+    // `okne` GREEN and `plain` in the DEFAULT color. The reset changes no cell,
+    // so returning the last cell's state leaked green onto every later line —
+    // and this exact shape (`\r\x1b[K\x1b[32m✓ built\x1b[0m`) is what every
+    // build tool writes.
+    expect(parseAnsiLines(`${ESC}[32mdone\rok${ESC}[0m\nplain`)).toEqual([
+      [{ text: 'okne', style: { color: 'var(--dsw-alias-state-success-primary)' } }],
+      [{ text: 'plain', style: undefined }],
+    ])
+  })
+
+  it('erases through the cursor column for 1K, not up to it', () => {
+    // Verified in a real terminal: `abcd\b\x1b[1K|` shows `   |` — the `d` under
+    // the cursor is erased too, which the CSI spec calls inclusive.
+    expect(onlySpan(`abcd${BS}${ESC}[1K|`)).toEqual({ text: '   |', style: undefined })
+  })
+
+  it('gives a combining mark no column of its own', () => {
+    // Verified in a real terminal: `é` (e + U+0301) then `x`, redrawn with `YZ`,
+    // shows `YZ`. Counting the mark as a column left the `x` standing.
+    expect(onlySpan('e\u0301x\rYZ')).toEqual({ text: 'YZ', style: undefined })
+  })
+
+  it('drops a combining mark left with no cell to attach to by a redraw', () => {
+    // Verified in a real terminal: `ab` then CR then U+0301 then `x` shows `xb`.
+    // The redraw puts the cursor at column 0, so the mark has no preceding cell
+    // and the terminal shows nothing for it rather than a lone accent.
+    expect(onlySpan(`ab\r${ACCENT}x`)).toEqual({ text: 'xb', style: undefined })
+    // A mark with no movement on its line never reaches the replay at all: it
+    // is width business, not a cursor move, so it stays as authored.
+    expect(onlySpan(`${ACCENT}abc`)).toEqual({ text: `${ACCENT}abc`, style: undefined })
+  })
+
+  it('carries a colour opened after the last write onto the next line', () => {
+    // The mirror of the reset case, verified in a real terminal: `ab` CR `X` then
+    // `\x1b[31m` with nothing after it shows `Xb` UNSTYLED and the next line red.
+    // The scan ends styled while the last cell is not, so the convergence has to
+    // open the run at the line end for it to reach the following line.
+    expect(parseAnsiLines(`ab\rX${ESC}[31m\nnext`)).toEqual([
+      [{ text: 'Xb', style: undefined }],
+      [{ text: 'next', style: { color: 'var(--dsw-alias-state-error-primary)' } }],
+    ])
+  })
+
+  it('blanks a wide character\'s spacer once its lead cell is overwritten', () => {
+    // Verified in a real terminal: `中x` redrawn with `A` shows `A x` — the wide
+    // glyph's second cell becomes a blank rather than closing the gap, so the
+    // `x` keeps column 3.
+    expect(onlySpan('中x\rA')).toEqual({ text: 'A x', style: undefined })
+    // Covering both of its columns leaves no spacer behind.
+    expect(onlySpan('中x\rab')).toEqual({ text: 'abx', style: undefined })
+  })
+
+  it('replays an erase whose parameters carry a semicolon', () => {
+    // The replay guard has to match the same CSI shape the parser accepts, or a
+    // form like `\x1b[1;2K` skips the replay and its erase never happens.
+    expect(onlySpan(`abcd${ESC}[1;2K|`)).toEqual({ text: '    |', style: undefined })
+  })
+})
+
+describe('parseAnsiLines: bounded state and true widths', () => {
+  it('emits one canonical sequence per boundary however the state was reached', () => {
+    // Colors that never fully reset used to accumulate raw sequence history per
+    // cell, so every boundary re-emitted the whole chain: 3200 such cells
+    // produced 25 MB and eventually a RangeError. The state is normalized now,
+    // so the emitted text stays linear in the number of cells.
+    let input = ''
+    for (let index = 0; index < 2000; index += 1) input += `${ESC}[3${index % 6 + 1}mx`
+    const emitted = parseAnsiLines(`${input}\rz`)[0] ?? []
+    expect(emitted.reduce((total, span) => total + span.text.length, 0)).toBe(2000)
+  })
+
+  it('closes an attribute with its closer instead of appending to the state', () => {
+    // `1` then `22` is bold then not-bold, which every chalk-based tool writes;
+    // appending both left the cell bold and grew the chain.
+    // Verified in a real terminal: the `22` closes the bold, so the `x` written
+    // after the redraw is PLAIN. Appending both left it bold and grew the chain.
+    expect(parseAnsiLines(`${ESC}[1mbold${ESC}[22mplain\r${ESC}[Kx`)).toEqual([[
+      { text: 'x', style: undefined },
+    ]])
+    expect(parseAnsiLines(`${ESC}[1mA${ESC}[22mB`)).toEqual([[
+      { text: 'A', style: { fontWeight: 700 } },
+      { text: 'B', style: undefined },
+    ]])
+  })
+
+  it('folds extended colors, backgrounds and every attribute closer', () => {
+    // The 256-palette and truecolor forms consume their own arguments, so the
+    // fold has to take them whole rather than as separate codes.
+    expect(parseAnsiLines(`${ESC}[38;5;208mA\r${ESC}[KB`)).toEqual([[
+      { text: 'B', style: { color: 'rgb(255, 135, 0)' } },
+    ]])
+    expect(parseAnsiLines(`${ESC}[38;2;10;20;30mA\r${ESC}[KB`)).toEqual([[
+      { text: 'B', style: { color: 'rgb(10, 20, 30)' } },
+    ]])
+    // A background survives the same way, and `49` closes it.
+    expect(parseAnsiLines(`${ESC}[41mA${ESC}[49mB\r${ESC}[KC`)).toEqual([[
+      { text: 'C', style: undefined },
+    ]])
+    // Each closer drops only its own attribute: `4` underline closed by `24`
+    // while the italic opened before it stays in force.
+    expect(parseAnsiLines(`${ESC}[3;4mA${ESC}[24mB\r${ESC}[KC`)).toEqual([[
+      { text: 'C', style: { fontStyle: 'italic' } },
+    ]])
+    // `39` closes a foreground without touching the background.
+    expect(parseAnsiLines(`${ESC}[31;42mA${ESC}[39mB\r${ESC}[KC`)).toEqual([[
+      { text: 'C', style: { backgroundColor: 'rgb(0, 187, 0)' } },
+    ]])
+  })
+
+  it('folds the remaining SGR shapes the model has to carry', () => {
+    // A 48-background in extended form, so the `48` arm and the `2`-span both run.
+    expect(parseAnsiLines(`${ESC}[48;2;1;2;3mA\r${ESC}[KB`)).toEqual([[
+      { text: 'B', style: { backgroundColor: 'rgb(1, 2, 3)' } },
+    ]])
+    // A bright foreground and a bright background, the 90-97 / 100-107 arms.
+    expect(parseAnsiLines(`${ESC}[91mA\r${ESC}[KB`)).toEqual([[
+      { text: 'B', style: { color: 'var(--dsw-alias-state-error-secondary)' } },
+    ]])
+    expect(parseAnsiLines(`${ESC}[101mA\r${ESC}[KB`)).toEqual([[
+      { text: 'B', style: { backgroundColor: 'rgb(255, 85, 85)' } },
+    ]])
+    // An extended form with no recognized kind byte consumes nothing extra.
+    expect(parseAnsiLines(`${ESC}[38mA\r${ESC}[KB`)).toEqual([[{ text: 'B', style: undefined }]])
+    // Re-opening an attribute already in force does not duplicate it, and a bare
+    // `\x1b[m` resets exactly as `\x1b[0m` does.
+    expect(parseAnsiLines(`${ESC}[1m${ESC}[1mA${ESC}[mB\r${ESC}[KC`)).toEqual([[
+      { text: 'C', style: undefined },
+    ]])
+  })
+
+  it('treats a text-presentation symbol as one column', () => {
+    // Verified in a real terminal: `A✓B` redrawn with `XY` shows `XYB`, so the
+    // check mark is ONE column. Taking the whole U+2600-U+27BF block as wide
+    // misaligned exactly the progress output this card exists to show.
+    expect(onlySpan('A\u2713B\rXY')).toEqual({ text: 'XYB', style: undefined })
+    // An emoji-presentation character is two, so the same redraw leaves a blank.
+    expect(onlySpan('A\u{1f600}B\rXY')).toEqual({ text: 'XY B', style: undefined })
+  })
+
+  it('clears a wide pair from either side, including through an erase', () => {
+    // Verified in a real terminal (`A x`): the redraw puts the cursor at column
+    // 0, the backspace clamps there, and writing `A` over the wide lead blanks
+    // its spacer rather than letting the `x` slide left.
+    expect(onlySpan(`\u4e2dx\r${BS}A`)).toEqual({ text: 'A x', style: undefined })
+    // An erase reaching the lead blanks its spacer through the same helper.
+    // Verified in a real terminal (`   |`): 1K blanks through the cursor column,
+    // so the wide glyph's two cells and the `x` all become blanks.
+    expect(onlySpan(`\u4e2dx${ESC}[1K|`)).toEqual({ text: '   |', style: undefined })
+  })
+
+  it('clears the lead when the write lands on the spacer itself', () => {
+    // Two backspaces from after `中x` stop ON the wide glyph's second cell;
+    // writing there blanks the lead through the spacer side of the pair clear,
+    // so the glyph cannot survive as half a character.
+    expect(onlySpan(`中x${BS}${BS}A`)).toEqual({ text: ' Ax', style: undefined })
+  })
+
+  it('keeps a surviving spacer as a blank when its lead was replaced by a spacer', () => {
+    // `好` written over the first glyph's spacer puts its own spacer on the
+    // second glyph's lead cell — a write that goes down without a pair clear.
+    // The second glyph's spacer survives with a dead lead and must emit a
+    // blank, or everything after it shifts one column left.
+    expect(onlySpan(`中中${BS}${BS}${BS}好`)).toEqual({ text: ' 好 ', style: undefined })
+  })
+
+  it('blanks both halves of a wide pair when either is overwritten', () => {
+    // A terminal cannot leave one cell of a two-cell glyph standing, so writing
+    // over the spacer clears the lead as well.
+    // Verified in a real terminal: two wide chars, CR, then `A` shows `A ` and
+    // the second glyph — writing the lead cell blanks its spacer, so the column
+    // stays occupied rather than collapsing.
+    expect(onlySpan('\u4e2d\u4e2d\rA')).toEqual({ text: 'A \u4e2d', style: undefined })
+  })
+})
+
+describe('parseAnsiLines: SGR across lines', () => {
+  it('carries active state past a newline, as a terminal does', () => {
+    // Verified in a real terminal: `\x1b[31mabc\rX\nnext` paints BOTH lines red.
+    // A newline does not reset the graphic state, so a replayed line must hand
+    // its state to the next one instead of closing it off.
+    expect(parseAnsiLines(`${ESC}[31mabc\rX\nnext`)).toEqual([
+      [{ text: 'Xbc', style: { color: 'var(--dsw-alias-state-error-primary)' } }],
+      [{ text: 'next', style: { color: 'var(--dsw-alias-state-error-primary)' } }],
+    ])
+  })
+
+  it('tracks state through a line that needs no replay', () => {
+    // The middle line has no movement, so it is not replayed — but its own SGR
+    // still has to reach the line after it.
+    expect(parseAnsiLines(`a\r${ESC}[32mb\nplain\nc`)).toEqual([
+      [{ text: 'b', style: { color: 'var(--dsw-alias-state-success-primary)' } }],
+      [{ text: 'plain', style: { color: 'var(--dsw-alias-state-success-primary)' } }],
+      [{ text: 'c', style: { color: 'var(--dsw-alias-state-success-primary)' } }],
+    ])
+  })
+})
+
+describe('parseAnsiLines: runs spanning lines', () => {
+  it('carries one run\'s style onto every line it covers', () => {
+    expect(parseAnsiLines(sgr('32', 'first\nsecond'))).toEqual([
+      [{ text: 'first', style: { color: 'var(--dsw-alias-state-success-primary)' } }],
+      [{ text: 'second', style: { color: 'var(--dsw-alias-state-success-primary)' } }],
+    ])
+  })
+
+  it('keeps several runs of one line in order', () => {
+    expect(parseAnsiLines(`plain${sgr('31', 'red')}tail`)).toEqual([[
+      { text: 'plain', style: undefined },
+      { text: 'red', style: { color: 'var(--dsw-alias-state-error-primary)' } },
+      { text: 'tail', style: undefined },
+    ]])
+  })
+})
diff --git a/packages/client/ui-primitives/tests/atoms.spec.tsx b/packages/client/ui-primitives/tests/atoms.spec.tsx
index 7724afa493..51b14faa17 100644
--- a/packages/client/ui-primitives/tests/atoms.spec.tsx
+++ b/packages/client/ui-primitives/tests/atoms.spec.tsx
@@ -123,6 +123,7 @@ describe('Menu', () => {
     render(
       <Menu
         open
+        compact
         anchor={<span>trigger</span>}
         items={[
           { id: 'a', label: 'Alpha', icon: <svg data-testid="ic" /> },
@@ -186,6 +187,7 @@ describe('Menu', () => {
     render(
       <Menu
         open
+        compact
         anchor={<span>trigger</span>}
         items={[
           { id: 'plain', label: 'Plain' },
diff --git a/packages/client/ui-primitives/tests/icons.spec.tsx b/packages/client/ui-primitives/tests/icons.spec.tsx
index 487fbb696e..e43d10666a 100644
--- a/packages/client/ui-primitives/tests/icons.spec.tsx
+++ b/packages/client/ui-primitives/tests/icons.spec.tsx
@@ -14,8 +14,8 @@ const icons = Object.fromEntries(
 const iconNames = Object.keys(icons)
 
 describe('ic_ds_ icon set', () => {
-  it('exports the full P-I set (43 deepsuite + 13 figma extracts + the hand-authored sparkle)', () => {
-    expect(iconNames.length).toBe(57)
+  it('exports the full P-I set (44 deepsuite + 13 figma extracts + the hand-authored sparkle)', () => {
+    expect(iconNames.length).toBe(58)
   })
 
   it.each(iconNames)('%s renders an svg with currentColor fills and no hardcoded palette', (name) => {
diff --git a/packages/client/ui-primitives/tests/json-tree.spec.tsx b/packages/client/ui-primitives/tests/json-tree.spec.tsx
new file mode 100644
index 0000000000..b9dd962ad3
--- /dev/null
+++ b/packages/client/ui-primitives/tests/json-tree.spec.tsx
@@ -0,0 +1,339 @@
+// @vitest-environment jsdom
+
+import { act, cleanup, fireEvent, render, screen, waitFor, within } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import { JsonTree } from '@deepseek-ai/dsh-client-ui-primitives'
+
+let writeText: ReturnType<typeof vi.fn>
+
+beforeEach(() => {
+  writeText = vi.fn().mockResolvedValue(undefined)
+  Object.defineProperty(navigator, 'clipboard', {
+    configurable: true,
+    value: { writeText },
+  })
+})
+
+afterEach(() => {
+  cleanup()
+  vi.useRealTimers()
+})
+
+describe('JsonTree', () => {
+  it('keeps the top level open and renders expandable value previews', () => {
+    render(
+      <JsonTree
+        label="Payload"
+        data={{
+          nested: { answer: 42 },
+          list: ['alpha', 'beta'],
+        }}
+      />,
+    )
+
+    const tree = screen.getByRole('tree', { name: 'Payload' })
+    const rows = within(tree).getAllByRole('treeitem')
+    expect(rows).toHaveLength(2)
+    expect(rows[0]?.textContent).toBe('nested:{answer: 42},')
+    expect(rows[1]?.textContent).toBe('list:["alpha", "beta"]')
+
+    const expanders = within(tree).getAllByRole('button', { name: 'Expand JSON node' })
+    expect(expanders[0]?.tabIndex).toBe(0)
+    expect(expanders[1]?.tabIndex).toBe(-1)
+
+    fireEvent.click(expanders[0] as HTMLElement)
+    expect(within(tree).getAllByRole('treeitem')).toHaveLength(3)
+    expect(screen.getByText('answer:')).toBeDefined()
+    expect(within(tree).getByRole('button', { name: 'Collapse JSON node' })).toBeDefined()
+  })
+
+  it('moves the single tab stop between visible expanders with arrow keys', () => {
+    render(
+      <JsonTree
+        expandTopLevel={false}
+        data={{
+          first: { nested: 1 },
+          second: { nested: 2 },
+        }}
+      />,
+    )
+
+    const tree = screen.getByRole('tree', { name: 'JSON' })
+    const root = within(tree).getByRole('button', { name: 'Collapse JSON node' })
+    const children = within(tree).getAllByRole('button', { name: 'Expand JSON node' })
+
+    expect(root.tabIndex).toBe(0)
+    fireEvent.keyDown(root, { key: 'ArrowDown' })
+    expect(document.activeElement).toBe(children[0])
+    expect(root.tabIndex).toBe(-1)
+    expect(children[0]?.tabIndex).toBe(0)
+
+    fireEvent.keyDown(children[0] as HTMLElement, { key: 'ArrowRight' })
+    expect(children[0]?.getAttribute('aria-expanded')).toBe('true')
+    fireEvent.keyDown(children[0] as HTMLElement, { key: 'ArrowLeft' })
+    expect(children[0]?.getAttribute('aria-expanded')).toBe('false')
+    fireEvent.keyDown(children[0] as HTMLElement, { key: 'Enter' })
+
+    fireEvent.keyDown(children[0] as HTMLElement, { key: 'ArrowUp' })
+    expect(document.activeElement).toBe(root)
+    fireEvent.keyDown(root, { key: 'ArrowUp' })
+    expect(document.activeElement).toBe(children[1])
+  })
+
+  it('copies an array element path without recovering data from rendered labels', async () => {
+    render(<JsonTree data={{ list: [{ value: 'x' }, 'tail'] }} />)
+
+    const tree = screen.getByRole('tree')
+    fireEvent.click(within(tree).getByRole('button', { name: 'Expand JSON node' }))
+    const arrayRow = within(tree).getAllByRole('treeitem')
+      .find(row => row.textContent?.startsWith('0:'))
+    expect(arrayRow).toBeDefined()
+
+    fireEvent.mouseOver(arrayRow as HTMLElement)
+    const copyButton = screen.getByRole('button', { name: 'Copy pretty JSON' })
+    fireEvent.contextMenu(copyButton)
+    fireEvent.click(screen.getByRole('menuitem', { name: 'Copy property path' }))
+
+    await waitFor(() => {
+      expect(writeText).toHaveBeenCalledWith('$.list[0]')
+    })
+  })
+
+  it('renders empty containers, JSON-adjacent primitives, and bounded deep previews', () => {
+    const anonymous = Object.defineProperty(() => {}, 'name', { value: '' })
+    const date = new Date('2026-07-28T00:00:00.000Z')
+    const data = {
+      '': 'empty key',
+      nil: null,
+      text: 'quoted',
+      flag: true,
+      count: 3,
+      big: 4n,
+      date,
+      named: function named() {},
+      missing: undefined,
+      symbol: Symbol('token'),
+      emptyObject: {},
+      emptyArray: [],
+      primitivePreview: {
+        nil: null,
+        flag: false,
+        big: 9n,
+        missing: undefined,
+      },
+      exoticPreview: {
+        symbol: Symbol(),
+        named: function sample() {},
+        anonymous,
+        date,
+      },
+      wideObject: { a: 1, b: 2, c: 3, d: 4, e: 5 },
+      wideArray: [1, 2, 3, 4, 5, 6],
+      deep: { a: { b: { c: 1 } } },
+    }
+    render(<JsonTree copyable={false} data={data} />)
+
+    const text = screen.getByRole('tree').textContent
+    expect(text).toContain('"":\"empty key\"')
+    expect(text).toContain('nil:null')
+    expect(text).toContain('flag:true')
+    expect(text).toContain('count:3')
+    expect(text).toContain('big:4n')
+    expect(text).toContain('date:2026-07-28T00:00:00.000Z')
+    expect(text).toContain('named:function() { }')
+    expect(text).toContain('missing:undefined')
+    expect(text).toContain('symbol:Symbol(token)')
+    expect(text).toContain('emptyObject:{}')
+    expect(text).toContain('emptyArray:[]')
+    expect(text).toContain('primitivePreview:{nil: null, flag: false, big: 9, missing: undefined}')
+    expect(text).toContain('exoticPreview:{symbol: Symbol, named: sample, anonymous: Function, date: }')
+    expect(text).toContain('wideObject:{a: 1, b: 2, c: 3, d: 4, …}')
+    expect(text).toContain('wideArray:[1, 2, 3, 4, 5, …]')
+    expect(text).toContain('deep:{a: {b: {…}}}')
+    expect(screen.queryByRole('button', { name: /Copy/ })).toBeNull()
+    fireEvent.mouseOver(screen.getByRole('tree').parentElement as HTMLElement)
+  })
+
+  it('renders child commas and lets a clickable property label toggle its node', () => {
+    render(<JsonTree data={{ parent: { emptyObject: {}, emptyArray: [], scalar: 1, last: 2 } }} />)
+
+    fireEvent.click(screen.getByText('parent:'))
+    const tree = screen.getByRole('tree')
+    const rows = within(tree).getAllByRole('treeitem')
+    expect(rows.find(row => row.textContent === 'emptyObject:{},')).toBeDefined()
+    expect(rows.find(row => row.textContent === 'emptyArray:[],')).toBeDefined()
+    expect(rows.find(row => row.textContent === 'scalar:1,')).toBeDefined()
+    expect(rows.find(row => row.textContent === 'last:2')).toBeDefined()
+
+    fireEvent.click(screen.getByText('parent:'))
+    expect(within(tree).getAllByRole('treeitem')).toHaveLength(1)
+  })
+
+  it('assigns the initial array tab stop and supports an empty collapsible root', () => {
+    const first = render(<JsonTree data={['plain', { nested: true }]} />)
+    const tree = screen.getByRole('tree')
+    expect(tree.textContent).toContain('0:"plain"')
+    expect(within(tree).getByRole('button', { name: 'Expand JSON node' }).tabIndex).toBe(0)
+    first.unmount()
+
+    render(<JsonTree expandTopLevel={false} data={{}} />)
+    expect(screen.getByRole('tree').textContent).toBe('{}')
+    expect(screen.queryByRole('button', { name: /JSON node/ })).toBeNull()
+  })
+
+  it('copies primitive and object values in every menu mode', async () => {
+    const anonymous = Object.defineProperty(() => {}, 'name', { value: '' })
+    render(
+      <JsonTree
+        data={{
+          plain: 'hello',
+          'odd-key': 3,
+          object: { a: 1 },
+          missing: undefined,
+          big: 7n,
+          symbol: Symbol(),
+          symbolNamed: Symbol('token'),
+          named: function named() {},
+          anonymous,
+        }}
+      />,
+    )
+
+    const tree = screen.getByRole('tree')
+    const row = (prefix: string) => {
+      const match = within(tree).getAllByRole('treeitem')
+        .find(item => item.textContent?.startsWith(prefix))
+      expect(match).toBeDefined()
+      return match as HTMLElement
+    }
+    const hover = (prefix: string) => {
+      fireEvent.mouseOver(row(prefix))
+      return screen.getByRole('button', { name: /Cop/ })
+    }
+    const select = (name: string) => {
+      const button = screen.getByRole('button', { name: /Cop/ })
+      fireEvent.contextMenu(button)
+      fireEvent.click(screen.getByRole('menuitem', { name }))
+    }
+
+    fireEvent.click(hover('plain:'))
+    await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith('hello') })
+
+    hover('odd-key:')
+    select('Copy property path')
+    await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith('$["odd-key"]') })
+    select('Copy JSON')
+    await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith('3') })
+    fireEvent.click(hover('odd-key:'))
+    await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith('3') })
+
+    fireEvent.click(hover('object:'))
+    await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith('{\n  "a": 1\n}') })
+    select('Copy compact JSON')
+    await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith('{"a":1}') })
+
+    for (const [prefix, expected] of [
+      ['missing:', 'undefined'],
+      ['big:', '7'],
+      ['symbol:', 'Symbol'],
+      ['symbolNamed:', 'token'],
+      ['named:', 'named'],
+      ['anonymous:', 'Function'],
+    ] as const) {
+      fireEvent.click(hover(prefix))
+      await waitFor(() => { expect(writeText).toHaveBeenLastCalledWith(expected) })
+    }
+  })
+
+  it('reports clipboard failure, resets feedback, and clears a prior timer', async () => {
+    vi.useFakeTimers()
+    writeText.mockRejectedValue(new Error('denied'))
+    const view = render(<JsonTree data={{ value: 'x' }} />)
+    const row = screen.getByRole('treeitem')
+    fireEvent.mouseOver(row)
+    fireEvent.click(screen.getByRole('button', { name: 'Copy value' }))
+    await act(async () => { await Promise.resolve() })
+    expect(screen.getByRole('button', { name: 'Copy failed' })).toBeDefined()
+
+    fireEvent.click(screen.getByRole('button', { name: 'Copy failed' }))
+    await act(async () => { await Promise.resolve() })
+    act(() => { vi.advanceTimersByTime(1_500) })
+    expect(screen.getByRole('button', { name: 'Copy value' })).toBeDefined()
+    view.unmount()
+  })
+
+  it('keeps copy placement synchronized and clears stale targets', () => {
+    const view = render(<JsonTree data={{ first: { a: 1 }, second: 2 }} />)
+    const root = view.container.firstElementChild as HTMLElement
+    const tree = screen.getByRole('tree')
+    const firstRow = within(tree).getAllByRole('treeitem')[0] as HTMLElement
+    const secondRow = within(tree).getAllByRole('treeitem')[1] as HTMLElement
+
+    Object.defineProperty(root, 'clientHeight', { configurable: true, value: 100 })
+    Object.defineProperty(root, 'clientWidth', { configurable: true, value: 300 })
+    vi.spyOn(root, 'getBoundingClientRect').mockReturnValue({
+      bottom: 100,
+      height: 100,
+      left: 10,
+      right: 310,
+      top: 0,
+      width: 300,
+      x: 10,
+      y: 0,
+      toJSON: () => ({}),
+    })
+    vi.spyOn(firstRow, 'getBoundingClientRect').mockReturnValue({
+      bottom: 91,
+      height: 16,
+      left: 10,
+      right: 200,
+      top: 75,
+      width: 190,
+      x: 10,
+      y: 75,
+      toJSON: () => ({}),
+    })
+
+    fireEvent.mouseOver(firstRow)
+    const copyButton = screen.getByRole('button', { name: 'Copy pretty JSON' })
+    expect((copyButton.closest('span')?.parentElement as HTMLElement).style.left).toBe('284px')
+    fireEvent.mouseOver(copyButton)
+    expect(screen.getByRole('button', { name: 'Copy pretty JSON' })).toBeDefined()
+    fireEvent.mouseOver(firstRow)
+
+    fireEvent.scroll(root)
+    fireEvent.scroll(window)
+    fireEvent.resize(window)
+
+    fireEvent.contextMenu(copyButton)
+    fireEvent.mouseOver(secondRow)
+    fireEvent.mouseOver(root)
+    fireEvent.mouseLeave(root)
+    expect(screen.getByRole('menu')).toBeDefined()
+    fireEvent.keyDown(document, { key: 'Escape' })
+    expect(screen.queryByRole('button', { name: /Copy/ })).toBeNull()
+
+    fireEvent.mouseOver(secondRow)
+    expect(screen.getByRole('button', { name: 'Copy value' })).toBeDefined()
+    fireEvent.mouseOver(root)
+    expect(screen.queryByRole('button', { name: /Copy/ })).toBeNull()
+
+    fireEvent.scroll(root)
+    view.rerender(<JsonTree data={{ replacement: 3 }} />)
+    expect(screen.queryByRole('button', { name: /Copy/ })).toBeNull()
+  })
+
+  it('copies the fixed root and clears it when the pointer leaves', async () => {
+    const view = render(<JsonTree data={{ value: 1 }} />)
+    const root = view.container.firstElementChild as HTMLElement
+    const openingBracket = root.querySelector<HTMLElement>('[data-json-root-row]')
+    expect(openingBracket).not.toBeNull()
+
+    fireEvent.mouseOver(openingBracket as HTMLElement)
+    fireEvent.click(screen.getByRole('button', { name: 'Copy pretty JSON' }))
+    await waitFor(() => { expect(writeText).toHaveBeenCalledWith('{\n  "value": 1\n}') })
+
+    fireEvent.mouseLeave(root)
+    expect(screen.queryByRole('button', { name: /Copy/ })).toBeNull()
+  })
+})
diff --git a/packages/client/ui-primitives/tests/markdown-plain-text.spec.ts b/packages/client/ui-primitives/tests/markdown-plain-text.spec.ts
new file mode 100644
index 0000000000..010d5803ab
--- /dev/null
+++ b/packages/client/ui-primitives/tests/markdown-plain-text.spec.ts
@@ -0,0 +1,71 @@
+import { describe, expect, it } from 'vitest'
+import { extractMarkdownPlainText } from '@deepseek-ai/dsh-client-ui-primitives'
+
+const MARKDOWN = [
+  '# Release notes',
+  '',
+  'First **paragraph** with [a link](https://example.com) and ![diagram](diagram.png).',
+  '',
+  '- shipped',
+  '- `verified`',
+  '',
+  '```ts',
+  'const ready = true',
+  '```',
+].join('\n')
+
+describe('extractMarkdownPlainText', () => {
+  it('projects the complete GFM document without presentation syntax', () => {
+    expect(extractMarkdownPlainText(MARKDOWN)).toBe([
+      'Release notes',
+      '',
+      'First paragraph with a link and diagram.',
+      '',
+      'shipped',
+      'verified',
+      '',
+      'const ready = true',
+    ].join('\n'))
+  })
+
+  it('selects the first visible line or first semantic paragraph', () => {
+    expect(extractMarkdownPlainText(MARKDOWN, { mode: 'first-line' })).toBe('Release notes')
+    expect(extractMarkdownPlainText(MARKDOWN, { mode: 'first-paragraph' }))
+      .toBe('First paragraph with a link and diagram.')
+  })
+
+  it('preserves raw HTML while removing Markdown presentation markup', () => {
+    const block = [
+      '<background-task-complete id="trajectory-ui-watch">',
+      'Command: pnpm test',
+      'Exit code: 0',
+      '</background-task-complete>',
+    ].join('\n')
+    expect(extractMarkdownPlainText(block)).toBe(block)
+    expect(extractMarkdownPlainText('**Status:** <span data-state="ok">ready</span>'))
+      .toBe('Status: <span data-state="ok">ready</span>')
+    expect(extractMarkdownPlainText(block, { mode: 'first-paragraph' }))
+      .toBe('<background-task-complete id="trajectory-ui-watch">')
+  })
+
+  it('projects GFM tables, references, hard breaks, and block structure', () => {
+    const markdown = [
+      '> first\\',
+      '> second with ![diagram][asset] and <span>visible</span>',
+      '',
+      '---',
+      '',
+      '| Name | Value |',
+      '| --- | --- |',
+      '| alpha | `1` |',
+      '',
+      '[asset]: diagram.png',
+    ].join('\n')
+    expect(extractMarkdownPlainText(markdown)).toBe([
+      'first second with diagram and <span>visible</span>',
+      '',
+      'Name\tValue',
+      'alpha\t1',
+    ].join('\n'))
+  })
+})
diff --git a/packages/client/ui-primitives/tests/terminal-block.spec.tsx b/packages/client/ui-primitives/tests/terminal-block.spec.tsx
new file mode 100644
index 0000000000..13121351a8
--- /dev/null
+++ b/packages/client/ui-primitives/tests/terminal-block.spec.tsx
@@ -0,0 +1,430 @@
+// @vitest-environment jsdom
+// TerminalBlock: the prompt label's cwd shortening, the running/empty/settled
+// arms, the prompt line's run-state dot, the exit-status pill, the head/tail height cap and its expand control,
+// and the copy control writing the raw output on both the accepted and the
+// refused clipboard paths. writeClipboard's own return contract is pinned here
+// too, since it is the seam both copy controls in this package share; the
+// resolution of ANSI runs into styles is pinned in ansi.spec.ts, so only its
+// DOM consequence (which runs get a span wrapper) is asserted here.
+
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
+import { DEFAULT_TERMINAL_MAX_LINES, TerminalBlock } from '../src/index.ts'
+import { writeClipboard } from '../src/clipboard.ts'
+
+const ESC = '\u001b'
+
+afterEach(cleanup)
+
+beforeEach(() => {
+  vi.useRealTimers()
+})
+
+/** The rendered output rows, one string per visible line (CSS-module class prefix). */
+function outputLines(container: HTMLElement): string[] {
+  return [...container.querySelectorAll('[class^="_line_"]')].map(row => row.textContent ?? '')
+}
+
+/** The prompt line's run-state dot: its StateDot state plus the hidden text label beside it. */
+function runStateOf(container: HTMLElement): { state: string | null; label: string | undefined } {
+  const dot = container.querySelector('[class*="_runState_"][data-state]')
+  return {
+    state: dot?.getAttribute('data-state') ?? null,
+    label: container.querySelector('[class^="_runStateLabel_"]')?.textContent ?? undefined,
+  }
+}
+
+/** The prompt rows as `<label><command>`, one per command line (the visual gap is CSS). */
+function promptRows(container: HTMLElement): string[] {
+  return [...container.querySelectorAll('[class^="_promptLine_"]')].map(row => (row.textContent ?? '').trim())
+}
+
+/** `count` numbered output lines, without the terminating newline. */
+function body(count: number): string {
+  return Array.from({ length: count }, (_value, index) => `line ${index + 1}`).join('\n')
+}
+
+describe('TerminalBlock prompt label', () => {
+  it('collapses the home directory itself to ~', () => {
+    render(<TerminalBlock command="ls" cwd="/Users/me" home="/Users/me" />)
+    expect(screen.getByText('~')).toBeTruthy()
+  })
+
+  it('shows only the last segment below home', () => {
+    render(<TerminalBlock command="ls" cwd="/Users/me/Documents" home="/Users/me" />)
+    expect(screen.getByText('Documents')).toBeTruthy()
+  })
+
+  it('ignores trailing separators on both the cwd and home', () => {
+    const view = render(<TerminalBlock command="ls" cwd="/Users/me/" home="/Users/me" />)
+    expect(view.getByText('~')).toBeTruthy()
+    view.rerender(<TerminalBlock command="ls" cwd="/Users/me" home="/Users/me/" />)
+    expect(view.getByText('~')).toBeTruthy()
+  })
+
+  it('drops trailing separators before taking the last segment', () => {
+    render(<TerminalBlock command="ls" cwd="/Users/me/Documents///" home="/Users/me" />)
+    expect(screen.getByText('Documents')).toBeTruthy()
+  })
+
+  it('takes the last segment when no home is known', () => {
+    render(<TerminalBlock command="ls" cwd="C:\\Users\\me\\Projects" />)
+    expect(screen.getByText('Projects')).toBeTruthy()
+  })
+
+  it('collapses a backslash home path to ~', () => {
+    render(<TerminalBlock command="ls" cwd="C:\\Users\\me" home="C:\\Users\\me" />)
+    expect(screen.getByText('~')).toBeTruthy()
+  })
+
+  it('falls back to the raw path when it has no segment', () => {
+    render(<TerminalBlock command="ls" cwd="/" home="/Users/me" />)
+    expect(screen.getByText('/')).toBeTruthy()
+  })
+
+  it('renders a plain $ with no cwd', () => {
+    render(<TerminalBlock command="ls" />)
+    expect(screen.getByText('$')).toBeTruthy()
+  })
+
+  it('renders the command verbatim after the label', () => {
+    render(<TerminalBlock command="git log --oneline | head -3" cwd="/Users/me/app" />)
+    expect(screen.getByText('git log --oneline | head -3')).toBeTruthy()
+  })
+})
+
+describe('TerminalBlock states', () => {
+  it('running shows the command line only: no output, no placeholder, no copy', () => {
+    const view = render(<TerminalBlock command="sleep 5" running output="partial" />)
+    expect(view.getByText('sleep 5')).toBeTruthy()
+    expect(view.queryByText('partial')).toBeNull()
+    expect(view.queryByText('无输出')).toBeNull()
+    expect(view.queryByRole('button')).toBeNull()
+    expect(view.container.firstElementChild?.getAttribute('data-running')).toBe('')
+  })
+
+  it('running still shows a settled-looking status pill when one is supplied', () => {
+    render(<TerminalBlock command="sleep 5" running signal="SIGINT" />)
+    expect(screen.getByText('信号 SIGINT')).toBeTruthy()
+  })
+
+  it('settled with whitespace-only output shows the dimmed placeholder', () => {
+    const view = render(<TerminalBlock command="true" output={'  \n '} exitCode={0} />)
+    expect(view.getByText('无输出')).toBeTruthy()
+    expect(view.queryByRole('button', { name: '复制' })).toBeNull()
+  })
+
+  it('settled with absent output shows the placeholder', () => {
+    render(<TerminalBlock command="true" exitCode={0} />)
+    expect(screen.getByText('无输出')).toBeTruthy()
+  })
+
+  it('settled with an empty string shows the placeholder', () => {
+    render(<TerminalBlock command="true" output="" exitCode={0} />)
+    expect(screen.getByText('无输出')).toBeTruthy()
+  })
+
+  it('treats output that renders nothing visible as empty', () => {
+    // A lone reset, an OSC title, an erase: all survive `text.trim()` yet parse
+    // to nothing. Judging emptiness on the raw text drew a box of blank rows
+    // plus a copy control for invisible bytes, and hid the placeholder.
+    const view = render(<TerminalBlock command="true" output={`${ESC}[0m`} exitCode={0} />)
+    expect(view.getByText('无输出')).toBeTruthy()
+    expect(view.queryByText('复制')).toBeNull()
+    view.rerender(<TerminalBlock command="true" output={`${ESC}]0;title${ESC}\\`} exitCode={0} />)
+    expect(view.getByText('无输出')).toBeTruthy()
+  })
+
+  it('merges className onto the wrapper', () => {
+    const view = render(<TerminalBlock command="ls" className="x" output="a" />)
+    expect(view.container.firstElementChild?.classList.contains('x')).toBe(true)
+    expect(view.container.firstElementChild?.hasAttribute('data-running')).toBe(false)
+  })
+
+  it('drops the output text terminator instead of drawing a blank line', () => {
+    const view = render(<TerminalBlock command="ls" output={'a\nb\n'} />)
+    expect(outputLines(view.container)).toEqual(['a', 'b'])
+  })
+
+  it('drops the output terminator even when a reset follows the final newline', () => {
+    // `line\n\x1b[0m` does not end in a newline as a string, yet its last parsed
+    // line holds nothing visible — a common shape, since tools close their color
+    // after the last line. Judging the terminator on the raw text added a blank
+    // row and inflated both the card height and the collapse count.
+    const view = render(<TerminalBlock command="ls" output={`a\nb\n${ESC}[0m`} />)
+    expect(outputLines(view.container)).toEqual(['a', 'b'])
+  })
+
+  it('keeps a genuinely blank final line when the output ends with two newlines', () => {
+    const view = render(<TerminalBlock command="ls" output={'a\nb\n\n'} />)
+    expect(outputLines(view.container)).toEqual(['a', 'b', ''])
+  })
+
+  it('renders ANSI runs as styled spans and plain text bare', () => {
+    const view = render(<TerminalBlock command="ls" output={`${ESC}[31mbad${ESC}[39m ok`} />)
+    // Scoped to a line: the prompt line's run-state dot is a styled span too.
+    const span = view.container.querySelector('[class^="_line_"] span[style]')
+    expect(span?.textContent).toBe('bad')
+    expect(span?.getAttribute('style')).toContain('--dsw-alias-state-error-primary')
+    expect(outputLines(view.container)).toEqual(['bad ok'])
+  })
+
+  it('renders uncolored output with no span wrappers at all', () => {
+    const view = render(<TerminalBlock command="ls" output={'plain one\nplain two\n'} />)
+    expect(view.container.querySelectorAll('[class^="_line_"] span')).toHaveLength(0)
+  })
+})
+
+describe('TerminalBlock status pill', () => {
+  it('renders no pill for a clean exit', () => {
+    const view = render(<TerminalBlock command="true" output="a" exitCode={0} />)
+    expect(view.queryByText(/退出码|信号/u)).toBeNull()
+  })
+
+  it('renders no pill while the exit status is unknown', () => {
+    const view = render(<TerminalBlock command="ls" output="a" />)
+    expect(view.queryByText(/退出码|信号/u)).toBeNull()
+  })
+
+  it('renders the exit-code pill for a non-zero exit', () => {
+    render(<TerminalBlock command="false" output="a" exitCode={1} />)
+    expect(screen.getByText('退出码 1')).toBeTruthy()
+  })
+
+  it('renders the signal pill, which outranks the exit code', () => {
+    render(<TerminalBlock command="sleep 9" output="a" exitCode={0} signal="SIGKILL" />)
+    expect(screen.getByText('信号 SIGKILL')).toBeTruthy()
+    expect(screen.queryByText(/退出码/u)).toBeNull()
+  })
+})
+
+describe('TerminalBlock run-state dot', () => {
+  it('shows the running chase and its running label while the command runs', () => {
+    const view = render(<TerminalBlock command="sleep 5" running />)
+    expect(runStateOf(view.container)).toEqual({ state: 'ongoing', label: '运行中' })
+  })
+
+  it('shows the done dot for a clean settled exit', () => {
+    const view = render(<TerminalBlock command="true" output="a" exitCode={0} />)
+    expect(runStateOf(view.container)).toEqual({ state: 'done', label: '已完成' })
+  })
+
+  it('counts a settled command with no exit status as a clean settle', () => {
+    const view = render(<TerminalBlock command="ls" output="a" />)
+    expect(runStateOf(view.container)).toEqual({ state: 'done', label: '已完成' })
+  })
+
+  it('shows the error dot for a non-zero exit', () => {
+    const view = render(<TerminalBlock command="false" output="a" exitCode={1} />)
+    expect(runStateOf(view.container)).toEqual({ state: 'error', label: '失败' })
+  })
+
+  it('shows the error dot for a signal, whatever the exit code says', () => {
+    const view = render(<TerminalBlock command="sleep 9" output="a" exitCode={0} signal="SIGKILL" />)
+    expect(runStateOf(view.container)).toEqual({ state: 'error', label: '失败' })
+  })
+
+  // The dot precedes the prompt label, which is what makes it read as the
+  // state OF this command rather than of the card's chrome.
+  it('places the dot ahead of the prompt label and the command', () => {
+    const view = render(<TerminalBlock command="ls" cwd="/srv/app" output="a" />)
+    const row = view.container.querySelector('[class^="_promptLine_"]')
+    expect([...row!.children].map(node => node.textContent)).toEqual(['', 'app', 'ls'])
+  })
+
+  // The cwd labels the call, not each line: a `cd` in the command moves later
+  // lines elsewhere, so repeating the label would state a directory per line
+  // that the view does not know.
+  it('labels only the first row with the cwd, leaving later rows a bare $', () => {
+    const view = render(<TerminalBlock command={'cd ~\nls'} cwd="/srv/app" output="a" exitCode={0} />)
+    expect(promptRows(view.container)).toEqual(['appcd ~', '$ls'])
+  })
+
+  it('gives a multi-line command one row per line', () => {
+    const view = render(<TerminalBlock command={'echo one\necho two'} output="a" exitCode={0} />)
+    expect(promptRows(view.container)).toEqual(['$echo one', '$echo two'])
+  })
+
+  // A heredoc or an editor-authored command commonly ends in a newline; that
+  // terminator is not a further, empty command to draw a row for.
+  it('drops a trailing newline instead of drawing an empty final row', () => {
+    const view = render(<TerminalBlock command={'echo one\necho two\n'} output="a" exitCode={0} />)
+    expect(promptRows(view.container)).toEqual(['$echo one', '$echo two'])
+  })
+
+  it('keeps a genuinely blank command line when the command ends with two newlines', () => {
+    const view = render(<TerminalBlock command={'echo one\n\n'} output="a" exitCode={0} />)
+    expect(promptRows(view.container)).toEqual(['$echo one', '$'])
+  })
+
+  // The exit status the view carries is the whole call's — bash reports no
+  // per-command status — so exactly one dot and one label are correct however
+  // many lines the command spans. A dot per row would assert, of a line that
+  // succeeded inside a failing call, that the line itself failed.
+  it('marks the call once, on the first row, never per line', () => {
+    const view = render(<TerminalBlock command={'true\nfalse\ntrue'} output="x" exitCode={1} />)
+    expect(view.container.querySelectorAll('[class*="_runState_"][data-state]')).toHaveLength(1)
+    expect(view.container.querySelectorAll('[class^="_runStateLabel_"]')).toHaveLength(1)
+    expect(runStateOf(view.container)).toEqual({ state: 'error', label: '失败' })
+    const rows = view.container.querySelectorAll('[class^="_promptLine_"]')
+    expect(rows[0]!.querySelector('[data-state]')).not.toBeNull()
+    expect(rows[1]!.querySelector('[data-state]')).toBeNull()
+    expect(rows[2]!.querySelector('[data-state]')).toBeNull()
+  })
+
+  it('keeps the running dot even while a settled-looking status pill is supplied', () => {
+    const view = render(<TerminalBlock command="sleep 5" running signal="SIGINT" />)
+    expect(runStateOf(view.container)).toEqual({ state: 'ongoing', label: '运行中' })
+  })
+})
+
+describe('TerminalBlock height cap', () => {
+  it('renders every line and no expand control under the cap', () => {
+    const view = render(<TerminalBlock command="ls" output={body(4)} maxLines={4} />)
+    expect(outputLines(view.container)).toHaveLength(4)
+    expect(view.container.querySelector('[aria-expanded]')).toBeNull()
+  })
+
+  it('does not count the output terminator against the cap', () => {
+    const view = render(<TerminalBlock command="ls" output={`${body(4)}\n`} maxLines={4} />)
+    expect(outputLines(view.container)).toHaveLength(4)
+    expect(view.container.querySelector('[aria-expanded]')).toBeNull()
+  })
+
+  it('slices head and tail over the cap and expands on click', () => {
+    const view = render(<TerminalBlock command="ls" output={body(10)} maxLines={4} />)
+    // maxLines 4: head = ceil(4/2) = 2, tail = 4 - 2 = 2, 6 hidden.
+    expect(outputLines(view.container)).toEqual(['line 1', 'line 2', 'line 9', 'line 10'])
+    const toggle = view.getByRole('button', { name: '展开其余 6 行输出' })
+    expect(toggle.getAttribute('aria-expanded')).toBe('false')
+    expect(toggle.textContent).toBe('… 其余 6 行')
+
+    fireEvent.click(toggle)
+    expect(outputLines(view.container)).toHaveLength(10)
+    const collapse = view.getByRole('button', { name: '收起输出' })
+    expect(collapse.getAttribute('aria-expanded')).toBe('true')
+    expect(collapse.textContent).toBe('收起')
+
+    fireEvent.click(collapse)
+    expect(outputLines(view.container)).toEqual(['line 1', 'line 2', 'line 9', 'line 10'])
+  })
+
+  it('renders the head slice alone when the cap leaves no tail', () => {
+    const view = render(<TerminalBlock command="ls" output={body(5)} maxLines={1} />)
+    expect(outputLines(view.container)).toEqual(['line 1'])
+    expect(view.getByRole('button', { name: '展开其余 4 行输出' })).toBeTruthy()
+  })
+
+  it('caps at the documented default when maxLines is absent', () => {
+    const view = render(<TerminalBlock command="ls" output={body(DEFAULT_TERMINAL_MAX_LINES + 1)} />)
+    expect(outputLines(view.container)).toHaveLength(DEFAULT_TERMINAL_MAX_LINES)
+    expect(view.getByRole('button', { name: '展开其余 1 行输出' })).toBeTruthy()
+  })
+})
+
+describe('TerminalBlock copy', () => {
+  it('copies the raw output, never the prompt line or the pill', async () => {
+    vi.useFakeTimers()
+    const writeText = vi.fn().mockResolvedValue(undefined)
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: { writeText } })
+    const output = `${ESC}[31mbad${ESC}[39m\n`
+    render(<TerminalBlock command="make" cwd="/Users/me/app" output={output} exitCode={2} />)
+    fireEvent.click(screen.getByRole('button', { name: '复制' }))
+    // Escape codes, the newline terminator, and nothing of the chrome around them.
+    expect(writeText).toHaveBeenCalledWith(output)
+    await act(async () => {
+      await Promise.resolve()
+    })
+    expect(screen.getByRole('button', { name: '复制成功' })).toBeTruthy()
+    // While the ok label is showing, further clicks are no-ops.
+    fireEvent.click(screen.getByRole('button', { name: '复制成功' }))
+    expect(writeText).toHaveBeenCalledTimes(1)
+    await vi.advanceTimersByTimeAsync(1000)
+    expect(screen.getByRole('button', { name: '复制' })).toBeTruthy()
+  })
+
+  it('copies the whole output while the height cap hides its middle', async () => {
+    const writeText = vi.fn().mockResolvedValue(undefined)
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: { writeText } })
+    const output = `${body(10)}\n`
+    render(<TerminalBlock command="ls" output={output} maxLines={4} exitCode={0} />)
+    fireEvent.click(screen.getByRole('button', { name: '复制' }))
+    expect(writeText).toHaveBeenCalledWith(output)
+    expect(await screen.findByRole('button', { name: '复制成功' })).toBeTruthy()
+  })
+
+  it('does not claim success when the host refuses the write', async () => {
+    Object.defineProperty(navigator, 'clipboard', {
+      configurable: true,
+      value: { writeText: vi.fn().mockRejectedValue(new Error('denied')) },
+    })
+    render(<TerminalBlock command="ls" output="a" />)
+    fireEvent.click(screen.getByRole('button', { name: '复制' }))
+    await act(async () => {
+      await Promise.resolve()
+    })
+    expect(screen.getByRole('button', { name: '复制' })).toBeTruthy()
+    expect(screen.queryByRole('button', { name: '复制成功' })).toBeNull()
+  })
+})
+
+describe('writeClipboard', () => {
+  it('reports true after the async Clipboard API accepts the exact text', async () => {
+    const writeText = vi.fn().mockResolvedValue(undefined)
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: { writeText } })
+    await expect(writeClipboard('payload')).resolves.toBe(true)
+    expect(writeText).toHaveBeenCalledWith('payload')
+  })
+
+  it('reports false when the Clipboard API rejects', async () => {
+    Object.defineProperty(navigator, 'clipboard', {
+      configurable: true,
+      value: { writeText: vi.fn().mockRejectedValue(new Error('denied')) },
+    })
+    await expect(writeClipboard('payload')).resolves.toBe(false)
+  })
+
+  it('selects a detached textarea for the execCommand fallback and removes it after', async () => {
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: undefined })
+    let selected: string | undefined
+    const exec = vi.fn(() => {
+      selected = document.querySelector<HTMLTextAreaElement>('textarea[readonly]')?.value
+      return true
+    })
+    Object.defineProperty(document, 'execCommand', { configurable: true, value: exec })
+    await expect(writeClipboard('payload')).resolves.toBe(true)
+    expect(exec).toHaveBeenCalledWith('copy')
+    expect(selected).toBe('payload')
+    expect(document.querySelector('textarea')).toBeNull()
+  })
+
+  it('reports execCommand\'s own refusal verbatim', async () => {
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: undefined })
+    Object.defineProperty(document, 'execCommand', { configurable: true, value: vi.fn(() => false) })
+    await expect(writeClipboard('payload')).resolves.toBe(false)
+  })
+
+  it('reports false and still removes the textarea when execCommand throws', async () => {
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: undefined })
+    Object.defineProperty(document, 'execCommand', {
+      configurable: true,
+      value: () => {
+        throw new Error('denied')
+      },
+    })
+    await expect(writeClipboard('payload')).resolves.toBe(false)
+    expect(document.querySelector('textarea')).toBeNull()
+  })
+
+  it('reports false on a host with neither clipboard path', async () => {
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: undefined })
+    Object.defineProperty(document, 'execCommand', { configurable: true, value: undefined })
+    await expect(writeClipboard('payload')).resolves.toBe(false)
+  })
+
+  it('reports false when navigator.clipboard exists without writeText', async () => {
+    Object.defineProperty(navigator, 'clipboard', { configurable: true, value: {} })
+    Object.defineProperty(document, 'execCommand', { configurable: true, value: undefined })
+    await expect(writeClipboard('payload')).resolves.toBe(false)
+  })
+})
diff --git a/packages/client/ui-settings/README.i18n.yaml b/packages/client/ui-settings/README.i18n.yaml
index 91f8103288..946f1f6c1f 100644
--- a/packages/client/ui-settings/README.i18n.yaml
+++ b/packages/client/ui-settings/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/ui-settings/README.md
 README.md: bb99f9b37927eec57650aa4025deb043b369c78e
-README.zh.md: fce11e2cf44fe6c1debe850df644b0114dbde5e3
+README.zh.md: 64b207aadfbcd7d25c005b3dcd5358013d53c173
diff --git a/packages/client/ui-settings/README.zh.md b/packages/client/ui-settings/README.zh.md
index fce11e2cf4..64b207aadf 100644
--- a/packages/client/ui-settings/README.zh.md
+++ b/packages/client/ui-settings/README.zh.md
@@ -10,7 +10,7 @@
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
diff --git a/packages/client/ui-sidebar/README.i18n.yaml b/packages/client/ui-sidebar/README.i18n.yaml
index b9072c3787..6c5f1735e3 100644
--- a/packages/client/ui-sidebar/README.i18n.yaml
+++ b/packages/client/ui-sidebar/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/ui-sidebar/README.md
 README.md: 93a1f15a5802f94a0ebe930dda1dbd4fbc7343c9
-README.zh.md: 1ef636dbd00c894c8312ab1fbfa9a96af45956f0
+README.zh.md: 8c8545a5d7d8cb4d58772abf867d7ee82c31bf1d
diff --git a/packages/client/ui-sidebar/README.zh.md b/packages/client/ui-sidebar/README.zh.md
index 1ef636dbd0..8c8545a5d7 100644
--- a/packages/client/ui-sidebar/README.zh.md
+++ b/packages/client/ui-sidebar/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-侧边栏插件：真实 Host Workspace 按稳定的 Host 顺序排列；每个 Workspace 按自身顺序包含其 `sessionIds`，并以 `parentId` 嵌套；不属于任何 Workspace 的 Session 显示在末尾的 `Ungrouped` 分区。搜索、状态点以及折叠到布局拥有的 56px 轨道，都只属于呈现层。契约：[slot 系统标准](../../../.agents/notes/implemented/architecture/2026-07-22-slot-type-chain-implementation.md)。
+侧边栏插件：真实 Host Workspace 按稳定的 Host 顺序排列；每个 Workspace 按自身顺序包含其 `sessionIds`，并以 `parentId` 嵌套；不属于任何 Workspace 的会话显示在末尾的 `Ungrouped` 分区。搜索、状态点以及折叠到布局拥有的 56px 轨道，都只属于呈现层。契约：[slot 系统标准](../../../.agents/notes/implemented/architecture/2026-07-22-slot-type-chain-implementation.md)。
 
-New Session 会启动运行时的页面局部前端 Session Intent；真实 Workspace 的「+」会启动一项以该 Workspace 为目标的 Intent。Workspace 标题栏的「+」打开 ui-workspace 的共享选择器，选择结果同样以一个前端 Session 为目标。Workspace Intent 不会出现在侧边栏中。
+New Session 会启动运行时的页面局部前端 Session Intent；真实 Workspace 的「+」会启动一项以该 Workspace 为目标的 Intent。Workspace 标题栏的「+」打开 ui-workspace 的共享选择器，选择结果同样以一个前端会话为目标。Workspace Intent 不会出现在侧边栏中。
 
-`SidebarRootComponentProps` 组合布局 owner share、全局 `useSessions` 和 `useWorkspaces` hook、已声明的 `sidebar.workspace` 与 `sidebar.settings` 子 slot，以及注入的 `startSession`、`open` 和侧边栏切换回调。这里没有插件 store：`deriveGroups` 消费对象层快照与组件局部的展开／搜索状态。
+`SidebarRootComponentProps` 组合布局 owner share、全局 `useSessions` 和 `useWorkspaces` 钩子、已声明的 `sidebar.workspace` 与 `sidebar.settings` 子 slot，以及注入的 `startSession`、`open` 和侧边栏切换回调。这里没有插件 store：`deriveGroups` 消费对象层快照与组件局部的展开／搜索状态。
 
 页脚承载 `sidebar.settings`：侧边栏只渲染固定在底部的布局 slot，并共享其栏状态（`wide`）；ui-settings 在此注册触发行和设置面板。
 
@@ -18,10 +18,10 @@ New Session 会启动运行时的页面局部前端 Session Intent；真实 Work
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
-- **状态点只有两种实时数据状态（running/none）**：done/error/amber 数据源随 P-II 审批与通知到来；四色原语已经接线。
+- **状态点只有两种实时数据状态（running/none）**：done/error/amber 的数据源将随 P-II 审批与通知功能一并提供；四色原语已接入。
 - **分组选单只提供按 Workspace 分组**：Update/Status 分组策略只有图稿而没有规范，暂缓实现。
-- **「New task completed」未读标记是本地查看状态**：完成时间 > 上次查看时间这一事实永远不会到达主机。
+- **「New task completed」未读标记是本地查看状态**：完成时间 > 上次查看时间这一事实永远不会到达宿主。
diff --git a/packages/client/ui-skill/README.i18n.yaml b/packages/client/ui-skill/README.i18n.yaml
index 543e3797a1..3481eb7d7d 100644
--- a/packages/client/ui-skill/README.i18n.yaml
+++ b/packages/client/ui-skill/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/ui-skill/README.md
 README.md: 4838be893c1d5422cc707cb0d7542a056be41fa7
-README.zh.md: 368171a43ef3a449049542cd227459f82ec43086
+README.zh.md: ed582128246a62297f555f8abe09f427cb9d256a
diff --git a/packages/client/ui-skill/README.zh.md b/packages/client/ui-skill/README.zh.md
index 368171a43e..ed58212824 100644
--- a/packages/client/ui-skill/README.zh.md
+++ b/packages/client/ui-skill/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-skill（技能）引用 source 的浏览器半侧：把 `/` 触发的 `skill` source 注册进 `ctx.slash`。候选来自 `skill.list` RPC，以每次调用的 `ClientSessionContext` 投影中的 `{sessionId}` 寻址——每个会话恒为 agent-backed，host 从会话 header 解析 `cwd`。目录按会话缓存，拉取走 single-flight；scope 出生的 `warm` 钩子预热该会话的缓存项，`connection/reset` 清空全部缓存。结果按 `startsWith(query)` 过滤；pick 一个候选会把字面文本 `/name ` 经 slash 管线落进草稿（决策 21 的纯文本引用），source 的 `codec` 拥有该引用的两种投影：`clipboardText` → `/name`，`serialize` → 提交时生成的模型形式 `<skill>name</skill>`。RPC 使用插件注册时捕获的根上下文连接——source 绝不从每次调用的参数上读取服务。source 不实现 `matchSpace`／`matchEnter` 钩子——skill 引用永不进入命令裁决，随普通提示词落入 default sink。
+skill（技能）引用 source 的浏览器端：把 `/` 触发的 `skill` source 注册进 `ctx.slash`。候选来自 `skill.list` RPC，以每次调用的 `ClientSessionContext` 投影中的 `{sessionId}` 寻址——每个会话始终由 agent（智能体）支撑，host 从会话 header 解析 `cwd`。目录按会话缓存，拉取走 single-flight；scope 创建时的 `warm` 钩子预热该会话的缓存项，`connection/reset` 清空全部缓存。结果按 `startsWith(query)` 过滤；pick 一个候选会把字面文本 `/name ` 经 slash 管线落进草稿（决策 21 的纯文本引用），source 的 `codec` 拥有该引用的两种投影：`clipboardText` → `/name`，`serialize` → 提交时生成的模型形式 `<skill>name</skill>`。RPC 使用插件注册时捕获的根上下文连接——source 绝不从每次调用的参数上读取服务。source 不实现 `matchSpace`／`matchEnter` 钩子——skill 引用永不进入命令裁决，随普通提示词落入 default sink。
 
 `skill.list` 失败时 `candidates` 抛出异常，slash 壳层记录日志并折叠为静默的菜单组丢弃——菜单只显示 pending／ready 状态。
 
@@ -12,20 +12,20 @@ skill（技能）引用 source 的浏览器半侧：把 `/` 触发的 `skill` so
 
 ### 用户提示词中的 skill 引用文本
 
-#### 模型所见
+#### 模型看到的内容
 
 被 pick 的候选会把字面文本 `/name ` 落进草稿（决策 21：纯文本，无 `<skill>` 标签）；该文本原样进入普通用户消息（`session.prompt`）到达模型，没有专用内容块、提示词 section 或 host 侧展开。与实际 skill 的关联在模型侧建立且不确定：会话前缀已携带 skill 目录（由 `dsh-tool-skill` 渲染），引用名称与目录条目匹配，正是这一点引导模型去加载它。
 
 #### Token 影响
 
-有条件且极小：只有 pick（或手动键入相同文本）会把引用的字符加进那一条用户消息。浏览菜单和候选拉取增加零模型 token。
+有条件且极小：只有 pick（或手动键入相同文本）会把引用的字符加进那一条用户消息。浏览菜单和拉取候选不会增加任何模型 token。
 
 #### KV Cache 影响
 
-仅追加：引用是追加在可复用历史前缀之后的新用户消息的一部分。该包绝不改写较早的请求 token。
+仅追加：引用是追加在可复用历史前缀之后的新用户消息的一部分。该包（package）绝不改写较早的请求 token。
 
 ## 已知限制与暂缓事项
 
-- **skill 加载不确定**：引用是协作线索，不是保证；模型可能忽略它。命中率被证明不足时的返工路径（host 侧 `context/skill-reference` 引导包，或全文注入）记录在设计台账中；wire 上的文本形状不会改变。
-- **首次击键可能与预热竞速**：scope 出生的预热会启动目录拉取，但目录落定之前打开的菜单，在那次击键下不会显示 skill 候选。这是设计上接受的取舍：skill 引用不参与回车裁决，因此没有任何攸关正确性的环节等待目录。
-- **文本即真身**：引用是普通的草稿文本；手动键入的相同 token 就是同一个引用。chip 视觉由 lexicon 扫描派生；没有 occurrence 身份或位置跟踪（组件化 chip 是台账事项）。
+- **skill 加载不确定**：引用是协作线索，不是保证；模型可能忽略它。针对命中率不足情况的返工路径（host 侧 `context/skill-reference` 引导包，或全文注入）记录在设计台账中；协议中的文本形态不会改变。
+- **首次击键可能与预热竞速**：scope 创建时的预热会启动目录拉取，但目录落定之前打开的菜单，在那次击键下不会显示 skill 候选。这是设计上接受的取舍：skill 引用不参与回车裁决，因此没有任何攸关正确性的环节等待目录。
+- **文本是唯一依据**：引用是普通的草稿文本；手动键入的相同 token 就是同一个引用。chip 视觉由 lexicon 扫描派生；没有 occurrence 身份或位置跟踪（组件化 chip 是台账事项）。
diff --git a/packages/client/ui-skill/src/client/index.ts b/packages/client/ui-skill/src/client/index.ts
index d34c6ba96d..b0f55fd3e5 100644
--- a/packages/client/ui-skill/src/client/index.ts
+++ b/packages/client/ui-skill/src/client/index.ts
@@ -100,6 +100,7 @@ export function apply(ctx: ClientContext): void {
   const source: SlashSource = {
     trigger: '/',
     name: 'skill',
+    order: 2,
     async candidates(session, { query, signal }) {
       const skills = await fetchCatalog(session.sessionId)
       // Superseded keystroke: the shared fetch stays warm, this caller yields.
diff --git a/packages/client/ui-slash/README.i18n.yaml b/packages/client/ui-slash/README.i18n.yaml
index 1053e205b0..494ccdad20 100644
--- a/packages/client/ui-slash/README.i18n.yaml
+++ b/packages/client/ui-slash/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-slash/README.md
-README.md: 4e363c2682bf91862ec40f3f2174831451fb9b0d
-README.zh.md: 76d39673cb853d1889ee84cb9f3595708eae2db3
+README.md: 29f1a71ce20f898ffe2ab3a1c6f3d73a7aecfe38
+README.zh.md: 03dac56870de5b083124716825001009b4293736
diff --git a/packages/client/ui-slash/README.md b/packages/client/ui-slash/README.md
index 4e363c2682..29f1a71ce2 100644
--- a/packages/client/ui-slash/README.md
+++ b/packages/client/ui-slash/README.md
@@ -6,7 +6,7 @@ Input trigger pipeline plugin: `/` and `@` detection under the caret (word-bound
 
 Layering: `src/core/` (T2) is the pure core — `detectTrigger`, `menuReduce`/`seedGroups`/`MENU_CLOSED`, `exactMatch`, zero React/DOM/cordis; `src/client/service.ts` is the shell wiring the core to the menu snapshot store, the per-hit candidate fetch (generation-gated, `AbortSignal`-superseded, failed sources drop silently with a console record), and the three pick paths. `src/types.ts` and the two `contract.ts` files are the frozen cross-package contract (design v4 §5.1); changes require main-thread arbitration.
 
-MenuView renders the menu store into the `conversation.input.overlay` slot (list kind, session scope) and renders null while closed. The slot is owned by ui-conversation's composer entry (anchor, children declaration, lifecycle); its SlotMap type merge lives in this package's `src/client/slots.ts` because the dependency direction (ui-conversation → ui-slash) admits no reverse type import. Combobox pattern: focus stays in the textarea, rows pick on mousedown, the highlight rides `aria-activedescendant`.
+MenuView renders the menu store into the `conversation.input.overlay` slot (list kind, session scope) and renders null while closed. Groups sort by the optional `SlashSource.order` (lower first, default 0, ties keep registration order) under title rows localized through the `slash.menu` locale namespace (an unknown source shows its raw name); the list height clamps to the space above the composer, and a pointer down outside both the menu and the surrounding composer card dismisses it. The slot is owned by ui-conversation's composer entry (anchor, children declaration, lifecycle); its SlotMap type merge lives in this package's `src/client/slots.ts` because the dependency direction (ui-conversation → ui-slash) admits no reverse type import. Combobox pattern: focus stays in the textarea, rows pick on mousedown, the highlight rides `aria-activedescendant`.
 
 The `/client` export surface is the plugin body (`apply`/`inject`), `SlashService`, `MenuViewInjected`, and the contract types. MenuView itself is internal — the slot registration closes over it.
 
@@ -23,4 +23,3 @@ None; this package neither assembles nor sends a provider request.
 - **Global source layer only** — session-scope source registration (per-session shadowing, ScopedLayers-alike) is designed but not enabled; the ledger tracks the trigger condition (a real per-session source need).
 - **`SlashCandidate.icon` renders as text** — MenuView drops the string into the icon slot verbatim; wiring to the design-system icon enum (iconFile five-variant family) lands when that enum ships.
 - **Overlay SlotMap merge home is split from slot ownership** — the `conversation.input.overlay` merge lives here (sole copy) while the slot's owner semantics (anchor, children declaration, lifecycle) stay with ui-conversation; the dependency direction (ui-conversation → ui-slash) forces the split, so a future dependency reshuffle should revisit it.
-- **Menu group order is registration order** — no explicit ordering seam across sources; acceptable while the roster is command/skill/subagent, revisit if business sources join.
diff --git a/packages/client/ui-slash/README.zh.md b/packages/client/ui-slash/README.zh.md
index 76d39673cb..03dac56870 100644
--- a/packages/client/ui-slash/README.zh.md
+++ b/packages/client/ui-slash/README.zh.md
@@ -2,25 +2,24 @@
 
 [English](README.md) | 中文
 
-输入触发管线插件：光标处的 `/` 与 `@` 检测（词边界 + guard tier 规则）、分组候选菜单，以及把 pick 路由到已注册 source。`ctx.slash` 拥有 source roster，并按会话 scope（`sessionOf`）各解析一个 `SlashController`；会话领域的接线层在 controller 上驱动 `track`／`arbitrate`／`onSpace`／`adjudicate`。source 每次调用收到一个 `ClientSessionContext` 投影——会话恒为 agent-backed，因此投影只含会话身份。source 在它能触达的每个会话 controller 中都会被预热：scope 出生时在场的 roster 随 controller 构造预热，晚于此注册的 source 由注册动作本身预热进每个活 controller。`lexicon` 名录在预热后仍会变化的 source 实现 `subscribeLexicon(session, listener)`；controller 每收到通知就重拉，并把聚合结果经其 `lexicon` snapshot store 发布。管线对命令零知识：空格／回车裁决按注册序轮询可选的 `matchSpace`／`matchEnter` 钩子，第一个非 undefined 的应答胜出。
+输入触发流水线插件：光标处的 `/` 与 `@` 检测（词边界 + guard tier 规则）、分组候选菜单，以及把 pick 路由到已注册 source。`ctx.slash` 拥有 source roster，并按会话 scope（`sessionOf`）各解析一个 `SlashController`；对话接线层在 controller 上驱动 `track`／`arbitrate`／`onSpace`／`adjudicate`。source 每次调用收到一个 `ClientSessionContext` 投影——会话始终由 agent（智能体）支撑，因此投影只含会话身份。source 在它能触达的每个会话 controller 中都会被预热：scope 出生时在场的 roster 随 controller 构造预热，晚于此注册的 source 由注册动作本身预热进每个活 controller。`lexicon` 名录在预热后仍会变化的 source 实现 `subscribeLexicon(session, listener)`；controller 每收到通知就重拉，并把聚合结果经其 `lexicon` 快照 store 发布。流水线与命令无关：空格／回车裁决按注册序轮询可选的 `matchSpace`／`matchEnter` 钩子，第一个非 undefined 的应答胜出。
 
 分层：`src/core/`（T2）是纯内核——`detectTrigger`、`menuReduce`／`seedGroups`／`MENU_CLOSED`、`exactMatch`，零 React／DOM／cordis；`src/client/service.ts` 是壳层，把内核接到菜单快照 store、逐 hit 候选拉取（以 generation 把关、后继请求经 `AbortSignal` 取代旧请求、失败的 source 静默丢弃并留一条 console 记录）和三条 pick 路径上。`src/types.ts` 与两个 `contract.ts` 文件是冻结的跨包契约（设计 v4 §5.1）；变更需经主线程仲裁。
 
-MenuView 把菜单 store 渲染进 `conversation.input.overlay` slot（列表类，会话 scope），菜单关闭期间渲染 null。该 slot 由 ui-conversation 的编辑器配置项拥有（锚点、children 声明、生命周期）；其 SlotMap 类型合并放在本包的 `src/client/slots.ts`，因为依赖方向（ui-conversation → ui-slash）不允许反向的类型导入。combobox 模式：焦点始终留在 textarea，行在 mousedown 时完成 pick，高亮由 `aria-activedescendant` 承载。
+MenuView 把菜单 store 渲染进 `conversation.input.overlay` slot（列表类，会话 scope），菜单关闭期间渲染 null。分组按可选的 `SlashSource.order` 排序（越小越靠前，默认 0，同值保持注册序），组标题行经 `slash.menu` locale 命名空间本地化（未知 source 显示其原名）；列表高度收敛到 composer 上方的可用空间，指针落在菜单与所在 composer 卡片之外即关闭菜单。该 slot 由 ui-conversation 的组合器条目拥有（锚点、children 声明、生命周期）；其 SlotMap 类型合并放在本包的 `src/client/slots.ts`，因为依赖方向（ui-conversation → ui-slash）不允许反向的类型导入。combobox 模式：焦点始终留在 textarea，行在 mousedown 时完成 pick，高亮由 `aria-activedescendant` 承载。
 
 `/client` 导出表层是插件主体（`apply`／`inject`）、`SlashService`、`MenuViewInjected` 与契约类型。MenuView 本身是内部实现——slot 注册以闭包持有它。
 
 ## 模型体验
 
-无。触发管线只是浏览器呈现——pick 产出 `CommandClaim`／`ReferenceInsert` 数据，其模型可见后果（host 命令执行；插入的引用文本随普通提示词发送）由消费方的 host 包与输入状态机包拥有。
+无。触发流水线只是浏览器呈现——pick 产出 `CommandClaim`／`ReferenceInsert` 数据，其模型可见后果（宿主命令执行；插入的引用文本随普通提示词发送）由消费方的 host 包与输入状态机包拥有。
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
 - **只有全局 source 层**：会话 scope 的 source 注册（逐会话遮蔽、类 ScopedLayers 机制）已有设计但未启用；台账记录着触发条件（出现真实的逐会话 source 需求）。
-- **`SlashCandidate.icon` 以文本渲染**：MenuView 把该字符串原样放进图标位；接到设计系统图标枚举（iconFile 五变体家族）的接线等该枚举交付后落地。
+- **`SlashCandidate.icon` 以文本渲染**：MenuView 把该字符串原样放进图标位；与设计系统图标枚举（iconFile 五变体家族）的接入将在该枚举交付后完成。
 - **overlay 的 SlotMap 合并归属与 slot 所有权分离**：`conversation.input.overlay` 的合并放在本包（唯一副本），而该 slot 的 owner 语义（锚点、children 声明、生命周期）留在 ui-conversation；依赖方向（ui-conversation → ui-slash）迫使这一拆分，未来依赖关系调整时应重新审视。
-- **菜单组顺序即注册顺序**：source 之间没有显式排序 seam；roster 还是 command／skill／subagent 时可以接受，业务 source 加入后需重新审视。
diff --git a/packages/client/ui-slash/package.json b/packages/client/ui-slash/package.json
index 1c376c5492..d39d49f09d 100644
--- a/packages/client/ui-slash/package.json
+++ b/packages/client/ui-slash/package.json
@@ -24,7 +24,8 @@
   },
   "dshClient": {
     "inject": [
-      "@deepseek-ai/dsh-client-runtime"
+      "@deepseek-ai/dsh-client-runtime",
+      "@deepseek-ai/dsh-client-locale"
     ],
     "platform": "web"
   },
@@ -37,14 +38,18 @@
     "clsx": "^2.0.0"
   },
   "peerDependencies": {
+    "@deepseek-ai/dsh-client-locale": "^0.0.1",
     "@deepseek-ai/dsh-client-runtime": "^0.0.1",
+    "@deepseek-ai/dsh-client-ui-primitives": "^0.0.1",
     "@deepseek-ai/dsh-client-ui-slots": "^0.0.1",
     "@deepseek-ai/dsh-invariants": "^0.0.1",
     "cordis": "^4.0.0-rc.7",
     "react": "^18.2.0"
   },
   "devDependencies": {
+    "@deepseek-ai/dsh-client-locale": "workspace:^",
     "@deepseek-ai/dsh-client-runtime": "workspace:^",
+    "@deepseek-ai/dsh-client-ui-primitives": "workspace:^",
     "@deepseek-ai/dsh-client-ui-slots": "workspace:^",
     "@deepseek-ai/dsh-invariants": "workspace:^",
     "@types/react": "~18.3.1",
diff --git a/packages/client/ui-slash/src/client/MenuView.module.css b/packages/client/ui-slash/src/client/MenuView.module.css
index 7bcd95b2de..9a781e6b14 100644
--- a/packages/client/ui-slash/src/client/MenuView.module.css
+++ b/packages/client/ui-slash/src/client/MenuView.module.css
@@ -11,8 +11,10 @@
   z-index: 100;
   min-width: 260px;
   max-width: 537px;
+  /* Height cap: the 320px design maximum, clamped at runtime to the space
+   * above the composer (inline max-height set in MenuView.tsx). */
   max-height: 320px;
-  overflow-y: auto;
+  overflow: hidden;
   /* Elevated surface: the scrollbar thumb takes the l2 elevation tokens
      (see ui-theme styles/scrollbar.css for the rebinding contract). */
   --dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2);
@@ -26,6 +28,13 @@
   box-shadow: var(--dsw-shadow-lv3);
 }
 
+.viewport {
+  display: flex;
+  flex-direction: column;
+  min-height: 0;
+  overflow-y: auto;
+}
+
 .item {
   display: flex;
   align-items: center;
@@ -75,6 +84,15 @@
   color: var(--dsw-alias-label-tertiary);
 }
 
+/* Heading row above a source group: non-interactive small grey text,
+ * padding aligned with items (mirrors ui-primitives Menu .label). */
+.groupTitle {
+  padding: 8px 10px;
+  font-size: 12px;
+  line-height: 16px;
+  color: var(--dsw-alias-label-tertiary);
+}
+
 /* Pending-source row: same cell metrics, dimmed label. */
 .loading {
   display: flex;
diff --git a/packages/client/ui-slash/src/client/MenuView.tsx b/packages/client/ui-slash/src/client/MenuView.tsx
index a6336e71e1..ecb69bf7fa 100644
--- a/packages/client/ui-slash/src/client/MenuView.tsx
+++ b/packages/client/ui-slash/src/client/MenuView.tsx
@@ -1,16 +1,21 @@
 /**
  * Trigger candidate menu: renders the SlashService menu store into the
  * conversation.input.overlay anchor. Closed state renders null (the overlay
- * slot stays mounted); groups render in roster order, pending groups as a
- * loading row; pointer picks route back through the service (combobox
- * pattern — focus never leaves the textarea, so rows are mousedown-handled
- * and the highlight is exposed via aria-activedescendant on the listbox).
+ * slot stays mounted); groups render in roster order under localized title
+ * rows, pending groups as a loading row; pointer picks route back through
+ * the service (combobox pattern — focus never leaves the textarea, so rows
+ * are mousedown-handled and the highlight is exposed via
+ * aria-activedescendant on the listbox).
  */
-import { useSyncExternalStore } from 'react'
+import { Fragment, useEffect, useRef, useSyncExternalStore } from 'react'
 import clsx from 'clsx'
+import { useAnchoredMaxHeight } from '@deepseek-ai/dsh-client-ui-primitives'
 import css from './MenuView.module.css'
 import type { MenuViewInjected } from './slots.ts'
 
+/** Design cap on the list height (figma SLASH 39:26572 MenuDropdown). */
+const MAX_HEIGHT = 320
+
 /** DOM id of one option row (the aria-activedescendant target). */
 function optionId(source: string, index: number): string {
   return `dsh-slash-option-${source}-${index}`
@@ -18,49 +23,86 @@ function optionId(source: string, index: number): string {
 
 /**
  * Render the candidate menu overlay entry.
- * @param props - injected face: the menu store and the pick route.
+ * @param props - injected face: the menu store, the pick route, and the menu-namespace translator.
  * @returns the dropdown while open; null while closed.
  */
-export function MenuView({ menu, onPick }: MenuViewInjected) {
+export function MenuView({ menu, onPick, onDismiss, t }: MenuViewInjected) {
   const state = useSyncExternalStore(
     fn => menu.subscribe(fn),
     () => menu.getSnapshot(),
   )
+  const listRef = useRef<HTMLDivElement>(null)
+  // The list is bottom-anchored above the composer; clamp the design cap to
+  // the space above it, re-measured on every store update (the anchor moves
+  // when the composer grows).
+  const maxHeight = useAnchoredMaxHeight(listRef, MAX_HEIGHT, state)
+  const highlight = state.open ? state.highlight : null
+  // Focus stays in the textarea (combobox pattern), so the browser never
+  // scrolls the active option into view on keyboard moves — do it here.
+  useEffect(() => {
+    if (highlight === null) return
+    document.getElementById(optionId(highlight.source, highlight.index))
+      ?.scrollIntoView({ block: 'nearest' })
+  }, [highlight])
+  // Dismiss on pointer outside the menu AND outside the composer card
+  // (clicking the textarea or bottom bar must not close the menu).
+  useEffect(() => {
+    if (!state.open) return
+    const onPointerDown = (ev: PointerEvent): void => {
+      if (!(ev.target instanceof Node)) return
+      if (listRef.current?.contains(ev.target)) return
+      const composerCard = listRef.current?.closest('[data-composer-card]')
+      if (composerCard?.contains(ev.target)) return
+      onDismiss()
+    }
+    document.addEventListener('pointerdown', onPointerDown, true)
+    return () => { document.removeEventListener('pointerdown', onPointerDown, true) }
+  }, [state.open, onDismiss])
   if (!state.open) return null
-  const { highlight } = state
   return (
     <div
+      ref={listRef}
       className={css.menu}
+      style={{ maxHeight }}
       role="listbox"
       aria-label="Trigger suggestions"
       aria-activedescendant={highlight !== null ? optionId(highlight.source, highlight.index) : undefined}
     >
-      {state.groups.map(group => group.status === 'pending'
-        ? <div key={group.source} className={css.loading} data-source={group.source}>Loading {group.source}…</div>
-        : group.items.map((item, index) => {
-          const active = highlight !== null && highlight.source === group.source && highlight.index === index
-          return (
-            <button
-              key={`${group.source}:${item.name}`}
-              id={optionId(group.source, index)}
-              type="button"
-              role="option"
-              aria-selected={active}
-              className={clsx(css.item, active && css.active)}
-              // mousedown, not click: the textarea keeps focus (combobox
-              // pattern) — preventing default stops the focus steal, and the
-              // pick runs before any blur-driven teardown.
-              onMouseDown={(ev) => {
-                ev.preventDefault()
-                onPick(group.source, index)
-              }}
-            >
-              {item.icon !== undefined && <span className={css.itemIcon} aria-hidden>{item.icon}</span>}
-              <span className={css.itemName}>{item.name}</span>
-              {item.description !== undefined && <span className={css.itemDescription}>{item.description}</span>}
-            </button>
-          )
-        }))}
+      <div className={css.viewport}>
+        {state.groups.map(group => (group.status === 'ready' && group.items.length === 0)
+          ? null
+          : (
+            <Fragment key={group.source}>
+              <div className={css.groupTitle} role="presentation" data-source={group.source}>{t(group.source)}</div>
+              {group.status === 'pending'
+                ? <div className={css.loading} data-source={group.source}>{t('loading')}</div>
+                : group.items.map((item, index) => {
+                  const active = highlight !== null && highlight.source === group.source && highlight.index === index
+                  return (
+                    <button
+                      key={`${group.source}:${item.name}`}
+                      id={optionId(group.source, index)}
+                      type="button"
+                      role="option"
+                      aria-selected={active}
+                      className={clsx(css.item, active && css.active)}
+                      // mousedown, not click: the textarea keeps focus (combobox
+                      // pattern) — preventing default stops the focus steal, and the
+                      // pick runs before any blur-driven teardown.
+                      onMouseDown={(ev) => {
+                        ev.preventDefault()
+                        onPick(group.source, index)
+                      }}
+                    >
+                      {item.icon !== undefined && <span className={css.itemIcon} aria-hidden>{item.icon}</span>}
+                      <span className={css.itemName}>{item.name}</span>
+                      {item.description !== undefined && <span className={css.itemDescription}>{item.description}</span>}
+                    </button>
+                  )
+                })}
+            </Fragment>
+          ))}
+      </div>
     </div>
   )
 }
diff --git a/packages/client/ui-slash/src/client/controller.ts b/packages/client/ui-slash/src/client/controller.ts
index ab0b26da54..3a8e85afc3 100644
--- a/packages/client/ui-slash/src/client/controller.ts
+++ b/packages/client/ui-slash/src/client/controller.ts
@@ -256,6 +256,13 @@ export class SlashController {
     this.refreshLexicon()
   }
 
+  /** External dismiss (e.g. pointer outside the composer area). */
+  dismiss(): void {
+    if (this.disposed) return
+    this.stopFetch()
+    this.reduce({ type: 'close' })
+  }
+
   /** Scope teardown: close and abort (the service deletes the map entry). */
   dispose(): void {
     this.disposed = true
diff --git a/packages/client/ui-slash/src/client/index.ts b/packages/client/ui-slash/src/client/index.ts
index f1c1d1953e..0ef751462a 100644
--- a/packages/client/ui-slash/src/client/index.ts
+++ b/packages/client/ui-slash/src/client/index.ts
@@ -4,6 +4,8 @@
  * self-registers into the conversation.input.overlay slot. Frozen pipeline
  * contract in ./contract.ts; sources register through ctx.slash alone.
  */
+// Type-only: pulls the locale plugin's Context merge (ctx.locale).
+import type {} from '@deepseek-ai/dsh-client-locale/client'
 import type { ClientContext } from '@deepseek-ai/dsh-client-runtime/client'
 import { SlashService } from './service.ts'
 import type { MenuViewInjected } from './slots.ts'
@@ -29,8 +31,11 @@ declare module 'cordis' {
   }
 }
 
-/** Required services: controller resolution reads the session scope tree. */
-export const inject = ['sessions']
+/** Namespace owning the candidate-menu copy: group titles keyed by source name plus the pending row. */
+const MENU_NS = 'slash.menu'
+
+/** Required services: controller resolution reads the session scope tree; the menu copy is localized. */
+export const inject = ['sessions', 'locale']
 
 /**
  * Client plugin body: mount the service, then register MenuView into the
@@ -39,6 +44,13 @@ export const inject = ['sessions']
  */
 export function apply(ctx: ClientContext): void {
   ctx.plugin(SlashService)
+  ctx.effect(() => {
+    const disposers = [
+      ctx.locale.register(MENU_NS, 'zh', { command: '命令', skill: '技能', subagent: '子智能体', loading: '正在加载…' }),
+      ctx.locale.register(MENU_NS, 'en', { command: 'Commands', skill: 'Skills', subagent: 'Subagents', loading: 'Loading…' }),
+    ]
+    return () => { for (const dispose of disposers) dispose() }
+  }, 'ui-slash: menu dictionaries')
   // Conditional mount: 'conversation.input.overlay' is declared by the
   // conversation composer entry, and the conversation service is mounted
   // after that declaration lands on the ledger — its presence is the
@@ -59,6 +71,8 @@ export function apply(ctx: ClientContext): void {
         return {
           menu: controller.menu,
           onPick: (source, index) => { controller.pick(source, index) },
+          onDismiss: () => { controller.dismiss() },
+          t: scope.locale.bind(MENU_NS),
         }
       },
     }, MenuView), 'ui-slash: MenuView overlay registration')
diff --git a/packages/client/ui-slash/src/client/service.ts b/packages/client/ui-slash/src/client/service.ts
index c5448a3d50..d9af10e887 100644
--- a/packages/client/ui-slash/src/client/service.ts
+++ b/packages/client/ui-slash/src/client/service.ts
@@ -87,7 +87,7 @@ export class SlashService extends Service implements SlashServiceContract {
       actx,
       sessionId: id,
       roster: {
-        sources: trigger => live.sources.filter(s => s.trigger === trigger),
+        sources: trigger => live.sources.filter(s => s.trigger === trigger).sort((a, b) => (a.order ?? 0) - (b.order ?? 0)),
         all: () => live.sources,
       },
     })
diff --git a/packages/client/ui-slash/src/client/slots.ts b/packages/client/ui-slash/src/client/slots.ts
index f74ec29457..f69be9f28a 100644
--- a/packages/client/ui-slash/src/client/slots.ts
+++ b/packages/client/ui-slash/src/client/slots.ts
@@ -9,6 +9,7 @@
  */
 // Type-only edge: the SlotMap augmentation below merges into this package's interface.
 import type {} from '@deepseek-ai/dsh-client-ui-slots'
+import type { Translate } from '@deepseek-ai/dsh-client-locale/client'
 import type { SnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
 import type { MenuState } from '../core/contract.ts'
 
@@ -35,4 +36,12 @@ export interface MenuViewInjected {
    * @param index - candidate index within the group.
    */
   onPick: (source: string, index: number) => void
+  /** Dismiss the menu (external pointer outside the composer area). */
+  onDismiss: () => void
+  /**
+   * Bound translator for the menu namespace: group titles keyed by source
+   * name (the locale fallback chain returns the key itself, so an unknown
+   * source shows its raw name) plus the pending-row text.
+   */
+  t: Translate
 }
diff --git a/packages/client/ui-slash/src/types.ts b/packages/client/ui-slash/src/types.ts
index 4b9bd64408..2b63381efb 100644
--- a/packages/client/ui-slash/src/types.ts
+++ b/packages/client/ui-slash/src/types.ts
@@ -138,6 +138,8 @@ export interface SlashSource {
   readonly trigger: TriggerChar
   /** Menu group label; unique per trigger — duplicate registration throws. */
   readonly name: string
+  /** Menu group display order (lower = higher in the list; default 0). */
+  readonly order?: number
   candidates(session: ClientSessionContext, req: CandidateRequest): Promise<readonly SlashCandidate[]>
   /** Every pick lands here; claim/insert outcomes are executed by the pipeline via the scoped input events. */
   onPick(pick: SlashPick): PickOutcome
diff --git a/packages/client/ui-slash/tests/apply.spec.ts b/packages/client/ui-slash/tests/apply.spec.ts
index 637f18f102..2e20dc96f8 100644
--- a/packages/client/ui-slash/tests/apply.spec.ts
+++ b/packages/client/ui-slash/tests/apply.spec.ts
@@ -7,6 +7,7 @@
  */
 import { Context } from 'cordis'
 import { describe, expect, it, vi } from 'vitest'
+import { LocaleService } from '@deepseek-ai/dsh-client-locale/client'
 import { createScope, scopeOf, SlotsService } from '@deepseek-ai/dsh-client-runtime/client'
 import type { SessionId } from '@deepseek-ai/dsh-client-runtime/client'
 import { apply, inject, SlashService } from '@deepseek-ai/dsh-client-ui-slash/client'
@@ -31,12 +32,25 @@ async function bench() {
     scope: (id: SessionId) => (id === sid('a') ? scope.ctx : undefined),
     scopeOf: (c: Context) => scopeOf(c),
   })
-  return { ctx, slots }
+  const locale = new LocaleService(ctx)
+  ctx.provide('locale', locale)
+  return { ctx, slots, locale }
 }
 
 describe('apply', () => {
-  it('declares the sessions dependency (controller resolution reads the scope tree)', () => {
-    expect(inject).toEqual(['sessions'])
+  it('declares the sessions and locale dependencies (scope tree + localized menu copy)', () => {
+    expect(inject).toEqual(['sessions', 'locale'])
+  })
+
+  it('registers the bilingual menu dictionaries (group titles by source name + the pending row)', async () => {
+    const { ctx, locale } = await bench()
+    await ctx.plugin({ inject: [...inject], apply }).await()
+    const t = locale.bind('slash.menu')
+    expect(t('command')).toBe('命令')
+    locale.setLocale('en')
+    expect(t('skill')).toBe('Skills')
+    expect(t('subagent')).toBe('Subagents')
+    expect(t('loading')).toBe('Loading…')
   })
 
   it('mounts ctx.slash once sessions is up, before any conversation service exists', async () => {
@@ -65,9 +79,14 @@ describe('apply', () => {
       (ctx.get('sessions') as { scope(id: SessionId): Context }).scope(sid('a')),
     )
     expect(injected.menu).toBe(controller.menu)
+    // The injected translator is the menu-namespace binding.
+    expect(injected.t('command')).toBe('命令')
     // The pick face routes into the controller pipeline (closed menu → no-op).
     injected.onPick('command', 0)
     expect(controller.menu.getSnapshot().open).toBe(false)
+    // The dismiss face routes into the controller too (closed menu → no-op).
+    injected.onDismiss()
+    expect(controller.menu.getSnapshot().open).toBe(false)
     // An unknown session id fails loud (no silent scope miss).
     expect(() => injectEntry(sid('ghost'))).toThrow(/resolved no scope/)
   })
diff --git a/packages/client/ui-slash/tests/menu-view.spec.tsx b/packages/client/ui-slash/tests/menu-view.spec.tsx
index d9b6a08a88..1c74340575 100644
--- a/packages/client/ui-slash/tests/menu-view.spec.tsx
+++ b/packages/client/ui-slash/tests/menu-view.spec.tsx
@@ -1,11 +1,13 @@
 // @vitest-environment jsdom
 /**
  * MenuView rendering spec, props-direct (slot-parity doctrine): closed store
- * renders null, groups render in roster order with pending rows as loading,
- * pointer picks route (source, index) back without stealing focus, and the
- * highlight is exposed through aria-activedescendant + aria-selected.
+ * renders null, groups render in roster order under localized title rows
+ * (unknown sources fall back to the raw name) with pending rows as loading,
+ * pointer picks route (source, index) back without stealing focus, the
+ * highlight is exposed through aria-activedescendant + aria-selected, and
+ * the list height clamps to the space above the composer.
  */
-import { afterEach, describe, expect, it, vi } from 'vitest'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
 import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
 import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
 import type { MenuState, TriggerHit } from '@deepseek-ai/dsh-client-ui-slash/client'
@@ -34,13 +36,35 @@ function openState(partial?: Partial<MenuState>): MenuState {
   }
 }
 
-afterEach(cleanup)
+// jsdom has no scrollIntoView; the view calls it on the highlighted option.
+const scrollIntoView = vi.fn()
+beforeEach(() => {
+  Element.prototype.scrollIntoView = scrollIntoView
+  scrollIntoView.mockClear()
+})
+
+afterEach(() => {
+  cleanup()
+  vi.restoreAllMocks()
+})
+
+// Dictionary-backed fake mirroring the LocaleService key fallback (an
+// unknown key comes back verbatim, so unknown sources show their raw name).
+const DICT: Record<string, string> = { command: 'Commands', skill: 'Skills', loading: 'Loading…' }
+const t = (key: string) => DICT[key] ?? key
 
 function mount(state: MenuState) {
   const menu = createSnapshotStore<MenuState>(state)
   const onPick = vi.fn()
-  const view = render(<MenuView menu={menu} onPick={onPick} />)
-  return { menu, onPick, view }
+  const onDismiss = vi.fn()
+  const view = render(<MenuView menu={menu} onPick={onPick} onDismiss={onDismiss} t={t} />)
+  return { menu, onPick, onDismiss, view }
+}
+
+/** The non-interactive group title rows (role=presentation), in document order. */
+function titles(container: HTMLElement): string[] {
+  return [...container.querySelectorAll('div[role="presentation"][data-source]')]
+    .map(el => el.textContent ?? '')
 }
 
 describe('MenuView', () => {
@@ -57,7 +81,19 @@ describe('MenuView', () => {
     mount(openState())
     const options = screen.getAllByRole('option')
     expect(options.map(o => o.textContent)).toEqual(['⚑goalSet up a goal', 'plan'])
-    expect(screen.queryByText('Loading skill…')).not.toBeNull()
+    expect(screen.queryByText('Loading…')).not.toBeNull()
+  })
+
+  it('titles each group with the localized source name, raw name for unknown sources, none for empty ready groups', () => {
+    const { view } = mount(openState({
+      groups: [
+        { source: 'command', status: 'ready', items: [{ name: 'goal' }] },
+        { source: 'hollow', status: 'ready', items: [] },
+        { source: 'mystery', status: 'ready', items: [{ name: 'x' }] },
+        { source: 'skill', status: 'pending', items: [] },
+      ],
+    }))
+    expect(titles(view.container)).toEqual(['Commands', 'mystery', 'Skills'])
   })
 
   it('exposes the highlight via aria-activedescendant and aria-selected', () => {
@@ -75,6 +111,79 @@ describe('MenuView', () => {
     expect(screen.getByRole('listbox').getAttribute('aria-activedescendant')).toBeNull()
   })
 
+  it('scrolls the highlighted option into view when the highlight moves', () => {
+    const { menu } = mount(openState())
+    scrollIntoView.mockClear()
+    act(() => { menu.set(openState({ highlight: { source: 'command', index: 1 } })) })
+    const options = screen.getAllByRole('option')
+    expect(scrollIntoView).toHaveBeenCalledWith({ block: 'nearest' })
+    expect(scrollIntoView.mock.instances.at(-1)).toBe(options[1])
+  })
+
+  it('caps the list height at the design maximum when the composer sits low enough', () => {
+    vi.spyOn(Element.prototype, 'getBoundingClientRect').mockReturnValue({ bottom: 800 } as DOMRect)
+    mount(openState())
+    expect(screen.getByRole('listbox').style.maxHeight).toBe('320px')
+  })
+
+  it('clamps the list height to the space above the composer minus the safe margin', () => {
+    vi.spyOn(Element.prototype, 'getBoundingClientRect').mockReturnValue({ bottom: 200 } as DOMRect)
+    mount(openState())
+    expect(screen.getByRole('listbox').style.maxHeight).toBe('188px')
+  })
+
+  it('re-fits the height when the window resizes', () => {
+    const rect = vi.spyOn(Element.prototype, 'getBoundingClientRect')
+    rect.mockReturnValue({ bottom: 800 } as DOMRect)
+    mount(openState())
+    expect(screen.getByRole('listbox').style.maxHeight).toBe('320px')
+    rect.mockReturnValue({ bottom: 100 } as DOMRect)
+    act(() => { window.dispatchEvent(new Event('resize')) })
+    expect(screen.getByRole('listbox').style.maxHeight).toBe('88px')
+  })
+
+  it('pointerdown outside the menu (no composer card ancestor) dismisses', () => {
+    const { onDismiss } = mount(openState())
+    fireEvent.pointerDown(document.body)
+    expect(onDismiss).toHaveBeenCalledTimes(1)
+  })
+
+  it('pointerdown inside the list does not dismiss', () => {
+    const { onDismiss } = mount(openState())
+    fireEvent.pointerDown(screen.getAllByRole('option')[0]!)
+    expect(onDismiss).not.toHaveBeenCalled()
+  })
+
+  it('pointerdown inside the surrounding composer card does not dismiss; outside it does', () => {
+    const menu = createSnapshotStore<MenuState>(openState())
+    const onDismiss = vi.fn()
+    render(
+      <div data-composer-card="">
+        <MenuView menu={menu} onPick={vi.fn()} onDismiss={onDismiss} t={t} />
+        <button type="button" data-testid="composer-button" />
+      </div>,
+    )
+    fireEvent.pointerDown(screen.getByTestId('composer-button'))
+    expect(onDismiss).not.toHaveBeenCalled()
+    fireEvent.pointerDown(document.body)
+    expect(onDismiss).toHaveBeenCalledTimes(1)
+  })
+
+  it('ignores a pointerdown whose target is not a DOM node', () => {
+    const { onDismiss } = mount(openState())
+    const ev = new Event('pointerdown', { bubbles: true })
+    Object.defineProperty(ev, 'target', { value: {} })
+    document.dispatchEvent(ev)
+    expect(onDismiss).not.toHaveBeenCalled()
+  })
+
+  it('closing the menu removes the dismiss listener', () => {
+    const { menu, onDismiss } = mount(openState())
+    act(() => { menu.set(CLOSED) })
+    fireEvent.pointerDown(document.body)
+    expect(onDismiss).not.toHaveBeenCalled()
+  })
+
   it('mousedown on a row picks (source, index) and prevents the focus steal', () => {
     const { onPick } = mount(openState())
     const options = screen.getAllByRole('option')
diff --git a/packages/client/ui-slash/tsconfig.json b/packages/client/ui-slash/tsconfig.json
index a3002d4981..deca328a0a 100644
--- a/packages/client/ui-slash/tsconfig.json
+++ b/packages/client/ui-slash/tsconfig.json
@@ -11,9 +11,15 @@
     {
       "path": "../../../vendor/cordis"
     },
+    {
+      "path": "../locale"
+    },
     {
       "path": "../runtime"
     },
+    {
+      "path": "../ui-primitives"
+    },
     {
       "path": "../ui-slots"
     },
diff --git a/packages/client/ui-slots/README.i18n.yaml b/packages/client/ui-slots/README.i18n.yaml
index 4fb34675c7..66d309f045 100644
--- a/packages/client/ui-slots/README.i18n.yaml
+++ b/packages/client/ui-slots/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/ui-slots/README.md
 README.md: ed6f052b3a47e08d693928b6763e32427b829467
-README.zh.md: 8f15352d09a33a862203507ac89841da91a43e59
+README.zh.md: 17c3cbb28defe0c9bc66df417976984be3955b53
diff --git a/packages/client/ui-slots/README.zh.md b/packages/client/ui-slots/README.zh.md
index 8f15352d09..17c3cbb28d 100644
--- a/packages/client/ui-slots/README.zh.md
+++ b/packages/client/ui-slots/README.zh.md
@@ -2,24 +2,24 @@
 
 [English](README.md) | 中文
 
-Slot 注册表纯核心、slot 终端设计：SlotMap 声明合并、SlotCore 上唯一的 `register` 组合 API、四 share 组件 props 类型家族、store seat 类型家族，以及 renderer 安装 seam 契约。React 类型仅在运行时使用，该包不依赖 React，也不依赖 cordis。
+Slot 注册表纯核心、slot 终端设计：SlotMap 声明合并、SlotCore 上唯一的 `register` 组合 API、四 share 组件 props 类型家族、store seat 类型家族，以及 renderer 安装 seam 契约。只使用 React 类型；该包（package）不依赖 React，也不依赖 Cordis。
 
 一次 `register({ name, children?, store?, inject?, ...kind }, Component)` 调用会向已声明 slot 贡献一个组件，同时声明子 slot（声明 = 渲染授权 = 运行时规范，三者共用一张表）、store seat 以及注册方的业务表层。组件会在调用点依据 `ComposedProps` 接受检查；该类型是四个 share 的交集，每个 share 都从各自的唯一真源派生：
 
 | share | 类型 | 来源 |
 |---|---|---|
-| runtime | `PropsRuntime<K>` | SlotMap 配置项：`owner`（父级 renderSlot 调用点）+ Session 标准工具包 + 全局 seat |
+| runtime | `PropsRuntime<K>` | SlotMap 条目：`owner`（父级 renderSlot 调用点）+ Session 标准工具包 + 全局 seat |
 | child render | `PropsRenderSlots<S>` | register 调用的 `children` key 集合（静态缩窄的 `renderSlot`） |
 | store | `PropsStore<H>` | 已声明 handle：`useStore` selector hook + 移除 draft 的 `actions` |
 | business | `I` | 从 `inject` factory 返回值推断 |
 
-chain-kind slot 会反转键控路由：配置项自行提名，而不是由分发点选择 `entryKey`。每次注册都携带一个纯 `ChainSelect` selector（另有可选的升序 `priority`，相同值按注册顺序处理）；第一个非 null 返回值选中其配置项，并成为组件的 `matched` prop；全部返回 null 时则使用 owner 的 `renderSlotChain` fallback（`ChainRenderOpts`）。
+chain-kind slot 会反转键控路由：条目自行提名，而不是由分发点选择 `entryKey`。每次注册都携带一个纯 `ChainSelect` selector（另有可选的升序 `priority`，相同值按注册顺序处理）；第一个非 null 返回值选中其条目，并成为组件的 `matched` prop；全部返回 null 时则使用 owner 的 `renderSlotChain` fallback（`ChainRenderOpts`）。
 
 标准工具包接口（`SessionStandardProps`、`GlobalStandardProps`）在这里声明为空，由 runtime 包合并（与 SlotMap key 相同的 declare-merge 模式）。renderer 会把运行时 Session 和 Workspace observable source 绑定为 selector hook。Inject factory 参数从声明派生（`InjectParams`）：Session slot 获得 `sessionId`；声明 store 时追加 baked `actions`；没有其他参数，数据访问位于 apply 闭包的 ctx 中。
 
 store 家族（输入 `defineStore` 规范／输出 `StoreHandle<T, A>`）为 store seat 建模：`init` 推断状态 schema；`actions` 是完整的 draft-transform 写入集合；`BakedActions` 移除 draft 参数，成为组件和 inject factory 收到的回调。`defineStore` 值实现位于 runtime 包（引擎所属位置），并满足这里导出的 `DefineStore` 契约。引擎产物与 renderer host 契约携带裸快照 source（`getSnapshot`／`subscribe`），绝不携带 React hook；hook 绑定属于渲染机制这一侧的 seam，只有 props 契约 hook 类型（`SnapshotSelectorHook`）位于这里。
 
-`SlotCore` 在构造时播种先验的 `'root'` slot，并强制执行加载时验证（注册未声明 slot、重复声明子项、在两个 scope 下使用同一个共享 handle、chain 注册缺少 `select`，这些情况都在 register 时抛出）。配置项的 disposer 会递归折叠其声明的子 slot：账本行、贡献和 store 挂载都沿同一生命周期轴消失。`renderer.ts` 携带安装 seam（`SlotRenderer`、`SlotRendererHost`）以及 `StaleAuthorizationError`／`SlotOwnershipError`；实现在 web-react 中，安装则在外壳启动中完成。
+`SlotCore` 在构造时预置 `'root'` slot，并强制执行加载时验证（注册未声明 slot、重复声明子项、在两个 scope 下使用同一个共享 handle、chain 注册缺少 `select`，这些情况都在 register 时抛出）。条目的 disposer 会递归移除其声明的子 slot：账本行、贡献和 store 挂载都会随同一生命周期结束而移除。`renderer.ts` 携带安装 seam（`SlotRenderer`、`SlotRendererHost`）以及 `StaleAuthorizationError`／`SlotOwnershipError`；实现在 web-react 中，安装则在外壳启动中完成。
 
 ## 模型体验
 
@@ -31,5 +31,5 @@ store 家族（输入 `defineStore` 规范／输出 `StoreHandle<T, A>`）为 st
 
 ## 已知限制与暂缓事项
 
-- **`isLive` 会线性扫描所有记录**：在 UI 插件的注册规模（数十项）下没有问题；如果账本变得频繁访问，再使用配置项→记录反向引用改进。
+- **`isLive` 会线性扫描所有记录**：在 UI 插件的注册规模（数十项）下没有问题；如果账本变得频繁访问，再使用条目→记录反向引用改进。
 - **`__renders` 幻象锚点在 `PropsRenderSlots` 上可见**：这是与类型链设计的 `__accepts` 相同且已接受的噪声；泛型方法签名在 key 联合之间比较宽松，因此必须依靠逆变标记强制执行「组件 key 集合 ⊆ children 声明」。
diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml
index 86995fc65c..39c7c6cd5b 100644
--- a/packages/client/ui-subagent/README.i18n.yaml
+++ b/packages/client/ui-subagent/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md
 README.md: 7a70add139eae7bc507469b4fe7170359efdec31
-README.zh.md: 2d8ee677c71179df88211d90120a6017ceac8f6a
+README.zh.md: 4ff79780fd33a47a0a45695ab9feda15cd1763cd
diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md
index 2d8ee677c7..4ff79780fd 100644
--- a/packages/client/ui-subagent/README.zh.md
+++ b/packages/client/ui-subagent/README.zh.md
@@ -4,15 +4,15 @@
 
 subagent 引用 source 的浏览器半侧：把 `@` 触发的 `subagent` source 注册进 `ctx.slash`。候选零 RPC——从注册时捕获的根 `ctx.sessions.list` 快照过滤（每次调用的投影所指会话的子会话：`parentId` 匹配、`running`、`displayTitle` 包含 query）；pick 一个候选会把字面文本 `@label ` 经 slash 管线落进草稿（决策 21 的纯文本引用），source 的 `codec` 把两种投影都产出为 `@label`——在 `@` 消费功能定义模型表示之前，模型序列化保持原始 label。source 不实现 `matchSpace`／`matchEnter` 钩子——subagent 引用永不进入命令裁决，随普通提示词落入 default sink。
 
-没有运行中子会话的会话就是没有候选。本阶段只交付「菜单 + 引用文本」；消费一个 `@label` 意味着什么（对子会话做 steering（中途引导）、恢复已 dispose 的子会话）是未来的业务工作。
+没有运行中子会话的会话就是没有候选。本阶段只交付「菜单 + 引用文本」；消费一个 `@label` 意味着什么（对子会话做 steering（中途引导）、恢复已 dispose（资源释放）的子会话）是未来的业务工作。
 
-`/client` 导出表层只有插件主体（`apply`／`inject`）；source 对象是注册 effect 的内部实现。
+`/client` 的导出内容只有插件主体（`apply`／`inject`）；source 对象是注册 effect 的内部实现。
 
 ## 模型体验
 
 ### 用户提示词中的 subagent label 文本
 
-#### 模型所见
+#### 模型看到的内容
 
 被 pick 的候选会把字面文本 `@label`（子会话的显示标题）落进草稿；该文本原样进入普通用户消息（`session.prompt`）到达模型，没有专用内容块、提示词 section 或 host 侧解析。目前不存在任何消费语义：模型看到的是纯文本，只能自行解读。
 
@@ -26,6 +26,6 @@ subagent 引用 source 的浏览器半侧：把 `@` 触发的 `subagent` source
 
 ## 已知限制与暂缓事项
 
-- **`@` 消费语义尚未构建**：引用只是惰性文本；把它接到对指名子会话的 steering／发消息（以及是否允许恢复已 dispose 的子会话），等待台账中它自己的设计决策。
-- **候选只有运行中的子会话**：已完成或已 dispose 的 subagent 永不出现，roster 只含 scope 所指会话的直接子会话（不含孙辈，不含跨会话 agent）。
-- **label 是显示标题，不是稳定 id**：两个子会话共用一个显示标题时，产生的引用无法区分；标题变更会使先前插入的文本失去指向。引用还是惰性文本时可以接受；消费功能必须绑定到会话 id。
+- **`@` 消费语义尚未构建**：引用只是不具消费语义的纯文本；将其接入对指名子会话进行 steering／发送消息的机制（以及是否允许恢复已 dispose 的子会话），仍有待台账中的专门设计决策。
+- **候选只有运行中的子会话**：已完成或已 dispose 的 subagent 永不出现；roster 只含 scope 所指会话的直接子会话，不含孙辈，也不含跨会话 agent（智能体）。
+- **label 是显示标题，不是稳定 id**：两个子会话共用一个显示标题时，产生的引用无法区分；标题变更会使先前插入的文本失去指向。引用仍是不具消费语义的纯文本时尚可接受；消费功能必须绑定到会话 id。
diff --git a/packages/client/ui-theme/README.i18n.yaml b/packages/client/ui-theme/README.i18n.yaml
index cc5c4ec07b..cac7485444 100644
--- a/packages/client/ui-theme/README.i18n.yaml
+++ b/packages/client/ui-theme/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-theme/README.md
 README.md: a1ff7d840dae86f5da98de1208ecda3b8b62026b
-README.zh.md: 49b52bcb1e07527e98c602086404228c5513091a
+README.zh.md: 2e034f3173baabb3eea6b4ae670d5070c95480be
diff --git a/packages/client/ui-theme/README.zh.md b/packages/client/ui-theme/README.zh.md
index 49b52bcb1e..2e034f3173 100644
--- a/packages/client/ui-theme/README.zh.md
+++ b/packages/client/ui-theme/README.zh.md
@@ -6,9 +6,9 @@
 
 `src/styles/` 下有五张样式表，全部由 web 壳的 `base.css` 导入：`base.css`、`design-platform.css`、`scrollbar.css`、`gradient-shadow-text.css` 与 `shiki.css`。`scrollbar.css` 是 `--dsw-alias-scrollbar-*` token 的唯一消费方，必须排在声明这些 token 的 `design-platform.css` 之后。
 
-滚动条重新绑定契约：`scrollbar.css` 在 `body` 上把 `--dsh-scrollbar-thumb` 与 `--dsh-scrollbar-thumb-hover` 绑定到 l1（基础表面）token，两条渲染路径都读取这一组变量。抬升表面（菜单、浮层、对话框）在自己的容器上设置 `--dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2)` 与 `--dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2)`；一次重新绑定即可为引擎实际走的那条路径换色。
+滚动条重新绑定契约：`scrollbar.css` 在 `body` 上把 `--dsh-scrollbar-thumb` 与 `--dsh-scrollbar-thumb-hover` 绑定到 l1（基础表面）token，两条渲染路径都读取这一组变量。高层级表面（菜单、浮层、对话框）在自己的容器上设置 `--dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2)` 与 `--dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2)`；一次重新绑定即可为引擎实际走的那条路径换色。
 
-两条路径在构造上互斥。`scrollbar-width`／`scrollbar-color` 写在 `@supports not selector(::-webkit-scrollbar)` 之内，因为这两个属性只要取非 `auto` 值，Chromium 与 Safari 就会丢弃该元素上的全部 `::-webkit-scrollbar*` 规则，`::-webkit-scrollbar-thumb:hover` 也在其中——若无条件地同时声明，`--dsh-scrollbar-thumb-hover` 在任何引擎上都不会被渲染。因此 Firefox 走标准属性，WebKit 系引擎走伪元素，hover token 只经由伪元素这条路径渲染。推理过程与实测计算值见[滚动条 Agent Note](../../../.agents/notes/implemented/bug-fix/2026-07-28-themed-scrollbars-and-reserved-gutter.md)。
+两条路径在构造上互斥。`scrollbar-width`／`scrollbar-color` 写在 `@supports not selector(::-webkit-scrollbar)` 之内，因为这两个属性中的任一个只要取非 `auto` 值，Chromium 与 Safari 就会丢弃该元素上的全部 `::-webkit-scrollbar*` 规则，`::-webkit-scrollbar-thumb:hover` 也在其中——若无条件地同时声明，`--dsh-scrollbar-thumb-hover` 在任何引擎上都不会被渲染。因此 Firefox 走标准属性，WebKit 系引擎走伪元素，hover token 只经由伪元素这条路径渲染。推理过程与实测计算值见[滚动条 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/bug-fix/2026-07-28-themed-scrollbars-and-reserved-gutter.md)。
 
 ## 模型体验
 
@@ -16,9 +16,9 @@
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
 - **第三方主题是表层，不是产品**：注册主题意味着覆盖同名别名变量；目前不会验证一组覆盖是否完整。
-- **token 样式表是颜色的唯一权威**：不会追加 cssdesign 中缺失的值（例如设计中的 #4176E6 标签页蓝色）；应使用最接近的语义 token（裁定于 2026-07-22）。
+- **token 样式表是颜色值的唯一权威来源**：会有意不补入 cssdesign 中缺失的值（例如设计中的 #4176E6 标签页蓝色）；一律采用最接近的语义 token（裁定于 2026-07-22）。
diff --git a/packages/client/ui-trajectory/README.i18n.yaml b/packages/client/ui-trajectory/README.i18n.yaml
index b07fe7a1fd..dcaa1020a7 100644
--- a/packages/client/ui-trajectory/README.i18n.yaml
+++ b/packages/client/ui-trajectory/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 9e7dee8d5572baad13d7598d1fbde7b042dd5ab4
-README.zh.md: da14d5265c16acb4e75d21bc445ef9702c83e697
+#   pnpm run verify-translation-pairing --write packages/client/ui-trajectory/README.md
+README.md: b9c8b849b3454fe46e1fc37713d9d3b9449734cf
+README.zh.md: 19ae5050a4c4f7dfe80de0ab58e772e9d26e3f6a
diff --git a/packages/client/ui-trajectory/README.md b/packages/client/ui-trajectory/README.md
index 9e7dee8d55..b9c8b849b3 100644
--- a/packages/client/ui-trajectory/README.md
+++ b/packages/client/ui-trajectory/README.md
@@ -2,7 +2,7 @@
 
 English | [中文](README.zh.md)
 
-Trajectory turn-list chrome (sticky Turn / Message·Step groups / step cells) plus Waterfall placeholder; the pure-consumer minimal plugin exemplar (registers two view tabs into the conversation's `'conversation.view'` slot ring, provides no service, declares no Context merge). Contract: api-contracts v3 §8.
+Trajectory renders a turn-aware event ledger with selectable User, Assistant, Tool, and nested Subtool records. Thick rules mark Turn boundaries, compact inline markers identify Steps, and the main ledger keeps only index, event, and content; selection opens a local inspector for token usage, duration, Input, Output, and Timing. A fixed Overview above the ledger projects real record start/duration timing from left to right; dragging an interval focuses the ledger on every record active at any point in that inclusive range, while clearing the selection restores the full branch. The runtime's independent history source supplies raw context lineage and projects cancellation-frozen Assistant and Tool records, so Trajectory neither reads nor changes the Chat conversation snapshot. The package remains a pure-consumer plugin (registers one view tab into the conversation's `'conversation.view'` slot ring, provides no service, declares no Context merge). Contract: api-contracts v3 §8.
 
 ## Model Experience
 
@@ -14,4 +14,4 @@ None; this package neither assembles nor sends a provider request.
 
 ## Known Limitations and Deferred Work
 
-- **In-flight Time stays blank** — `partial` / `runningCalls` rows render with `—` until a live clock policy lands; selected styling is local-only (not wired to chat details); anchor deep-linking remains deferred.
+- **In-flight Time stays blank** — `partial` / `runningCalls` rows show their running state without a fabricated duration until a live clock policy lands, so the Overview renders a start marker rather than inventing a live span; record and timeline selection are intentionally local to Trajectory; anchor deep-linking remains deferred.
diff --git a/packages/client/ui-trajectory/README.zh.md b/packages/client/ui-trajectory/README.zh.md
index da14d5265c..19ae5050a4 100644
--- a/packages/client/ui-trajectory/README.zh.md
+++ b/packages/client/ui-trajectory/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-轨迹轮次列表 chrome（吸顶 Turn／Message·Step 分组／步骤单元格）及 Waterfall 占位符；这是纯消费方最小插件范例（向会话的 `'conversation.view'` slot 环注册两个视图标签页，不提供服务，也不声明 Context 合并）。契约：api-contracts v3 §8。
+Trajectory 渲染按轮次组织的事件记录表，其中可选择用户、助手、工具和嵌套子工具记录。较粗的分割线标示轮次边界，紧凑的行内标记标识步骤，主记录表仅保留索引、事件和内容；选择记录则会打开局部检查器，查看 token 用量、耗时、输入、输出和计时。固定在记录表上方的 Overview 区域从左到右投影记录的真实开始时间与耗时；拖选一个区间会将记录表聚焦到活动区间与该闭区间有重叠的所有记录，清除选择则恢复完整分支。runtime 的独立历史数据源提供原始上下文谱系，并投影因取消而冻结的助手和工具记录，因此 Trajectory 既不读取也不改变 Chat 会话快照。该包（package）保持为纯消费方插件（向会话的 `'conversation.view'` slot 环注册一个视图标签页，不提供服务，也不声明 Context 合并）。契约：api-contracts v3 §8。
 
 ## 模型体验
 
@@ -10,8 +10,8 @@
 
 #### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
 ## 已知限制与暂缓事项
 
-- **进行中的 Time 保持空白**：`partial`／`runningCalls` 行在实时钟策略落地前渲染为 `—`；选中样式只在本地生效（未连接到聊天详情）；锚点深链接仍暂缓实现。
+- **进行中时，Time 保持空白**：`partial`／`runningCalls` 行会显示运行状态，但在实时钟策略落地前不会虚构耗时，因此 Overview 区域只渲染开始标记，而不会杜撰实时跨度；记录选择与时间线选择有意保持在 Trajectory 内部；锚点深链接仍暂缓实现。
diff --git a/packages/client/ui-trajectory/package.json b/packages/client/ui-trajectory/package.json
index 28f6620f2c..fc853d19d1 100644
--- a/packages/client/ui-trajectory/package.json
+++ b/packages/client/ui-trajectory/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@deepseek-ai/dsh-client-ui-trajectory",
-  "description": "Trajectory/Waterfall placeholder views: pure-consumer plugin registering into the conversation ViewMap (no service)",
+  "description": "Trajectory event ledger with an interactive timing overview: pure-consumer plugin registering into the conversation ViewMap (no service)",
   "version": "0.0.1",
   "private": true,
   "type": "module",
@@ -33,13 +33,18 @@
     "watch": "tsdown --watch"
   },
   "license": "BSD-3-Clause",
+  "dependencies": {
+    "diff": "^9.0.0"
+  },
   "peerDependencies": {
     "@deepseek-ai/dsh-invariants": "^0.0.1",
+    "@deepseek-ai/dsh-client-ui-primitives": "^0.0.1",
     "cordis": "^4.0.0-rc.7",
     "react": "^18.2.0"
   },
   "devDependencies": {
     "@deepseek-ai/dsh-client-runtime": "workspace:^",
+    "@deepseek-ai/dsh-client-ui-primitives": "workspace:^",
     "@deepseek-ai/dsh-client-ui-conversation": "workspace:^",
     "@deepseek-ai/dsh-client-ui-slots": "workspace:^",
     "@deepseek-ai/dsh-invariants": "workspace:^",
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryCell.module.css b/packages/client/ui-trajectory/src/client/TrajectoryCell.module.css
index c5efc232d1..013b1fd222 100644
--- a/packages/client/ui-trajectory/src/client/TrajectoryCell.module.css
+++ b/packages/client/ui-trajectory/src/client/TrajectoryCell.module.css
@@ -46,14 +46,40 @@
   white-space: nowrap;
 }
 
+.tagSystem {
+  color: var(--dsw-alias-label-secondary);
+  background: var(--dsw-alias-bg-module-platform);
+}
+
 .tagUser {
   color: var(--dsw-alias-state-success-primary);
   background: var(--dsw-alias-state-success-tertiary);
 }
 
+.tagContext {
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-state-success-primary) 68%,
+    var(--dsw-alias-label-secondary)
+  );
+  background: var(--dsw-alias-state-success-tertiary);
+}
+
 .tagMessage {
-  color: var(--dsw-alias-brand-primary-new-colorprimary-new-color);
-  background: var(--dsw-specific-bubble);
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-brand-primary-new-colorprimary-new-color) 60%,
+    var(--dsw-alias-state-error-secondary)
+  );
+  background: color-mix(
+    in srgb,
+    color-mix(
+      in srgb,
+      var(--dsw-alias-brand-primary-new-colorprimary-new-color) 55%,
+      var(--dsw-alias-state-error-secondary)
+    ) 15%,
+    var(--dsw-alias-bg-layer-1)
+  );
 }
 
 .tagTool {
@@ -64,8 +90,16 @@
 /* run_code sub-dispatch cells: the business tint plus an indent so the
    nesting under the parent Tool cell reads at a glance. */
 .tagSubtool {
-  color: var(--dsw-alias-state-business-primary);
-  background: var(--dsw-alias-state-business-tertiary);
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-state-warn-label) 62%,
+    var(--dsw-alias-label-tertiary)
+  );
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-warn-tertiary) 58%,
+    var(--dsw-alias-bg-layer-1)
+  );
 }
 
 .root[data-kind='subtool'] {
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryCell.tsx b/packages/client/ui-trajectory/src/client/TrajectoryCell.tsx
index bc868f64c0..51ec2a0ec3 100644
--- a/packages/client/ui-trajectory/src/client/TrajectoryCell.tsx
+++ b/packages/client/ui-trajectory/src/client/TrajectoryCell.tsx
@@ -1,62 +1,40 @@
-// TrajectoryCell: one step row in the trajectory list — index, kind tag,
-// ellipsis text, optional Message token metrics, and own-duration time.
+// Legacy standalone trajectory cell retained for direct consumers and specs.
 
-import type { HTMLAttributes } from 'react'
+import {
+  formatElapsedSeconds,
+  type TrajectoryCellKind,
+  type TrajectoryCellProps,
+} from './trajectory-record.ts'
 import css from './TrajectoryCell.module.css'
 
-/** Closed set of trajectory step kinds (call+result fold into Tool; no Think;
- *  subtool = one run_code sub-dispatch nested under its Tool cell). */
-export type TrajectoryCellKind = 'user' | 'message' | 'tool' | 'subtool'
+export { formatElapsedSeconds }
+export type {
+  AssistantMetricDetail,
+  TrajectoryCellKind,
+  TrajectoryCellProps,
+} from './trajectory-record.ts'
 
 /** Display label per kind (matches the design tags). */
 const KIND_LABEL: Record<TrajectoryCellKind, string> = {
+  system: 'System',
   user: 'User',
+  context: 'Context',
+  compacted: 'Compacted',
   message: 'Message',
   tool: 'Tool',
   subtool: 'Sub',
 }
 
 const TAG_CLASS: Record<TrajectoryCellKind, string | undefined> = {
+  system: css.tagSystem,
   user: css.tagUser,
+  context: css.tagContext,
+  compacted: css.tagSystem,
   message: css.tagMessage,
   tool: css.tagTool,
   subtool: css.tagSubtool,
 }
 
-export interface TrajectoryCellProps extends HTMLAttributes<HTMLDivElement> {
-  /** 1-based step index shown as `#N`. */
-  index: number
-  kind: TrajectoryCellKind
-  /** Single-line summary; CSS ellipsis when it overflows. */
-  text: string
-  /**
-   * Own duration in seconds. `null` means no duration to show (em dash) —
-   * used for in-flight tools and tools missing callTime.
-   */
-  timeSeconds: number | null
-  /** Message-only: prompt token count. */
-  input?: number
-  /** Message-only: completion token count. */
-  output?: number
-  /** Message-only: reasoning token count (usage column, not a Think cell). */
-  think?: number
-  /** Selected: 2px inset brand-primary-new-color ring (not wired to chat selection yet). */
-  selected?: boolean
-}
-
-/**
- * Format own-duration for the trailing time column: `—` when unknown, `+Ns`
- * or `+N.1s` otherwise.
- * @param seconds - duration seconds, or null when absent.
- * @returns display string.
- */
-export function formatElapsedSeconds(seconds: number | null): string {
-  if (seconds === null || !Number.isFinite(seconds)) return '—'
-  const rounded = Math.round(seconds * 10) / 10
-  if (Number.isInteger(rounded)) return `+${rounded}s`
-  return `+${rounded.toFixed(1)}s`
-}
-
 /**
  * Render one trajectory step cell.
  * @param props - index, kind, text, time, and optional Message metrics.
@@ -66,7 +44,20 @@ export function TrajectoryCell({
   index,
   kind,
   text,
+  inputDetail: _inputDetail,
+  promptDetail: _promptDetail,
+  previousPromptDetail: _previousPromptDetail,
+  outputDetail: _outputDetail,
+  thinkingDetail: _thinkingDetail,
+  sourceBlocks: _sourceBlocks,
+  outputBlocks: _outputBlocks,
+  schemaDetail: _schemaDetail,
+  assistantMetrics: _assistantMetrics,
+  result: _result,
+  callId: _callId,
+  isError: _isError,
   timeSeconds,
+  startedAt: _startedAt,
   input,
   output,
   think,
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryGroupHeader.tsx b/packages/client/ui-trajectory/src/client/TrajectoryGroupHeader.tsx
index 90252ce373..1b7b03e78e 100644
--- a/packages/client/ui-trajectory/src/client/TrajectoryGroupHeader.tsx
+++ b/packages/client/ui-trajectory/src/client/TrajectoryGroupHeader.tsx
@@ -5,7 +5,7 @@ import css from './TrajectoryGroupHeader.module.css'
 export interface TrajectoryGroupHeaderProps {
   /** Group title (`Message`, `Step 1`, …). */
   title: string
-  /** Secondary summary (`49s`, `2.2s skill`, …). */
+  /** Secondary summary (`49 s`, `2.2 s skill`, …). */
   description?: string
 }
 
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryStatsHeader.module.css b/packages/client/ui-trajectory/src/client/TrajectoryStatsHeader.module.css
deleted file mode 100644
index 9def8790a9..0000000000
--- a/packages/client/ui-trajectory/src/client/TrajectoryStatsHeader.module.css
+++ /dev/null
@@ -1,7 +0,0 @@
-.root {
-  padding: 4px 16px;
-  font-size: 12px;
-  line-height: 18px;
-  color: var(--dsw-alias-label-secondary);
-  border-bottom: 1px solid var(--dsw-alias-border-l2);
-}
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryStatsHeader.tsx b/packages/client/ui-trajectory/src/client/TrajectoryStatsHeader.tsx
deleted file mode 100644
index 089b56acbc..0000000000
--- a/packages/client/ui-trajectory/src/client/TrajectoryStatsHeader.tsx
+++ /dev/null
@@ -1,21 +0,0 @@
-// TrajectoryStatsHeader: span totals row rendered at the top of both
-// placeholder view bodies (chrome dissolved into the views — the header is
-// part of what these views ARE, not registration metadata). Subscribes to
-// `nodes` only: chunk batches never swap that reference, so the row is quiet
-// during streaming.
-
-import { memo, useMemo } from 'react'
-import type { ConversationSnapshot } from '@deepseek-ai/dsh-client-runtime/client'
-import type { SnapshotSelectorHook } from '@deepseek-ai/dsh-client-ui-slots'
-import { deriveSpans, deriveSpanStats } from './spans.ts'
-import css from './TrajectoryStatsHeader.module.css'
-
-/** Props: the conversation-snapshot selector hook (handed down by the view body). */
-export interface TrajectoryStatsHeaderProps { useSession: SnapshotSelectorHook<ConversationSnapshot> }
-
-export const TrajectoryStatsHeader = memo(function TrajectoryStatsHeader({ useSession }: TrajectoryStatsHeaderProps) {
-  const nodes = useSession(s => s.nodes)
-  const stats = useMemo(() => deriveSpanStats(deriveSpans(nodes)), [nodes])
-  if (stats.turns === 0) return null
-  return <div className={css.root}>{`${stats.turns} turns · ${stats.steps} steps · ${stats.calls} tool calls`}</div>
-})
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css
new file mode 100644
index 0000000000..1b30318f68
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css
@@ -0,0 +1,1487 @@
+.split {
+  --dsh-scrollbar-thumb: var(--dsw-alias-scrollbar-bg-l2);
+  --dsh-scrollbar-thumb-hover: var(--dsw-alias-scrollbar-hover-l2);
+
+  position: relative;
+  display: flex;
+  flex: 1;
+  width: 100%;
+  min-height: 0;
+  overflow: hidden;
+  background: var(--dsw-alias-bg-layer-1);
+  container-type: inline-size;
+}
+
+.tablePane {
+  flex: 1;
+  min-width: 0;
+  overflow: auto;
+}
+
+.table {
+  --trajectory-turn-accent: color-mix(
+    in srgb,
+    var(--dsw-static-blue-500) 22%,
+    var(--dsw-alias-bg-layer-1)
+  );
+
+  width: 100%;
+  min-width: 480px;
+  border-spacing: 0;
+  table-layout: fixed;
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-bg-layer-1);
+  font: var(--dsw-font-xxs-12);
+}
+
+.eventColumn {
+  width: 122px;
+}
+
+.contentColumn {
+  width: auto;
+}
+
+.table th {
+  position: sticky;
+  top: 0;
+  z-index: 3;
+  box-sizing: border-box;
+  height: 30px;
+  padding: 0 8px;
+  overflow: hidden;
+  border-bottom: 1px solid var(--dsw-alias-border-l2);
+  color: var(--dsw-alias-label-tertiary);
+  background: var(--dsw-specific-sidebar-fill);
+  font: var(--dsw-font-xxs-12);
+  font-weight: 500;
+  text-align: left;
+  text-overflow: ellipsis;
+  user-select: none;
+  white-space: nowrap;
+}
+
+.eventHeader {
+  padding-right: 4px !important;
+  text-align: right !important;
+}
+
+.table td {
+  box-sizing: border-box;
+  height: 30px;
+  padding: 0 8px;
+  overflow: hidden;
+  border-bottom: 1px solid var(--dsw-alias-border-l1);
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.table tbody tr:not([data-collapsed-summary]) {
+  cursor: default;
+  outline: none;
+  transition:
+    background-color 120ms var(--ds-ease-in-out),
+    opacity 120ms var(--ds-ease-in-out);
+}
+
+.table tbody tr[data-timeline-focus='outside'] {
+  opacity: 0.24;
+}
+
+.table tbody tr:not([data-collapsed-summary]):not([data-selected='true']):hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.table tbody tr[data-request-only='true']:hover {
+  background: transparent;
+}
+
+.table tbody tr[data-request-only='true'] td {
+  height: 1px;
+  padding-top: 0;
+  padding-bottom: 0;
+  border-bottom: 0;
+}
+
+.table tbody tr[data-request-only='true'] .turnRail {
+  top: -15px;
+  bottom: 0;
+}
+
+.table tbody tr:not([data-collapsed-summary]):focus-visible {
+  box-shadow: inset 0 0 0 1px var(--dsw-alias-state-business-primary);
+}
+
+.table tbody tr[data-selected='true'] {
+  background: var(--dsw-alias-interactive-bg-active);
+}
+
+.requestBoundaryControl {
+  position: absolute;
+  z-index: 6;
+  top: -8px;
+  left: 12px;
+  width: 16px;
+  height: 16px;
+  padding: 0;
+  border: 0;
+  background: transparent;
+  cursor: pointer;
+}
+
+.requestBoundaryControl::before {
+  position: absolute;
+  top: 5.5px;
+  left: 5.5px;
+  width: 5px;
+  height: 5px;
+  border-radius: 50%;
+  background: var(--dsw-alias-label-caption);
+  box-shadow:
+    0 0 0 2px var(--dsw-alias-bg-layer-1),
+    0 0 0 3px transparent;
+  content: '';
+  transition:
+    background 120ms var(--ds-ease-in-out),
+    box-shadow 120ms var(--ds-ease-in-out);
+}
+
+.requestBoundaryControl::after {
+  position: absolute;
+  top: 2px;
+  left: 17px;
+  width: max-content;
+  padding: 0 4px;
+  border: 1px solid var(--dsw-alias-border-l1);
+  border-radius: 2px;
+  color: var(--dsw-alias-label-secondary);
+  background: var(--dsw-alias-bg-layer-1);
+  box-shadow: 0 2px 6px rgba(0, 0, 0, 0.12);
+  content: attr(data-label);
+  font: 9px/12px var(--ds-font-family-code);
+  opacity: 0;
+  pointer-events: none;
+  transform: translateX(-2px);
+  transition:
+    opacity 120ms var(--ds-ease-in-out),
+    transform 120ms var(--ds-ease-in-out);
+  user-select: none;
+  white-space: nowrap;
+}
+
+.requestBoundaryControl:hover::before,
+.requestBoundaryControl:focus-visible::before {
+  background: var(--dsw-alias-brand-primary-new-colorprimary-new-color);
+}
+
+.requestBoundaryControlActive::before,
+.requestBoundaryControlActive:hover::before,
+.requestBoundaryControlActive:focus-visible::before {
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-brand-primary-new-colorprimary-new-color) 18%,
+    var(--dsw-alias-bg-layer-1)
+  );
+  box-shadow: 0 0 0 1.5px var(--dsw-alias-brand-primary-new-colorprimary-new-color);
+}
+
+.requestBoundaryControl:hover::after,
+.requestBoundaryControl:focus-visible::after {
+  opacity: 1;
+  transform: translateX(0);
+}
+
+.requestBoundaryControl:focus-visible {
+  outline: none;
+}
+
+.table tbody tr:has(.requestBoundaryControl:hover):not([data-selected='true']) {
+  background: transparent;
+}
+
+.event {
+  position: relative;
+}
+
+.turnRail,
+.selectionRail {
+  position: absolute;
+  left: 0;
+  background: var(--dsw-alias-brand-primary-new-colorprimary-new-color);
+  pointer-events: none;
+}
+
+.turnRail {
+  z-index: 4;
+  top: -1px;
+  bottom: -1px;
+  width: 2px;
+  background: var(--trajectory-turn-accent);
+}
+
+.table tbody tr[data-turn-end='true'] .turnRail {
+  bottom: 0;
+}
+
+.selectionRail {
+  z-index: 5;
+  top: 0;
+  bottom: 0;
+  width: 3px;
+}
+
+.table tbody tr[data-turn-start='true'] td {
+  position: relative;
+  overflow: visible;
+}
+
+.table tbody tr[data-turn-start='true']:not(:first-child) td::before {
+  position: absolute;
+  z-index: 1;
+  top: 0;
+  right: 0;
+  left: 0;
+  height: 2px;
+  background: var(--dsw-alias-border-l1);
+  content: '';
+  pointer-events: none;
+  transform: translateY(-50%);
+}
+
+.event {
+  overflow: visible !important;
+  padding-right: 4px !important;
+  padding-left: 36px !important;
+}
+
+.turnLabel {
+  position: absolute;
+  z-index: 3;
+  top: 0;
+  left: 0;
+  display: inline-flex;
+  flex: none;
+  align-items: center;
+  box-sizing: border-box;
+  width: max-content;
+  padding: 1px 5px;
+  border-radius: 0 0 2px;
+  color: var(--dsw-alias-label-tertiary);
+  background: var(--dsw-alias-bg-module-platform);
+  font: 8px/10px var(--ds-font-family-code);
+  font-variant-numeric: tabular-nums;
+  user-select: none;
+  white-space: nowrap;
+}
+
+.turnLabelActive {
+  color: color-mix(
+    in srgb,
+    var(--dsw-static-blue-500) 55%,
+    var(--dsw-alias-label-tertiary)
+  );
+  background: var(--trajectory-turn-accent);
+}
+
+.eventInner {
+  display: flex;
+  align-items: center;
+  justify-content: flex-start;
+  min-width: 0;
+  height: 100%;
+}
+
+.kindSlot {
+  display: flex;
+  flex: none;
+  align-items: flex-end;
+  justify-content: flex-end;
+  width: 76px;
+}
+
+.content {
+  padding-left: 4px !important;
+  color: var(--dsw-alias-label-primary);
+}
+
+.kindTag {
+  display: inline-flex;
+  flex: none;
+  align-items: center;
+  box-sizing: border-box;
+  height: 19px;
+  padding: 0 5px;
+  border: 1px solid transparent;
+  border-radius: 4px;
+  font-size: 10px;
+  line-height: 16px;
+  font-weight: 650;
+  letter-spacing: 0.035em;
+  user-select: none;
+}
+
+.table .kindSlot .message {
+  justify-content: center;
+  width: 100%;
+}
+
+.user {
+  color: var(--dsw-alias-state-business-primary);
+  background: var(--dsw-alias-state-business-tertiary);
+}
+
+.systemNeutral {
+  color: var(--dsw-alias-label-secondary);
+  background: var(--dsw-alias-bg-module-platform);
+}
+
+.contextGreen {
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-state-success-primary) 68%,
+    var(--dsw-alias-label-secondary)
+  );
+  background: var(--dsw-alias-state-success-tertiary);
+}
+
+.compacted {
+  color: var(--dsw-alias-label-secondary);
+  background: var(--dsw-alias-bg-module-platform);
+}
+
+.compactedSummary {
+  margin-top: 12px;
+  padding: 0 0 14px;
+  font: var(--dsw-font-xs-13);
+}
+
+.promptDiffSections {
+  display: flex;
+  max-height: 100%;
+  flex-direction: column;
+  padding: 10px 14px 14px;
+  gap: 14px;
+  overflow: auto;
+}
+
+.promptDiffSection {
+  min-width: 0;
+}
+
+.promptDiffTitle {
+  margin: 0 0 6px;
+  color: var(--dsw-alias-label-secondary);
+  font: var(--dsw-font-xs-strong-13);
+  user-select: none;
+}
+
+.promptDiff {
+  margin: 0;
+  overflow: auto;
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-bg-layer-1);
+  font: 11px/17px var(--ds-font-family-code);
+  white-space: pre;
+}
+
+.promptDiff span {
+  display: block;
+  min-width: max-content;
+  padding: 0 6px;
+}
+
+.promptDiffLinemeta {
+  color: var(--dsw-alias-label-caption);
+  background: var(--dsw-alias-bg-module-platform);
+  user-select: none;
+}
+
+.promptDiffLinecontext {
+  color: var(--dsw-alias-label-secondary);
+}
+
+.promptDiffLineadded {
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-state-success-primary) 72%,
+    var(--dsw-alias-label-primary)
+  );
+  background: var(--dsw-alias-state-success-tertiary);
+}
+
+.promptDiffLineremoved {
+  color: var(--dsw-alias-state-error-primary);
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-error-primary) 12%,
+    var(--dsw-alias-bg-layer-1)
+  );
+}
+
+.assistantVioletBright {
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-brand-primary-new-colorprimary-new-color) 60%,
+    var(--dsw-alias-state-error-secondary)
+  );
+  background: color-mix(
+    in srgb,
+    color-mix(
+      in srgb,
+      var(--dsw-alias-brand-primary-new-colorprimary-new-color) 55%,
+      var(--dsw-alias-state-error-secondary)
+    ) 15%,
+    var(--dsw-alias-bg-layer-1)
+  );
+}
+
+.toolAmber {
+  color: var(--dsw-alias-state-warn-label);
+  background: var(--dsw-alias-state-warn-tertiary);
+}
+
+.subtoolAmber {
+  color: color-mix(
+    in srgb,
+    var(--dsw-alias-state-warn-label) 62%,
+    var(--dsw-alias-label-tertiary)
+  );
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-warn-tertiary) 58%,
+    var(--dsw-alias-bg-layer-1)
+  );
+}
+
+.contentText {
+  display: block;
+  min-width: 0;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.table tbody tr[data-collapsed-summary='turn'] td,
+.table tbody tr[data-collapsed-summary='assistant'] td {
+  height: 20px;
+}
+
+.table tbody tr[data-collapsed-summary] {
+  cursor: pointer;
+  outline: none;
+  transition: background-color 120ms var(--ds-ease-in-out);
+}
+
+.table tbody tr[data-collapsed-summary]:hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.table tbody tr[data-collapsed-summary]:focus-visible {
+  box-shadow: inset 0 0 0 1px var(--dsw-alias-state-business-primary);
+}
+
+.collapsedTurnContent {
+  display: flex;
+  align-items: center;
+  min-width: 0;
+  color: var(--dsw-alias-label-secondary);
+  font-size: 12px;
+  line-height: 16px;
+}
+
+.collapsedTurnEllipsis {
+  flex: none;
+  margin-right: 6px;
+  color: var(--dsw-alias-label-tertiary);
+  font-weight: 600;
+  user-select: none;
+}
+
+.collapsedTurnText {
+  min-width: 0;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.resultPreview {
+  display: grid;
+  grid-template-columns:
+    clamp(
+      180px,
+      var(--trajectory-tool-request-width, calc(36cqw - 56px)),
+      480px
+    )
+    minmax(0, 1fr);
+  align-items: center;
+  min-width: 0;
+  gap: 8px;
+}
+
+.resultRequest,
+.inlineResultText {
+  min-width: 0;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.toolCallNameTypeface {
+  color: var(--dsw-alias-label-primary);
+  font: 400 12px/18px Menlo, Consolas, 'Liberation Mono', 'PingFang SC',
+    'Microsoft YaHei';
+}
+
+.toolCallPayload {
+  margin-left: 7px;
+  color: var(--dsw-alias-label-secondary);
+  font: 400 12px/18px var(--ds-font-family-code);
+}
+
+.table tbody tr[data-kind='tool'] .contentText,
+.table tbody tr[data-kind='subtool'] .contentText,
+.table tbody tr[data-kind='tool'] .resultPreview,
+.table tbody tr[data-kind='subtool'] .resultPreview {
+  font-family: var(--ds-font-family-code);
+  font-size: 12px;
+}
+
+.table tbody tr[data-kind='subtool'] .content {
+  padding-left: 26px;
+}
+
+.inlineResult {
+  display: flex;
+  align-items: center;
+  min-width: 0;
+  color: var(--dsw-alias-label-secondary);
+}
+
+.noOutputText {
+  color: var(--dsw-alias-label-caption);
+}
+
+.arrow {
+  flex: none;
+  margin-right: 8px;
+  color: var(--dsw-alias-label-caption);
+}
+
+.error {
+  color: var(--dsw-alias-state-error-primary);
+}
+
+.details {
+  position: relative;
+  display: flex;
+  flex: none;
+  flex-direction: column;
+  width: clamp(320px, 38%, 440px);
+  max-width: calc(100% - 280px);
+  min-width: 0;
+  min-height: 0;
+  border-left: 1px solid var(--dsw-alias-border-l2);
+  background: var(--dsw-alias-bg-layer-1);
+}
+
+.detailsResizeHandle {
+  position: absolute;
+  z-index: 6;
+  top: 0;
+  bottom: 0;
+  left: -4px;
+  width: 8px;
+  padding: 0;
+  border: 0;
+  background: transparent;
+  cursor: col-resize;
+  touch-action: none;
+  user-select: none;
+}
+
+.detailsResizeHandle:focus-visible {
+  outline: none;
+}
+
+.detailsHeader {
+  display: flex;
+  flex: none;
+  align-items: center;
+  justify-content: space-between;
+  box-sizing: border-box;
+  height: 42px;
+  padding: 0 8px 0 12px;
+  border-bottom: 1px solid var(--dsw-alias-border-l2);
+}
+
+.detailsTitle {
+  display: flex;
+  align-items: center;
+  min-width: 0;
+  gap: 8px;
+  color: var(--dsw-alias-label-primary);
+}
+
+.requestDetailsDot {
+  flex: none;
+  width: 5px;
+  height: 5px;
+  border-radius: 50%;
+  background: var(--dsw-alias-label-secondary);
+}
+
+.requestDetailsName {
+  flex: none;
+  font: 500 12px/16px var(--ds-font-family-code);
+}
+
+.detailsLocation {
+  min-width: 0;
+  overflow: hidden;
+  color: var(--dsw-alias-label-tertiary);
+  font: 11px/16px var(--ds-font-family-code);
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.close {
+  display: inline-flex;
+  flex: none;
+  align-items: center;
+  justify-content: center;
+  width: 28px;
+  height: 28px;
+  padding: 0;
+  border: 0;
+  border-radius: 6px;
+  color: var(--dsw-alias-label-secondary);
+  background: transparent;
+  cursor: pointer;
+  font-size: 18px;
+  line-height: 18px;
+}
+
+.close:hover {
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.detailTabs {
+  display: flex;
+  flex: none;
+  box-sizing: border-box;
+  width: 100%;
+  min-width: 0;
+  max-width: 100%;
+  height: 34px;
+  padding: 0 8px;
+  overflow-x: auto;
+  overflow-y: hidden;
+  gap: 1px;
+  border-bottom: 1px solid var(--dsw-alias-border-l2);
+  overscroll-behavior-x: contain;
+  scrollbar-width: none;
+  white-space: nowrap;
+}
+
+.detailTabs::-webkit-scrollbar {
+  display: none;
+}
+
+.detailTab {
+  position: relative;
+  flex: none;
+  padding: 0 9px;
+  border: 0;
+  color: var(--dsw-alias-label-tertiary);
+  background: transparent;
+  cursor: pointer;
+  font: var(--dsw-font-xs-13);
+}
+
+.detailTab:hover {
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.detailTabActive {
+  color: var(--dsw-alias-state-business-primary);
+}
+
+.detailTabActive::after {
+  position: absolute;
+  right: 9px;
+  bottom: 0;
+  left: 9px;
+  height: 2px;
+  border-radius: 1px 1px 0 0;
+  background: var(--dsw-alias-state-business-primary);
+  content: '';
+}
+
+.close:focus-visible,
+.detailTab:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: -1px;
+}
+
+.detailBody {
+  flex: 1;
+  min-height: 0;
+  overflow: auto;
+}
+
+.detailBodySummary {
+  display: flex;
+  box-sizing: border-box;
+  flex-direction: column;
+  padding-bottom: 12px;
+  overflow: hidden;
+}
+
+.detailBodySummary > .overview {
+  flex: 0 1 auto;
+  min-height: 0;
+  overflow: auto;
+  overscroll-behavior: contain;
+}
+
+.detailBodySummary > .compactedSummary {
+  flex: 1;
+  min-height: 0;
+  overflow: auto;
+}
+
+.compactedSummary .markdownPayload {
+  padding-right: 18px;
+}
+
+.overview {
+  margin: 0;
+  padding: 8px 0;
+  font: var(--dsw-font-xs-13);
+}
+
+.overviewParentLinks {
+  display: flex;
+  gap: 14px;
+}
+
+.overviewHierarchyNavLink {
+  all: unset;
+  display: inline-flex;
+  align-items: center;
+  color: var(--dsw-alias-label-secondary);
+  cursor: pointer;
+  font: inherit;
+  gap: 1px;
+  opacity: 1;
+}
+
+.overviewHierarchyNavLink:hover {
+  color: var(--dsw-alias-label-primary);
+}
+
+.overviewHierarchyJumpIconTight {
+  flex: none;
+  color: var(--dsw-alias-label-caption);
+}
+
+.overviewHierarchyNavLink:hover .overviewHierarchyJumpIconTight,
+.overviewHierarchyNavLink:focus-visible .overviewHierarchyJumpIconTight {
+  color: var(--dsw-alias-label-primary);
+}
+
+.overviewHierarchyNavLink:focus-visible {
+  color: var(--dsw-alias-label-primary);
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 2px;
+}
+
+.overview > div {
+  display: grid;
+  grid-template-columns: 94px minmax(0, 1fr);
+  min-height: 22px;
+  padding: 0 14px;
+  align-items: center;
+}
+
+.overview > .requestTokenDetail dt {
+  padding-left: 12px;
+}
+
+.usagePanel {
+  padding: 4px 0 10px;
+}
+
+.usageGroup + .usageGroup {
+  margin-top: 8px;
+}
+
+.usageHeading {
+  margin: 0;
+  padding: 4px 14px 1px;
+  color: var(--dsw-alias-label-secondary);
+  font: var(--dsw-font-xs-strong-13);
+  user-select: none;
+}
+
+.usageGroup .overview {
+  padding: 0;
+}
+
+.overview dt {
+  color: var(--dsw-alias-label-tertiary);
+}
+
+.overview dd {
+  min-width: 0;
+  margin: 0;
+  overflow: hidden;
+  color: var(--dsw-alias-label-primary);
+  font-variant-numeric: tabular-nums;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.timestampToggle {
+  all: unset;
+  color: inherit;
+  cursor: pointer;
+  font: inherit;
+  font-variant-numeric: tabular-nums;
+  user-select: text;
+}
+
+.timestampToggle:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 2px;
+}
+
+.overviewSections {
+  display: flex;
+  flex: 1;
+  min-height: min-content;
+  flex-direction: column;
+  overflow: hidden;
+  border-top: 0;
+}
+
+.overviewSection {
+  display: flex;
+  flex: 1 1 0;
+  max-height: max-content;
+  min-height: 28px;
+  flex-direction: column;
+  overflow: hidden;
+}
+
+.overviewSection + .overviewSection {
+  padding-top: 8px;
+}
+
+.overviewHeading {
+  display: flex;
+  flex: none;
+  align-items: flex-end;
+  box-sizing: border-box;
+  width: 100%;
+  height: 28px;
+  margin: 0;
+  padding: 0 0 3px 14px;
+  color: var(--dsw-alias-label-secondary);
+  background: var(--dsw-alias-bg-base);
+  font: var(--dsw-font-xs-strong-13);
+  user-select: none;
+}
+
+.overviewTitle {
+  display: inline-flex;
+  flex: none;
+  width: max-content;
+  height: auto;
+  align-items: center;
+  padding: 0;
+  border: 0;
+  color: inherit;
+  background: transparent;
+  cursor: pointer;
+  font: inherit;
+  gap: 3px;
+  line-height: 1;
+  user-select: none;
+}
+
+.overviewTitleIcon {
+  flex: none;
+  color: var(--dsw-alias-label-caption);
+}
+
+.overviewTitle:hover .overviewTitleIcon {
+  color: var(--dsw-alias-label-primary);
+}
+
+.overviewTitle:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 2px;
+}
+
+.overviewTitle:focus-visible .overviewTitleIcon {
+  color: var(--dsw-alias-state-business-primary);
+}
+
+.overviewPreview {
+  flex: 1;
+  min-height: 0;
+  overflow: auto;
+  overscroll-behavior: contain;
+  background: var(--dsw-alias-bg-layer-1);
+}
+
+.overviewPreview > :last-child {
+  margin-bottom: 0;
+  padding-bottom: 0;
+}
+
+.overviewPreview .jsonPreview > :first-child,
+.overviewPreview .schemaTree > :first-child {
+  padding-bottom: 0;
+}
+
+.overviewPreview .schemaTree {
+  margin-bottom: 0;
+}
+
+.overviewPreview .overview {
+  padding: 2px 0 0;
+}
+
+.overviewPreview .overview > div {
+  min-height: 22px;
+}
+
+.markdownPreview,
+.markdownPayload {
+  color: var(--dsw-alias-label-primary);
+}
+
+.markdownPreview > div,
+.markdownPayload > div {
+  gap: 8px;
+  font: var(--dsw-font-xs-13);
+}
+
+.markdownPreview > div h1,
+.markdownPayload > div h1 {
+  font: 600 16px/22px var(--dsw-font-family);
+}
+
+.markdownPreview > div h2,
+.markdownPayload > div h2 {
+  font: 600 15px/22px var(--dsw-font-family);
+}
+
+.markdownPreview > div :where(h3, h4, h5, h6),
+.markdownPayload > div :where(h3, h4, h5, h6) {
+  font: 600 14px/20px var(--dsw-font-family);
+}
+
+.markdownPreview > div :not(pre) > code,
+.markdownPayload > div :not(pre) > code {
+  font: 12px/18px var(--ds-font-family-code);
+}
+
+.markdownPreview > div table,
+.markdownPayload > div table {
+  font: var(--dsw-font-xs-13);
+}
+
+.markdownPreview > div > :first-child {
+  margin-top: 0;
+}
+
+.markdownPreview > div > :last-child {
+  margin-bottom: 0;
+}
+
+.markdownPreview > div :where(h1, h2, h3) {
+  margin: 12px 0 6px;
+}
+
+.markdownPreview > div :where(h4, h5, h6) {
+  margin: 10px 0 5px;
+}
+
+.markdownPreview > div :where(p, ul, ol) {
+  margin: 8px 0;
+}
+
+.markdownPreview > div li:not(:first-child) {
+  margin-top: 3px;
+}
+
+.markdownPreview > div blockquote {
+  margin-top: 8px;
+}
+
+.markdownPreview > div hr {
+  margin: 14px 0;
+}
+
+.markdownPreview > div > :global(.md-code-block) {
+  margin: 10px 0;
+}
+
+.assistantContent .markdownPayload,
+.assistantContent .payload {
+  min-height: 0;
+}
+
+.thinkingQuote {
+  margin: 6px 14px 0 12px;
+  padding-left: 6px;
+  border-left: 2px solid var(--dsw-alias-markdown-citation);
+  color: var(--dsw-alias-label-secondary);
+  background: transparent;
+}
+
+.thinkingQuoteOnlyPreview {
+  margin-bottom: 8px;
+}
+
+.detailBody > .assistantContentRendered .thinkingQuote {
+  margin-top: 14px;
+}
+
+.thinkingToggle {
+  display: flex;
+  align-items: center;
+  width: max-content;
+  height: 18px;
+  padding: 0;
+  border: 0;
+  color: var(--dsw-alias-label-tertiary);
+  background: transparent;
+  cursor: pointer;
+  font: 600 12px/18px var(--dsw-font-family);
+  user-select: none;
+}
+
+.thinkingToggle:hover {
+  color: var(--dsw-alias-label-secondary);
+}
+
+.thinkingToggle:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 2px;
+}
+
+.thinkingQuote .markdownPreview,
+.thinkingQuote .markdownPayload,
+.thinkingQuote .payload {
+  min-height: 0;
+  padding: 2px 0;
+  color: var(--dsw-alias-label-secondary);
+  background: transparent;
+}
+
+.assistantOutput {
+  background: var(--dsw-alias-bg-layer-1);
+}
+
+.markdownPreview {
+  padding: 6px 14px 8px;
+}
+
+.markdownPayload {
+  min-height: 100%;
+  padding: 14px;
+}
+
+.jsonPayload {
+  min-height: 100%;
+}
+
+.jsonPreview {
+  min-height: 0;
+}
+
+.schema {
+  min-height: 100%;
+  background: var(--dsw-alias-bg-layer-1);
+}
+
+.schemaPreview {
+  min-height: 0;
+}
+
+.schemaIntro {
+  padding: 12px 14px 6px;
+}
+
+.schemaPreview .schemaIntro {
+  padding-top: 6px;
+}
+
+.schemaName {
+  margin: 0;
+  color: var(--dsw-alias-label-primary);
+  font: 600 12px/18px var(--ds-font-family-code);
+}
+
+.schemaDescription {
+  margin: 2px 0 0;
+  color: var(--dsw-alias-label-secondary);
+  font: var(--dsw-font-xs-13);
+  white-space: pre-wrap;
+}
+
+.schemaParameters {
+  min-width: 0;
+}
+
+.schemaParametersTitle {
+  margin: 0;
+  padding: 4px 14px 2px;
+  color: var(--dsw-alias-label-tertiary);
+  font: 600 11px/16px var(--dsw-font-family);
+  user-select: none;
+}
+
+.schemaTree {
+  margin: 0 6px 8px 0;
+}
+
+.payload {
+  box-sizing: border-box;
+  min-height: 100%;
+  margin: 0;
+  padding: 14px;
+  overflow-wrap: anywhere;
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-markdown-code-block);
+  font: 12px/19px var(--ds-font-family-code);
+  tab-size: 2;
+  white-space: pre-wrap;
+}
+
+.payloadPreview {
+  min-height: 0;
+  padding: 6px 14px 8px;
+}
+
+.sourceBlocks {
+  box-sizing: border-box;
+  min-height: 100%;
+  padding: 12px 14px;
+}
+
+.sourceBlock + .sourceBlock {
+  margin-top: 14px;
+}
+
+.sourceBlockHeader,
+.sourceBlockJumpTarget {
+  display: flex;
+  align-items: center;
+  width: max-content;
+  margin-bottom: 3px;
+  gap: 3px;
+  user-select: none;
+}
+
+.sourceBlockJumpTarget {
+  all: unset;
+  display: flex;
+  align-items: center;
+  width: max-content;
+  margin-bottom: 3px;
+  gap: 3px;
+  cursor: pointer;
+  user-select: none;
+}
+
+.sourceBlockLabel {
+  color: var(--dsw-alias-label-tertiary);
+  font: 11px/16px var(--ds-font-family-code);
+}
+
+.sourceBlockJumpIcon {
+  flex: none;
+  color: var(--dsw-alias-label-caption);
+}
+
+.sourceBlockJumpTarget:hover .sourceBlockJumpIcon {
+  color: var(--dsw-alias-label-primary);
+}
+
+.sourceBlockJumpTarget:focus-visible {
+  outline: none;
+}
+
+.sourceBlockJumpTarget:focus-visible .sourceBlockJumpIcon {
+  color: var(--dsw-alias-state-business-primary);
+}
+
+.sourceBlockContent {
+  margin: 0;
+  overflow-wrap: anywhere;
+  color: var(--dsw-alias-label-primary);
+  background: transparent;
+  font: 12px/19px var(--ds-font-family-code);
+  tab-size: 2;
+  white-space: pre-wrap;
+}
+
+.panelImageLink {
+  display: block;
+  width: auto;
+  max-width: 100%;
+  overflow: hidden;
+  border-radius: 4px;
+  cursor: zoom-in;
+}
+
+.panelImageLinkPreview {
+  max-height: 140px;
+}
+
+.panelImage {
+  display: block;
+  width: auto;
+  max-width: 100%;
+  height: auto;
+  max-height: 320px;
+  margin: 0;
+  border-radius: inherit;
+  background: var(--dsw-alias-bg-base);
+  object-fit: contain;
+}
+
+.panelImageLinkPreview .panelImage {
+  max-height: 140px;
+}
+
+.messageImages {
+  display: flex;
+  flex-direction: column;
+  margin: 8px 14px 14px;
+  gap: 8px;
+}
+
+.messageImagesPreview {
+  margin: 6px 14px 8px;
+  gap: 6px;
+}
+
+.assistantToolCalls {
+  box-sizing: border-box;
+  max-width: 100%;
+  margin: 2px 14px 12px;
+  padding: 0;
+  color: var(--dsw-alias-label-secondary);
+  font: 11px/17px var(--ds-font-family-code);
+  list-style: none;
+}
+
+.assistantToolCallsPreview {
+  margin-top: 2px;
+  margin-bottom: 8px;
+}
+
+.assistantToolCalls li {
+  min-width: 0;
+}
+
+.assistantToolCallButton {
+  all: unset;
+  display: flex;
+  box-sizing: border-box;
+  width: calc(100% + 4px);
+  min-width: 0;
+  height: 19px;
+  align-items: center;
+  margin-left: -4px;
+  padding: 0 4px;
+  border-radius: 3px;
+  cursor: pointer;
+}
+
+.assistantToolCallButton:hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.assistantToolCallButton:focus-visible {
+  background: var(--dsw-alias-interactive-bg-hover);
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: -1px;
+}
+
+.assistantToolCallIcon {
+  flex: none;
+  margin-right: 5px;
+  color: var(--dsw-alias-label-caption);
+}
+
+.assistantToolCallButton:hover .assistantToolCallIcon,
+.assistantToolCallButton:focus-visible .assistantToolCallIcon {
+  color: var(--dsw-alias-label-secondary);
+}
+
+.assistantToolCallText {
+  display: flex;
+  flex: 1;
+  min-width: 0;
+  overflow: hidden;
+  white-space: nowrap;
+}
+
+.assistantToolCallName {
+  flex: none;
+  margin-right: 5px;
+  color: var(--dsw-alias-label-secondary);
+  font-weight: 500;
+}
+
+.assistantToolCallArgs {
+  min-width: 0;
+  overflow: hidden;
+  color: var(--dsw-alias-label-tertiary);
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.systemPrompt {
+  box-sizing: border-box;
+}
+
+.toolCatalog {
+  min-height: 100%;
+  padding: 8px 0 16px;
+}
+
+.toolCatalogItem {
+  border-bottom: 1px solid var(--dsw-alias-border-l1);
+}
+
+.toolCatalogSummary {
+  display: grid;
+  grid-template-columns: 12px 12px max-content minmax(0, 1fr);
+  align-items: center;
+  box-sizing: border-box;
+  min-height: 30px;
+  padding: 4px 12px;
+  gap: 5px;
+  cursor: pointer;
+  list-style: none;
+  user-select: none;
+}
+
+.toolCatalogSummary::-webkit-details-marker {
+  display: none;
+}
+
+.toolCatalogSummary:hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.toolCatalogSummary:focus-visible {
+  background: var(--dsw-alias-interactive-bg-hover);
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: -1px;
+}
+
+.toolCatalogChevron,
+.toolCatalogIcon {
+  color: var(--dsw-alias-label-caption);
+}
+
+.toolCatalogChevron {
+  transition: transform 100ms var(--ds-ease-in-out);
+}
+
+.toolCatalogItem[open] .toolCatalogChevron {
+  transform: rotate(90deg);
+}
+
+.toolCatalogName {
+  color: var(--dsw-alias-label-primary);
+  font: 500 12px/18px var(--ds-font-family-code);
+}
+
+.toolCatalogDescription {
+  min-width: 0;
+  overflow: hidden;
+  color: var(--dsw-alias-label-tertiary);
+  font: var(--dsw-font-xs-13);
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.toolCatalogDefinition {
+  padding: 0 0 8px 29px;
+  background: var(--dsw-alias-bg-base);
+}
+
+.toolCatalogFullDescription {
+  margin: 0;
+  padding: 7px 14px 4px 0;
+  color: var(--dsw-alias-label-secondary);
+  font: var(--dsw-font-xs-13);
+  white-space: pre-wrap;
+}
+
+.toolCatalogTree {
+  margin-left: -14px;
+  margin-right: 6px;
+}
+
+.resultBlocks {
+  display: flex;
+  min-height: 100%;
+  flex-direction: column;
+  box-sizing: border-box;
+  padding: 14px;
+  gap: 10px;
+}
+
+.resultBlocksPreview {
+  min-height: 0;
+  padding: 6px 14px 8px;
+  gap: 6px;
+}
+
+.resultBlockText {
+  margin: 0;
+  overflow-wrap: anywhere;
+  color: var(--dsw-alias-label-primary);
+  background: transparent;
+  font: 12px/19px var(--ds-font-family-code);
+  tab-size: 2;
+  white-space: pre-wrap;
+}
+
+.overviewPreview .noPayload {
+  padding: 6px 14px 8px;
+}
+
+.noPayload {
+  margin: 0;
+  padding: 18px 14px;
+  color: var(--dsw-alias-label-tertiary);
+  font: var(--dsw-font-xs-13);
+}
+
+@media (max-width: 760px) {
+  .details {
+    position: absolute;
+    z-index: 5;
+    top: 0;
+    right: 0;
+    bottom: 0;
+    width: min(92%, 420px);
+    max-width: 92%;
+    border-left-color: var(--dsw-alias-border-l3);
+    box-shadow: -12px 0 32px rgba(0, 0, 0, 0.14);
+  }
+}
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx b/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx
new file mode 100644
index 0000000000..09ad60b427
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx
@@ -0,0 +1,2329 @@
+/** Turn-aware trajectory event ledger with a local record inspector. */
+
+import { useEffect, useRef, useState } from 'react'
+import type { CSSProperties, ReactNode } from 'react'
+import {
+  extractMarkdownPlainText, IconChevronRightOutline14, JsonTree, MarkdownText,
+} from '@deepseek-ai/dsh-client-ui-primitives'
+import { structuredPatch } from 'diff'
+import type {
+  AssistantRequestConfig, ConversationPromptSnapshot,
+} from '@deepseek-ai/dsh-client-runtime/client'
+import type {
+  AssistantMetricDetail, TrajectoryCellKind, TrajectoryCellProps, TrajectorySourceBlock,
+} from './trajectory-record.ts'
+import { formatElapsedSeconds } from './trajectory-record.ts'
+import type { TrajectoryTurnModel } from './layout.ts'
+import css from './TrajectoryTable.module.css'
+
+const KIND_LABEL: Record<TrajectoryCellKind, string> = {
+  system: 'SYSTEM',
+  user: 'USER',
+  context: 'CONTEXT',
+  compacted: 'COMPACTED',
+  message: 'ASSISTANT',
+  tool: 'TOOL',
+  subtool: 'SUBTOOL',
+}
+
+interface TableRecord {
+  turn: number
+  group: string
+  groupStart: boolean
+  turnStart: boolean
+  cell: TrajectoryCellProps
+  turnEnd: boolean
+  collapsedSummary?: string
+  collapsedSummaryKind?: 'turn' | 'assistant'
+}
+
+type DetailTab =
+  | 'system-prompt'
+  | 'tools'
+  | 'overview'
+  | 'rendered'
+  | 'source'
+  | 'origin'
+  | 'input'
+  | 'output'
+  | 'schema'
+  | 'options'
+  | 'usage'
+  | 'timing'
+  | 'diff'
+type RecordState = 'complete' | 'running' | 'error'
+
+interface DetailTabItem {
+  id: DetailTab
+  label: string
+}
+
+interface ParentRecords {
+  message?: TableRecord
+  tool?: TableRecord
+}
+
+interface ToolCallTextParts {
+  name: string
+  args?: string
+}
+
+interface SelectedRequest {
+  turn: number
+  number: number
+  group: string
+}
+
+interface DetailsResizeDrag {
+  pointerId: number
+  startX: number
+  startWidth: number
+  splitWidth: number
+  startToolRequestOffset: number
+}
+
+const DETAILS_MIN_WIDTH = 320
+const DETAILS_MAX_WIDTH = 720
+const TABLE_MIN_WIDTH = 280
+const DETAILS_RESIZE_STEP = 16
+const TOOL_REQUEST_SHARE = 0.58
+const TOOL_REQUEST_MIN_WIDTH = 180
+const TOOL_REQUEST_MAX_WIDTH = 480
+const DEFAULT_TOOL_REQUEST_SHARE = 0.36
+const DEFAULT_TOOL_REQUEST_OFFSET = 56
+const SYSTEM_PROMPT_TABS: readonly DetailTabItem[] = [
+  { id: 'system-prompt', label: 'System Prompt' },
+  { id: 'tools', label: 'Tools' },
+]
+const SYSTEM_UPDATE_TABS: readonly DetailTabItem[] = [
+  { id: 'diff', label: 'Diff' },
+  ...SYSTEM_PROMPT_TABS,
+]
+const REQUEST_TABS: readonly DetailTabItem[] = [
+  { id: 'overview', label: 'Summary' },
+  { id: 'options', label: 'Options' },
+  { id: 'usage', label: 'Usage' },
+  { id: 'timing', label: 'Timing' },
+]
+
+type TrajectorySplitStyle = CSSProperties & {
+  '--trajectory-tool-request-width': string
+}
+
+function clampDetailsWidth(width: number, splitWidth: number): number {
+  const maxWidth = Math.max(
+    DETAILS_MIN_WIDTH,
+    Math.min(DETAILS_MAX_WIDTH, splitWidth - TABLE_MIN_WIDTH),
+  )
+  return Math.round(Math.min(Math.max(width, DETAILS_MIN_WIDTH), maxWidth))
+}
+
+function defaultToolRequestWidth(splitWidth: number): number {
+  return Math.min(
+    Math.max(
+      splitWidth * DEFAULT_TOOL_REQUEST_SHARE - DEFAULT_TOOL_REQUEST_OFFSET,
+      TOOL_REQUEST_MIN_WIDTH,
+    ),
+    TOOL_REQUEST_MAX_WIDTH,
+  )
+}
+
+function formatDurationMs(milliseconds: number): string {
+  if (milliseconds < 1_000) return `${Math.round(milliseconds)} ms`
+  return `${(milliseconds / 1_000).toFixed(milliseconds < 10_000 ? 2 : 1)} s`
+}
+
+function formatStartedAt(timestamp: number | null): string {
+  if (timestamp === null || !Number.isFinite(timestamp)) return 'Not available'
+  const date = new Date(timestamp)
+  const two = (value: number) => String(value).padStart(2, '0')
+  const three = (value: number) => String(value).padStart(3, '0')
+  const time = `${two(date.getHours())}:${two(date.getMinutes())}:${two(date.getSeconds())}.${three(date.getMilliseconds())}`
+  const day = `${date.getFullYear()}-${two(date.getMonth() + 1)}-${two(date.getDate())}`
+  return `${day} ${time}`
+}
+
+function StartedAtValue({ timestamp }: { timestamp: number | null }) {
+  const [showUnix, setShowUnix] = useState(false)
+  if (timestamp === null || !Number.isFinite(timestamp)) return <dd>Not available</dd>
+  return (
+    <dd>
+      <button
+        type="button"
+        className={css.timestampToggle}
+        title={showUnix ? 'Show local time' : 'Show Unix timestamp'}
+        onClick={(event) => {
+          const selection = window.getSelection()
+          if (
+            selection !== null
+            && !selection.isCollapsed
+            && selection.rangeCount > 0
+            && selection.getRangeAt(0).intersectsNode(event.currentTarget)
+          ) return
+          setShowUnix(current => !current)
+        }}
+      >
+        {showUnix ? (timestamp / 1_000).toFixed(3) : formatStartedAt(timestamp)}
+      </button>
+    </dd>
+  )
+}
+
+function totalTime(metrics: AssistantMetricDetail): string {
+  if (!metrics.timingRecorded) return 'Not recorded'
+  if (metrics.stepStartTime === null) return 'Step start unavailable'
+  if (metrics.completedTime === null) return 'Pending'
+  return formatDurationMs(Math.max(0, metrics.completedTime - metrics.stepStartTime))
+}
+
+function ttft(metrics: AssistantMetricDetail): string {
+  if (!metrics.timingRecorded) return 'Not recorded'
+  if (metrics.stepStartTime === null) return 'Step start unavailable'
+  if (metrics.firstTokenTime === null) return 'First token unavailable'
+  return formatDurationMs(Math.max(0, metrics.firstTokenTime - metrics.stepStartTime))
+}
+
+function generationTime(metrics: AssistantMetricDetail): string {
+  if (!metrics.timingRecorded || metrics.firstTokenTime === null) return 'First token unavailable'
+  if (metrics.completedTime === null) return 'Pending'
+  return formatDurationMs(Math.max(0, metrics.completedTime - metrics.firstTokenTime))
+}
+
+function throughput(metrics: AssistantMetricDetail): string {
+  if (!metrics.usageProvided) return 'Usage unavailable'
+  if (metrics.outputTokens === null) return 'Output tokens unavailable'
+  if (!metrics.timingRecorded || metrics.firstTokenTime === null) return 'First token unavailable'
+  if (metrics.completedTime === null) return 'Pending'
+  const generationSeconds = (metrics.completedTime - metrics.firstTokenTime) / 1_000
+  if (generationSeconds <= 0) return 'Duration too short'
+  return `${(metrics.outputTokens / generationSeconds).toFixed(1)} tok/s`
+}
+
+function AssistantTimingPanel({ metrics }: { metrics: AssistantMetricDetail }) {
+  return (
+    <dl className={css.overview}>
+      <div><dt>Started</dt><StartedAtValue timestamp={metrics.stepStartTime} /></div>
+      <div><dt>Total duration</dt><dd>{totalTime(metrics)}</dd></div>
+      <div><dt>TTFT</dt><dd>{ttft(metrics)}</dd></div>
+      <div><dt>Generation</dt><dd>{generationTime(metrics)}</dd></div>
+      <div><dt>Throughput</dt><dd>{throughput(metrics)}</dd></div>
+    </dl>
+  )
+}
+
+/** Props for the trajectory ledger. */
+export interface TrajectoryTableProps {
+  /** Session-global request numbers for the request groups visible in this context. */
+  requestNumbers?: readonly TrajectoryRequestNumber[]
+  /** Grouped records in display order. */
+  turns: readonly TrajectoryTurnModel[]
+  /** Record indexes emphasized by the active timeline focus. */
+  timelineFocusIndexes?: ReadonlySet<number> | null
+  /** Record indexes retained by the active live search, or null without a query. */
+  searchMatchIndexes?: ReadonlySet<number> | null
+  /** Report the record currently selected in the local inspector. */
+  onSelectedIndexChange?: (index: number | null) => void
+  /** Report a direct user selection from a ledger row. */
+  onRecordSelect?: (index: number) => void
+  /** Clear selection state owned by the ledger host. */
+  onClearSelection?: () => void
+  /** Turn ids whose rows after the first are folded into a summary. */
+  collapsedTurns: ReadonlySet<number>
+  /** Toggle one turn between folded and expanded. */
+  onToggleTurn: (turn: number) => void
+  /** Assistant record indexes whose tool calls are folded. */
+  collapsedAssistants: ReadonlySet<number>
+  /** Toggle tool calls under one assistant record. */
+  onToggleAssistant: (index: number) => void
+}
+
+/** One request identity paired with its session-global number. */
+export interface TrajectoryRequestNumber {
+  /** Request anchor event sequence; absent for the currently streaming ordinary request. */
+  seq?: number
+  turn: number
+  step: number
+  group: string
+  number: number
+  purpose?: 'compaction'
+  status?: 'complete' | 'running' | 'error'
+  startedAt?: number
+  completedAt?: number | null
+  error?: string
+  retry?: number
+  maxRetries?: number
+  retryDelayMs?: number
+  resultSeq?: number
+  provider?: string
+  model?: string
+  requestConfig?: AssistantRequestConfig
+  usage?: TrajectoryUsage
+  cumulativeUsage?: TrajectoryUsage
+}
+
+/** Disjoint provider token buckets for one request or a session prefix. */
+export interface TrajectoryUsage {
+  input?: number
+  cacheRead?: number
+  cacheWrite?: number
+  output?: number
+  reasoning?: number
+}
+
+function flattenRecords(turns: readonly TrajectoryTurnModel[]): TableRecord[] {
+  return turns.flatMap((turn) => {
+    let firstInTurn = true
+    const records = turn.groups.flatMap((group) => {
+      return group.cells.map((cell, index) => {
+        const turnStart = firstInTurn
+          && cell.requestOnly !== true
+          && cell.kind !== 'system'
+          && cell.kind !== 'compacted'
+        if (turnStart) firstInTurn = false
+        return {
+          turn: turn.turn,
+          group: group.title,
+          groupStart: index === 0,
+          turnStart,
+          cell,
+          turnEnd: false,
+        }
+      })
+    })
+    const last = records.at(-1)
+    if (last !== undefined) last.turnEnd = true
+    return records
+  })
+}
+
+function filterRecords(
+  records: readonly TableRecord[],
+  matches: ReadonlySet<number>,
+): TableRecord[] {
+  const filtered = records
+    .filter(record =>
+      record.cell.requestOnly !== true && matches.has(record.cell.index),
+    )
+    .map(record => ({ ...record, groupStart: false, turnStart: false, turnEnd: false }))
+  const startedTurns = new Set<number>()
+  for (const [index, record] of filtered.entries()) {
+    const previous = filtered[index - 1]
+    const next = filtered[index + 1]
+    record.groupStart = previous === undefined
+      || previous.turn !== record.turn
+      || previous.group !== record.group
+    record.turnStart = !startedTurns.has(record.turn)
+      && record.cell.kind !== 'system'
+      && record.cell.kind !== 'compacted'
+    if (record.turnStart) startedTurns.add(record.turn)
+    record.turnEnd = next === undefined || next.turn !== record.turn
+  }
+  return filtered
+}
+
+function requestStep(group: string): number | undefined {
+  if (!group.startsWith('Step ')) return undefined
+  const value = Number(group.slice('Step '.length))
+  return Number.isInteger(value) && value > 0 ? value : undefined
+}
+
+function requestKey(turn: number, group: string): string {
+  return `${turn}\u0000${group}`
+}
+
+function indexRequestNumbers(
+  records: readonly TableRecord[],
+  sessionNumbers: readonly TrajectoryRequestNumber[] | undefined,
+): ReadonlyMap<string, number> {
+  const numbers = new Map<string, number>()
+  for (const request of sessionNumbers ?? []) {
+    numbers.set(requestKey(request.turn, request.group), request.number)
+  }
+  let next = Math.max(0, ...numbers.values()) + 1
+  const boundaries = records
+    .filter(record => record.groupStart && requestStep(record.group) !== undefined)
+    .sort((left, right) => left.cell.index - right.cell.index)
+  for (const record of boundaries) {
+    const key = requestKey(record.turn, record.group)
+    if (!numbers.has(key)) numbers.set(key, next++)
+  }
+  return numbers
+}
+
+function summarizeTurn(records: readonly TableRecord[]): string {
+  const steps = new Set(
+    records
+      .map(record => record.group)
+      .filter(group => group.startsWith('Step ')),
+  ).size
+  const toolCalls = records.filter(record =>
+    record.cell.kind === 'tool' || record.cell.kind === 'subtool',
+  ).length
+  return [
+    `${steps} ${steps === 1 ? 'step' : 'steps'}`,
+    `${toolCalls} tool ${toolCalls === 1 ? 'call' : 'calls'}`,
+  ].join(' · ')
+}
+
+function collapseTurnRecords(
+  records: readonly TableRecord[],
+  collapsedTurns: ReadonlySet<number>,
+): TableRecord[] {
+  if (collapsedTurns.size === 0) return [...records]
+  const recordsByTurn = new Map<number, TableRecord[]>()
+  for (const record of records) {
+    const turnRecords = recordsByTurn.get(record.turn) ?? []
+    turnRecords.push(record)
+    recordsByTurn.set(record.turn, turnRecords)
+  }
+  return records.flatMap((record) => {
+    if (!collapsedTurns.has(record.turn)) return [record]
+    const turnRecords = recordsByTurn.get(record.turn) ?? [record]
+    if (record.cell.requestOnly === true || record.cell.kind === 'system') return [record]
+    const contentRecords = turnRecords.filter(candidate =>
+      candidate.cell.requestOnly !== true && candidate.cell.kind !== 'system')
+    if (contentRecords.length <= 1) return [record]
+    if (record.cell.index !== contentRecords[0]?.cell.index) return []
+    return [
+      { ...record, turnEnd: false },
+      {
+        ...record,
+        groupStart: false,
+        turnStart: false,
+        turnEnd: true,
+        collapsedSummary: summarizeTurn(contentRecords.slice(1)),
+        collapsedSummaryKind: 'turn',
+      },
+    ]
+  })
+}
+
+function assistantToolCalls(
+  records: readonly TableRecord[],
+  assistantIndex: number,
+): readonly TableRecord[] {
+  const at = records.findIndex(record => record.cell.index === assistantIndex)
+  if (at === -1 || records[at]?.cell.kind !== 'message') return []
+  const calls: TableRecord[] = []
+  for (let i = at + 1; i < records.length; i++) {
+    const record = records[i]
+    if (record === undefined) break
+    if (record.cell.kind !== 'tool' && record.cell.kind !== 'subtool') break
+    calls.push(record)
+  }
+  return calls
+}
+
+function summarizeAssistantTools(records: readonly TableRecord[]): string {
+  const names = [...new Set(records.map((record) => {
+    const separator = record.cell.text.indexOf(' · ')
+    return separator === -1 ? record.cell.text : record.cell.text.slice(0, separator)
+  }).filter(name => name !== ''))]
+  const count = records.length
+  const summary = `${count} tool ${count === 1 ? 'call' : 'calls'}`
+  return names.length > 0 ? `${summary} · ${names.join(', ')}` : summary
+}
+
+function collapseAssistantRecords(
+  records: readonly TableRecord[],
+  collapsedAssistants: ReadonlySet<number>,
+): TableRecord[] {
+  if (collapsedAssistants.size === 0) return [...records]
+  const out: TableRecord[] = []
+  for (let i = 0; i < records.length; i++) {
+    const record = records[i]
+    if (record === undefined) continue
+    out.push(record)
+    if (record.cell.kind !== 'message' || !collapsedAssistants.has(record.cell.index)) continue
+    const calls: TableRecord[] = []
+    for (let j = i + 1; j < records.length; j++) {
+      const candidate = records[j]
+      if (
+        candidate === undefined
+        || candidate.collapsedSummary !== undefined
+        || (candidate.cell.kind !== 'tool' && candidate.cell.kind !== 'subtool')
+      ) break
+      calls.push(candidate)
+    }
+    if (calls.length === 0) continue
+    const last = calls.at(-1)
+    out[out.length - 1] = { ...record, turnEnd: false }
+    out.push({
+      ...record,
+      groupStart: false,
+      turnStart: false,
+      turnEnd: last?.turnEnd ?? false,
+      collapsedSummary: summarizeAssistantTools(calls),
+      collapsedSummaryKind: 'assistant',
+    })
+    i += calls.length
+  }
+  return out
+}
+
+function stateOf(record: TableRecord): RecordState {
+  if (record.cell.isError) return 'error'
+  if (record.cell.kind === 'compacted' && record.cell.timeSeconds === null) return 'running'
+  if (
+    (record.cell.kind === 'tool' || record.cell.kind === 'subtool')
+    && record.cell.outputDetail === undefined
+  ) return 'running'
+  return 'complete'
+}
+
+function statusLabel(state: RecordState): string {
+  if (state === 'error') return 'Failed'
+  if (state === 'running') return 'Pending'
+  return 'Completed'
+}
+
+function TokenRows({ cell }: { cell: TrajectoryCellProps }) {
+  const content = cell.output !== undefined && cell.think !== undefined
+    ? Math.max(0, cell.output - cell.think)
+    : undefined
+  return (
+    <>
+      <div>
+        <dt>Tokens</dt>
+        <dd>{cell.output === undefined ? '—' : `${cell.output} tok`}</dd>
+      </div>
+      {cell.think !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Reasoning</dt>
+          <dd>{cell.think} tok</dd>
+        </div>
+      )}
+      {content !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Content</dt>
+          <dd>{content} tok</dd>
+        </div>
+      )}
+    </>
+  )
+}
+
+function inputTotal(usage: TrajectoryUsage): number | undefined {
+  if (
+    usage.input === undefined
+    && usage.cacheRead === undefined
+    && usage.cacheWrite === undefined
+  ) return undefined
+  return (usage.input ?? 0) + (usage.cacheRead ?? 0) + (usage.cacheWrite ?? 0)
+}
+
+function UsageRows({ usage }: { usage: TrajectoryUsage | undefined }) {
+  if (usage === undefined) return <p className={css.noPayload}>Usage not reported</p>
+  const totalInput = inputTotal(usage)
+  const otherOutput = usage.output !== undefined && usage.reasoning !== undefined
+    ? usage.output - usage.reasoning
+    : undefined
+  return (
+    <dl className={css.overview}>
+      {totalInput !== undefined && (
+        <div><dt>Input</dt><dd>{totalInput} tok</dd></div>
+      )}
+      {usage.cacheRead !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Cached</dt>
+          <dd>{usage.cacheRead} tok</dd>
+        </div>
+      )}
+      {usage.cacheWrite !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Cache created</dt>
+          <dd>{usage.cacheWrite} tok</dd>
+        </div>
+      )}
+      {usage.input !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Other</dt>
+          <dd>{usage.input} tok</dd>
+        </div>
+      )}
+      {usage.output !== undefined && (
+        <div><dt>Output</dt><dd>{usage.output} tok</dd></div>
+      )}
+      {usage.reasoning !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Reasoning</dt>
+          <dd>{usage.reasoning} tok</dd>
+        </div>
+      )}
+      {otherOutput !== undefined && (
+        <div className={css.requestTokenDetail}>
+          <dt>Content</dt>
+          <dd>{otherOutput} tok</dd>
+        </div>
+      )}
+    </dl>
+  )
+}
+
+function RequestUsagePanel({
+  usage,
+  cumulative,
+}: {
+  usage: TrajectoryUsage | undefined
+  cumulative: TrajectoryUsage | undefined
+}) {
+  return (
+    <div className={css.usagePanel}>
+      <section className={css.usageGroup}>
+        <h4 className={css.usageHeading}>This request</h4>
+        <UsageRows usage={usage} />
+      </section>
+      <section className={css.usageGroup}>
+        <h4 className={css.usageHeading}>Session cumulative</h4>
+        <UsageRows usage={cumulative} />
+      </section>
+    </div>
+  )
+}
+
+function RequestOptions({
+  options,
+  preview = false,
+}: {
+  options: AssistantRequestConfig | undefined
+  preview?: boolean
+}) {
+  if (options === undefined) {
+    return <p className={css.noPayload}>Options not recorded</p>
+  }
+  return (
+    <JsonTree
+      data={options}
+      label="Request options JSON"
+      className={preview ? css.jsonPreview : css.jsonPayload}
+    />
+  )
+}
+
+function messageOriginLabel(source: unknown): string {
+  if (typeof source !== 'object' || source === null || Array.isArray(source)) {
+    return 'Unknown'
+  }
+  const properties = source as Record<string, unknown>
+  const kind = properties.kind
+  if (kind === 'user') return 'User'
+  if (kind === 'plugin') {
+    const plugin = properties.plugin
+    return typeof plugin === 'string' && plugin !== ''
+      ? `Plugin · ${plugin}`
+      : 'Plugin'
+  }
+  if (kind === 'goal') {
+    const round = properties.round
+    return typeof round === 'number' && round > 0
+      ? `Goal · Round ${round}`
+      : 'Goal'
+  }
+  if (typeof kind !== 'string' || kind === '') return 'Unknown'
+  return `${kind[0]?.toUpperCase() ?? ''}${kind.slice(1)}`
+}
+
+function MessageOrigin({ record }: { record: TableRecord }) {
+  const source = record.cell.messageSource
+  if (source === undefined) return <p className={css.noPayload}>Origin not recorded</p>
+  const data = typeof source === 'object' && source !== null
+    ? source
+    : { value: source }
+  return (
+    <JsonTree
+      data={data}
+      label="Message origin JSON"
+      className={css.jsonPayload}
+    />
+  )
+}
+
+function isMarkdownRecord(record: TableRecord): boolean {
+  return record.cell.kind === 'user'
+    || record.cell.kind === 'context'
+    || record.cell.kind === 'message'
+}
+
+function parentRecords(
+  records: readonly TableRecord[],
+  record: TableRecord,
+): ParentRecords {
+  if (record.cell.kind !== 'tool' && record.cell.kind !== 'subtool') return {}
+  const at = records.findIndex(candidate => candidate.cell.index === record.cell.index)
+  if (at === -1) return {}
+  let tool: TableRecord | undefined
+  if (record.cell.kind === 'subtool') {
+    for (let i = at - 1; i >= 0; i--) {
+      const candidate = records[i]
+      if (
+        candidate === undefined
+        || candidate.turn !== record.turn
+        || candidate.group !== record.group
+      ) break
+      if (candidate.cell.kind === 'tool') {
+        tool = candidate
+        break
+      }
+    }
+  }
+  const parentCallId = tool?.cell.callId ?? record.cell.callId
+  let message: TableRecord | undefined
+  if (parentCallId !== undefined) {
+    message = records.find(candidate =>
+      candidate.turn === record.turn
+      && candidate.cell.kind === 'message'
+      && candidate.cell.sourceBlocks?.some(block => block.callId === parentCallId) === true,
+    )
+  }
+  return { ...(message === undefined ? {} : { message }), ...(tool === undefined ? {} : { tool }) }
+}
+
+function markdownSource(record: TableRecord): string | undefined {
+  if (record.cell.kind === 'user' || record.cell.kind === 'context') {
+    return record.cell.inputDetail
+  }
+  if (record.cell.kind === 'message' || record.cell.kind === 'compacted') {
+    return record.cell.outputDetail
+  }
+  return undefined
+}
+
+function detailTabs(record: TableRecord): readonly DetailTabItem[] {
+  if (record.cell.kind === 'system') {
+    return record.cell.previousPromptDetail === undefined
+      ? SYSTEM_PROMPT_TABS
+      : SYSTEM_UPDATE_TABS
+  }
+  if (record.cell.kind === 'compacted') {
+    return [
+      { id: 'overview', label: 'Summary' },
+      { id: 'source', label: 'Raw Output' },
+    ]
+  }
+  if (isMarkdownRecord(record)) {
+    return [
+      { id: 'overview', label: 'Summary' },
+      { id: 'rendered', label: 'Preview' },
+      { id: 'source', label: 'Source' },
+      ...(record.cell.messageSource === undefined
+        ? []
+        : [{ id: 'origin', label: 'Origin' } as const]),
+    ]
+  }
+  return [
+    { id: 'overview', label: 'Summary' },
+    ...(record.cell.inputDetail ? [{ id: 'input', label: 'Payload' } as const] : []),
+    ...(record.cell.outputDetail ? [{ id: 'output', label: 'Result' } as const] : []),
+    { id: 'schema', label: 'Schema' },
+    { id: 'timing', label: 'Timing' },
+  ]
+}
+
+function recordDisplayText(cell: TrajectoryCellProps): string {
+  if (isToolCallOnly(cell)) return ''
+  const markdown = cell.kind === 'user' || cell.kind === 'context'
+    ? cell.inputDetail
+    : cell.kind === 'message'
+      ? cell.outputDetail ?? cell.thinkingDetail
+      : undefined
+  if (!markdown) return cell.text
+  return extractMarkdownPlainText(markdown).replace(/\s+/g, ' ').trim()
+}
+
+function toolCallTextParts(
+  kind: TrajectoryCellKind,
+  text: string,
+): ToolCallTextParts | undefined {
+  if (kind !== 'tool' && kind !== 'subtool') return undefined
+  const separator = text.indexOf(' · ')
+  if (separator === -1) return { name: text }
+  return {
+    name: text.slice(0, separator),
+    args: text.slice(separator + 3),
+  }
+}
+
+function isToolCallOnly(cell: TrajectoryCellProps): boolean {
+  return cell.kind === 'message'
+    && !cell.outputDetail
+    && !cell.thinkingDetail
+    && cell.text === 'Tool call only'
+}
+
+function MarkdownFragment({
+  text,
+  rendered,
+  preview,
+}: {
+  text: string
+  rendered: boolean
+  preview: boolean
+}) {
+  if (rendered) {
+    return (
+      <div className={preview ? css.markdownPreview : css.markdownPayload}>
+        <MarkdownText text={text} />
+      </div>
+    )
+  }
+  return (
+    <pre className={`${css.payload} ${preview ? css.payloadPreview : ''}`}>
+      {text}
+    </pre>
+  )
+}
+
+function SourceBlocks({
+  blocks,
+  onOpenCall,
+}: {
+  blocks: readonly TrajectorySourceBlock[]
+  onOpenCall: (callId: string) => void
+}) {
+  return (
+    <div className={css.sourceBlocks}>
+      {blocks.map((block, index) => (
+        <section className={css.sourceBlock} key={index}>
+          {block.callId !== undefined
+            ? (
+              <button
+                type="button"
+                className={css.sourceBlockJumpTarget}
+                aria-label={`Open Block #${index + 1} tool call summary`}
+                title="Open tool call summary"
+                onClick={() => {
+                  if (block.callId !== undefined) onOpenCall(block.callId)
+                }}
+              >
+                <span className={css.sourceBlockLabel}>
+                  {`Block #${index + 1} ${block.type}`}
+                </span>
+                <IconChevronRightOutline14 className={css.sourceBlockJumpIcon} size={12} />
+              </button>
+            )
+            : (
+              <div className={css.sourceBlockHeader}>
+                <span className={css.sourceBlockLabel}>
+                  {`Block #${index + 1} ${block.type}`}
+                </span>
+              </div>
+            )}
+          {block.imageSrc !== undefined
+            ? <PanelImage block={block} />
+            : <pre className={css.sourceBlockContent}>{block.content}</pre>}
+        </section>
+      ))}
+    </div>
+  )
+}
+
+function PanelImage({
+  block,
+  preview = false,
+}: {
+  block: TrajectorySourceBlock
+  preview?: boolean
+}) {
+  if (block.imageSrc === undefined) return null
+  return (
+    <a
+      className={preview ? `${css.panelImageLink} ${css.panelImageLinkPreview}` : css.panelImageLink}
+      href={block.imageSrc}
+      target="_blank"
+      rel="noopener noreferrer"
+      title="Open image"
+    >
+      <img
+        className={css.panelImage}
+        src={block.imageSrc}
+        alt={block.imageAlt ?? ''}
+      />
+    </a>
+  )
+}
+
+function MessageImages({
+  blocks,
+  preview,
+}: {
+  blocks: readonly TrajectorySourceBlock[] | undefined
+  preview: boolean
+}) {
+  const images = blocks?.filter(block => block.imageSrc !== undefined) ?? []
+  if (images.length === 0) return null
+  return (
+    <div className={preview ? `${css.messageImages} ${css.messageImagesPreview}` : css.messageImages}>
+      {images.map((block, index) => <PanelImage block={block} preview={preview} key={index} />)}
+    </div>
+  )
+}
+
+function AssistantToolCalls({
+  blocks,
+  preview,
+  onOpenCall,
+}: {
+  blocks: readonly TrajectorySourceBlock[] | undefined
+  preview: boolean
+  onOpenCall: (callId: string) => void
+}) {
+  const calls = blocks?.filter(block => block.type === 'tool-call') ?? []
+  if (calls.length === 0) return null
+  return (
+    <ul className={preview
+      ? `${css.assistantToolCalls} ${css.assistantToolCallsPreview}`
+      : css.assistantToolCalls}
+    >
+      {calls.map((call, index) => (
+        <li key={call.callId ?? index}>
+          <button
+            type="button"
+            className={css.assistantToolCallButton}
+            title="Open tool call summary"
+            onClick={() => {
+              if (call.callId !== undefined) onOpenCall(call.callId)
+            }}
+          >
+            <svg
+              className={css.assistantToolCallIcon}
+              width="12"
+              height="12"
+              viewBox="0 0 24 24"
+              fill="none"
+              aria-hidden="true"
+            >
+              <path
+                d="M14.7 6.3a1 1 0 0 0 0 1.4l1.6 1.6a1 1 0 0 0 1.4 0l3.77-3.77a6 6 0 0 1-7.94 7.94l-6.91 6.91a2.12 2.12 0 0 1-3-3l6.91-6.91a6 6 0 0 1 7.94-7.94z"
+                stroke="currentColor"
+                strokeWidth="1.8"
+                strokeLinecap="round"
+                strokeLinejoin="round"
+              />
+            </svg>
+            <span className={css.assistantToolCallText}>
+              <span className={css.assistantToolCallName}>
+                {call.toolName ?? 'tool-call'}
+              </span>
+              {call.content !== '' && (
+                <span className={css.assistantToolCallArgs}>{call.content}</span>
+              )}
+            </span>
+          </button>
+        </li>
+      ))}
+    </ul>
+  )
+}
+
+function ToolGlyph() {
+  return (
+    <svg
+      className={css.toolCatalogIcon}
+      width="12"
+      height="12"
+      viewBox="0 0 24 24"
+      fill="none"
+      aria-hidden="true"
+    >
+      <path
+        d="M14.7 6.3a1 1 0 0 0 0 1.4l1.6 1.6a1 1 0 0 0 1.4 0l3.77-3.77a6 6 0 0 1-7.94 7.94l-6.91 6.91a2.12 2.12 0 0 1-3-3l6.91-6.91a6 6 0 0 1 7.94-7.94z"
+        stroke="currentColor"
+        strokeWidth="1.8"
+        strokeLinecap="round"
+        strokeLinejoin="round"
+      />
+    </svg>
+  )
+}
+
+function ToolCatalog({ tools }: { tools: ConversationPromptSnapshot['tools'] }) {
+  if (tools.length === 0) return <p className={css.noPayload}>No tools in this request</p>
+  return (
+    <div className={css.toolCatalog}>
+      {tools.map((tool, index) => (
+        <details className={css.toolCatalogItem} key={`${tool.name}:${index}`}>
+          <summary className={css.toolCatalogSummary}>
+            <IconChevronRightOutline14 className={css.toolCatalogChevron} size={12} />
+            <ToolGlyph />
+            <span className={css.toolCatalogName}>{tool.name}</span>
+            <span className={css.toolCatalogDescription}>{tool.description}</span>
+          </summary>
+          <div className={css.toolCatalogDefinition}>
+            {tool.description !== '' && (
+              <p className={css.toolCatalogFullDescription}>{tool.description}</p>
+            )}
+            <JsonTree
+              data={tool.parameters}
+              label={`${tool.name} parameters JSON`}
+              className={css.toolCatalogTree}
+            />
+          </div>
+        </details>
+      ))}
+    </div>
+  )
+}
+
+interface PromptDiffLine {
+  kind: 'meta' | 'context' | 'added' | 'removed'
+  text: string
+}
+
+function promptDiffLines(before: string, after: string): readonly PromptDiffLine[] {
+  const patch = structuredPatch('', '', before, after, undefined, undefined, { context: 3 })
+  return patch.hunks.flatMap((hunk, hunkIndex) => [
+    ...(hunkIndex === 0 ? [] : [{ kind: 'meta' as const, text: '' }]),
+    {
+      kind: 'meta' as const,
+      text: `@@ -${hunk.oldStart},${hunk.oldLines} +${hunk.newStart},${hunk.newLines} @@`,
+    },
+    ...hunk.lines.flatMap((line): PromptDiffLine[] => {
+      if (line.startsWith('\\')) return []
+      if (line.startsWith('+')) return [{ kind: 'added', text: line }]
+      if (line.startsWith('-')) return [{ kind: 'removed', text: line }]
+      return [{ kind: 'context', text: line }]
+    }),
+  ])
+}
+
+function PromptDiffSection({
+  title,
+  before,
+  after,
+}: {
+  title: string
+  before: string
+  after: string
+}) {
+  const lines = promptDiffLines(before, after)
+  if (lines.length === 0) return null
+  return (
+    <section className={css.promptDiffSection}>
+      <h3 className={css.promptDiffTitle}>{title}</h3>
+      <pre className={css.promptDiff}>
+        {lines.map((line, index) => (
+          <span className={css[`promptDiffLine${line.kind}`]} key={index}>
+            {line.text || ' '}
+            {'\n'}
+          </span>
+        ))}
+      </pre>
+    </section>
+  )
+}
+
+function SystemPromptDiff({
+  before,
+  after,
+}: {
+  before: ConversationPromptSnapshot
+  after: ConversationPromptSnapshot
+}) {
+  const toolsBefore = JSON.stringify(before.tools, null, 2)
+  const toolsAfter = JSON.stringify(after.tools, null, 2)
+  return (
+    <div className={css.promptDiffSections}>
+      {before.system !== after.system && (
+        <PromptDiffSection
+          title="System Prompt"
+          before={before.system}
+          after={after.system}
+        />
+      )}
+      {toolsBefore !== toolsAfter && (
+        <PromptDiffSection
+          title="Tools"
+          before={toolsBefore}
+          after={toolsAfter}
+        />
+      )}
+    </div>
+  )
+}
+
+function ToolOutputBlocks({
+  blocks,
+  preview,
+}: {
+  blocks: readonly TrajectorySourceBlock[]
+  preview: boolean
+}) {
+  return (
+    <div className={preview ? `${css.resultBlocks} ${css.resultBlocksPreview}` : css.resultBlocks}>
+      {blocks.map((block, index) => (
+        block.imageSrc !== undefined
+          ? <PanelImage block={block} preview={preview} key={index} />
+          : block.content !== ''
+            ? <pre className={css.resultBlockText} key={index}>{block.content}</pre>
+            : null
+      ))}
+    </div>
+  )
+}
+
+function MarkdownRecordContent({
+  record,
+  rendered,
+  preview = false,
+  thinkingExpanded,
+  onThinkingExpandedChange,
+  onOpenCall,
+}: {
+  record: TableRecord
+  rendered: boolean
+  preview?: boolean
+  thinkingExpanded: boolean
+  onThinkingExpandedChange: (expanded: boolean) => void
+  onOpenCall: (callId: string) => void
+}) {
+  if (!rendered && record.cell.sourceBlocks && record.cell.sourceBlocks.length > 0) {
+    return <SourceBlocks blocks={record.cell.sourceBlocks} onOpenCall={onOpenCall} />
+  }
+  if (record.cell.thinkingDetail) {
+    if (!rendered) {
+      const source = [
+        record.cell.thinkingDetail,
+        record.cell.outputDetail,
+      ].filter((value): value is string => value !== undefined && value !== '').join('\n\n')
+      return <MarkdownFragment text={source} rendered={false} preview={preview} />
+    }
+    return (
+      <div className={`${css.assistantContent} ${css.assistantContentRendered}`}>
+        <div className={
+          preview && !record.cell.outputDetail
+            ? `${css.thinkingQuote} ${css.thinkingQuoteOnlyPreview}`
+            : css.thinkingQuote
+        }
+        >
+          <button
+            type="button"
+            className={css.thinkingToggle}
+            aria-expanded={thinkingExpanded}
+            onClick={() => { onThinkingExpandedChange(!thinkingExpanded) }}
+          >
+            {thinkingExpanded ? 'Thinking' : 'Thinking ...'}
+          </button>
+          {thinkingExpanded && (
+            <MarkdownFragment
+              text={record.cell.thinkingDetail}
+              rendered={rendered}
+              preview={preview}
+            />
+          )}
+        </div>
+        {record.cell.outputDetail && (
+          <div className={css.assistantOutput}>
+            <MarkdownFragment
+              text={record.cell.outputDetail}
+              rendered={rendered}
+              preview={preview}
+            />
+          </div>
+        )}
+        <AssistantToolCalls
+          blocks={record.cell.sourceBlocks}
+          preview={preview}
+          onOpenCall={onOpenCall}
+        />
+        <MessageImages
+          blocks={record.cell.sourceBlocks}
+          preview={preview}
+        />
+      </div>
+    )
+  }
+  const source = markdownSource(record)
+  const hasImages = record.cell.sourceBlocks?.some(block => block.imageSrc !== undefined) === true
+  const hasToolCalls = record.cell.kind === 'message'
+    && record.cell.sourceBlocks?.some(block => block.type === 'tool-call') === true
+  if (!source && !hasImages && !hasToolCalls) {
+    const emptyLabel = isToolCallOnly(record.cell)
+      ? 'Tool call only'
+      : record.cell.text || 'No content'
+    return <p className={css.noPayload}>{emptyLabel}</p>
+  }
+  if (!rendered || (!hasImages && !hasToolCalls)) {
+    return <MarkdownFragment text={source ?? ''} rendered={rendered} preview={preview} />
+  }
+  return (
+    <div>
+      {source && <MarkdownFragment text={source} rendered preview={preview} />}
+      {record.cell.kind === 'message' && (
+        <AssistantToolCalls
+          blocks={record.cell.sourceBlocks}
+          preview={preview}
+          onOpenCall={onOpenCall}
+        />
+      )}
+      <MessageImages blocks={record.cell.sourceBlocks} preview={preview} />
+    </div>
+  )
+}
+
+function RecordTiming({ record }: { record: TableRecord }) {
+  return record.cell.kind === 'message' && record.cell.assistantMetrics !== undefined
+    ? <AssistantTimingPanel metrics={record.cell.assistantMetrics} />
+    : (
+      <dl className={css.overview}>
+        <div><dt>Started</dt><StartedAtValue timestamp={record.cell.startedAt ?? null} /></div>
+        <div><dt>Duration</dt><dd>{formatElapsedSeconds(record.cell.timeSeconds)}</dd></div>
+        <div><dt>Timing source</dt><dd>{record.cell.timeSeconds === null ? 'Not available' : 'Session timestamps'}</dd></div>
+      </dl>
+    )
+}
+
+function RequestTiming({
+  assistant,
+  anchor,
+  request,
+}: {
+  assistant: TableRecord | undefined
+  anchor: TableRecord | undefined
+  request: TrajectoryRequestNumber | undefined
+}) {
+  if (assistant !== undefined) return <RecordTiming record={assistant} />
+  if (request?.startedAt !== undefined) {
+    const duration = request.completedAt === null || request.completedAt === undefined
+      ? null
+      : Math.max(0, (request.completedAt - request.startedAt) / 1000)
+    return (
+      <dl className={css.overview}>
+        <div><dt>Started</dt><StartedAtValue timestamp={request.startedAt} /></div>
+        <div><dt>Duration</dt><dd>{formatElapsedSeconds(duration)}</dd></div>
+        <div>
+          <dt>Timing source</dt>
+          <dd>{duration === null ? 'Session timestamps (running)' : 'Session timestamps'}</dd>
+        </div>
+      </dl>
+    )
+  }
+  return (
+    <dl className={css.overview}>
+      <div>
+        <dt>Started</dt>
+        <StartedAtValue timestamp={anchor?.cell.startedAt ?? null} />
+      </div>
+      <div><dt>Duration</dt><dd>—</dd></div>
+    </dl>
+  )
+}
+
+function RecordPayload({
+  record,
+  direction,
+  preview = false,
+}: {
+  record: TableRecord
+  direction: 'input' | 'output'
+  preview?: boolean
+}) {
+  const value = direction === 'input' ? record.cell.inputDetail : record.cell.outputDetail
+  const missing = direction === 'input'
+    ? 'No payload captured'
+    : 'No result captured'
+  if (!value) return <p className={css.noPayload}>{missing}</p>
+
+  const json = parseJsonContainer(value)
+  const singleTextResult = direction === 'output'
+    && record.cell.outputBlocks?.length === 1
+    && record.cell.outputBlocks[0]?.type === 'text'
+  if (singleTextResult && json !== undefined) {
+    return (
+      <JsonTree
+        data={json}
+        label="Result JSON"
+        className={preview ? css.jsonPreview : css.jsonPayload}
+      />
+    )
+  }
+
+  if (
+    direction === 'output'
+    && record.cell.outputBlocks?.some(block =>
+      block.imageSrc !== undefined || block.content !== '') === true
+  ) {
+    return (
+      <ToolOutputBlocks
+        blocks={record.cell.outputBlocks}
+        preview={preview}
+      />
+    )
+  }
+
+  const markdown = (
+    direction === 'input'
+    && (record.cell.kind === 'user' || record.cell.kind === 'context')
+  ) || (
+    direction === 'output' && record.cell.kind === 'message'
+  )
+  if (markdown) {
+    return (
+      <div className={preview ? css.markdownPreview : css.markdownPayload}>
+        <MarkdownText text={value} />
+      </div>
+    )
+  }
+  if (json !== undefined) {
+    return (
+      <JsonTree
+        data={json}
+        label={`${direction === 'input' ? 'Payload' : 'Result'} JSON`}
+        className={preview ? css.jsonPreview : css.jsonPayload}
+      />
+    )
+  }
+  return (
+    <pre className={[
+      css.payload,
+      preview ? css.payloadPreview : undefined,
+      record.cell.isError ? css.error : undefined,
+      value === 'No output' ? css.noOutputText : undefined,
+    ].filter((value): value is string => value !== undefined).join(' ')}
+    >
+      {value}
+    </pre>
+  )
+}
+
+function RecordSchema({
+  record,
+  preview = false,
+}: {
+  record: TableRecord
+  preview?: boolean
+}) {
+  if (!record.cell.schemaDetail) {
+    return <p className={css.noPayload}>Schema unavailable</p>
+  }
+  const schema = parseToolSchema(record.cell.schemaDetail)
+  if (schema !== undefined) {
+    return (
+      <div className={preview ? `${css.schema} ${css.schemaPreview}` : css.schema}>
+        <header className={css.schemaIntro}>
+          <h3 className={css.schemaName}>{schema.name}</h3>
+          <p className={css.schemaDescription}>{schema.description}</p>
+        </header>
+        <section className={css.schemaParameters}>
+          <h4 className={css.schemaParametersTitle}>Parameters</h4>
+          <JsonTree
+            data={schema.parameters}
+            label={`${schema.name} parameters JSON`}
+            className={css.schemaTree}
+          />
+        </section>
+      </div>
+    )
+  }
+  return (
+    <pre className={`${css.payload} ${preview ? css.payloadPreview : ''}`}>
+      {record.cell.schemaDetail}
+    </pre>
+  )
+}
+
+interface ParsedToolSchema {
+  name: string
+  description: string
+  parameters: object
+}
+
+function parseToolSchema(value: string): ParsedToolSchema | undefined {
+  try {
+    const parsed: unknown = JSON.parse(value)
+    if (typeof parsed !== 'object' || parsed === null || Array.isArray(parsed)) return undefined
+    const schema = parsed as Record<string, unknown>
+    if (
+      typeof schema.name !== 'string'
+      || typeof schema.description !== 'string'
+      || typeof schema.parameters !== 'object'
+      || schema.parameters === null
+      || Array.isArray(schema.parameters)
+    ) return undefined
+    return {
+      name: schema.name,
+      description: schema.description,
+      parameters: schema.parameters,
+    }
+  } catch {
+    return undefined
+  }
+}
+
+function parseJsonContainer(value: string): object | undefined {
+  try {
+    const parsed: unknown = JSON.parse(value)
+    return typeof parsed === 'object' && parsed !== null ? parsed : undefined
+  } catch {
+    return undefined
+  }
+}
+
+function OverviewSection({
+  label,
+  onOpen,
+  children,
+}: {
+  label: string
+  onOpen: () => void
+  children: ReactNode
+}) {
+  return (
+    <section className={css.overviewSection}>
+      <h3 className={css.overviewHeading}>
+        <button
+          type="button"
+          className={css.overviewTitle}
+          onClick={onOpen}
+        >
+          <span>{label}</span>
+          <IconChevronRightOutline14 className={css.overviewTitleIcon} size={12} />
+        </button>
+      </h3>
+      <div className={css.overviewPreview}>{children}</div>
+    </section>
+  )
+}
+
+/**
+ * Render trajectory events as a dense ledger with turn and step separators.
+ * Clicking ledger whitespace clears the active record or request selection.
+ * @param props - Grouped trajectory data and whole-ledger fold state.
+ * @returns The ledger and an optional local record inspector.
+ */
+export function TrajectoryTable({
+  requestNumbers: sessionRequestNumbers,
+  turns,
+  timelineFocusIndexes = null,
+  searchMatchIndexes = null,
+  onSelectedIndexChange,
+  onRecordSelect,
+  onClearSelection,
+  collapsedTurns,
+  onToggleTurn,
+  collapsedAssistants,
+  onToggleAssistant,
+}: TrajectoryTableProps) {
+  const [selectedIndex, setSelectedIndex] = useState<number | null>(null)
+  const [selectedRequest, setSelectedRequest] = useState<SelectedRequest | null>(null)
+  const [activeTab, setActiveTab] = useState<DetailTab>('overview')
+  const [thinkingExpanded, setThinkingExpanded] = useState(true)
+  const [detailsWidth, setDetailsWidth] = useState<number | null>(null)
+  const [toolRequestOffset, setToolRequestOffset] = useState<number | null>(null)
+  const detailsResizeDrag = useRef<DetailsResizeDrag | null>(null)
+  const tabHistory = useRef<Set<DetailTab>>(new Set(['overview']))
+  useEffect(() => {
+    onSelectedIndexChange?.(selectedIndex)
+  }, [onSelectedIndexChange, selectedIndex])
+  const allRecords = flattenRecords(turns)
+  const requestNumbers = indexRequestNumbers(allRecords, sessionRequestNumbers)
+  const records = searchMatchIndexes === null
+    ? collapseAssistantRecords(
+      collapseTurnRecords(allRecords, collapsedTurns),
+      collapsedAssistants,
+    )
+    : filterRecords(allRecords, searchMatchIndexes)
+  const selected = allRecords.find(record => record.cell.index === selectedIndex)
+  const selectedPrompt = selected?.cell.kind === 'system'
+    ? selected.cell.promptDetail
+    : undefined
+  const selectedPreviousPrompt = selected?.cell.kind === 'system'
+    ? selected.cell.previousPromptDetail
+    : undefined
+  const promptSelected = selectedPrompt !== undefined
+  const selectedState = selected === undefined ? undefined : stateOf(selected)
+  const selectedRequestRecords = selectedRequest === null
+    ? []
+    : allRecords.filter(record =>
+      record.turn === selectedRequest.turn
+        && record.group === selectedRequest.group,
+    )
+  const selectedRequestAssistant = selectedRequestRecords.find(
+    record => record.cell.kind === 'message',
+  )
+  const selectedRequestAnchor = selectedRequestAssistant ?? selectedRequestRecords[0]
+  const selectedRequestInfo = selectedRequest === null
+    ? undefined
+    : sessionRequestNumbers?.find(request => request.number === selectedRequest.number)
+  const selectedRequestState: RecordState | undefined = selectedRequest === null
+    ? undefined
+    : selectedRequestInfo?.status
+      ?? (selectedRequestAssistant?.cell.assistantMetrics?.completedTime === null
+        ? 'running'
+        : selectedRequestAssistant === undefined
+          && selectedRequestRecords.some(record => stateOf(record) === 'running')
+          ? 'running'
+          : 'complete')
+  const selectedRequestToolCalls = selectedRequestRecords.filter(
+    record => record.cell.kind === 'tool',
+  ).length
+  const selectedRequestSubtoolCalls = selectedRequestRecords.filter(
+    record => record.cell.kind === 'subtool',
+  ).length
+  const selectedRequestResult = selectedRequestInfo?.resultSeq === undefined
+    ? selectedRequestAssistant
+    : allRecords.find(record => record.cell.sourceSeq === selectedRequestInfo.resultSeq)
+  const selectedRequestUsage = selectedRequestInfo?.usage ?? (
+    selectedRequestAssistant === undefined
+      ? undefined
+      : {
+        ...(selectedRequestAssistant.cell.input === undefined
+          ? {}
+          : { input: selectedRequestAssistant.cell.input }),
+        ...(selectedRequestAssistant.cell.cacheRead === undefined
+          ? {}
+          : { cacheRead: selectedRequestAssistant.cell.cacheRead }),
+        ...(selectedRequestAssistant.cell.cacheWrite === undefined
+          ? {}
+          : { cacheWrite: selectedRequestAssistant.cell.cacheWrite }),
+        ...(selectedRequestAssistant.cell.output === undefined
+          ? {}
+          : { output: selectedRequestAssistant.cell.output }),
+        ...(selectedRequestAssistant.cell.think === undefined
+          ? {}
+          : { reasoning: selectedRequestAssistant.cell.think }),
+      }
+  )
+  const selectedRequestCumulativeUsage =
+    selectedRequestInfo?.cumulativeUsage ?? selectedRequestUsage
+  const selectedRequestOptions = selectedRequestInfo?.requestConfig
+  const activeTurn = selectedRequest?.turn ?? selected?.turn
+  const selectedTabs = selectedRequest !== null
+    ? REQUEST_TABS.filter(tab => tab.id !== 'options' || selectedRequestOptions !== undefined)
+    : selected === undefined ? [] : detailTabs(selected)
+  const selectedParents: ParentRecords = selected === undefined
+    ? {}
+    : parentRecords(allRecords, selected)
+  const selectedParentMessage = selectedParents.message
+  const selectedParentTool = selectedParents.tool
+  const selectedAssistantRequest = selected?.cell.kind === 'message'
+    ? requestNumbers.get(requestKey(selected.turn, selected.group))
+    : undefined
+  const selectedAssistantRequestTarget: SelectedRequest | undefined =
+    selected !== undefined && selectedAssistantRequest !== undefined
+      ? {
+        turn: selected.turn,
+        number: selectedAssistantRequest,
+        group: selected.group,
+      }
+      : undefined
+  const hasSelectedHierarchy = selectedAssistantRequestTarget !== undefined
+    || selectedParents.message !== undefined
+    || selectedParents.tool !== undefined
+  const splitStyle: TrajectorySplitStyle | undefined = toolRequestOffset === null
+    ? undefined
+    : {
+      '--trajectory-tool-request-width': `calc(58cqw - ${toolRequestOffset}px)`,
+    }
+
+  const activateTab = (tab: DetailTab) => {
+    tabHistory.current.delete(tab)
+    tabHistory.current.add(tab)
+    setActiveTab(tab)
+  }
+
+  const clearInspectorSelection = () => {
+    setSelectedIndex(null)
+    setSelectedRequest(null)
+  }
+
+  const clearAllSelections = () => {
+    clearInspectorSelection()
+    onClearSelection?.()
+  }
+
+  const selectRecord = (index: number) => {
+    const record = allRecords.find(candidate => candidate.cell.index === index)
+    onRecordSelect?.(index)
+    setSelectedRequest(null)
+    setSelectedIndex(index)
+    if (record === undefined) return
+    const tabs = detailTabs(record)
+    const available = new Set(tabs.map(tab => tab.id))
+    const recent = [...tabHistory.current].reverse().find(tab => available.has(tab))
+    setActiveTab(recent ?? tabs[0]?.id ?? 'overview')
+  }
+
+  const selectRequest = (
+    request: SelectedRequest,
+    tab: 'overview' | 'timing' = 'overview',
+  ) => {
+    setSelectedIndex(null)
+    setSelectedRequest(request)
+    activateTab(tab)
+  }
+
+  const openRecordSummary = (target: TableRecord) => {
+    const targetAt = allRecords.findIndex(record => record.cell.index === target.cell.index)
+    if (collapsedTurns.has(target.turn)) onToggleTurn(target.turn)
+    if (target.cell.kind === 'tool' || target.cell.kind === 'subtool') {
+      for (let i = targetAt - 1; i >= 0; i--) {
+        const candidate = allRecords[i]
+        if (candidate === undefined || candidate.turn !== target.turn) break
+        if (candidate.cell.kind !== 'message') continue
+        if (collapsedAssistants.has(candidate.cell.index)) onToggleAssistant(candidate.cell.index)
+        break
+      }
+    }
+    setSelectedRequest(null)
+    setSelectedIndex(target.cell.index)
+    activateTab('overview')
+  }
+
+  const openCallSummary = (callId: string) => {
+    const target = allRecords.find(record => record.cell.callId === callId)
+    if (target !== undefined) openRecordSummary(target)
+  }
+
+  return (
+    <div className={css.split} style={splitStyle}>
+      <div
+        className={css.tablePane}
+        onClick={(event) => {
+          if (event.target === event.currentTarget) clearAllSelections()
+        }}
+      >
+        <table className={css.table}>
+          <colgroup>
+            <col className={css.eventColumn} />
+            <col className={css.contentColumn} />
+          </colgroup>
+          <tbody>
+            {records.map((record) => {
+              const displayText = recordDisplayText(record.cell)
+              const toolCallText = toolCallTextParts(record.cell.kind, displayText)
+              const listDisplayText = toolCallText === undefined
+                ? displayText
+                : [toolCallText.name, toolCallText.args].filter(Boolean).join(' ')
+              const isCollapsedSummary = record.collapsedSummary !== undefined
+              const isRequestOnly = record.cell.requestOnly === true
+              const isInitialSystem = record.cell.kind === 'system'
+                && record.cell.index === allRecords[0]?.cell.index
+              const request = record.groupStart
+                && !isCollapsedSummary
+                && !collapsedTurns.has(record.turn)
+                ? requestNumbers.get(requestKey(record.turn, record.group))
+                : undefined
+              const requestInfo = request === undefined
+                ? undefined
+                : sessionRequestNumbers?.find(candidate => candidate.number === request)
+              const requestLabel = request === undefined
+                ? undefined
+                : `Request #${request}${requestInfo?.purpose === 'compaction' ? ' · Compaction' : ''}`
+              const requestSelected = request !== undefined
+                && selectedRequest?.turn === record.turn
+                && selectedRequest.number === request
+              return (
+                <tr
+                  key={`${record.cell.index}:${record.collapsedSummaryKind ?? 'record'}`}
+                  tabIndex={isRequestOnly ? -1 : 0}
+                  aria-label={isCollapsedSummary
+                    ? `Collapsed ${record.collapsedSummaryKind} summary, ${record.collapsedSummary}`
+                    : isRequestOnly
+                      ? `Request ${request ?? ''}, compaction`
+                      : `${request === undefined ? '' : `Request ${request}, `}${KIND_LABEL[record.cell.kind]}, ${listDisplayText || 'no content'}`}
+                  aria-selected={!isCollapsedSummary && !isRequestOnly && selectedIndex === record.cell.index}
+                  data-kind={record.cell.kind}
+                  data-record-index={!isCollapsedSummary && !isRequestOnly
+                    ? record.cell.index
+                    : undefined}
+                  data-request-only={isRequestOnly || undefined}
+                  data-group-start={record.groupStart || undefined}
+                  data-turn-start={record.turnStart || undefined}
+                  data-error={record.cell.isError || undefined}
+                  data-running={stateOf(record) === 'running' || undefined}
+                  data-turn-end={record.turnEnd || undefined}
+                  data-collapsed-summary={record.collapsedSummaryKind}
+                  data-selected={!isCollapsedSummary && selectedIndex === record.cell.index || undefined}
+                  data-timeline-focus={isCollapsedSummary || timelineFocusIndexes === null
+                    ? undefined
+                    : timelineFocusIndexes.has(record.cell.index) ? 'inside' : 'outside'}
+                  onClick={isRequestOnly
+                    ? undefined
+                    : isCollapsedSummary
+                      ? () => {
+                        if (record.collapsedSummaryKind === 'turn') onToggleTurn(record.turn)
+                        else onToggleAssistant(record.cell.index)
+                      }
+                      : () => { selectRecord(record.cell.index) }}
+                  onDoubleClick={(event) => {
+                    if (isCollapsedSummary || isRequestOnly) return
+                    if (collapsedTurns.has(record.turn)) {
+                      event.preventDefault()
+                      onToggleTurn(record.turn)
+                      return
+                    }
+                    if (
+                      record.cell.kind === 'message'
+                      && assistantToolCalls(allRecords, record.cell.index).length > 0
+                    ) {
+                      event.preventDefault()
+                      onToggleAssistant(record.cell.index)
+                      return
+                    }
+                    if (!record.turnStart) return
+                    if (allRecords.filter(candidate =>
+                      candidate.turn === record.turn
+                      && candidate.cell.requestOnly !== true
+                      && candidate.cell.kind !== 'system').length <= 1) return
+                    event.preventDefault()
+                    onToggleTurn(record.turn)
+                  }}
+                  onKeyDown={(event) => {
+                    if (isRequestOnly) return
+                    if (event.key !== 'Enter' && event.key !== ' ') return
+                    event.preventDefault()
+                    if (isCollapsedSummary) {
+                      if (record.collapsedSummaryKind === 'turn') onToggleTurn(record.turn)
+                      else onToggleAssistant(record.cell.index)
+                      return
+                    }
+                    selectRecord(record.cell.index)
+                  }}
+                >
+                  <td className={css.event}>
+                    {request !== undefined && (
+                      <button
+                        type="button"
+                        className={requestSelected
+                          ? `${css.requestBoundaryControl} ${css.requestBoundaryControlActive}`
+                          : css.requestBoundaryControl}
+                        aria-label={requestLabel}
+                        aria-pressed={requestSelected}
+                        data-label={requestLabel}
+                        onClick={(event) => {
+                          event.stopPropagation()
+                          selectRequest({
+                            turn: record.turn,
+                            number: request,
+                            group: record.group,
+                          })
+                        }}
+                        onDoubleClick={(event) => { event.stopPropagation() }}
+                      />
+                    )}
+                    {activeTurn === record.turn && !isInitialSystem && (
+                      <span className={css.turnRail} aria-hidden="true" />
+                    )}
+                    {!isCollapsedSummary && selectedIndex === record.cell.index && (
+                      <span className={css.selectionRail} aria-hidden="true" />
+                    )}
+                    {!isCollapsedSummary
+                    && !isRequestOnly
+                    && record.turnStart && (
+                      <span
+                        className={activeTurn === record.turn
+                          ? `${css.turnLabel} ${css.turnLabelActive}`
+                          : css.turnLabel}
+                      >
+                        Turn {record.turn}
+                      </span>
+                    )}
+                    <div className={css.eventInner}>
+                      {!isCollapsedSummary && !isRequestOnly && (
+                        <span
+                          className={css.kindSlot}
+                        >
+                          <span className={`${css.kindTag} ${
+                            record.cell.kind === 'system'
+                              ? css.systemNeutral
+                              : record.cell.kind === 'context'
+                                ? css.contextGreen
+                                : record.cell.kind === 'compacted'
+                                  ? css.compacted
+                                  : record.cell.kind === 'tool'
+                                    ? css.toolAmber
+                                    : record.cell.kind === 'message'
+                                      ? css.assistantVioletBright
+                                      : record.cell.kind === 'subtool'
+                                        ? css.subtoolAmber
+                                        : css[record.cell.kind]
+                          }`}
+                          >
+                            {KIND_LABEL[record.cell.kind]}
+                          </span>
+                        </span>
+                      )}
+                    </div>
+                  </td>
+                  <td className={css.content}>
+                    {isRequestOnly
+                      ? null
+                      : record.collapsedSummary !== undefined
+                        ? (
+                          <span className={css.collapsedTurnContent} title={record.collapsedSummary}>
+                            <span className={css.collapsedTurnEllipsis}>…</span>
+                            <span className={css.collapsedTurnText}>{record.collapsedSummary}</span>
+                          </span>
+                        )
+                        : (
+                          <span
+                            className={record.cell.result === undefined ? css.contentText : css.resultPreview}
+                            title={record.cell.result === undefined
+                              ? listDisplayText
+                              : `${listDisplayText} → ${record.cell.result}`}
+                          >
+                            <span className={record.cell.result === undefined ? undefined : css.resultRequest}>
+                              {isToolCallOnly(record.cell)
+                                ? null
+                                : toolCallText === undefined
+                                  ? listDisplayText || '—'
+                                  : (
+                                    <>
+                                      <span className={css.toolCallNameTypeface}>
+                                        {toolCallText.name || '—'}
+                                      </span>
+                                      {toolCallText.args !== undefined && (
+                                        <span className={css.toolCallPayload}>
+                                          {toolCallText.args}
+                                        </span>
+                                      )}
+                                    </>
+                                  )}
+                            </span>
+                            {record.cell.result !== undefined && (
+                              <span className={record.cell.isError ? `${css.inlineResult} ${css.error}` : css.inlineResult}>
+                                <span className={css.arrow}>→</span>
+                                <span className={record.cell.result === 'No output'
+                                  ? `${css.inlineResultText} ${css.noOutputText}`
+                                  : css.inlineResultText}
+                                >
+                                  {record.cell.result}
+                                </span>
+                              </span>
+                            )}
+                          </span>
+                        )}
+                  </td>
+                </tr>
+              )
+            })}
+          </tbody>
+        </table>
+      </div>
+      {(selectedRequest !== null
+        || promptSelected
+        || (selected !== undefined && selectedState !== undefined)) && (
+        <aside
+          className={css.details}
+          aria-label="Event details"
+          style={detailsWidth === null ? undefined : { width: detailsWidth }}
+        >
+          <div
+            className={css.detailsResizeHandle}
+            role="separator"
+            aria-label="Resize event details"
+            aria-controls="trajectory-detail-panel"
+            aria-orientation="vertical"
+            tabIndex={0}
+            title="Drag to resize. Double-click to reset."
+            onDoubleClick={() => {
+              setDetailsWidth(null)
+              setToolRequestOffset(null)
+            }}
+            onPointerDown={(event) => {
+              if (event.button !== 0) return
+              const details = event.currentTarget.parentElement
+              if (details === null) return
+              const split = details.parentElement
+              if (split === null) return
+              const splitWidth = split.getBoundingClientRect().width
+              detailsResizeDrag.current = {
+                pointerId: event.pointerId,
+                startX: event.clientX,
+                startWidth: details.getBoundingClientRect().width,
+                splitWidth,
+                startToolRequestOffset: toolRequestOffset ?? (
+                  splitWidth * TOOL_REQUEST_SHARE - defaultToolRequestWidth(splitWidth)
+                ),
+              }
+              event.currentTarget.setPointerCapture(event.pointerId)
+              event.preventDefault()
+            }}
+            onPointerMove={(event) => {
+              const drag = detailsResizeDrag.current
+              if (drag === null || drag.pointerId !== event.pointerId) return
+              const nextDetailsWidth = clampDetailsWidth(
+                drag.startWidth + drag.startX - event.clientX,
+                drag.splitWidth,
+              )
+              setDetailsWidth(nextDetailsWidth)
+              setToolRequestOffset(
+                drag.startToolRequestOffset
+                + (nextDetailsWidth - drag.startWidth) * TOOL_REQUEST_SHARE,
+              )
+            }}
+            onPointerUp={(event) => {
+              if (detailsResizeDrag.current?.pointerId !== event.pointerId) return
+              detailsResizeDrag.current = null
+              event.currentTarget.releasePointerCapture(event.pointerId)
+            }}
+            onPointerCancel={() => {
+              detailsResizeDrag.current = null
+            }}
+            onKeyDown={(event) => {
+              if (event.key !== 'ArrowLeft' && event.key !== 'ArrowRight') return
+              const details = event.currentTarget.parentElement
+              if (details === null) return
+              const split = details.parentElement
+              if (split === null) return
+              const direction = event.key === 'ArrowLeft' ? 1 : -1
+              const currentDetailsWidth = details.getBoundingClientRect().width
+              const splitWidth = split.getBoundingClientRect().width
+              const nextDetailsWidth = clampDetailsWidth(
+                currentDetailsWidth + direction * DETAILS_RESIZE_STEP,
+                splitWidth,
+              )
+              const currentToolRequestOffset = toolRequestOffset ?? (
+                splitWidth * TOOL_REQUEST_SHARE - defaultToolRequestWidth(splitWidth)
+              )
+              setDetailsWidth(nextDetailsWidth)
+              setToolRequestOffset(
+                currentToolRequestOffset
+                + (nextDetailsWidth - currentDetailsWidth) * TOOL_REQUEST_SHARE,
+              )
+              event.preventDefault()
+            }}
+          />
+          <div className={css.detailsHeader}>
+            <div className={css.detailsTitle}>
+              {selectedRequest !== null
+                ? (
+                  <>
+                    <span className={css.requestDetailsDot} aria-hidden="true" />
+                    <span className={css.requestDetailsName}>
+                      Request #{selectedRequest.number}
+                    </span>
+                    <span className={css.detailsLocation}>
+                      {selectedRequestInfo?.purpose === 'compaction'
+                        ? `Compaction · Turn ${selectedRequest.turn}`
+                        : `Turn ${selectedRequest.turn}`}
+                    </span>
+                  </>
+                )
+                : promptSelected
+                  ? (
+                    <>
+                      <span className={`${css.kindTag} ${css.systemNeutral}`}>SYSTEM</span>
+                      <span className={css.detailsLocation}>{selected?.cell.text}</span>
+                    </>
+                  )
+                  : selected !== undefined && (
+                    <>
+                      <span className={`${css.kindTag} ${
+                        selected.cell.kind === 'context'
+                          ? css.contextGreen
+                          : selected.cell.kind === 'compacted'
+                            ? css.compacted
+                            : selected.cell.kind === 'tool'
+                              ? css.toolAmber
+                              : selected.cell.kind === 'message'
+                                ? css.assistantVioletBright
+                                : selected.cell.kind === 'subtool'
+                                  ? css.subtoolAmber
+                                  : css[selected.cell.kind]
+                      }`}
+                      >
+                        {KIND_LABEL[selected.cell.kind]}
+                      </span>
+                      <span className={css.detailsLocation}>
+                        {selected.cell.kind === 'compacted'
+                          ? `Turn ${selected.turn}`
+                          : `Turn ${selected.turn} · ${selected.group}`}
+                      </span>
+                    </>
+                  )}
+            </div>
+            <button
+              type="button"
+              className={css.close}
+              aria-label="Close details"
+              onClick={clearInspectorSelection}
+            >
+              <span aria-hidden="true">×</span>
+            </button>
+          </div>
+          <div className={css.detailTabs} role="tablist" aria-label="Event details">
+            {selectedTabs.map(tab => (
+              <button
+                key={tab.id}
+                id={`trajectory-detail-${tab.id}`}
+                type="button"
+                role="tab"
+                aria-controls="trajectory-detail-panel"
+                aria-selected={activeTab === tab.id}
+                className={activeTab === tab.id ? `${css.detailTab} ${css.detailTabActive}` : css.detailTab}
+                onClick={() => { activateTab(tab.id) }}
+              >
+                {tab.label}
+              </button>
+            ))}
+          </div>
+          <div
+            id="trajectory-detail-panel"
+            className={activeTab === 'overview'
+              ? `${css.detailBody} ${css.detailBodySummary}`
+              : css.detailBody}
+            role="tabpanel"
+            aria-labelledby={`trajectory-detail-${activeTab}`}
+          >
+            {selectedRequest !== null
+              && selectedRequestState !== undefined
+              && activeTab === 'overview' && (
+              <>
+                <dl className={css.overview}>
+                  <div>
+                    <dt>Status</dt>
+                    <dd>{statusLabel(selectedRequestState)}</dd>
+                  </div>
+                  {selectedRequestInfo?.purpose === 'compaction' && (
+                    <div>
+                      <dt>Purpose</dt>
+                      <dd>Compaction</dd>
+                    </div>
+                  )}
+                  {(selectedRequestInfo?.provider
+                    ?? selectedRequestInfo?.requestConfig?.provider) !== undefined && (
+                    <div>
+                      <dt>Provider</dt>
+                      <dd>
+                        {selectedRequestInfo?.provider
+                          ?? selectedRequestInfo?.requestConfig?.provider}
+                      </dd>
+                    </div>
+                  )}
+                  {(selectedRequestInfo?.model
+                    ?? selectedRequestInfo?.requestConfig?.model) !== undefined && (
+                    <div>
+                      <dt>Model</dt>
+                      <dd>
+                        {selectedRequestInfo?.model
+                          ?? selectedRequestInfo?.requestConfig?.model}
+                      </dd>
+                    </div>
+                  )}
+                  <div>
+                    <dt>Tool calls</dt>
+                    <dd>{selectedRequestToolCalls}</dd>
+                  </div>
+                  {selectedRequestSubtoolCalls > 0 && (
+                    <div>
+                      <dt>Subtool calls</dt>
+                      <dd>{selectedRequestSubtoolCalls}</dd>
+                    </div>
+                  )}
+                  {selectedRequestInfo?.error !== undefined && (
+                    <div>
+                      <dt>Error</dt>
+                      <dd>{selectedRequestInfo.error}</dd>
+                    </div>
+                  )}
+                  {selectedRequestInfo?.retry !== undefined && (
+                    <div>
+                      <dt>Retry</dt>
+                      <dd>
+                        Scheduled {selectedRequestInfo.retry}
+                        {selectedRequestInfo.maxRetries === undefined
+                          ? ''
+                          : ` of ${selectedRequestInfo.maxRetries}`}
+                      </dd>
+                    </div>
+                  )}
+                  {selectedRequestInfo?.retryDelayMs !== undefined && (
+                    <div>
+                      <dt>Retry delay</dt>
+                      <dd>{formatDurationMs(selectedRequestInfo.retryDelayMs)}</dd>
+                    </div>
+                  )}
+                  {selectedRequestResult !== undefined && (
+                    <div>
+                      <dt>Result</dt>
+                      <dd className={css.overviewParentLinks}>
+                        <button
+                          type="button"
+                          className={css.overviewHierarchyNavLink}
+                          onClick={() => {
+                            openRecordSummary(selectedRequestResult)
+                          }}
+                        >
+                          <span>
+                            {selectedRequestInfo?.purpose === 'compaction'
+                              ? 'Compacted'
+                              : 'Assistant Message'}
+                          </span>
+                          <IconChevronRightOutline14
+                            className={css.overviewHierarchyJumpIconTight}
+                            size={11}
+                          />
+                        </button>
+                      </dd>
+                    </div>
+                  )}
+                </dl>
+                <div className={css.overviewSections}>
+                  {selectedRequestOptions !== undefined && (
+                    <OverviewSection label="Options" onOpen={() => { activateTab('options') }}>
+                      <RequestOptions options={selectedRequestOptions} preview />
+                    </OverviewSection>
+                  )}
+                  <OverviewSection label="Usage" onOpen={() => { activateTab('usage') }}>
+                    <UsageRows usage={selectedRequestUsage} />
+                  </OverviewSection>
+                  <OverviewSection label="Timing" onOpen={() => { activateTab('timing') }}>
+                    <RequestTiming
+                      assistant={selectedRequestAssistant}
+                      anchor={selectedRequestAnchor}
+                      request={selectedRequestInfo}
+                    />
+                  </OverviewSection>
+                </div>
+              </>
+            )}
+            {selectedRequest !== null && activeTab === 'options' && (
+              <RequestOptions options={selectedRequestOptions} />
+            )}
+            {selectedRequest !== null && activeTab === 'usage' && (
+              <RequestUsagePanel
+                usage={selectedRequestUsage}
+                cumulative={selectedRequestCumulativeUsage}
+              />
+            )}
+            {selectedRequest !== null && activeTab === 'timing' && (
+              <RequestTiming
+                assistant={selectedRequestAssistant}
+                anchor={selectedRequestAnchor}
+                request={selectedRequestInfo}
+              />
+            )}
+            {promptSelected
+              && selectedPreviousPrompt !== undefined
+              && activeTab === 'diff' && (
+              <SystemPromptDiff
+                before={selectedPreviousPrompt}
+                after={selectedPrompt}
+              />
+            )}
+            {promptSelected && activeTab === 'system-prompt' && (
+              selectedPrompt.system === ''
+                ? <p className={css.noPayload}>No system prompt in this request</p>
+                : (
+                  <div className={`${css.markdownPayload} ${css.systemPrompt}`}>
+                    <MarkdownText text={selectedPrompt.system} />
+                  </div>
+                )
+            )}
+            {promptSelected && activeTab === 'tools' && (
+              <ToolCatalog tools={selectedPrompt.tools} />
+            )}
+            {!promptSelected
+              && selected?.cell.kind === 'compacted'
+              && selectedState !== undefined
+              && activeTab === 'overview' && (
+              <>
+                <dl className={css.overview}>
+                  <div>
+                    <dt>Status</dt>
+                    <dd>{statusLabel(selectedState)}</dd>
+                  </div>
+                  <div>
+                    <dt>Duration</dt>
+                    <dd>{formatElapsedSeconds(selected.cell.timeSeconds)}</dd>
+                  </div>
+                  <div>
+                    <dt>Tokens</dt>
+                    <dd>—</dd>
+                  </div>
+                </dl>
+                {selected.cell.outputDetail !== undefined && (
+                  <div className={css.compactedSummary}>
+                    <MarkdownRecordContent
+                      record={selected}
+                      rendered
+                      thinkingExpanded={thinkingExpanded}
+                      onThinkingExpandedChange={setThinkingExpanded}
+                      onOpenCall={openCallSummary}
+                    />
+                  </div>
+                )}
+              </>
+            )}
+            {!promptSelected
+              && selected !== undefined
+              && selected.cell.kind !== 'compacted'
+              && selectedState !== undefined
+              && activeTab === 'overview' && (
+              <>
+                <dl className={css.overview}>
+                  {selected.cell.messageSource !== undefined && (
+                    <div>
+                      <dt>Origin</dt>
+                      <dd className={css.overviewParentLinks}>
+                        <button
+                          type="button"
+                          className={css.overviewHierarchyNavLink}
+                          onClick={() => { activateTab('origin') }}
+                        >
+                          <span>{messageOriginLabel(selected.cell.messageSource)}</span>
+                          <IconChevronRightOutline14
+                            className={css.overviewHierarchyJumpIconTight}
+                            size={11}
+                          />
+                        </button>
+                      </dd>
+                    </div>
+                  )}
+                  {hasSelectedHierarchy && (
+                    <div>
+                      <dt>
+                        {selectedAssistantRequestTarget !== undefined
+                          ? 'Origin'
+                          : 'Hierarchy'}
+                      </dt>
+                      <dd className={css.overviewParentLinks}>
+                        {selectedAssistantRequestTarget !== undefined && (
+                          <button
+                            type="button"
+                            className={css.overviewHierarchyNavLink}
+                            onClick={() => {
+                              selectRequest(selectedAssistantRequestTarget)
+                            }}
+                          >
+                            <span>Request #{selectedAssistantRequestTarget.number}</span>
+                            <IconChevronRightOutline14
+                              className={css.overviewHierarchyJumpIconTight}
+                              size={11}
+                            />
+                          </button>
+                        )}
+                        {selectedParentMessage !== undefined && (
+                          <button
+                            type="button"
+                            className={css.overviewHierarchyNavLink}
+                            onClick={() => { openRecordSummary(selectedParentMessage) }}
+                          >
+                            <span>Assistant Message</span>
+                            <IconChevronRightOutline14
+                              className={css.overviewHierarchyJumpIconTight}
+                              size={11}
+                            />
+                          </button>
+                        )}
+                        {selectedParentTool !== undefined && (
+                          <button
+                            type="button"
+                            className={css.overviewHierarchyNavLink}
+                            onClick={() => { openRecordSummary(selectedParentTool) }}
+                          >
+                            <span>Tool Call</span>
+                            <IconChevronRightOutline14
+                              className={css.overviewHierarchyJumpIconTight}
+                              size={11}
+                            />
+                          </button>
+                        )}
+                      </dd>
+                    </div>
+                  )}
+                  <div>
+                    <dt>Status</dt>
+                    <dd>{statusLabel(selectedState)}</dd>
+                  </div>
+                  {selected.cell.kind === 'message' && (
+                    <TokenRows cell={selected.cell} />
+                  )}
+                  {(selected.cell.kind === 'user' || selected.cell.kind === 'context') && (
+                    <div>
+                      <dt>Duration</dt>
+                      <dd>{formatElapsedSeconds(selected.cell.timeSeconds)}</dd>
+                    </div>
+                  )}
+                </dl>
+                <div className={css.overviewSections}>
+                  {isMarkdownRecord(selected)
+                    ? (
+                      <>
+                        <OverviewSection label="Preview" onOpen={() => { activateTab('rendered') }}>
+                          <MarkdownRecordContent
+                            record={selected}
+                            rendered
+                            preview
+                            thinkingExpanded={thinkingExpanded}
+                            onThinkingExpandedChange={setThinkingExpanded}
+                            onOpenCall={openCallSummary}
+                          />
+                        </OverviewSection>
+                      </>
+                    )
+                    : (
+                      <>
+                        {selected.cell.inputDetail && (
+                          <OverviewSection label="Payload" onOpen={() => { activateTab('input') }}>
+                            <RecordPayload record={selected} direction="input" preview />
+                          </OverviewSection>
+                        )}
+                        {selected.cell.outputDetail && (
+                          <OverviewSection label="Result" onOpen={() => { activateTab('output') }}>
+                            <RecordPayload record={selected} direction="output" preview />
+                          </OverviewSection>
+                        )}
+                        <OverviewSection label="Schema" onOpen={() => { activateTab('schema') }}>
+                          <RecordSchema record={selected} preview />
+                        </OverviewSection>
+                      </>
+                    )}
+                  {selectedAssistantRequestTarget !== undefined && (
+                    <OverviewSection
+                      label="Timing"
+                      onOpen={() => {
+                        selectRequest(selectedAssistantRequestTarget, 'timing')
+                      }}
+                    >
+                      <RecordTiming record={selected} />
+                    </OverviewSection>
+                  )}
+                  {(selected.cell.kind === 'tool' || selected.cell.kind === 'subtool') && (
+                    <OverviewSection label="Timing" onOpen={() => { activateTab('timing') }}>
+                      <RecordTiming record={selected} />
+                    </OverviewSection>
+                  )}
+                </div>
+              </>
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'rendered' && (
+              <MarkdownRecordContent
+                record={selected}
+                rendered
+                thinkingExpanded={thinkingExpanded}
+                onThinkingExpandedChange={setThinkingExpanded}
+                onOpenCall={openCallSummary}
+              />
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'source' && (
+              <MarkdownRecordContent
+                record={selected}
+                rendered={false}
+                thinkingExpanded={thinkingExpanded}
+                onThinkingExpandedChange={setThinkingExpanded}
+                onOpenCall={openCallSummary}
+              />
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'origin' && (
+              <MessageOrigin record={selected} />
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'input' && (
+              <RecordPayload record={selected} direction="input" />
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'output' && (
+              <RecordPayload record={selected} direction="output" />
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'schema' && (
+              <RecordSchema record={selected} />
+            )}
+            {!promptSelected && selected !== undefined && activeTab === 'timing' && (
+              <RecordTiming record={selected} />
+            )}
+          </div>
+        </aside>
+      )}
+    </div>
+  )
+}
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css
new file mode 100644
index 0000000000..a2cac22a71
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css
@@ -0,0 +1,228 @@
+.root {
+  flex: none;
+  border-bottom: 1px solid var(--dsw-alias-border-l2);
+  user-select: none;
+}
+
+.plot {
+  display: grid;
+  grid-template-columns: 44px minmax(0, 1fr);
+  height: 50px;
+  overflow: hidden;
+  background: var(--dsw-alias-bg-layer-2);
+}
+
+.labels {
+  position: relative;
+  border-right: 1px solid var(--dsw-alias-border-l1);
+  color: var(--dsw-alias-label-caption);
+  font: var(--dsw-font-xs-13);
+  font-size: 10px;
+  line-height: 1;
+}
+
+.labels span {
+  position: absolute;
+  right: 3px;
+  display: flex;
+  align-items: center;
+  justify-content: flex-end;
+  height: 8px;
+  text-align: right;
+}
+
+.labels span:nth-child(1) {
+  top: 7px;
+}
+
+.labels span:nth-child(2) {
+  top: 21px;
+}
+
+.labels span:nth-child(3) {
+  top: 35px;
+}
+
+.track {
+  position: relative;
+  overflow: hidden;
+  cursor: crosshair;
+  touch-action: none;
+}
+
+.empty {
+  position: absolute;
+  top: 50%;
+  left: 50%;
+  transform: translate(-50%, -50%);
+  color: var(--dsw-alias-label-caption);
+  font: var(--dsw-font-xs-13);
+}
+
+.track:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: -1px;
+}
+
+.lanes {
+  position: absolute;
+  z-index: 2;
+  inset: 7px 0;
+}
+
+.turnBoundaries {
+  position: absolute;
+  z-index: 3;
+  inset: 0;
+  pointer-events: none;
+}
+
+.turnBoundary {
+  position: absolute;
+  top: 0;
+  bottom: 0;
+  left: var(--trajectory-turn-left);
+  width: 1px;
+  background: var(--dsw-alias-border-l2);
+}
+
+.span {
+  position: absolute;
+  top: calc(var(--trajectory-span-lane) * 14px);
+  left: calc(var(--trajectory-span-left) + 1px);
+  width: max(2px, calc(var(--trajectory-span-width) - 2px));
+  height: 8px;
+  min-width: 2px;
+  border-radius: 1px;
+  background: var(--dsw-alias-label-secondary);
+  opacity: 0.78;
+}
+
+.span[data-timeline-span='user'] {
+  background: var(--dsw-alias-state-business-primary);
+}
+
+.span[data-timeline-span='context'] {
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-success-primary) 68%,
+    var(--dsw-alias-label-secondary)
+  );
+}
+
+.span[data-timeline-span='message'] {
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-brand-primary-new-colorprimary-new-color) 60%,
+    var(--dsw-alias-state-error-secondary)
+  );
+}
+
+.span[data-timeline-span='tool'] {
+  background: var(--dsw-alias-state-warn-label);
+}
+
+.span[data-timeline-span='subtool'] {
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-warn-label) 62%,
+    var(--dsw-alias-label-tertiary)
+  );
+}
+
+.span[data-equal-duration='true'] {
+  width: 8px;
+  min-width: 8px;
+}
+
+.span[data-selected='false'] {
+  opacity: 0.2;
+}
+
+.span[data-current='true'] {
+  z-index: 1;
+  opacity: 1;
+  box-shadow:
+    0 0 0 1px var(--dsw-alias-bg-layer-2),
+    0 0 0 2px var(--dsw-alias-state-business-primary);
+}
+
+.span[data-search-match='false'] {
+  opacity: 0.14;
+}
+
+.selection {
+  position: absolute;
+  z-index: 1;
+  top: 0;
+  bottom: 0;
+  left: var(--trajectory-selection-left);
+  width: var(--trajectory-selection-width);
+  min-width: 1px;
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-business-primary) 12%,
+    transparent
+  );
+  box-shadow:
+    -100vw 0 0 100vw color-mix(in srgb, var(--dsw-alias-bg-layer-1) 58%, transparent),
+    100vw 0 0 100vw color-mix(in srgb, var(--dsw-alias-bg-layer-1) 58%, transparent);
+  pointer-events: none;
+}
+
+.selectionEdges {
+  position: absolute;
+  z-index: 4;
+  top: 0;
+  bottom: 0;
+  left: var(--trajectory-selection-left);
+  width: var(--trajectory-selection-width);
+  min-width: 1px;
+  pointer-events: none;
+}
+
+.hoverLine {
+  position: absolute;
+  z-index: 4;
+  top: 0;
+  bottom: 0;
+  left: clamp(
+    0px,
+    calc(var(--trajectory-hover-left) - 1px),
+    calc(100% - 2px)
+  );
+  width: 2px;
+  background: var(--dsw-alias-state-business-primary);
+  pointer-events: none;
+}
+
+.selectionEdges::before,
+.selectionEdges::after {
+  position: absolute;
+  top: 0;
+  bottom: 0;
+  width: 3px;
+  background: var(--dsw-alias-state-business-primary);
+  content: '';
+}
+
+.selectionEdges::before {
+  left: 0;
+}
+
+.selectionEdges::after {
+  right: 0;
+}
+
+.selectionEdges[data-dragging='true']::before,
+.selectionEdges[data-dragging='true']::after {
+  width: 2px;
+}
+
+.selection[data-dragging='true'] {
+  background: color-mix(
+    in srgb,
+    var(--dsw-alias-state-business-primary) 18%,
+    transparent
+  );
+}
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx
new file mode 100644
index 0000000000..8ad1afefc1
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx
@@ -0,0 +1,372 @@
+/** Chrome-Network-style overview timeline for focusing the trajectory ledger. */
+
+import {
+  memo, useEffect, useMemo, useRef, useState, type CSSProperties, type KeyboardEvent,
+  type PointerEvent, type WheelEvent,
+} from 'react'
+import type { TrajectoryTurnModel } from './layout.ts'
+import {
+  deriveTrajectoryTimeline,
+  formatTimelineOffset,
+  type TrajectoryTimelineMode,
+  type TrajectoryTimeRange,
+} from './timeline.ts'
+import css from './TrajectoryTimeline.module.css'
+
+const MINIMUM_DRAG_PX = 3
+const MINIMUM_ZOOM_OPERATIONS = 4
+
+interface FractionRange {
+  start: number
+  end: number
+}
+
+/** Props for the fixed full-domain overview above the trajectory ledger. */
+export interface TrajectoryTimelineProps {
+  turns: readonly TrajectoryTurnModel[]
+  mode: TrajectoryTimelineMode
+  range: TrajectoryTimeRange | null
+  selectedIndex?: number | null
+  /** Record indexes matching the active ledger search, or null without a query. */
+  searchMatchIndexes?: ReadonlySet<number> | null
+  onRangeChange: (range: TrajectoryTimeRange | null) => void
+  onRecordFocus?: (index: number) => void
+}
+
+function orderedRange(left: number, right: number): FractionRange {
+  return left <= right ? { start: left, end: right } : { start: right, end: left }
+}
+
+function clampFraction(value: number): number {
+  return Math.min(1, Math.max(0, value))
+}
+
+function centeredRange(center: number, width: number): FractionRange {
+  const clampedWidth = Math.min(1, Math.max(0, width))
+  const start = Math.min(
+    Math.max(center - clampedWidth / 2, 0),
+    1 - clampedWidth,
+  )
+  return { start, end: start + clampedWidth }
+}
+
+function rangeFraction(
+  range: TrajectoryTimeRange,
+  start: number,
+  duration: number,
+): FractionRange {
+  return orderedRange(
+    clampFraction((range.start - start) / duration),
+    clampFraction((range.end - start) / duration),
+  )
+}
+
+function LaneLabels() {
+  return (
+    <div className={css.labels} aria-hidden="true">
+      <span>Input</span>
+      <span>Model</span>
+      <span>Tools</span>
+    </div>
+  )
+}
+
+/** Overview renderer with drag ranges, click-sized focus, and Escape reset. */
+export const TrajectoryTimeline = memo(function TrajectoryTimeline({
+  turns,
+  mode,
+  range,
+  selectedIndex = null,
+  searchMatchIndexes = null,
+  onRangeChange,
+  onRecordFocus,
+}: TrajectoryTimelineProps) {
+  const model = useMemo(() => deriveTrajectoryTimeline(turns, mode), [mode, turns])
+  const durationByIndex = useMemo(
+    () => new Map(turns.flatMap(turn =>
+      turn.groups.flatMap(group =>
+        group.cells.flatMap(cell =>
+          cell.timeSeconds === null || !Number.isFinite(cell.timeSeconds)
+            ? []
+            : [[cell.index, Math.max(0, cell.timeSeconds * 1_000)] as const],
+        ),
+      ),
+    )),
+    [turns],
+  )
+  const dragRef = useRef<{ pointerId: number; anchor: number; width: number } | null>(null)
+  const [draft, setDraft] = useState<FractionRange | null>(null)
+  const [hover, setHover] = useState<number | null>(null)
+  const [viewport, setViewport] = useState<TrajectoryTimeRange | null>(null)
+  useEffect(() => {
+    if (
+      model !== null
+      && range !== null
+      && (range.end < model.start || range.start > model.end)
+    ) {
+      onRangeChange(null)
+    }
+  }, [model, onRangeChange, range])
+  useEffect(() => {
+    if (model === null) return
+    setViewport(current =>
+      current !== null && (current.end < model.start || current.start > model.end)
+        ? null
+        : current)
+  }, [model])
+  const fullDuration = Math.max(1, (model?.end ?? 0) - (model?.start ?? 0))
+  const viewportDuration = Math.min(
+    fullDuration,
+    Math.max(1, (viewport?.end ?? 0) - (viewport?.start ?? 0)),
+  )
+  const viewportStart = model === null || viewport === null
+    ? model?.start ?? 0
+    : Math.min(
+      Math.max(viewport.start, model.start),
+      model.end - viewportDuration,
+    )
+  const domainDuration = viewport === null ? fullDuration : viewportDuration
+  const domainStart = viewport === null ? model?.start ?? 0 : viewportStart
+  const committed = model === null || range === null
+    ? null
+    : rangeFraction(range, domainStart, domainDuration)
+  const visibleRange = draft ?? committed
+  const activeRange = draft === null
+    ? range
+    : {
+      start: domainStart + draft.start * domainDuration,
+      end: domainStart + draft.end * domainDuration,
+    }
+
+  if (model === null) {
+    return (
+      <section className={css.root} aria-label="Trajectory timeline">
+        <div className={css.plot}>
+          <LaneLabels />
+          <div className={css.track}>
+            <span className={css.empty}>No timing data</span>
+          </div>
+        </div>
+      </section>
+    )
+  }
+
+  const minimumSelectionFraction = Math.min(
+    1,
+    fullDuration / domainDuration / model.spans.length,
+  )
+
+  const fractionAt = (event: PointerEvent<HTMLDivElement>): number => {
+    const rect = event.currentTarget.getBoundingClientRect()
+    return clampFraction((event.clientX - rect.left) / Math.max(1, rect.width))
+  }
+
+  const commit = (fraction: FractionRange) => {
+    onRangeChange({
+      start: domainStart + fraction.start * domainDuration,
+      end: domainStart + fraction.end * domainDuration,
+    })
+  }
+
+  const onPointerDown = (event: PointerEvent<HTMLDivElement>) => {
+    if (event.button !== 0) return
+    const rect = event.currentTarget.getBoundingClientRect()
+    const anchor = fractionAt(event)
+    setHover(anchor)
+    dragRef.current = { pointerId: event.pointerId, anchor, width: Math.max(1, rect.width) }
+    if (typeof event.currentTarget.setPointerCapture === 'function') {
+      event.currentTarget.setPointerCapture(event.pointerId)
+    }
+    setDraft({ start: anchor, end: anchor })
+  }
+
+  const onPointerMove = (event: PointerEvent<HTMLDivElement>) => {
+    const drag = dragRef.current
+    const fraction = fractionAt(event)
+    setHover(fraction)
+    if (drag === null || drag.pointerId !== event.pointerId) return
+    setDraft(orderedRange(drag.anchor, fraction))
+  }
+
+  const onPointerEnd = (event: PointerEvent<HTMLDivElement>) => {
+    const drag = dragRef.current
+    if (drag === null || drag.pointerId !== event.pointerId) return
+    const point = fractionAt(event)
+    const selected = orderedRange(drag.anchor, point)
+    setHover(point)
+    dragRef.current = null
+    setDraft(null)
+    const click = (selected.end - selected.start) * drag.width < MINIMUM_DRAG_PX
+    const committedRange = selected.end - selected.start < minimumSelectionFraction
+      ? centeredRange(
+        click ? selected.start : (selected.start + selected.end) / 2,
+        minimumSelectionFraction,
+      )
+      : selected
+    commit(committedRange)
+    if (click) {
+      const timelinePoint = domainStart + selected.start * domainDuration
+      const nearest = model.spans.reduce((candidate, span) => {
+        const candidateDistance = timelinePoint < candidate.start
+          ? candidate.start - timelinePoint
+          : timelinePoint > candidate.end ? timelinePoint - candidate.end : 0
+        const spanDistance = timelinePoint < span.start
+          ? span.start - timelinePoint
+          : timelinePoint > span.end ? timelinePoint - span.end : 0
+        return spanDistance < candidateDistance ? span : candidate
+      })
+      onRecordFocus?.(nearest.index)
+    }
+  }
+
+  const onKeyDown = (event: KeyboardEvent<HTMLDivElement>) => {
+    if (event.key !== 'Escape' || range === null) return
+    event.preventDefault()
+    onRangeChange(null)
+  }
+
+  const onPointerCancel = () => {
+    dragRef.current = null
+    setDraft(null)
+    setHover(null)
+  }
+
+  const onWheel = (event: WheelEvent<HTMLDivElement>) => {
+    event.preventDefault()
+    const rect = event.currentTarget.getBoundingClientRect()
+    const anchorFraction =
+      clampFraction((event.clientX - rect.left) / Math.max(1, rect.width))
+    const nextDuration = Math.min(
+      fullDuration,
+      Math.max(
+        Math.min(mode === 'sequence' ? MINIMUM_ZOOM_OPERATIONS : 20, fullDuration),
+        domainDuration * Math.exp(event.deltaY * 0.0015),
+      ),
+    )
+    if (nextDuration >= fullDuration * 0.999) {
+      setViewport(null)
+      return
+    }
+    const anchorTime = domainStart + anchorFraction * domainDuration
+    const nextStart = Math.min(
+      Math.max(anchorTime - anchorFraction * nextDuration, model.start),
+      model.end - nextDuration,
+    )
+    setViewport({ start: nextStart, end: nextStart + nextDuration })
+  }
+
+  return (
+    <section className={css.root} aria-label="Trajectory timeline">
+      <div className={css.plot}>
+        <LaneLabels />
+        <div
+          className={css.track}
+          aria-label="Timeline overview; drag horizontally to focus events"
+          tabIndex={0}
+          onKeyDown={onKeyDown}
+          onPointerDown={onPointerDown}
+          onPointerMove={onPointerMove}
+          onPointerUp={onPointerEnd}
+          onPointerCancel={onPointerCancel}
+          onPointerLeave={() => {
+            if (dragRef.current === null) setHover(null)
+          }}
+          onDoubleClick={(event) => {
+            event.preventDefault()
+            onRangeChange(null)
+          }}
+          onWheel={onWheel}
+          onContextMenu={(event) => {
+            event.preventDefault()
+            onRangeChange(null)
+            setViewport(null)
+          }}
+        >
+          {hover !== null && draft === null && (
+            <div
+              className={css.hoverLine}
+              aria-hidden="true"
+              style={{
+                '--trajectory-hover-left': `${hover * 100}%`,
+              } as CSSProperties}
+            />
+          )}
+          {visibleRange !== null && (
+            <>
+              <div
+                className={css.selection}
+                data-dragging={draft === null ? undefined : 'true'}
+                aria-hidden="true"
+                style={{
+                  '--trajectory-selection-left': `${visibleRange.start * 100}%`,
+                  '--trajectory-selection-width': `${(visibleRange.end - visibleRange.start) * 100}%`,
+                } as CSSProperties}
+              />
+              <div
+                className={css.selectionEdges}
+                data-dragging={draft === null ? undefined : 'true'}
+                aria-hidden="true"
+                style={{
+                  '--trajectory-selection-left': `${visibleRange.start * 100}%`,
+                  '--trajectory-selection-width': `${(visibleRange.end - visibleRange.start) * 100}%`,
+                } as CSSProperties}
+              />
+            </>
+          )}
+          <div className={css.turnBoundaries} aria-hidden="true">
+            {model.turnBoundaries
+              .slice(1)
+              .filter(boundary =>
+                boundary.time >= domainStart
+                && boundary.time <= domainStart + domainDuration)
+              .map(boundary => (
+                <span
+                  className={css.turnBoundary}
+                  data-turn={boundary.turn}
+                  key={boundary.turn}
+                  style={{
+                    '--trajectory-turn-left':
+                      `${(boundary.time - domainStart) / domainDuration * 100}%`,
+                  } as CSSProperties}
+                />
+              ))}
+          </div>
+          <div className={css.lanes} aria-hidden="true">
+            {model.spans
+              .filter(span => span.end >= domainStart && span.start <= domainStart + domainDuration)
+              .map((span) => {
+                const left = (span.start - domainStart) / domainDuration
+                const width = (span.end - span.start) / domainDuration
+                const durationMs = durationByIndex.get(span.index)
+                return (
+                  <span
+                    className={css.span}
+                    data-timeline-span={span.kind}
+                    data-equal-duration={mode === 'time' || undefined}
+                    data-current={span.index === selectedIndex || undefined}
+                    data-search-match={searchMatchIndexes === null
+                      ? undefined
+                      : searchMatchIndexes.has(span.index) ? 'true' : 'false'}
+                    data-selected={activeRange === null
+                      ? undefined
+                      : span.start <= activeRange.end && span.end >= activeRange.start
+                        ? 'true'
+                        : 'false'}
+                    key={span.index}
+                    title={durationMs === undefined
+                      ? span.label
+                      : `${span.label} · ${formatTimelineOffset(durationMs)}`}
+                    style={{
+                      '--trajectory-span-left': `${left * 100}%`,
+                      '--trajectory-span-width': `${Math.max(width * 100, 0.35)}%`,
+                      '--trajectory-span-lane': span.lane,
+                    } as CSSProperties}
+                  />
+                )
+              })}
+          </div>
+        </div>
+      </div>
+    </section>
+  )
+})
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryToolbar.module.css b/packages/client/ui-trajectory/src/client/TrajectoryToolbar.module.css
new file mode 100644
index 0000000000..6e3dd27916
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/TrajectoryToolbar.module.css
@@ -0,0 +1,219 @@
+.root {
+  position: sticky;
+  top: 0;
+  z-index: 4;
+  box-sizing: border-box;
+  width: 100%;
+  height: var(--dsh-trajectory-toolbar-height);
+  border-bottom: 1px solid var(--dsw-alias-border-l2);
+  background: var(--dsw-alias-bg-layer-1);
+}
+
+.inner {
+  display: flex;
+  align-items: center;
+  box-sizing: border-box;
+  width: 100%;
+  height: 100%;
+  padding: 0 6px;
+  gap: 8px;
+}
+
+.actions {
+  display: flex;
+  flex: none;
+  align-items: center;
+  gap: 2px;
+}
+
+.toggle {
+  display: inline-flex;
+  flex: none;
+  align-items: center;
+  height: 20px;
+  padding: 0 7px;
+  gap: 4px;
+  border: 0;
+  border-radius: 3px;
+  color: var(--dsw-alias-label-tertiary);
+  background: transparent;
+  cursor: pointer;
+  font: var(--dsw-font-xxs-12);
+}
+
+.toggle:hover {
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.toggle[aria-pressed='true'] {
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.toggle:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 1px;
+}
+
+.toggleIcon {
+  flex: none;
+  width: 12px;
+  height: 12px;
+  stroke: currentColor;
+  stroke-width: 1.25;
+  stroke-linecap: round;
+  stroke-linejoin: round;
+}
+
+.control {
+  display: inline-flex;
+  flex: none;
+  align-items: center;
+  justify-content: center;
+  box-sizing: border-box;
+  width: 88px;
+  height: 20px;
+  padding: 0 5px;
+  gap: 4px;
+  border: 0;
+  border-radius: 0;
+  color: var(--dsw-alias-label-tertiary);
+  background: transparent;
+  cursor: pointer;
+  font: var(--dsw-font-xxs-12);
+}
+
+.control[hidden] {
+  display: none;
+}
+
+.control:hover:not(:disabled),
+.control[aria-checked='true'],
+.control[aria-pressed='true'] {
+  color: var(--dsw-alias-label-primary);
+}
+
+.control:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 1px;
+}
+
+.control:disabled {
+  color: var(--dsw-alias-label-dimmed);
+  cursor: not-allowed;
+}
+
+.controlTrack {
+  position: relative;
+  display: inline-block;
+  flex: none;
+  width: 20px;
+  height: 10px;
+  border-radius: 5px;
+  background: var(--dsw-alias-border-l2);
+  transition: background-color 120ms var(--ds-ease-in-out);
+}
+
+.controlThumb {
+  position: absolute;
+  top: 2px;
+  left: 2px;
+  width: 6px;
+  height: 6px;
+  border-radius: 50%;
+  background: var(--dsw-alias-bg-layer-1);
+  transition: transform 120ms var(--ds-ease-in-out);
+}
+
+.controlTrack[data-on='true'] {
+  background: var(--dsw-alias-state-business-primary);
+}
+
+.controlTrack[data-on='true'] .controlThumb {
+  transform: translateX(10px);
+}
+
+.action {
+  display: inline-flex;
+  flex: none;
+  align-items: center;
+  height: 20px;
+  padding: 0 5px;
+  gap: 4px;
+  border: 0;
+  border-radius: 3px;
+  color: var(--dsw-alias-label-tertiary);
+  background: transparent;
+  cursor: pointer;
+  font: var(--dsw-font-xxs-12);
+}
+
+.action:hover:not(:disabled) {
+  color: var(--dsw-alias-label-primary);
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.action:focus-visible {
+  outline: 1px solid var(--dsw-alias-state-business-primary);
+  outline-offset: 1px;
+}
+
+.action:disabled {
+  color: var(--dsw-alias-label-dimmed);
+  cursor: not-allowed;
+}
+
+.actionIcon {
+  color: var(--dsw-alias-label-tertiary);
+  font: 14px/14px var(--ds-font-family-code);
+}
+
+.search {
+  display: flex;
+  flex: 0 1 164px;
+  align-items: center;
+  min-width: 84px;
+  height: 22px;
+  margin-left: auto;
+  padding: 0 6px;
+  gap: 4px;
+  border: 1px solid var(--dsw-alias-border-l2);
+  border-radius: 4px;
+  color: var(--dsw-alias-label-caption);
+  background: var(--dsw-alias-bg-layer-2);
+}
+
+.search:hover {
+  border-color: var(--dsw-alias-label-caption);
+}
+
+.search:focus-within {
+  border-color: var(--dsw-alias-state-business-primary);
+  background: var(--dsw-alias-bg-layer-1);
+}
+
+.searchIcon {
+  flex: none;
+}
+
+.searchInput {
+  min-width: 0;
+  width: 100%;
+  padding: 0;
+  border: 0;
+  outline: 0;
+  color: var(--dsw-alias-label-primary);
+  background: transparent;
+  font: var(--dsw-font-xxs-12);
+}
+
+.searchInput::placeholder {
+  color: var(--dsw-alias-label-caption);
+}
+
+.searchInput::-webkit-search-cancel-button {
+  width: 12px;
+  height: 12px;
+  cursor: pointer;
+}
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryToolbar.tsx b/packages/client/ui-trajectory/src/client/TrajectoryToolbar.tsx
new file mode 100644
index 0000000000..09d213c36e
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/TrajectoryToolbar.tsx
@@ -0,0 +1,131 @@
+/** Trajectory toolbar: timeline and ledger fold controls. */
+
+import { IconSearchOutline16 } from '@deepseek-ai/dsh-client-ui-primitives'
+import css from './TrajectoryToolbar.module.css'
+
+export interface TrajectoryToolbarProps {
+  /** Whether timeline blocks use recorded durations instead of equal widths. */
+  actualDuration: boolean
+  /** Select recorded-duration or equal-width blocks. */
+  onActualDurationChange: (actualDuration: boolean) => void
+  /** Whether recorded timing retains idle gaps between user turns. */
+  actualTime: boolean
+  /** Select complete wall-clock timing or idle-compressed timing. */
+  onActualTimeChange: (actualTime: boolean) => void
+  /** Number of turns containing more than one row. */
+  collapsibleTurns: number
+  /** Whether every collapsible turn is currently folded. */
+  allTurnsCollapsed: boolean
+  /** Fold or expand every collapsible turn. */
+  onToggleAllTurns: () => void
+  /** Number of assistant messages followed by tool calls. */
+  collapsibleAssistants: number
+  /** Whether every collapsible assistant's tool calls are currently folded. */
+  allAssistantsCollapsed: boolean
+  /** Fold or expand tool calls under every collapsible assistant. */
+  onToggleAllAssistants: () => void
+  /** Current live ledger search query. */
+  searchQuery: string
+  /** Update the live ledger search query. */
+  onSearchQueryChange: (query: string) => void
+}
+
+/**
+ * Render the sticky trajectory toolbar.
+ * @param props - rendered counts and whole-list fold state.
+ * @returns the toolbar element.
+ */
+export function TrajectoryToolbar({
+  actualDuration,
+  onActualDurationChange,
+  actualTime,
+  onActualTimeChange,
+  collapsibleTurns,
+  allTurnsCollapsed,
+  onToggleAllTurns,
+  collapsibleAssistants,
+  allAssistantsCollapsed,
+  onToggleAllAssistants,
+  searchQuery,
+  onSearchQueryChange,
+}: TrajectoryToolbarProps) {
+  return (
+    <div className={css.root} role="toolbar" aria-label="Trajectory toolbar">
+      <div className={css.inner}>
+        <div className={css.actions}>
+          <button
+            type="button"
+            className={css.toggle}
+            aria-label="Use actual duration"
+            aria-pressed={actualDuration}
+            title={actualDuration ? 'Use equal-width operations' : 'Use actual duration'}
+            onClick={() => { onActualDurationChange(!actualDuration) }}
+          >
+            <svg
+              className={css.toggleIcon}
+              viewBox="0 0 16 16"
+              fill="none"
+              aria-hidden="true"
+            >
+              <circle cx="8" cy="8" r="5.25" />
+              <path d="M8 4.75V8l2.25 1.5" />
+            </svg>
+            Duration
+          </button>
+          <button
+            type="button"
+            className={css.control}
+            role="switch"
+            aria-checked={actualTime}
+            hidden
+            onClick={() => { onActualTimeChange(!actualTime) }}
+          >
+            <span>Actual time</span>
+            <span className={css.controlTrack} data-on={actualTime || undefined} aria-hidden="true">
+              <span className={css.controlThumb} />
+            </span>
+          </button>
+          <button
+            type="button"
+            className={css.action}
+            aria-label={allTurnsCollapsed ? 'Expand turns' : 'Collapse turns'}
+            aria-pressed={allTurnsCollapsed}
+            title={allTurnsCollapsed ? 'Expand turns' : 'Collapse turns'}
+            disabled={collapsibleTurns === 0}
+            onClick={onToggleAllTurns}
+          >
+            <span className={css.actionIcon} aria-hidden="true">
+              {allTurnsCollapsed ? '⊞' : '⊟'}
+            </span>
+            Turns
+          </button>
+          <button
+            type="button"
+            className={css.action}
+            aria-label={allAssistantsCollapsed ? 'Expand calls' : 'Collapse calls'}
+            aria-pressed={allAssistantsCollapsed}
+            title={allAssistantsCollapsed ? 'Expand calls' : 'Collapse calls'}
+            disabled={collapsibleAssistants === 0}
+            onClick={onToggleAllAssistants}
+          >
+            <span className={css.actionIcon} aria-hidden="true">
+              {allAssistantsCollapsed ? '⊞' : '⊟'}
+            </span>
+            Calls
+          </button>
+        </div>
+        <div className={css.search}>
+          <IconSearchOutline16 size={11} className={css.searchIcon} />
+          <input
+            type="search"
+            className={css.searchInput}
+            aria-label="Search trajectory"
+            placeholder="Search"
+            value={searchQuery}
+            onChange={(event) => { onSearchQueryChange(event.currentTarget.value) }}
+          />
+        </div>
+      </div>
+    </div>
+  )
+}
diff --git a/packages/client/ui-trajectory/src/client/TrajectoryView.tsx b/packages/client/ui-trajectory/src/client/TrajectoryView.tsx
index 193b7e9dce..c7f4e193da 100644
--- a/packages/client/ui-trajectory/src/client/TrajectoryView.tsx
+++ b/packages/client/ui-trajectory/src/client/TrajectoryView.tsx
@@ -1,41 +1,509 @@
-// TrajectoryView: sticky Turn sections with Message/Step groups and step cells.
+/** Trajectory view: compact summary over a turn-aware event ledger. */
 
-import { useMemo } from 'react'
+import { useCallback, useEffect, useMemo, useRef, useState } from 'react'
 import type { ConvViewProps } from '@deepseek-ai/dsh-client-ui-conversation/client'
-import { TrajectoryCell } from './TrajectoryCell.tsx'
-import { TrajectoryGroupHeader } from './TrajectoryGroupHeader.tsx'
-import { TrajectoryTurn } from './TrajectoryTurn.tsx'
+import type { InjectFace } from '@deepseek-ai/dsh-client-ui-slots'
+import type {
+  AssistantMessageNode, ConversationContext,
+  SessionHistoryFace,
+} from '@deepseek-ai/dsh-client-runtime/client'
+import {
+  deriveTrajectoryContextBranches, trajectoryBranchContainsRequest,
+} from './context-branches.ts'
+import {
+  TrajectoryTable,
+  type TrajectoryRequestNumber,
+  type TrajectoryUsage,
+} from './TrajectoryTable.tsx'
+import { TrajectoryToolbar } from './TrajectoryToolbar.tsx'
+import { TrajectoryTimeline } from './TrajectoryTimeline.tsx'
 import { deriveTrajectoryLayout } from './layout.ts'
+import {
+  trajectoryTimelineFocusIndexes,
+  type TrajectoryTimelineMode,
+  type TrajectoryTimeRange,
+} from './timeline.ts'
 import css from './views.module.css'
 
-export function TrajectoryView({ useSession }: ConvViewProps) {
-  const nodes = useSession(s => s.nodes)
-  const partial = useSession(s => s.partial)
-  const runningCalls = useSession(s => s.runningCalls)
-  const codeDispatches = useSession(s => s.codeDispatches)
-  const turns = useMemo(
-    () => deriveTrajectoryLayout({ nodes, partial, runningCalls, codeDispatches }),
-    [nodes, partial, runningCalls, codeDispatches],
-  )
-  if (turns.length === 0) {
-    return <div className={css.root}><p className={css.empty}>暂无轨迹数据</p></div>
+const EMPTY_IDS: ReadonlySet<number> = new Set()
+
+/** Session-history paging needed by the event-complete trajectory view. */
+export interface TrajectoryViewInjected {
+  hooks: { history: SessionHistoryFace }
+  loadAllHistory: (signal: AbortSignal) => Promise<void>
+}
+
+interface UsageLike {
+  inputTokens?: number
+  cacheReadTokens?: number
+  cacheWriteTokens?: number
+  outputTokens?: number
+  reasoningTokens?: number
+}
+
+function requestUsage(value: unknown): TrajectoryUsage | undefined {
+  const usage = value as UsageLike | undefined
+  if (usage === undefined) return undefined
+  return {
+    ...(usage.inputTokens === undefined ? {} : { input: usage.inputTokens }),
+    ...(usage.cacheReadTokens === undefined ? {} : { cacheRead: usage.cacheReadTokens }),
+    ...(usage.cacheWriteTokens === undefined ? {} : { cacheWrite: usage.cacheWriteTokens }),
+    ...(usage.outputTokens === undefined ? {} : { output: usage.outputTokens }),
+    ...(usage.reasoningTokens === undefined ? {} : { reasoning: usage.reasoningTokens }),
   }
+}
+
+function addUsage(
+  total: TrajectoryUsage | undefined,
+  usage: TrajectoryUsage | undefined,
+): TrajectoryUsage | undefined {
+  if (usage === undefined) return total
+  return {
+    ...(total?.input === undefined && usage.input === undefined
+      ? {}
+      : { input: (total?.input ?? 0) + (usage.input ?? 0) }),
+    ...(total?.cacheRead === undefined && usage.cacheRead === undefined
+      ? {}
+      : { cacheRead: (total?.cacheRead ?? 0) + (usage.cacheRead ?? 0) }),
+    ...(total?.cacheWrite === undefined && usage.cacheWrite === undefined
+      ? {}
+      : { cacheWrite: (total?.cacheWrite ?? 0) + (usage.cacheWrite ?? 0) }),
+    ...(total?.output === undefined && usage.output === undefined
+      ? {}
+      : { output: (total?.output ?? 0) + (usage.output ?? 0) }),
+    ...(total?.reasoning === undefined && usage.reasoning === undefined
+      ? {}
+      : { reasoning: (total?.reasoning ?? 0) + (usage.reasoning ?? 0) }),
+  }
+}
+
+function searchableJson(value: unknown): string {
+  if (value === undefined) return ''
+  try {
+    return JSON.stringify(value)
+  } catch {
+    return ''
+  }
+}
+
+function searchMatches(
+  turns: ReturnType<typeof deriveTrajectoryLayout>,
+  query: string,
+): ReadonlySet<number> | null {
+  const terms = query.trim().toLocaleLowerCase().split(/\s+/).filter(Boolean)
+  if (terms.length === 0) return null
+  const matches = new Set<number>()
+  for (const turn of turns) {
+    for (const group of turn.groups) {
+      for (const cell of group.cells) {
+        if (cell.requestOnly === true) continue
+        const blocks = [
+          ...(cell.sourceBlocks ?? []),
+          ...(cell.outputBlocks ?? []),
+        ]
+        const text = [
+          `turn ${turn.turn}`,
+          group.title,
+          cell.kind,
+          cell.kind === 'message' ? 'assistant' : undefined,
+          cell.text,
+          cell.inputDetail,
+          cell.outputDetail,
+          cell.thinkingDetail,
+          cell.schemaDetail,
+          cell.result,
+          cell.callId,
+          ...blocks.flatMap(block => [
+            block.type,
+            block.content,
+            block.callId,
+            block.toolName,
+            block.imageAlt,
+          ]),
+          searchableJson(cell.messageSource),
+          searchableJson(cell.promptDetail),
+          searchableJson(cell.previousPromptDetail),
+        ].filter((value): value is string => typeof value === 'string')
+          .join('\n')
+          .toLocaleLowerCase()
+        if (terms.every(term => text.includes(term))) matches.add(cell.index)
+      }
+    }
+  }
+  return matches
+}
+
+export function TrajectoryView({
+  useHistory, loadAllHistory,
+}: ConvViewProps & InjectFace<TrajectoryViewInjected>) {
+  const [collapsedTurns, setCollapsedTurns] = useState<ReadonlySet<number>>(EMPTY_IDS)
+  const [collapsedAssistants, setCollapsedAssistants] =
+    useState<ReadonlySet<number>>(EMPTY_IDS)
+  const [timelineSelection, setTimelineSelection] = useState<{
+    branchId: number
+    range: TrajectoryTimeRange
+  } | null>(null)
+  const [actualDuration, setActualDuration] = useState(false)
+  const [actualTime, setActualTime] = useState(false)
+  const [searchQuery, setSearchQuery] = useState('')
+  const [selectedTimelineIndex, setSelectedTimelineIndex] = useState<number | null>(null)
+  const ledgerRef = useRef<HTMLDivElement>(null)
+  const inspection = useHistory(snapshot => snapshot.inspection)
+  const nodes = inspection.eventNodes
+  const partial = inspection.partial
+  const runningCalls = inspection.runningCalls
+  const codeDispatches = inspection.codeDispatches
+  const loadAllHistoryRef = useRef(loadAllHistory)
+  loadAllHistoryRef.current = loadAllHistory
+  useEffect(() => {
+    const controller = new AbortController()
+    void loadAllHistoryRef.current(controller.signal)
+    return () => { controller.abort() }
+  }, [])
+  const requests = inspection.requests
+  const callSchemas = inspection.callSchemas
+  const contexts = useMemo<readonly ConversationContext[]>(
+    () => inspection.contexts.length === 0
+      ? [{ id: 0, nodes }]
+      : inspection.contexts,
+    [inspection, nodes],
+  )
+  const branches = useMemo(
+    () => deriveTrajectoryContextBranches(contexts),
+    [contexts],
+  )
+  const currentBranch = branches.at(-1)
+  if (currentBranch === undefined) throw new Error('trajectory branch projection must not be empty')
+  const selectedNodes = useMemo(() => {
+    const selected = new Map(currentBranch.nodes.map(node => [node.seq, node]))
+    for (const node of inspection.interruptedNodes) {
+      selected.set(node.seq, node)
+    }
+    return [...selected.values()].sort((left, right) => left.seq - right.seq)
+  }, [currentBranch, inspection])
+  const selectedRequests = useMemo(
+    () => requests.filter(request =>
+      trajectoryBranchContainsRequest(currentBranch, request),
+    ),
+    [currentBranch, requests],
+  )
+  const globalRequestNumbers = useMemo<readonly TrajectoryRequestNumber[]>(() => {
+    const assistantsByStep = new Map<string, AssistantMessageNode>()
+    for (const context of contexts) {
+      for (const node of context.nodes) {
+        if (node.kind !== 'assistant' || node.step <= 0) continue
+        assistantsByStep.set(`${node.turn}\u0000${node.step}`, node)
+      }
+    }
+    for (const node of nodes) {
+      if (node.kind !== 'assistant' || node.step <= 0) continue
+      assistantsByStep.set(`${node.turn}\u0000${node.step}`, node)
+    }
+    const requestsByStep = new Map(
+      requests
+        .filter(request => request.purpose === 'assistant')
+        .map(request => [
+          `${request.turn}\u0000${request.step}`,
+          request,
+        ]),
+    )
+    const orderedRequests = [
+      ...requests.map(request => ({
+        seq: request.startSeq,
+        request,
+        node: request.purpose === 'assistant'
+          ? assistantsByStep.get(`${request.turn}\u0000${request.step}`)
+          : undefined,
+      })),
+      ...[...assistantsByStep.entries()].flatMap(([key, node]) =>
+        requestsByStep.has(key)
+          ? []
+          : [{
+            seq: node.seq,
+            request: undefined,
+            node,
+          }],
+      ),
+    ].sort((left, right) => left.seq - right.seq)
+    const numbered: TrajectoryRequestNumber[] = []
+    let cumulativeUsage: TrajectoryUsage | undefined
+    for (const [index, entry] of orderedRequests.entries()) {
+      const usage = requestUsage(entry.request?.usage ?? entry.node?.usage)
+      cumulativeUsage = addUsage(cumulativeUsage, usage)
+      if (entry.request?.purpose !== 'compaction') {
+        const request = entry.request
+        const node = entry.node
+        const turn = request?.turn ?? node?.turn
+        const step = request?.step ?? node?.step
+        if (turn === undefined || step === undefined) continue
+        const provider = request?.provenance?.provider ?? node?.provenance?.provider
+        const model = request?.provenance?.model ?? node?.provenance?.model
+        const requestConfig = request?.requestConfig ?? node?.requestConfig
+        numbered.push({
+          seq: entry.seq,
+          turn,
+          step,
+          group: `Step ${step}`,
+          number: index + 1,
+          ...(request?.status === undefined ? {} : { status: request.status }),
+          ...(request?.startedAt === undefined ? {} : { startedAt: request.startedAt }),
+          ...(request?.completedAt === undefined ? {} : { completedAt: request.completedAt }),
+          ...(request?.error === undefined ? {} : { error: request.error }),
+          ...(request?.resultSeq === undefined ? {} : { resultSeq: request.resultSeq }),
+          ...(request?.retry === undefined ? {} : { retry: request.retry }),
+          ...(request?.maxRetries === undefined ? {} : { maxRetries: request.maxRetries }),
+          ...(request?.retryDelayMs === undefined
+            ? {}
+            : { retryDelayMs: request.retryDelayMs }),
+          ...(provider === undefined ? {} : { provider }),
+          ...(model === undefined ? {} : { model }),
+          ...(requestConfig === undefined ? {} : { requestConfig }),
+          ...(usage === undefined ? {} : { usage }),
+          ...(cumulativeUsage === undefined ? {} : { cumulativeUsage }),
+        })
+        continue
+      }
+      const request = entry.request
+      numbered.push({
+        seq: request.startSeq,
+        turn: request.turn,
+        step: 0,
+        group: `Compaction ${request.startSeq}`,
+        number: index + 1,
+        purpose: 'compaction',
+        status: request.status,
+        startedAt: request.startedAt,
+        completedAt: request.completedAt,
+        ...(request.error === undefined ? {} : { error: request.error }),
+        resultSeq: request.startSeq,
+        ...(request.provenance?.provider === undefined
+          ? {}
+          : { provider: request.provenance.provider }),
+        ...(request.provenance?.model === undefined
+          ? {}
+          : { model: request.provenance.model }),
+        ...(request.requestConfig === undefined ? {} : { requestConfig: request.requestConfig }),
+        ...(usage === undefined ? {} : { usage }),
+        ...(cumulativeUsage === undefined ? {} : { cumulativeUsage }),
+      })
+    }
+
+    if (partial !== null && partial.step > 0) {
+      const key = `${partial.turn}\u0000${partial.step}`
+      const recorded = numbered.some(request =>
+        `${request.turn}\u0000${request.step}` === key,
+      )
+      if (!recorded) {
+        numbered.push({
+          turn: partial.turn,
+          step: partial.step,
+          group: `Step ${partial.step}`,
+          number: orderedRequests.length + 1,
+          ...(currentBranch.latest.prompt?.config.provider === undefined
+            ? {}
+            : { provider: currentBranch.latest.prompt.config.provider }),
+          ...(currentBranch.latest.prompt?.config.model === undefined
+            ? {}
+            : { model: currentBranch.latest.prompt.config.model }),
+          ...(currentBranch.latest.prompt?.config === undefined
+            ? {}
+            : { requestConfig: currentBranch.latest.prompt.config }),
+          ...(cumulativeUsage === undefined ? {} : { cumulativeUsage }),
+        })
+      }
+    }
+    return numbered
+  }, [
+    contexts, currentBranch.latest.prompt, nodes, partial, requests,
+  ])
+  const requestNumbers = globalRequestNumbers
+  const turns = useMemo(
+    () => deriveTrajectoryLayout({
+      nodes: selectedNodes,
+      partial,
+      runningCalls,
+      requests: selectedRequests,
+      callSchemas,
+      codeDispatches,
+    }),
+    [
+      selectedNodes, partial, runningCalls, selectedRequests, callSchemas, codeDispatches,
+    ],
+  )
+  const timelineMode: TrajectoryTimelineMode = actualDuration
+    ? actualTime ? 'actual' : 'duration'
+    : actualTime ? 'time' : 'sequence'
+  const searchMatchIndexes = useMemo(
+    () => searchMatches(turns, searchQuery),
+    [searchQuery, turns],
+  )
+  const timelineRange = timelineSelection?.branchId === currentBranch.id
+    ? timelineSelection.range
+    : null
+  const timelineFocusIndexes = useMemo(
+    () => timelineRange === null
+      ? null
+      : trajectoryTimelineFocusIndexes(turns, timelineRange, timelineMode),
+    [timelineMode, timelineRange, turns],
+  )
+  const handleRecordSelect = useCallback((index: number) => {
+    if (
+      timelineFocusIndexes !== null
+      && !timelineFocusIndexes.has(index)
+    ) {
+      setTimelineSelection(null)
+    }
+  }, [timelineFocusIndexes])
+  useEffect(() => {
+    if (timelineFocusIndexes === null || timelineFocusIndexes.size === 0) return
+    const ledger = ledgerRef.current
+    if (ledger === null) return
+    const focusedRows = [
+      ...ledger.querySelectorAll<HTMLElement>('tr[data-timeline-focus="inside"]'),
+    ]
+    const first = focusedRows.at(0)
+    const last = focusedRows.at(-1)
+    if (first === undefined || last === undefined) return
+    const focusHeight =
+      last.getBoundingClientRect().bottom - first.getBoundingClientRect().top
+    if (focusHeight > ledger.clientHeight) {
+      if (typeof first.scrollIntoView === 'function') {
+        first.scrollIntoView({ behavior: 'smooth', block: 'start' })
+      }
+      return
+    }
+    const middle = focusedRows[Math.floor((focusedRows.length - 1) / 2)]
+    if (middle !== undefined && typeof middle.scrollIntoView === 'function') {
+      middle.scrollIntoView({ behavior: 'smooth', block: 'center' })
+    }
+  }, [timelineFocusIndexes])
+  const collapsibleTurnIds = useMemo(
+    () => turns
+      .filter(turn =>
+        turn.groups.reduce(
+          (count, group) =>
+            count + group.cells.filter(cell =>
+              cell.requestOnly !== true && cell.kind !== 'system').length,
+          0,
+        ) > 1)
+      .map(turn => turn.turn),
+    [turns],
+  )
+  const allTurnsCollapsed = collapsibleTurnIds.length > 0
+    && collapsibleTurnIds.every(turn => collapsedTurns.has(turn))
+  const collapsibleAssistantIds = useMemo(() => {
+    const ids: number[] = []
+    for (const turn of turns) {
+      const cells = turn.groups.flatMap(group => group.cells)
+      for (let i = 0; i < cells.length; i++) {
+        const cell = cells[i]
+        if (cell?.kind !== 'message') continue
+        const next = cells[i + 1]
+        if (next?.kind === 'tool' || next?.kind === 'subtool') ids.push(cell.index)
+      }
+    }
+    return ids
+  }, [turns])
+  const allAssistantsCollapsed = collapsibleAssistantIds.length > 0
+    && collapsibleAssistantIds.every(index => collapsedAssistants.has(index))
+
+  const toggleTurn = (turn: number) => {
+    setCollapsedTurns((current) => {
+      const collapsed = new Set(current)
+      if (collapsed.has(turn)) collapsed.delete(turn)
+      else collapsed.add(turn)
+      return collapsed
+    })
+  }
+
+  const toggleAllTurns = () => {
+    setCollapsedTurns((current) => {
+      const collapsed = new Set(current)
+      if (allTurnsCollapsed) {
+        for (const turn of collapsibleTurnIds) collapsed.delete(turn)
+      } else {
+        for (const turn of collapsibleTurnIds) collapsed.add(turn)
+      }
+      return collapsed
+    })
+  }
+
+  const toggleAssistant = (index: number) => {
+    setCollapsedAssistants((current) => {
+      const collapsed = new Set(current)
+      if (collapsed.has(index)) collapsed.delete(index)
+      else collapsed.add(index)
+      return collapsed
+    })
+  }
+
+  const toggleAllAssistants = () => {
+    setCollapsedAssistants((current) => {
+      const collapsed = new Set(current)
+      if (allAssistantsCollapsed) {
+        for (const index of collapsibleAssistantIds) collapsed.delete(index)
+      } else {
+        for (const index of collapsibleAssistantIds) collapsed.add(index)
+      }
+      return collapsed
+    })
+  }
+
   return (
     <div className={css.root}>
-      {turns.map(turn => (
-        <TrajectoryTurn key={turn.turn} turn={turn.turn}>
-          {turn.groups.flatMap(group => [
-            <TrajectoryGroupHeader
-              key={`${group.title}-h`}
-              title={group.title}
-              {...(group.description !== undefined ? { description: group.description } : {})}
-            />,
-            ...group.cells.map(cell => (
-              <TrajectoryCell key={cell.index} {...cell} />
-            )),
-          ])}
-        </TrajectoryTurn>
-      ))}
+      <TrajectoryToolbar
+        actualDuration={actualDuration}
+        onActualDurationChange={(nextActualDuration) => {
+          setActualDuration(nextActualDuration)
+          setTimelineSelection(null)
+        }}
+        actualTime={actualTime}
+        onActualTimeChange={(nextActualTime) => {
+          setActualTime(nextActualTime)
+          setTimelineSelection(null)
+        }}
+        collapsibleTurns={collapsibleTurnIds.length}
+        allTurnsCollapsed={allTurnsCollapsed}
+        onToggleAllTurns={toggleAllTurns}
+        collapsibleAssistants={collapsibleAssistantIds.length}
+        allAssistantsCollapsed={allAssistantsCollapsed}
+        onToggleAllAssistants={toggleAllAssistants}
+        searchQuery={searchQuery}
+        onSearchQueryChange={setSearchQuery}
+      />
+      <TrajectoryTimeline
+        turns={turns}
+        mode={timelineMode}
+        range={timelineRange}
+        selectedIndex={selectedTimelineIndex}
+        searchMatchIndexes={searchMatchIndexes}
+        onRangeChange={(range) => {
+          setTimelineSelection(range === null ? null : { branchId: currentBranch.id, range })
+        }}
+        onRecordFocus={(index) => {
+          const row = ledgerRef.current
+            ?.querySelector<HTMLElement>(`tr[data-record-index="${index}"]`)
+          if (row !== undefined && row !== null && typeof row.scrollIntoView === 'function') {
+            row.scrollIntoView({ behavior: 'smooth', block: 'center' })
+          }
+        }}
+      />
+      <div ref={ledgerRef} className={css.ledger}>
+        <TrajectoryTable
+          key={currentBranch.id}
+          requestNumbers={requestNumbers}
+          turns={turns}
+          timelineFocusIndexes={timelineFocusIndexes}
+          searchMatchIndexes={searchMatchIndexes}
+          onSelectedIndexChange={setSelectedTimelineIndex}
+          onRecordSelect={handleRecordSelect}
+          onClearSelection={() => { setTimelineSelection(null) }}
+          collapsedTurns={collapsedTurns}
+          onToggleTurn={toggleTurn}
+          collapsedAssistants={collapsedAssistants}
+          onToggleAssistant={toggleAssistant}
+        />
+      </div>
     </div>
   )
 }
diff --git a/packages/client/ui-trajectory/src/client/WaterfallView.tsx b/packages/client/ui-trajectory/src/client/WaterfallView.tsx
deleted file mode 100644
index a380c8378d..0000000000
--- a/packages/client/ui-trajectory/src/client/WaterfallView.tsx
+++ /dev/null
@@ -1,75 +0,0 @@
-// WaterfallView: span stats header over per-turn node-count lanes (P-I
-// stand-in for duration lanes; deviation ledger #3). run_code turns
-// additionally draw TRUTHFUL sub-call lanes: the dispatch start/settle pair
-// carries per-sub-call wall time, so each sub-span's width is its real
-// duration against the parent turn's dispatch window.
-
-import { useMemo } from 'react'
-import type { ConvViewProps } from '@deepseek-ai/dsh-client-ui-conversation/client'
-import { deriveSpans, deriveSubSpans } from './spans.ts'
-import { TrajectoryStatsHeader } from './TrajectoryStatsHeader.tsx'
-import css from './views.module.css'
-
-/** Bar width scale: px per node, clamped so tiny windows still show a bar. */
-const PX_PER_NODE = 14
-const MIN_BAR_PX = 8
-/** Sub-span lane width budget (the parent window scales into this). */
-const SUB_LANE_PX = 220
-
-/** Optional density override (test/standalone knob; the register site passes nothing). */
-export interface WaterfallExtraProps {
-  /** Bar-lane density in px per node; defaults to 14. */
-  pxPerNode?: number
-}
-
-export function WaterfallView({ useSession, pxPerNode }: ConvViewProps & WaterfallExtraProps) {
-  const scale = pxPerNode ?? PX_PER_NODE
-  const nodes = useSession(s => s.nodes)
-  const codeDispatches = useSession(s => s.codeDispatches)
-  const spans = useMemo(() => deriveSpans(nodes), [nodes])
-  const subSpans = useMemo(() => deriveSubSpans(nodes, codeDispatches), [nodes, codeDispatches])
-  if (spans.length === 0) return <div className={css.root}><p className={css.empty}>暂无瀑布数据</p></div>
-  return (
-    <>
-      <TrajectoryStatsHeader useSession={useSession} />
-      <div className={css.root}>
-        {spans.map((span, i) => (
-          <div key={span.turn}>
-            <div className={css.row} style={{ paddingLeft: i * 12 }}>
-              <span className={css.turnTag}>turn {span.turn}</span>
-              <span
-                className={css.bar}
-                style={{ width: Math.max(span.nodes * scale, MIN_BAR_PX) }}
-                title={`${span.nodes} nodes`}
-              />
-              {span.calls > 0 && (
-                <span
-                  className={`${css.bar} ${css.barCalls}`}
-                  style={{ width: Math.max(span.calls * scale, MIN_BAR_PX) }}
-                  title={`${span.calls} tool calls`}
-                />
-              )}
-            </div>
-            {(subSpans.get(span.turn) ?? []).map(lane => (
-              <div key={lane.callId} className={css.subRow} data-subspan style={{ paddingLeft: i * 12 + 24 }}>
-                <span className={css.subTag}>{lane.name}</span>
-                <span
-                  className={`${css.bar} ${css.barSub}`}
-                  data-timing={lane.timing}
-                  style={{
-                    marginLeft: Math.round(lane.offsetFraction * SUB_LANE_PX),
-                    width: Math.max(Math.round(lane.widthFraction * SUB_LANE_PX), 4),
-                  }}
-                  title={lane.timing === 'measured'
-                    /* durationMs is non-null exactly when timing is measured. */
-                    ? `${lane.name} · ${((lane.durationMs ?? 0) / 1000).toFixed(2)}s`
-                    : lane.timing === 'running' ? `${lane.name} · running` : `${lane.name} · duration unknown`}
-                />
-              </div>
-            ))}
-          </div>
-        ))}
-      </div>
-    </>
-  )
-}
diff --git a/packages/client/ui-trajectory/src/client/context-branches.ts b/packages/client/ui-trajectory/src/client/context-branches.ts
new file mode 100644
index 0000000000..239953b5fd
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/context-branches.ts
@@ -0,0 +1,113 @@
+/** Rewind-delimited trajectory branches assembled across surface rewrites. */
+
+import type {
+  ConversationContext, ConversationNode, RequestView,
+} from '@deepseek-ai/dsh-client-runtime/client'
+
+/** One continuous context branch; compactions stay inline while rewinds start a successor branch. */
+export interface TrajectoryContextBranch {
+  id: number
+  contexts: readonly ConversationContext[]
+  latest: ConversationContext
+  nodes: readonly ConversationNode[]
+  /** Seq that opened this branch; earlier requests require retained surface provenance. */
+  startSeq: number
+  /** Exact pre-rewind surface records inherited by this branch. */
+  retainedSurfaceSeqs: ReadonlySet<number>
+}
+
+interface MutableBranch {
+  id: number
+  contexts: ConversationContext[]
+  latest: ConversationContext
+  nodes: Map<number, ConversationNode>
+  startSeq: number
+  retainedSurfaceSeqs: Set<number>
+}
+
+function isCompactionCheckpoint(node: ConversationNode): boolean {
+  if (node.kind !== 'context') return false
+  const source = node.source
+  return typeof source === 'object'
+    && source !== null
+    && 'kind' in source
+    && source.kind === 'plugin'
+    && 'plugin' in source
+    && source.plugin === 'compact'
+}
+
+/**
+ * Join context generations across compaction/rewrite operations and split only at rewind.
+ * @param contexts - Append-only context generations from the runtime fold.
+ * @returns Rewind-delimited branches in creation order.
+ */
+export function deriveTrajectoryContextBranches(
+  contexts: readonly ConversationContext[],
+): readonly TrajectoryContextBranch[] {
+  const mutable: MutableBranch[] = []
+  for (const context of contexts) {
+    const startsBranch = mutable.length === 0 || context.origin === 'rewind'
+    if (startsBranch) {
+      const previous = mutable.at(-1)
+      const retainedSurfaceSeqs = new Set(
+        context.nodes
+          .filter(node =>
+            context.originSeq !== undefined && node.seq < context.originSeq,
+          )
+          .map(node => node.seq),
+      )
+      const inheritedNodes = previous === undefined
+        ? []
+        : [...previous.nodes.values()].filter(node =>
+          retainedSurfaceSeqs.has(node.seq),
+        )
+      mutable.push({
+        id: context.id,
+        contexts: [context],
+        latest: context,
+        nodes: new Map(
+          [...inheritedNodes, ...context.nodes.filter(node => !isCompactionCheckpoint(node))]
+            .map(node => [node.seq, node]),
+        ),
+        startSeq: context.originSeq ?? Number.NEGATIVE_INFINITY,
+        retainedSurfaceSeqs,
+      })
+      continue
+    }
+    const branch = mutable.at(-1)
+    if (branch === undefined) continue
+    branch.contexts.push(context)
+    branch.latest = context
+    for (const node of context.nodes) {
+      if (!isCompactionCheckpoint(node)) branch.nodes.set(node.seq, node)
+    }
+  }
+  return mutable.map(branch => ({
+    id: branch.id,
+    contexts: branch.contexts,
+    latest: branch.latest,
+    nodes: [...branch.nodes.values()].sort((left, right) => left.seq - right.seq),
+    startSeq: branch.startSeq,
+    retainedSurfaceSeqs: branch.retainedSurfaceSeqs,
+  }))
+}
+
+/**
+ * Test whether a provider request belongs to one rewind branch.
+ * @param branch - Branch carrying exact inherited surface provenance.
+ * @param request - Provider request to classify.
+ * @returns Whether the request began on this branch or produced a retained surface record.
+ */
+export function trajectoryBranchContainsRequest(
+  branch: TrajectoryContextBranch,
+  request: RequestView,
+): boolean {
+  if (request.startSeq >= branch.startSeq) return true
+  return (
+    request.resultSeq !== undefined
+    && branch.retainedSurfaceSeqs.has(request.resultSeq)
+  ) || (
+    request.replacementSeq !== undefined
+    && branch.retainedSurfaceSeqs.has(request.replacementSeq)
+  )
+}
diff --git a/packages/client/ui-trajectory/src/client/index.ts b/packages/client/ui-trajectory/src/client/index.ts
index c145fa3b3f..347b5c9cec 100644
--- a/packages/client/ui-trajectory/src/client/index.ts
+++ b/packages/client/ui-trajectory/src/client/index.ts
@@ -1,13 +1,13 @@
 /**
- * Browser trajectory plugin contributing two entries to the conversation
- * view slot without defining a service.
+ * Browser trajectory plugin contributing one entry to the conversation view
+ * slot without defining a service.
  */
 import type { Context } from 'cordis'
+import type { SessionId } from '@deepseek-ai/dsh-client-runtime/client'
 // Type-only: the 'conversation.view' SlotMap row (declared by the slot's
 // owning package) must be in the program for the register calls to type.
 import type {} from '@deepseek-ai/dsh-client-ui-conversation/client'
-import { TrajectoryView } from './TrajectoryView.tsx'
-import { WaterfallView } from './WaterfallView.tsx'
+import { TrajectoryView, type TrajectoryViewInjected } from './TrajectoryView.tsx'
 
 /**
  * Required services (cordis fiber inject). 'conversation' is an ordering
@@ -16,17 +16,25 @@ import { WaterfallView } from './WaterfallView.tsx'
  * into an undeclared slot throws — service waiting is what orders this
  * apply after the declaring one.
  */
-export const inject = ['slots', 'conversation']
+export const inject = ['slots', 'conversation', 'sessionHistory']
 
 /**
- * Client plugin body: register the trajectory and waterfall view tabs. The
- * registrations ride the slot service's effect wrapper (plugin unload
- * removes both tabs).
+ * Client plugin body: register the trajectory view tab. The registration
+ * rides the slot service's effect wrapper, so plugin unload removes the tab.
  * @param ctx - client root context.
  */
 export function apply(ctx: Context): void {
-  ctx.slots.register(
-    { name: 'conversation.view', id: 'trajectory', order: 10, label: 'Trajectory' }, TrajectoryView)
-  ctx.slots.register(
-    { name: 'conversation.view', id: 'waterfall', order: 20, label: 'Waterfall' }, WaterfallView)
+  ctx.slots.register({
+    name: 'conversation.view',
+    id: 'trajectory',
+    order: 10,
+    label: 'Trajectory',
+    inject: (sessionId: SessionId): TrajectoryViewInjected => {
+      const history = ctx.sessionHistory.source(sessionId)
+      return {
+        hooks: { history },
+        loadAllHistory: signal => history.loadAll(signal),
+      }
+    },
+  }, TrajectoryView)
 }
diff --git a/packages/client/ui-trajectory/src/client/layout.ts b/packages/client/ui-trajectory/src/client/layout.ts
index 37c86f6eb4..6e69d7ae31 100644
--- a/packages/client/ui-trajectory/src/client/layout.ts
+++ b/packages/client/ui-trajectory/src/client/layout.ts
@@ -3,12 +3,19 @@
  * own-duration times, in-flight partial/runningCalls, and group descriptions.
  */
 import type {
+  AssistantBlock,
   AssistantMessageNode,
   CodeSubCall,
   ConversationSnapshot,
+  RequestInspectionSnapshot,
+  RequestPromptChange,
+  RequestView,
   ToolResultNode,
 } from '@deepseek-ai/dsh-client-runtime/client'
-import type { TrajectoryCellProps } from './TrajectoryCell.tsx'
+import type {
+  TrajectoryCellProps,
+  TrajectorySourceBlock,
+} from './trajectory-record.ts'
 
 /** One Message or Step group inside a turn. */
 export interface TrajectoryGroupModel {
@@ -28,12 +35,16 @@ export interface TrajectoryLayoutInput {
   nodes: ConversationSnapshot['nodes']
   partial: ConversationSnapshot['partial']
   runningCalls: ConversationSnapshot['runningCalls']
+  requests?: readonly RequestView[]
+  callSchemas?: RequestInspectionSnapshot['callSchemas']
   /** run_code sub-dispatches by parent callId (sub-cells nest under the parent Tool cell). */
   codeDispatches: ConversationSnapshot['codeDispatches']
 }
 
 interface UsageLike {
   inputTokens?: number
+  cacheReadTokens?: number
+  cacheWriteTokens?: number
   outputTokens?: number
   reasoningTokens?: number
 }
@@ -46,15 +57,85 @@ interface LaidCell {
   callId?: string
 }
 
+interface LaidGroup {
+  title: string
+  laid: LaidCell[]
+}
+
+interface TurnBucket {
+  groups: LaidGroup[]
+}
+
+type InputNode = Extract<
+  ConversationSnapshot['nodes'][number],
+  { kind: 'user' | 'steering' | 'context' }
+>
+
+type OrderedLayoutEntry =
+  | {
+    kind: 'node'
+    seq: number
+    node: ConversationSnapshot['nodes'][number]
+    nodeIndex: number
+  }
+  | {
+    kind: 'compaction'
+    seq: number
+    request: RequestView
+  }
+  | {
+    kind: 'system'
+    seq: number
+    request: RequestView
+    change: RequestPromptChange
+  }
+  | {
+    kind: 'request'
+    seq: number
+    request: RequestView
+  }
+
+function layoutEntryOrder(entry: OrderedLayoutEntry): number {
+  return entry.kind === 'system' && entry.change.kind === 'initial'
+    ? Number.NEGATIVE_INFINITY
+    : entry.seq
+}
+
+function inputCellDetail(node: InputNode): Pick<
+  TrajectoryCellProps,
+  'text' | 'sourceSeq' | 'messageSource' | 'inputDetail' | 'sourceBlocks' | 'timeSeconds' | 'startedAt'
+> {
+  return {
+    text: summarizeContent(node.content),
+    sourceSeq: node.seq,
+    messageSource: node.source,
+    inputDetail: detailContent(node.content),
+    sourceBlocks: node.content.map(block => sourceBlock(block)),
+    timeSeconds: 0,
+    startedAt: finiteTime(node.time),
+  }
+}
+
 /**
  * Fold a snapshot into turn → Message/Step groups with expanded cells.
  * @param input - nodes plus in-flight partial/runningCalls.
  * @returns turns ordered by first appearance.
  */
 export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly TrajectoryTurnModel[] {
-  const { nodes, partial, runningCalls, codeDispatches } = input
+  const {
+    nodes, partial, runningCalls, requests = [], callSchemas, codeDispatches,
+  } = input
   const resultByCall = indexResults(nodes)
-  const turns = new Map<number, { message: LaidCell[]; steps: Map<number, LaidCell[]> }>()
+  const callStartById = new Map<string, number>()
+  for (const result of resultByCall.values()) {
+    const startedAt = finiteTime(result.callTime)
+    if (startedAt !== null) callStartById.set(result.callId, startedAt)
+  }
+  for (const call of runningCalls) {
+    const startedAt = finiteTime(call.time)
+    if (startedAt !== null) callStartById.set(call.callId, startedAt)
+  }
+  const turns = new Map<number, TurnBucket>()
   let index = 0
   let prevAbsTime: number | null = null
   let lastAssistantTurn: number | null = null
@@ -62,26 +143,171 @@ export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly T
   const bucket = (turn: number) => {
     let entry = turns.get(turn)
     if (entry === undefined) {
-      entry = { message: [], steps: new Map() }
+      entry = { groups: [] }
       turns.set(turn, entry)
     }
     return entry
   }
 
   const pushMessage = (turn: number, laid: LaidCell) => {
-    bucket(turn).message.push(laid)
+    const groups = bucket(turn).groups
+    const last = groups.at(-1)
+    if (last?.title === 'Message') {
+      last.laid.push(laid)
+      return
+    }
+    groups.push({ title: 'Message', laid: [laid] })
   }
-  const pushStep = (turn: number, step: number, laid: LaidCell) => {
-    const steps = bucket(turn).steps
-    const list = steps.get(step) ?? []
-    list.push(laid)
-    steps.set(step, list)
+  const pushStep = (turn: number, step: number, laid: readonly LaidCell[]) => {
+    if (laid.length === 0) return
+    const groups = bucket(turn).groups
+    const title = `Step ${step}`
+    const existing = groups.find(group => group.title === title)
+    if (existing !== undefined) {
+      existing.laid.push(...laid)
+      return
+    }
+    groups.push({ title, laid: [...laid] })
   }
 
-  for (let i = 0; i < nodes.length; i++) {
-    const node = nodes[i]
-    /* v8 ignore next -- dense-array guard: i stays within nodes.length, so the undefined arm needs a sparse array no caller builds. */
-    if (node === undefined) continue
+  const representedRequests = new Set<string>()
+  for (const node of nodes) {
+    if (node.kind === 'assistant' && node.step > 0) {
+      representedRequests.add(`${node.turn}\u0000${node.step}`)
+    }
+  }
+  if (partial !== null && partial.step > 0) {
+    representedRequests.add(`${partial.turn}\u0000${partial.step}`)
+  }
+  for (const call of runningCalls) {
+    if (call.step > 0) representedRequests.add(`${call.turn}\u0000${call.step}`)
+  }
+
+  const entries: OrderedLayoutEntry[] = [
+    ...nodes.map((node, nodeIndex) => ({
+      kind: 'node' as const,
+      seq: node.seq,
+      node,
+      nodeIndex,
+    })),
+    ...requests
+      .filter(request => request.purpose === 'compaction')
+      .map(request => ({
+        kind: 'compaction' as const,
+        seq: request.startSeq,
+        request,
+      })),
+    ...requests.flatMap(request => request.promptChange === undefined || request.prompt === undefined
+      ? []
+      : [{
+        kind: 'system' as const,
+        seq: request.promptChange.seq,
+        request,
+        change: request.promptChange,
+      }]),
+    ...requests
+      .filter(request => request.purpose === 'assistant')
+      .filter(request =>
+        !representedRequests.has(`${request.turn}\u0000${request.step}`),
+      )
+      .map(request => ({
+        kind: 'request' as const,
+        seq: request.startSeq,
+        request,
+      })),
+  ].sort((left, right) => layoutEntryOrder(left) - layoutEntryOrder(right))
+
+  for (const entry of entries) {
+    if (entry.kind === 'request') {
+      const { request } = entry
+      pushStep(request.turn, request.step, [{
+        absTime: finiteTime(request.startedAt),
+        cell: {
+          index: ++index,
+          kind: 'message',
+          text: '',
+          sourceSeq: request.startSeq,
+          requestOnly: true,
+          timeSeconds: request.completedAt === null
+            ? null
+            : durationSeconds(request.completedAt, request.startedAt),
+          startedAt: finiteTime(request.startedAt),
+          ...(request.status === 'error' ? { isError: true } : {}),
+        },
+      }])
+      prevAbsTime = finiteTime(request.completedAt)
+        ?? finiteTime(request.startedAt)
+        ?? prevAbsTime
+      continue
+    }
+    if (entry.kind === 'system') {
+      const { change, request } = entry
+      const turn = change.kind === 'initial'
+        ? firstVisibleTurn(nodes, partial)
+        : enclosingPromptTurn(nodes, change.seq, partial)
+      pushMessage(turn, {
+        absTime: finiteTime(change.time),
+        cell: {
+          index: ++index,
+          kind: 'system',
+          text: promptChangeLabel(change),
+          sourceSeq: change.seq,
+          ...(request.prompt === undefined ? {} : { promptDetail: request.prompt }),
+          ...(change.previous === undefined
+            ? {}
+            : { previousPromptDetail: change.previous }),
+          timeSeconds: 0,
+          startedAt: finiteTime(change.time),
+        },
+      })
+      prevAbsTime = finiteTime(change.time) ?? prevAbsTime
+      continue
+    }
+    if (entry.kind === 'compaction') {
+      const request = entry.request
+      const rawOutput = request.rawOutput ?? request.summary
+      const thinkingDetail = rawOutput === undefined
+        ? ''
+        : detailReasoning(rawOutput)
+      const cell: TrajectoryCellProps = {
+        index: ++index,
+        kind: 'compacted',
+        text: request.status === 'running'
+          ? 'Compacting context…'
+          : request.status === 'error'
+            ? request.error ?? 'Compaction failed'
+            : request.summary === undefined
+              ? 'Context compacted'
+              : summarizeContent(request.summary),
+        sourceSeq: request.startSeq,
+        ...(request.summary === undefined
+          ? {}
+          : {
+            outputDetail: detailContent(request.summary),
+            outputBlocks: request.summary.map(block => sourceBlock(block)),
+          }),
+        ...(thinkingDetail === '' ? {} : { thinkingDetail }),
+        ...(rawOutput === undefined
+          ? {}
+          : { sourceBlocks: rawOutput.map(block => sourceBlock(block)) }),
+        ...(request.status === 'error' ? { isError: true } : {}),
+        timeSeconds: request.completedAt === null
+          ? null
+          : durationSeconds(request.completedAt, request.startedAt),
+        startedAt: finiteTime(request.startedAt),
+      }
+      attachUsage(cell, request.usage as UsageLike | undefined)
+      bucket(request.turn).groups.push({
+        title: `Compaction ${request.startSeq}`,
+        laid: [{
+          absTime: finiteTime(request.startedAt),
+          cell,
+        }],
+      })
+      prevAbsTime = finiteTime(request.completedAt) ?? finiteTime(request.startedAt) ?? prevAbsTime
+      continue
+    }
+    const { node, nodeIndex: i } = entry
     if (node.kind === 'user' || node.kind === 'steering') {
       // user/message has no turn on the wire; enclose it in the next assistant
       // (or partial) turn, else open the turn after the last assistant.
@@ -91,19 +317,22 @@ export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly T
       pushMessage(turn, {
         absTime: finiteTime(node.time),
         cell: {
-          index: ++index, kind: 'user', text: summarizeContent(node.content),
-          timeSeconds: 0,
+          index: ++index,
+          kind: 'user',
+          ...inputCellDetail(node),
+          opensTurn: node.kind === 'user',
         },
       })
       prevAbsTime = finiteTime(node.time) ?? prevAbsTime
       continue
     }
     if (node.kind === 'assistant') {
-      const laidList = withSubCalls(expandAssistant(node, index + 1, prevAbsTime, resultByCall), codeDispatches)
-      for (const laid of laidList) {
-        if (node.step > 0) pushStep(node.turn, node.step, laid)
-        else pushMessage(node.turn, laid)
-      }
+      const laidList = withSubCalls(
+        expandAssistant(node, index + 1, prevAbsTime, resultByCall, callStartById),
+        codeDispatches,
+      )
+      if (node.step > 0) pushStep(node.turn, node.step, laidList)
+      else for (const laid of laidList) pushMessage(node.turn, laid)
       const last = laidList[laidList.length - 1]
       if (last !== undefined) index = last.cell.index
       prevAbsTime = finiteTime(node.time) ?? prevAbsTime
@@ -111,30 +340,47 @@ export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly T
       continue
     }
     if (node.kind === 'context') {
-      // No trajectory cell, but the surface still advances the duration cursor.
+      const turn = enclosingUserTurn(nodes, i, partial, lastAssistantTurn)
+      pushMessage(turn, {
+        absTime: finiteTime(node.time),
+        cell: {
+          index: ++index,
+          kind: 'context',
+          ...inputCellDetail(node),
+        },
+      })
       prevAbsTime = finiteTime(node.time) ?? prevAbsTime
       continue
     }
     if (node.kind === 'tool-result') {
       if (!callEmittedInAssistant(nodes, node.callId)) {
         const toolName = node.call?.name
-        pushStep(0, 1, {
+        const laidList: LaidCell[] = [{
           absTime: finiteTime(node.callTime ?? node.time),
           ...(toolName !== undefined ? { toolName } : {}),
           callId: node.callId,
           cell: {
             index: ++index,
             kind: 'tool',
+            sourceSeq: node.seq,
             text: node.call !== null
               ? summarizeCall(node.call.name, node.call.argsRaw)
               : summarizeResult(node),
+            ...(node.call !== null ? { inputDetail: node.call.argsRaw } : {}),
+            outputDetail: detailResult(node),
+            outputBlocks: node.content.map(block => sourceBlock(block)),
+            result: summarizeResult(node),
+            callId: node.callId,
+            isError: node.isError,
             timeSeconds: durationSeconds(node.time, node.callTime),
+            startedAt: finiteTime(node.callTime),
           },
-        })
+        }]
         for (const laid of expandSubCalls(codeDispatches.get(node.callId), index)) {
-          pushStep(0, 1, laid)
+          laidList.push(laid)
           index = laid.cell.index
         }
+        pushStep(0, 1, laidList)
       }
       prevAbsTime = finiteTime(node.time) ?? prevAbsTime
     }
@@ -145,11 +391,16 @@ export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly T
       kind: 'assistant', seq: Number.MAX_SAFE_INTEGER, time: 0,
       turn: partial.turn, step: partial.step, blocks: partial.blocks,
     }
-    const laidList = expandAssistant(fake, index + 1, prevAbsTime, resultByCall, { streaming: true })
-    for (const laid of laidList) {
-      if (partial.step > 0) pushStep(partial.turn, partial.step, laid)
-      else pushMessage(partial.turn, laid)
-    }
+    const laidList = expandAssistant(
+      fake,
+      index + 1,
+      prevAbsTime,
+      resultByCall,
+      callStartById,
+      { streaming: true },
+    )
+    if (partial.step > 0) pushStep(partial.turn, partial.step, laidList)
+    else for (const laid of laidList) pushMessage(partial.turn, laid)
     const last = laidList[laidList.length - 1]
     if (last !== undefined) index = last.cell.index
   }
@@ -157,7 +408,7 @@ export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly T
   const seenCalls = collectCallIds(turns)
   for (const call of runningCalls) {
     if (seenCalls.has(call.callId)) continue
-    pushStep(call.turn, call.step > 0 ? call.step : 1, {
+    const laidList: LaidCell[] = [{
       absTime: null,
       toolName: call.name,
       callId: call.callId,
@@ -165,63 +416,67 @@ export function deriveTrajectoryLayout(input: TrajectoryLayoutInput): readonly T
         index: ++index,
         kind: 'tool',
         text: summarizeCall(call.name, call.argsRaw),
+        inputDetail: call.argsRaw,
+        callId: call.callId,
         timeSeconds: null,
+        startedAt: finiteTime(call.time),
       },
-    })
+    }]
     for (const laid of expandSubCalls(codeDispatches.get(call.callId), index)) {
-      pushStep(call.turn, call.step > 0 ? call.step : 1, laid)
+      laidList.push(laid)
       index = laid.cell.index
     }
+    if (call.step > 0) pushStep(call.turn, call.step, laidList)
+    else for (const laid of laidList) pushMessage(call.turn, laid)
   }
 
   // Orphan turn-0 cells (orphaned tools / steering turn 0) fold into Turn 1.
   const prologue = turns.get(0)
   if (prologue !== undefined) {
     turns.delete(0)
-    const emptyTurn = (): { message: LaidCell[]; steps: Map<number, LaidCell[]> } => ({
-      message: [],
-      steps: new Map(),
-    })
+    const emptyTurn = (): TurnBucket => ({ groups: [] })
     const first = turns.get(1) ?? emptyTurn()
-    first.message = [...prologue.message, ...first.message]
-    for (const [step, cells] of prologue.steps) {
-      const existing = first.steps.get(step) ?? []
-      first.steps.set(step, [...cells, ...existing])
-    }
+    first.groups = [...prologue.groups, ...first.groups]
     turns.set(1, first)
   }
 
+  for (const entry of turns.values()) {
+    for (const group of entry.groups) {
+      for (const laid of group.laid) attachToolSchema(laid, callSchemas)
+    }
+  }
+
   return [...turns.entries()]
     .sort(([a], [b]) => a - b)
     .map(([turn, entry]) => toTurnModel(turn, entry))
 }
 
+function attachToolSchema(
+  laid: LaidCell,
+  callSchemas: RequestInspectionSnapshot['callSchemas'] | undefined,
+): void {
+  if (laid.callId === undefined || callSchemas === undefined) return
+  const schema = callSchemas.get(laid.callId)
+  if (schema === undefined) return
+  laid.cell.schemaDetail = JSON.stringify(schema, null, 2)
+}
+
 function toTurnModel(
   turn: number,
-  entry: { message: LaidCell[]; steps: Map<number, LaidCell[]> },
+  entry: TurnBucket,
 ): TrajectoryTurnModel {
-  const groups: TrajectoryGroupModel[] = []
-  if (entry.message.length > 0) {
-    const description = groupDescription(entry.message)
-    groups.push({
-      title: 'Message',
-      ...(description !== undefined ? { description } : {}),
-      cells: entry.message.map(l => l.cell),
-    })
-  }
-  for (const step of [...entry.steps.keys()].sort((a, b) => a - b)) {
-    const laid = entry.steps.get(step) ?? []
+  const groups = entry.groups.map(({ title, laid }): TrajectoryGroupModel => {
     const description = groupDescription(laid)
-    groups.push({
-      title: `Step ${step}`,
+    return {
+      title,
       ...(description !== undefined ? { description } : {}),
       cells: laid.map(l => l.cell),
-    })
-  }
+    }
+  })
   return { turn, groups }
 }
 
-/** Wall-span duration + tool histogram, e.g. `1.5s bash×6`. */
+/** Wall-span duration + tool histogram, e.g. `1.5 s bash×6`. */
 function groupDescription(laid: readonly LaidCell[]): string | undefined {
   const parts: string[] = []
   // Tool rows contribute start (absTime) and end (start + own duration) so a
@@ -256,8 +511,8 @@ function groupDescription(laid: readonly LaidCell[]): string | undefined {
 function formatGroupDuration(seconds: number): string | undefined {
   if (!Number.isFinite(seconds)) return undefined
   const rounded = Math.round(seconds * 10) / 10
-  if (Number.isInteger(rounded)) return `${rounded}s`
-  return `${rounded.toFixed(1)}s`
+  if (Number.isInteger(rounded)) return `${rounded} s`
+  return `${rounded.toFixed(1)} s`
 }
 
 /** Own-duration seconds from two epoch-ms stamps; null when either is unusable. */
@@ -267,8 +522,8 @@ function durationSeconds(later: number, earlier: number | null): number | null {
 }
 
 /** Epoch-ms usable as an absolute time, else null. */
-function finiteTime(time: number): number | null {
-  return Number.isFinite(time) ? time : null
+function finiteTime(time: number | null | undefined): number | null {
+  return typeof time === 'number' && Number.isFinite(time) ? time : null
 }
 
 function expandAssistant(
@@ -276,66 +531,175 @@ function expandAssistant(
   startIndex: number,
   prevAbsTime: number | null,
   results: Map<string, ToolResultNode>,
+  callStarts: ReadonlyMap<string, number>,
   opts?: { streaming?: boolean },
 ): LaidCell[] {
   const out: LaidCell[] = []
   let index = startIndex - 1
   const usage = node.usage as UsageLike | undefined
   const streaming = opts?.streaming === true
-  const messageDuration = streaming ? null : durationSeconds(node.time, prevAbsTime)
+  const recordedStart = finiteTime(node.timing?.stepStartTime)
+  const messageDuration = streaming
+    ? null
+    : durationSeconds(node.time, recordedStart ?? prevAbsTime)
   const nodeAbs = streaming ? null : finiteTime(node.time)
-  let usageAttached = false
+  const messageText = node.blocks
+    .filter(block => block.kind === 'text' && (!streaming || block.text !== ''))
+    .map(block => block.kind === 'text' ? block.text : '')
+    .join('\n\n')
+  const thinkingText = node.blocks
+    .filter(block => block.kind === 'reasoning' && (!streaming || block.text !== ''))
+    .map(block => block.kind === 'reasoning' ? block.text : '')
+    .join('\n\n')
+  const message: TrajectoryCellProps = {
+    index: ++index,
+    kind: 'message',
+    sourceSeq: node.seq,
+    text: messageText !== ''
+      ? summarizeText(messageText)
+      : thinkingText !== ''
+        ? summarizeText(thinkingText)
+        : summarizeAssistantActivity(node.blocks),
+    ...(messageText !== '' ? { outputDetail: messageText } : {}),
+    ...(thinkingText !== '' ? { thinkingDetail: thinkingText } : {}),
+    sourceBlocks: node.blocks.map(block => assistantSourceBlock(block)),
+    timeSeconds: messageDuration,
+    startedAt: recordedStart,
+  }
+  attachUsage(message, usage)
+  message.assistantMetrics = {
+    timingRecorded: node.timing !== undefined,
+    stepStartTime: node.timing?.stepStartTime ?? null,
+    firstTokenTime: node.timing?.firstTokenTime ?? null,
+    completedTime: streaming ? null : finiteTime(node.time),
+    usageProvided: usage !== undefined,
+    outputTokens: Number.isFinite(usage?.outputTokens) ? usage?.outputTokens ?? null : null,
+  }
+  out.push({ absTime: nodeAbs, cell: message })
 
   for (const block of node.blocks) {
-    // Reasoning blocks are skipped: no block-level clock, so no Think cell.
-    if (block.kind === 'reasoning') continue
-    if (block.kind === 'text') {
-      if (block.text === '' && streaming) continue
-      const cell: TrajectoryCellProps = {
-        index: ++index, kind: 'message', text: summarizeText(block.text),
-        timeSeconds: messageDuration,
-      }
-      if (!usageAttached) {
-        attachUsage(cell, usage)
-        usageAttached = usage !== undefined
-      }
-      out.push({ absTime: nodeAbs, cell })
-      continue
-    }
-    if (block.kind === 'tool-call') {
-      const result = results.get(block.callId)
-      const toolDuration = streaming || result === undefined
-        ? null
-        : durationSeconds(result.time, result.callTime)
-      const callAbs = streaming
-        ? null
-        : (result?.callTime !== null && result?.callTime !== undefined && Number.isFinite(result.callTime)
-          ? result.callTime
-          : nodeAbs)
-      out.push({
-        absTime: callAbs,
-        toolName: block.name,
+    // Text and reasoning belong to the one Assistant record emitted above.
+    if (block.kind !== 'tool-call') continue
+    const result = results.get(block.callId)
+    const toolDuration = streaming || result === undefined
+      ? null
+      : durationSeconds(result.time, result.callTime)
+    const callAbs = finiteTime(callStarts.get(block.callId))
+    out.push({
+      absTime: callAbs,
+      toolName: block.name,
+      callId: block.callId,
+      cell: {
+        index: ++index, kind: 'tool',
+        text: summarizeCall(block.name, block.argsRaw),
+        inputDetail: block.argsRaw,
         callId: block.callId,
-        cell: {
-          index: ++index, kind: 'tool',
-          text: summarizeCall(block.name, block.argsRaw),
-          timeSeconds: toolDuration,
-        },
-      })
-    }
-  }
-
-  if (out.length === 0 && !streaming) {
-    // Reasoning-only / empty success still owns provider usage on the Message row.
-    const cell: TrajectoryCellProps = {
-      index: ++index, kind: 'message', text: '', timeSeconds: messageDuration,
-    }
-    attachUsage(cell, usage)
-    out.push({ absTime: nodeAbs, cell })
+        ...(result !== undefined
+          ? {
+            outputDetail: detailResult(result),
+            outputBlocks: result.content.map(block => sourceBlock(block)),
+            result: summarizeResult(result),
+            isError: result.isError,
+          }
+          : {}),
+        timeSeconds: toolDuration,
+        startedAt: callAbs,
+      },
+    })
   }
   return out
 }
 
+function summarizeAssistantActivity(blocks: readonly AssistantBlock[]): string {
+  const tools = new Map<string, number>()
+  for (const block of blocks) {
+    if (block.kind !== 'tool-call') continue
+    tools.set(block.name, (tools.get(block.name) ?? 0) + 1)
+  }
+  if (tools.size > 0) {
+    return 'Tool call only'
+  }
+  return ''
+}
+
+function promptChangeLabel(change: RequestPromptChange): string {
+  if (change.kind === 'initial') return 'Initial System Prompt'
+  if (change.kind === 'system') return 'System Prompt Updated'
+  if (change.kind === 'tools') return 'Tools Updated'
+  return 'System Prompt and Tools Updated'
+}
+
+function assistantSourceBlock(block: AssistantBlock): TrajectorySourceBlock {
+  switch (block.kind) {
+    case 'text': return { type: 'text', content: block.text }
+    case 'reasoning': return { type: 'thinking', content: block.text }
+    case 'tool-call': return {
+      type: 'tool-call',
+      content: block.argsRaw,
+      callId: block.callId,
+      toolName: block.name,
+    }
+    case 'other': return sourceBlock(block.block)
+  }
+}
+
+function sourceBlock(value: unknown): TrajectorySourceBlock {
+  if (typeof value !== 'object' || value === null) {
+    return { type: 'unknown', content: stringifySourceValue(value) }
+  }
+  const block = value as Record<string, unknown>
+  const type = typeof block.type === 'string' ? block.type : 'unknown'
+  if (typeof block.text === 'string') {
+    return { type: type === 'reasoning' ? 'thinking' : type, content: block.text }
+  }
+  const imageSrc = sourceImage(block)
+  const imageAlt = typeof block.alt === 'string' ? block.alt : undefined
+  return {
+    type,
+    content: imageSrc === undefined ? stringifySourceValue(value) : '',
+    ...(imageSrc !== undefined ? { imageSrc } : {}),
+    ...(imageAlt !== undefined ? { imageAlt } : {}),
+  }
+}
+
+function sourceImage(block: Record<string, unknown>): string | undefined {
+  if (typeof block.type !== 'string' || !block.type.toLowerCase().includes('image')) return undefined
+  for (const candidate of [block.url, block.image_url]) {
+    if (typeof candidate === 'string') return safeImageSource(candidate)
+  }
+  if (typeof block.data === 'string') {
+    const mediaType = [block.mimeType, block.mediaType, block.media_type]
+      .find((candidate): candidate is string => typeof candidate === 'string')
+      ?? 'image/png'
+    return safeImageSource(
+      block.data.startsWith('data:')
+        ? block.data
+        : `data:${mediaType};base64,${block.data}`,
+    )
+  }
+  if (typeof block.source !== 'object' || block.source === null) return undefined
+  const source = block.source as Record<string, unknown>
+  if (typeof source.url === 'string') return safeImageSource(source.url)
+  if (typeof source.data !== 'string') return undefined
+  const mediaType = typeof source.media_type === 'string' ? source.media_type : 'image/png'
+  return safeImageSource(`data:${mediaType};base64,${source.data}`)
+}
+
+function safeImageSource(value: string): string | undefined {
+  if (value.startsWith('data:image/') || value.startsWith('blob:')) return value
+  try {
+    const protocol = new URL(value).protocol
+    return protocol === 'http:' || protocol === 'https:' ? value : undefined
+  } catch {
+    return undefined
+  }
+}
+
+function stringifySourceValue(value: unknown): string {
+  const json = JSON.stringify(value, null, 2)
+  return json || String(value)
+}
+
 /**
  * Turn that encloses a user/message: next assistant/steering turn, else the
  * in-flight partial, else the turn after the last finalized assistant (or 1).
@@ -357,10 +721,37 @@ function enclosingUserTurn(
   return 1
 }
 
+function enclosingPromptTurn(
+  nodes: ConversationSnapshot['nodes'],
+  seq: number,
+  partial: ConversationSnapshot['partial'],
+): number {
+  const next = nodes.find(node =>
+    node.seq > seq && node.kind === 'assistant' && node.step > 0)
+  if (next?.kind === 'assistant') return next.turn
+  return partial?.turn ?? 1
+}
+
+/** Earliest raw turn represented by the selected trajectory branch. */
+function firstVisibleTurn(
+  nodes: ConversationSnapshot['nodes'],
+  partial: ConversationSnapshot['partial'],
+): number {
+  const turns = nodes.flatMap(node =>
+    (node.kind === 'assistant' || node.kind === 'steering') && node.turn > 0
+      ? [node.turn]
+      : [],
+  )
+  if (partial !== null && partial.turn > 0) turns.push(partial.turn)
+  return turns.length === 0 ? 1 : Math.min(...turns)
+}
+
 /** Copy provider usage onto a Message cell when present. */
 function attachUsage(cell: TrajectoryCellProps, usage: UsageLike | undefined): void {
   if (usage === undefined) return
   if (usage.inputTokens !== undefined) cell.input = usage.inputTokens
+  if (usage.cacheReadTokens !== undefined) cell.cacheRead = usage.cacheReadTokens
+  if (usage.cacheWriteTokens !== undefined) cell.cacheWrite = usage.cacheWriteTokens
   if (usage.outputTokens !== undefined) cell.output = usage.outputTokens
   if (usage.reasoningTokens !== undefined) cell.think = usage.reasoningTokens
 }
@@ -382,15 +773,12 @@ function callEmittedInAssistant(nodes: ConversationSnapshot['nodes'], callId: st
 }
 
 function collectCallIds(
-  turns: Map<number, { message: LaidCell[]; steps: Map<number, LaidCell[]> }>,
+  turns: Map<number, TurnBucket>,
 ): Set<string> {
   const ids = new Set<string>()
   for (const entry of turns.values()) {
-    for (const laid of entry.message) {
-      if (laid.callId !== undefined) ids.add(laid.callId)
-    }
-    for (const list of entry.steps.values()) {
-      for (const laid of list) {
+    for (const group of entry.groups) {
+      for (const laid of group.laid) {
         if (laid.callId !== undefined) ids.add(laid.callId)
       }
     }
@@ -433,12 +821,27 @@ function expandSubCalls(
       cell: {
         index: ++index,
         kind: 'subtool',
+        callId: sub.callId,
         text: settled
           ? (sub.call !== null ? summarizeCall(sub.call.name, sub.call.argsRaw) : summarizeResult(sub))
           : summarizeCall(sub.name, sub.argsRaw),
+        ...(settled
+          ? (sub.call !== null ? { inputDetail: sub.call.argsRaw } : {})
+          : { inputDetail: sub.argsRaw }),
+        ...(settled
+          ? {
+            outputDetail: detailResult(sub),
+            outputBlocks: sub.content.map(block => sourceBlock(block)),
+            result: summarizeResult(sub),
+            isError: sub.isError,
+          }
+          : {}),
         // PR3's start/settle pair carries per-sub-call wall time; a running
         // (unsettled) or pre-pair log entry shows the em dash.
         timeSeconds: settled ? durationSeconds(sub.time, sub.callTime) : null,
+        startedAt: settled
+          ? finiteTime(sub.callTime)
+          : finiteTime(sub.time),
       },
     })
   }
@@ -448,8 +851,7 @@ function expandSubCalls(
 function summarizeCall(name: string, argsRaw: string): string {
   const args = argsRaw.replace(/\s+/g, ' ').trim()
   if (args === '') return name
-  const clipped = args.length > 72 ? `${args.slice(0, 71)}…` : args
-  return `${name} · ${clipped}`
+  return `${name} · ${args}`
 }
 
 function summarizeResult(node: ToolResultNode): string {
@@ -461,7 +863,40 @@ function summarizeResult(node: ToolResultNode): string {
       return summarizeText(block.text)
     }
   }
-  return node.call?.name ?? node.callId
+  return 'No output'
+}
+
+function detailResult(node: ToolResultNode): string {
+  if (node.isError) {
+    return node.error === undefined
+      ? 'error'
+      : `${node.error.name}: ${node.error.code}`
+  }
+  const text = node.content
+    .filter(block => block.type === 'text' && typeof block.text === 'string')
+    .map(block => block.type === 'text' ? block.text : '')
+    .join('\n')
+  if (text !== '') return text
+  if (
+    node.content.length === 0
+    || node.content.every(block =>
+      block.type === 'text' && (typeof block.text !== 'string' || block.text === ''))
+  ) return 'No output'
+  return JSON.stringify(node.content, null, 2)
+}
+
+function detailContent(content: readonly { type: string; text?: string }[]): string {
+  return content
+    .filter(block => block.type === 'text' && typeof block.text === 'string')
+    .map(block => block.text ?? '')
+    .join('\n')
+}
+
+function detailReasoning(content: readonly { type: string; text?: string }[]): string {
+  return content
+    .filter(block => block.type === 'reasoning' && typeof block.text === 'string')
+    .map(block => block.text ?? '')
+    .join('\n')
 }
 
 function summarizeContent(content: readonly { type: string; text?: string }[]): string {
diff --git a/packages/client/ui-trajectory/src/client/spans.ts b/packages/client/ui-trajectory/src/client/spans.ts
deleted file mode 100644
index 585a336333..0000000000
--- a/packages/client/ui-trajectory/src/client/spans.ts
+++ /dev/null
@@ -1,147 +0,0 @@
-/**
- * Rough per-turn span derivation shared by the two placeholder views and the
- * header stats bar. P-I ships no timing data, so a span's weight is its node
- * count, not wall time (deviation ledger #3 — real spans land in P-III).
- */
-import type { ConversationNode, ConversationSnapshot } from '@deepseek-ai/dsh-client-runtime/client'
-
-/** One run_code sub-dispatch lane in the waterfall: real timing off the start/settle pair. */
-export interface SubSpanLane {
-  callId: string
-  name: string
-  /** Wall duration in ms; null unless both endpoints were observed (`timing: 'measured'`). */
-  durationMs: number | null
-  /**
-   * Timing provenance: `measured` = start/settle pair observed; `running` =
-   * start seen, settle pending; `unknown` = settle-only replay window (the
-   * start fell outside), so no duration claim is possible.
-   */
-  timing: 'measured' | 'running' | 'unknown'
-  /** Start offset as a fraction of the parent turn's dispatch window [0, 1). */
-  offsetFraction: number
-  /** Width as a fraction of the window (running lanes extend to the window end). */
-  widthFraction: number
-}
-
-/** One turn's worth of activity, folded from the snapshot node window. */
-export interface TurnSpan {
-  turn: number
-  /** Assistant step messages inside the turn. */
-  steps: number
-  /** Tool results inside the turn (running calls are not folded in P-I). */
-  calls: number
-  /** Total nodes attributed to the turn (span weight stand-in). */
-  nodes: number
-}
-
-/** Aggregate totals for the header stats bar. */
-export interface SpanStats {
-  turns: number
-  steps: number
-  calls: number
-}
-
-/**
- * Fold snapshot nodes into per-turn spans. Only assistant nodes carry a turn
- * number; user/steering/context/tool nodes attach to the turn last seen in
- * sequence order (turn 0 collects the pre-assistant prologue).
- * @param nodes - snapshot nodes in surface order.
- * @returns spans ordered by first appearance.
- */
-export function deriveSpans(nodes: ConversationSnapshot['nodes']): readonly TurnSpan[] {
-  const spans = new Map<number, TurnSpan>()
-  let currentTurn = 0
-  const spanFor = (turn: number): TurnSpan => {
-    let span = spans.get(turn)
-    if (span === undefined) {
-      span = { turn, steps: 0, calls: 0, nodes: 0 }
-      spans.set(turn, span)
-    }
-    return span
-  }
-  for (const node of nodes) {
-    if (hasTurn(node)) currentTurn = node.turn
-    const span = spanFor(currentTurn)
-    span.nodes += 1
-    if (node.kind === 'assistant') span.steps += 1
-    if (node.kind === 'tool-result') span.calls += 1
-  }
-  return [...spans.values()]
-}
-
-/**
- * Aggregate spans into the header totals.
- * @param spans - deriveSpans product.
- * @returns turn/step/call totals.
- */
-export function deriveSpanStats(spans: readonly TurnSpan[]): SpanStats {
-  let steps = 0
-  let calls = 0
-  for (const span of spans) {
-    steps += span.steps
-    calls += span.calls
-  }
-  return { turns: spans.length, steps, calls }
-}
-
-function hasTurn(node: ConversationNode): node is ConversationNode & { turn: number } {
-  return node.kind === 'assistant' || node.kind === 'steering'
-}
-
-/**
- * Fold the dispatch index into per-turn sub-span lanes with REAL timing: each
- * lane's offset/width scale against its parent turn's dispatch window (first
- * start → last settle). Running (unsettled) lanes extend to the window end
- * with a null duration.
- * @param nodes - snapshot nodes (locates each parent run_code call's turn).
- * @param codeDispatches - the snapshot's dispatch index.
- * @returns lanes keyed by turn, in start order.
- */
-export function deriveSubSpans(
-  nodes: ConversationSnapshot['nodes'],
-  codeDispatches: ConversationSnapshot['codeDispatches'],
-): ReadonlyMap<number, readonly SubSpanLane[]> {
-  const out = new Map<number, SubSpanLane[]>()
-  if (codeDispatches.size === 0) return out
-  const turnByCall = new Map<string, number>()
-  let currentTurn = 0
-  for (const node of nodes) {
-    if (node.kind === 'assistant' || node.kind === 'steering') currentTurn = node.turn
-    if (node.kind === 'tool-result') turnByCall.set(node.callId, currentTurn)
-  }
-  for (const [parent, subs] of codeDispatches) {
-    if (subs.length === 0) continue
-    const turn = turnByCall.get(parent) ?? currentTurn
-    // A settle-only entry (callTime null: its start fell outside the replay
-    // window) anchors the window by its settle time — a real observation —
-    // but must never masquerade as a measured zero-duration span.
-    const starts: number[] = []
-    const ends: number[] = []
-    for (const sub of subs) {
-      const settled = 'kind' in sub
-      const start = settled ? sub.callTime ?? sub.time : sub.time
-      starts.push(start)
-      ends.push(settled ? sub.time : start)
-    }
-    const windowStart = Math.min(...starts)
-    const windowEnd = Math.max(...ends, windowStart + 1)
-    const windowSpan = windowEnd - windowStart
-    const lanes: SubSpanLane[] = subs.map((sub, i) => {
-      const settled = 'kind' in sub
-      const timing = settled ? (sub.callTime === null ? 'unknown' as const : 'measured' as const) : 'running' as const
-      const start = starts[i] ?? windowStart
-      const end = settled ? sub.time : windowEnd
-      return {
-        callId: sub.callId,
-        name: settled ? sub.call?.name ?? sub.callId : sub.name,
-        durationMs: timing === 'measured' ? Math.max(0, end - start) : null,
-        timing,
-        offsetFraction: (start - windowStart) / windowSpan,
-        widthFraction: Math.max((end - start) / windowSpan, 0.02),
-      }
-    })
-    const existing = out.get(turn) ?? []
-    out.set(turn, [...existing, ...lanes])
-  }
-  return out
-}
diff --git a/packages/client/ui-trajectory/src/client/timeline.ts b/packages/client/ui-trajectory/src/client/timeline.ts
new file mode 100644
index 0000000000..2e2227df39
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/timeline.ts
@@ -0,0 +1,186 @@
+/** Operation-sequence and recorded-time projections for the trajectory overview. */
+
+import type { TrajectoryTurnModel } from './layout.ts'
+import type { TrajectoryCellKind, TrajectoryCellProps } from './trajectory-record.ts'
+
+/** Horizontal projection used by the trajectory timeline. */
+export type TrajectoryTimelineMode = 'sequence' | 'duration' | 'time' | 'actual'
+
+/** Inclusive selection in the active timeline projection's domain. */
+export interface TrajectoryTimeRange {
+  start: number
+  end: number
+}
+
+/** One ledger record projected into the active timeline domain. */
+export interface TrajectoryTimelineSpan extends TrajectoryTimeRange {
+  index: number
+  kind: TrajectoryCellKind
+  label: string
+  lane: number
+}
+
+/** One turn boundary in the active timeline domain. */
+export interface TrajectoryTimelineTurnBoundary {
+  turn: number
+  time: number
+}
+
+/** Full-domain model used by the overview. */
+export interface TrajectoryTimelineModel extends TrajectoryTimeRange {
+  spans: readonly TrajectoryTimelineSpan[]
+  turnBoundaries: readonly TrajectoryTimelineTurnBoundary[]
+}
+
+/**
+ * Format a timeline duration with a compact unit.
+ * @param milliseconds - Non-negative duration in milliseconds.
+ * @returns Millisecond or second label.
+ */
+export function formatTimelineOffset(milliseconds: number): string {
+  if (milliseconds < 1_000) return `${Math.round(milliseconds)} ms`
+  const seconds = milliseconds / 1_000
+  return seconds >= 10 ? `${Math.round(seconds)} s` : `${seconds.toFixed(1)} s`
+}
+
+function laneFor(kind: TrajectoryCellKind): number {
+  if (kind === 'tool' || kind === 'subtool') return 2
+  if (kind === 'message' || kind === 'compacted') return 1
+  return 0
+}
+
+function finite(value: number | null | undefined): value is number {
+  return value !== null && value !== undefined && Number.isFinite(value)
+}
+
+function cellRange(cell: TrajectoryCellProps): TrajectoryTimeRange | null {
+  if (!finite(cell.startedAt)) return null
+  const durationMs = finite(cell.timeSeconds)
+    ? Math.max(0, cell.timeSeconds * 1_000)
+    : 0
+  return { start: cell.startedAt, end: cell.startedAt + durationMs }
+}
+
+/**
+ * Project every visible record into a stable three-lane timeline.
+ * @param turns - Unfiltered trajectory layout.
+ * @param mode - Independent equal/recorded duration and compressed/complete time projection.
+ * @returns Timeline model, or `null` when no record is visible.
+ */
+export function deriveTrajectoryTimeline(
+  turns: readonly TrajectoryTurnModel[],
+  mode: TrajectoryTimelineMode = 'sequence',
+): TrajectoryTimelineModel | null {
+  if (mode !== 'sequence') {
+    return deriveTimedTimeline(
+      turns,
+      mode === 'duration' || mode === 'actual',
+      mode === 'duration',
+    )
+  }
+  const spans: TrajectoryTimelineSpan[] = []
+  const turnBoundaries: TrajectoryTimelineTurnBoundary[] = []
+
+  for (const turn of turns) {
+    const cells = turn.groups.flatMap(group =>
+      group.cells.filter(cell => cell.requestOnly !== true),
+    )
+    if (cells.length === 0) continue
+    turnBoundaries.push({
+      turn: turn.turn,
+      time: spans.length,
+    })
+    spans.push(...cells.map((cell, offset): TrajectoryTimelineSpan => ({
+      start: spans.length + offset,
+      end: spans.length + offset + 1,
+      index: cell.index,
+      kind: cell.kind,
+      label: cell.text,
+      lane: laneFor(cell.kind),
+    })))
+  }
+
+  if (spans.length === 0) return null
+  return {
+    start: 0,
+    end: spans.length,
+    spans,
+    turnBoundaries,
+  }
+}
+
+function deriveTimedTimeline(
+  turns: readonly TrajectoryTurnModel[],
+  actualDuration: boolean,
+  removeUserIdle: boolean,
+): TrajectoryTimelineModel | null {
+  const spans: TrajectoryTimelineSpan[] = []
+  const turnBoundaries: TrajectoryTimelineTurnBoundary[] = []
+  let removedUserIdle = 0
+  let previousTurnEnd: number | null = null
+
+  for (const turn of turns) {
+    const rawSpans = turn.groups.flatMap(group =>
+      group.cells.flatMap((cell): TrajectoryTimelineSpan[] => {
+        if (cell.requestOnly === true) return []
+        const range = cellRange(cell)
+        return range === null
+          ? []
+          : [{
+            ...range,
+            index: cell.index,
+            kind: cell.kind,
+            label: cell.text,
+            lane: laneFor(cell.kind),
+          }]
+      }),
+    )
+    if (rawSpans.length === 0) continue
+
+    const turnStart = Math.min(...rawSpans.map(span => span.start))
+    const turnEnd = Math.max(...rawSpans.map(span => span.end))
+    if (removeUserIdle && previousTurnEnd !== null) {
+      removedUserIdle += Math.max(0, turnStart - previousTurnEnd)
+    }
+    spans.push(...rawSpans.map(span => ({
+      ...span,
+      start: span.start - removedUserIdle,
+      end: (actualDuration ? span.end : span.start) - removedUserIdle,
+    })))
+    turnBoundaries.push({
+      turn: turn.turn,
+      time: turnStart - removedUserIdle,
+    })
+    previousTurnEnd = previousTurnEnd === null
+      ? turnEnd
+      : Math.max(previousTurnEnd, turnEnd)
+  }
+
+  if (spans.length === 0) return null
+  return {
+    start: Math.min(...spans.map(span => span.start)),
+    end: Math.max(...spans.map(span => span.end)),
+    spans,
+    turnBoundaries,
+  }
+}
+
+/**
+ * Identify records active at any point inside an inclusive selected interval.
+ * @param turns - Unfiltered trajectory layout.
+ * @param range - Selected interval in the active projection.
+ * @param mode - Independent equal/recorded duration and compressed/complete time projection.
+ * @returns Record indexes inside the focus interval.
+ */
+export function trajectoryTimelineFocusIndexes(
+  turns: readonly TrajectoryTurnModel[],
+  range: TrajectoryTimeRange,
+  mode: TrajectoryTimelineMode = 'sequence',
+): ReadonlySet<number> {
+  const model = deriveTrajectoryTimeline(turns, mode)
+  return new Set(
+    model?.spans
+      .filter(span => span.start <= range.end && span.end >= range.start)
+      .map(span => span.index),
+  )
+}
diff --git a/packages/client/ui-trajectory/src/client/trajectory-record.ts b/packages/client/ui-trajectory/src/client/trajectory-record.ts
new file mode 100644
index 0000000000..e33bfb873c
--- /dev/null
+++ b/packages/client/ui-trajectory/src/client/trajectory-record.ts
@@ -0,0 +1,104 @@
+/** Shared trajectory record data and formatting contracts. */
+
+import type { HTMLAttributes } from 'react'
+import type { ConversationPromptSnapshot } from '@deepseek-ai/dsh-client-runtime/client'
+
+/** Closed set of trajectory record kinds. */
+export type TrajectoryCellKind =
+  | 'system'
+  | 'user'
+  | 'context'
+  | 'compacted'
+  | 'message'
+  | 'tool'
+  | 'subtool'
+
+/** Recorded inputs needed to derive assistant TTFT and decode throughput. */
+export interface AssistantMetricDetail {
+  timingRecorded: boolean
+  stepStartTime: number | null
+  firstTokenTime: number | null
+  completedTime: number | null
+  usageProvided: boolean
+  outputTokens: number | null
+}
+
+/** One source content block preserved in model order for the details panel. */
+export interface TrajectorySourceBlock {
+  type: string
+  content: string
+  imageSrc?: string
+  imageAlt?: string
+  callId?: string
+  toolName?: string
+}
+
+/** Data and optional presentation attributes for one trajectory record. */
+export interface TrajectoryCellProps extends HTMLAttributes<HTMLDivElement> {
+  /** 1-based record index shown as `#N`. */
+  index: number
+  kind: TrajectoryCellKind
+  /** Single-line summary; CSS ellipsis when it overflows. */
+  text: string
+  /** Whether this user record opens a new model turn. */
+  opensTurn?: boolean
+  /** Source session-event seq for cross-record navigation. */
+  sourceSeq?: number
+  /** Producer provenance from a user-role message or context injection. */
+  messageSource?: unknown
+  /** Producer-owned model-hidden metadata carried beside the message source. */
+  /** A separator-only anchor for an auxiliary request with no visible record. */
+  requestOnly?: boolean
+  /** Full request/message content for the details panel. */
+  inputDetail?: string
+  /** Complete system-prompt/tool-catalog state introduced by a SYSTEM record. */
+  promptDetail?: ConversationPromptSnapshot
+  /** System-prompt/tool-catalog state replaced by a SYSTEM update. */
+  previousPromptDetail?: ConversationPromptSnapshot
+  /** Full assistant/tool result content for the details panel. */
+  outputDetail?: string
+  /** Full assistant reasoning content for the details panel. */
+  thinkingDetail?: string
+  /** Original message blocks in source order for the details panel. */
+  sourceBlocks?: readonly TrajectorySourceBlock[]
+  /** Original tool result blocks in source order for the details panel. */
+  outputBlocks?: readonly TrajectorySourceBlock[]
+  /** Call-time model-visible tool schema for the details panel. */
+  schemaDetail?: string
+  /** Assistant-only timing and token facts for the details panel. */
+  assistantMetrics?: AssistantMetricDetail
+  /** Tool-only result summary paired with the call in the same record. */
+  result?: string
+  /** Tool call id used to link message source blocks to tool records. */
+  callId?: string
+  /** Tool-only result failure state. */
+  isError?: boolean
+  /** Own duration in seconds, or `null` when no duration is known. */
+  timeSeconds: number | null
+  /** Unix epoch milliseconds when this operation actually started, when known. */
+  startedAt?: number | null
+  /** Message-only prompt token count. */
+  input?: number
+  /** Message-only input tokens served from a provider cache. */
+  cacheRead?: number
+  /** Message-only input tokens written into a provider cache. */
+  cacheWrite?: number
+  /** Message-only completion token count. */
+  output?: number
+  /** Message-only reasoning token count. */
+  think?: number
+  /** Whether the legacy standalone cell renders its selection treatment. */
+  selected?: boolean
+}
+
+/**
+ * Format own-duration for the trailing time column.
+ * @param seconds - Duration seconds, or `null` when absent.
+ * @returns `—` when unknown, otherwise a seconds label.
+ */
+export function formatElapsedSeconds(seconds: number | null): string {
+  if (seconds === null || !Number.isFinite(seconds)) return '—'
+  const rounded = Math.round(seconds * 10) / 10
+  if (Number.isInteger(rounded)) return `${rounded} s`
+  return `${rounded.toFixed(1)} s`
+}
diff --git a/packages/client/ui-trajectory/src/client/views.module.css b/packages/client/ui-trajectory/src/client/views.module.css
index 16a853c441..14b4ce6ac2 100644
--- a/packages/client/ui-trajectory/src/client/views.module.css
+++ b/packages/client/ui-trajectory/src/client/views.module.css
@@ -1,82 +1,29 @@
-/* Full-bleed scroll host so Turn sticky bars can paint edge-to-edge;
- * cell content width is capped on the turn body (max 880). */
+/* Full-bleed, fixed-height host for the trajectory ledger. */
 .root {
-  overflow-y: auto;
+  --dsh-trajectory-toolbar-height: 32px;
+
+  display: flex;
+  flex-direction: column;
+  overflow: hidden;
   height: 100%;
   min-height: 0;
   width: 100%;
   box-sizing: border-box;
   color: var(--dsw-alias-label-primary);
-  background: var(--dsw-specific-sidebar-fill);
+  background: var(--dsw-alias-bg-layer-1);
 }
 
-.empty {
-  padding: 16px;
-  color: var(--dsw-alias-label-tertiary);
-  font: var(--dsw-font-xs-13);
+/* Under the active conversation column (`[data-conversation-scroll]`) the
+ * parent owns overflow so the sticky composer stays in the same scrollport. */
+:global([data-conversation-scroll]) .root {
+  overflow: visible;
+  height: auto;
 }
 
-/* Waterfall placeholder rows (shared module). */
-.row {
+.ledger {
   display: flex;
-  align-items: center;
-  gap: 8px;
-  padding: 4px 16px;
-}
-
-.turnTag {
-  flex: none;
-  width: 64px;
-  color: var(--dsw-alias-label-secondary);
-  font: var(--dsw-font-xs-13);
-}
-
-.bar {
-  height: 12px;
-  border-radius: 4px;
-  background: var(--dsw-alias-bg-skeleton);
-}
-
-.barCalls {
-  background: var(--dsw-alias-brand-primary-new-colorprimary-new-color);
-}
-
-.meta {
-  color: var(--dsw-alias-label-caption);
-  font: var(--dsw-font-xs-13);
-}
-
-/* run_code sub-span lanes: one row per sub-dispatch under its turn row,
-   offset/width scaled to the dispatch window (real wall time). A running
-   lane pulses via reduced opacity until its settle arrives. */
-.subRow {
-  display: flex;
-  align-items: center;
-  gap: 8px;
-  margin-top: 2px;
-}
-
-.subTag {
-  flex: none;
-  width: 88px;
+  flex: 1;
+  min-height: 0;
+  min-width: 0;
   overflow: hidden;
-  text-overflow: ellipsis;
-  white-space: nowrap;
-  color: var(--dsw-alias-label-tertiary);
-  font: var(--dsw-font-xs-13);
-}
-
-.barSub {
-  height: 8px;
-  background: var(--dsw-alias-state-business-primary);
-}
-
-.barSub[data-timing='running'] {
-  opacity: 0.45;
-}
-
-/* Settle-only replay entries: no measured span — hollow, not a solid bar. */
-.barSub[data-timing='unknown'] {
-  background: transparent;
-  border: 1px dashed var(--dsw-alias-state-business-primary);
 }
diff --git a/packages/client/ui-trajectory/src/invariant.ts b/packages/client/ui-trajectory/src/invariant.ts
index 8638145fbe..11e56bb058 100644
--- a/packages/client/ui-trajectory/src/invariant.ts
+++ b/packages/client/ui-trajectory/src/invariant.ts
@@ -16,8 +16,8 @@ export const inject = ['invariants']
 
 /**
  * No runtime invariant: a pure-consumer plugin — it emits no cordis events
- * and owns no mutable cross-plugin state; both view-slot registrations are
- * plain effects whose disposal the slot ledger's own specs and this
+ * and owns no mutable cross-plugin state; its view-slot registration is a
+ * plain effect whose disposal the slot ledger's own specs and this
  * package's behavior specs observe directly.
  */
 const install: InvariantInstaller = () => {}
diff --git a/packages/client/ui-trajectory/tests/cell.spec.tsx b/packages/client/ui-trajectory/tests/cell.spec.tsx
index d6bc1932b3..30f5814998 100644
--- a/packages/client/ui-trajectory/tests/cell.spec.tsx
+++ b/packages/client/ui-trajectory/tests/cell.spec.tsx
@@ -16,11 +16,11 @@ afterEach(cleanup)
 describe('formatElapsedSeconds', () => {
   it('formats known durations and uses an em dash when absent', () => {
     expect(formatElapsedSeconds(null)).toBe('—')
-    expect(formatElapsedSeconds(235)).toBe('+235s')
-    expect(formatElapsedSeconds(235.0)).toBe('+235s')
-    expect(formatElapsedSeconds(235.2)).toBe('+235.2s')
-    expect(formatElapsedSeconds(235.25)).toBe('+235.3s')
-    expect(formatElapsedSeconds(0)).toBe('+0s')
+    expect(formatElapsedSeconds(235)).toBe('235 s')
+    expect(formatElapsedSeconds(235.0)).toBe('235 s')
+    expect(formatElapsedSeconds(235.2)).toBe('235.2 s')
+    expect(formatElapsedSeconds(235.25)).toBe('235.3 s')
+    expect(formatElapsedSeconds(0)).toBe('0 s')
     expect(formatElapsedSeconds(Number.NaN)).toBe('—')
   })
 })
@@ -38,7 +38,7 @@ describe('TrajectoryCell', () => {
     expect(screen.getByText('#6')).toBeTruthy()
     expect(screen.getByText('Tool')).toBeTruthy()
     expect(screen.getByText('bash · Read src/index.ts')).toBeTruthy()
-    expect(screen.getByText('+5s')).toBeTruthy()
+    expect(screen.getByText('5 s')).toBeTruthy()
   })
 
   it('Message rows expose Input / Output / Think metric columns before time', () => {
@@ -57,11 +57,11 @@ describe('TrajectoryCell', () => {
     expect(screen.getByText('136')).toBeTruthy()
     expect(screen.getByText('381')).toBeTruthy()
     expect(screen.getByText('155')).toBeTruthy()
-    expect(screen.getByText('+235.2s')).toBeTruthy()
+    expect(screen.getByText('235.2 s')).toBeTruthy()
     const texts = [...container.querySelectorAll('span')].map(el => el.textContent)
     expect(texts.indexOf('136')).toBeLessThan(texts.indexOf('381'))
     expect(texts.indexOf('381')).toBeLessThan(texts.indexOf('155'))
-    expect(texts.indexOf('155')).toBeLessThan(texts.indexOf('+235.2s'))
+    expect(texts.indexOf('155')).toBeLessThan(texts.indexOf('235.2 s'))
   })
 
   it('selected marks the row for the brand-primary inset ring', () => {
diff --git a/packages/client/ui-trajectory/tests/client-bundle.spec.ts b/packages/client/ui-trajectory/tests/client-bundle.spec.ts
index edaf34a2cb..129f756aa6 100644
--- a/packages/client/ui-trajectory/tests/client-bundle.spec.ts
+++ b/packages/client/ui-trajectory/tests/client-bundle.spec.ts
@@ -3,7 +3,7 @@
  * Real tsdown artifact shape: lib/client.js hands off through
  * window.__ModuleLoader__.load, resolves externals through the injected
  * require, returns the export surface (apply + inject), and a mounted apply
- * registers both view tabs into a real SlotsService ring. Skips when dist/ is
+ * registers the view tab into a real SlotsService ring. Skips when dist/ is
  * not built (`pnpm --filter @deepseek-ai/dsh-client-ui-trajectory bundle`).
  */
 import { readFileSync } from 'node:fs'
@@ -47,6 +47,7 @@ describe('tsdown client artifact', () => {
     const modules = new Map<string, unknown>([
       ['react', await import('react')],
       ['react/jsx-runtime', await import('react/jsx-runtime')],
+      ['@deepseek-ai/dsh-client-ui-primitives', await import('@deepseek-ai/dsh-client-ui-primitives')],
     ])
     const surface = handoff!.factory((spec) => {
       if (!modules.has(spec)) throw new Error(`unexpected require: ${spec}`)
@@ -59,10 +60,10 @@ describe('tsdown client artifact', () => {
     const { handoff, surface } = await loadArtifact()
     expect(handoff.id).toBe(PLUGIN_ID)
     expect(surface.apply).toBeTypeOf('function')
-    expect(surface.inject).toEqual(['slots', 'conversation'])
+    expect(surface.inject).toEqual(['slots', 'conversation', 'sessions'])
   })
 
-  it.skipIf(code === undefined)('mounted as an object plugin, apply registers both view tabs on the real ring', async () => {
+  it.skipIf(code === undefined)('mounted as an object plugin, apply registers the view tab on the real ring', async () => {
     const { surface } = await loadArtifact()
     const ctx = new Context()
     const slots = new SlotsService(ctx)
@@ -71,13 +72,13 @@ describe('tsdown client artifact', () => {
       name: 'root',
       children: { 'conversation.view': { kind: 'list', scope: 'session' } },
     }, (_p: { renderSlot?: unknown }) => null)
-    // The plugin injects 'conversation' as an ordering edge (the declaring
-    // plugin provides it after declaring the ring); the bench declares the
-    // ring itself, so a stub satisfies the wait.
+    // The plugin injects 'conversation' as an ordering edge and 'sessions'
+    // for its per-session history callback; this bench supplies both.
     ctx.provide('conversation', {})
+    ctx.provide('sessions', {})
     const fiber = ctx.plugin(surface as { apply: (ctx: Context) => void })
     await fiber.await()
-    expect(slots.entries('conversation.view').map(e => e.options.id)).toEqual(['trajectory', 'waterfall'])
+    expect(slots.entries('conversation.view').map(e => e.options.id)).toEqual(['trajectory'])
     await fiber.dispose()
     expect(slots.entries('conversation.view')).toHaveLength(0)
   })
diff --git a/packages/client/ui-trajectory/tests/context-branches.spec.ts b/packages/client/ui-trajectory/tests/context-branches.spec.ts
new file mode 100644
index 0000000000..dc5068a471
--- /dev/null
+++ b/packages/client/ui-trajectory/tests/context-branches.spec.ts
@@ -0,0 +1,83 @@
+import { describe, expect, it } from 'vitest'
+import type {
+  ConversationContext, ConversationNode, RequestView,
+} from '@deepseek-ai/dsh-client-runtime/client'
+import {
+  deriveTrajectoryContextBranches,
+  trajectoryBranchContainsRequest,
+} from '../src/client/context-branches.ts'
+
+const checkpoint = {
+  kind: 'context',
+  seq: 100,
+  time: 100,
+  content: [],
+  source: { kind: 'plugin', plugin: 'compact' },
+} as ConversationNode
+
+const abandoned = {
+  kind: 'assistant',
+  seq: 20,
+  time: 20,
+  turn: 1,
+  step: 1,
+  blocks: [{ kind: 'text', text: 'abandoned' }],
+} as ConversationNode
+
+const current = {
+  kind: 'user',
+  seq: 110,
+  time: 110,
+  content: [{ type: 'text', text: 'rewound' }],
+  source: { kind: 'plugin', plugin: 'rewind' },
+} as ConversationNode
+
+function request(
+  purpose: RequestView['purpose'],
+  startSeq: number,
+  resultSeq?: number,
+  replacementSeq?: number,
+): RequestView {
+  return {
+    purpose,
+    startSeq,
+    turn: 1,
+    step: purpose === 'assistant' ? 1 : 0,
+    startedAt: startSeq,
+    completedAt: startSeq + 1,
+    status: 'complete',
+    ...(resultSeq === undefined ? {} : { resultSeq }),
+    ...(replacementSeq === undefined ? {} : { replacementSeq }),
+  }
+}
+
+describe('trajectory context branches', () => {
+  it('inherits nodes and requests by retained surface position rather than seq cutoff', () => {
+    const contexts: ConversationContext[] = [
+      { id: 0, nodes: [checkpoint, abandoned] },
+      {
+        id: 1,
+        parentId: 0,
+        origin: 'rewind',
+        originSeq: 110,
+        nodes: [checkpoint, current],
+      },
+    ]
+    const branches = deriveTrajectoryContextBranches(contexts)
+    const successor = branches[1]!
+
+    expect(successor.nodes.map(node => node.seq)).toEqual([110])
+    expect(trajectoryBranchContainsRequest(
+      successor,
+      request('assistant', 10, 20),
+    )).toBe(false)
+    expect(trajectoryBranchContainsRequest(
+      successor,
+      request('compaction', 90, 95, 100),
+    )).toBe(true)
+    expect(trajectoryBranchContainsRequest(
+      successor,
+      request('assistant', 111),
+    )).toBe(true)
+  })
+})
diff --git a/packages/client/ui-trajectory/tests/layout.spec.tsx b/packages/client/ui-trajectory/tests/layout.spec.tsx
index 5394d6ab09..c83cffb846 100644
--- a/packages/client/ui-trajectory/tests/layout.spec.tsx
+++ b/packages/client/ui-trajectory/tests/layout.spec.tsx
@@ -139,7 +139,7 @@ describe('deriveTrajectoryLayout', () => {
       },
     ] as unknown as ConversationSnapshot['nodes']
     const turns = deriveTrajectoryLayout({ codeDispatches: new Map(), nodes, partial: null, runningCalls: [] })
-    expect(turns[0]?.groups[0]?.description).toBe('2.9s bash×2')
+    expect(turns[0]?.groups[0]?.description).toBe('3 s bash×2')
   })
 
   it('assigns each user message to its enclosing turn instead of pooling into Turn 1', () => {
@@ -161,7 +161,7 @@ describe('deriveTrajectoryLayout', () => {
     expect(turns[1]?.groups.flatMap(g => g.cells.map(c => c.text))).toEqual(['second', 'ok2'])
   })
 
-  it('keeps usage on the fallback Message row when assistant has no text block', () => {
+  it('keeps usage and a meaningful summary when assistant has no text block', () => {
     const nodes = [
       {
         kind: 'assistant', seq: 1, time: 5_000, turn: 1, step: 0,
@@ -172,7 +172,7 @@ describe('deriveTrajectoryLayout', () => {
     const turns = deriveTrajectoryLayout({ codeDispatches: new Map(), nodes, partial: null, runningCalls: [] })
     const message = turns[0]?.groups.flatMap(g => g.cells).find(c => c.kind === 'message')
     expect(message).toMatchObject({
-      text: '', input: 11, output: 22, think: 3,
+      text: '…', input: 11, output: 22, think: 3,
     })
   })
 
@@ -204,6 +204,23 @@ describe('deriveTrajectoryLayout', () => {
     // From context at 9s, not from the earlier user/tool surfaces.
     expect(message?.timeSeconds).toBe(1)
   })
+
+  it('uses the recorded step start for assistant duration when timing exists', () => {
+    const nodes = [
+      { kind: 'user', seq: 1, time: 1_000, content: [{ type: 'text', text: 'hi' }], source: null },
+      {
+        kind: 'assistant', seq: 2, time: 4_000, turn: 1, step: 1,
+        blocks: [{ kind: 'text', text: 'done' }],
+        timing: { stepStartTime: 3_000, firstTokenTime: 3_500, completedTime: 4_000 },
+      },
+    ] as unknown as ConversationSnapshot['nodes']
+    const turns = deriveTrajectoryLayout({
+      codeDispatches: new Map(), nodes, partial: null, runningCalls: [],
+    })
+    const message = turns[0]?.groups.flatMap(group => group.cells)
+      .find(cell => cell.kind === 'message')
+    expect(message).toMatchObject({ startedAt: 3_000, timeSeconds: 1 })
+  })
 })
 
 describe('run_code sub-dispatch cells', () => {
@@ -235,11 +252,12 @@ describe('run_code sub-dispatch cells', () => {
     ]]]) as unknown as ConversationSnapshot['codeDispatches']
     const turns = deriveTrajectoryLayout({ codeDispatches, nodes: runCodeNodes, partial: null, runningCalls: [] })
     const cells = turns[0]!.groups.flatMap(g => g.cells)
-    expect(cells.map(c => c.kind)).toEqual(['tool', 'subtool', 'subtool'])
+    expect(cells.map(c => c.kind)).toEqual(['message', 'tool', 'subtool', 'subtool'])
+    expect(cells[0]?.text).toBe('Tool call only')
     // Sequential indexes across the interleave; durations from the pair times.
-    expect(cells.map(c => c.index)).toEqual([1, 2, 3])
-    expect(cells[1]).toMatchObject({ text: 'bash · {"x":1}', timeSeconds: 1 })
-    expect(cells[2]).toMatchObject({ timeSeconds: 0.5 })
+    expect(cells.map(c => c.index)).toEqual([1, 2, 3, 4])
+    expect(cells[2]).toMatchObject({ text: 'bash · {"x":1}', timeSeconds: 1 })
+    expect(cells[3]).toMatchObject({ timeSeconds: 0.5 })
   })
 
   it('a running (unsettled) sub-call renders a subtool cell with blank time', () => {
diff --git a/packages/client/ui-trajectory/tests/table.spec.tsx b/packages/client/ui-trajectory/tests/table.spec.tsx
new file mode 100644
index 0000000000..fb21ca45b7
--- /dev/null
+++ b/packages/client/ui-trajectory/tests/table.spec.tsx
@@ -0,0 +1,190 @@
+// @vitest-environment jsdom
+/** Trajectory ledger selection, details, status, and fold behavior. */
+
+import { afterEach, describe, expect, it, vi } from 'vitest'
+import { cleanup, fireEvent, render, screen } from '@testing-library/react'
+import { TrajectoryTable } from '../src/client/TrajectoryTable.tsx'
+import type { TrajectoryTurnModel } from '../src/client/layout.ts'
+
+afterEach(cleanup)
+
+const TURNS: readonly TrajectoryTurnModel[] = [{
+  turn: 1,
+  groups: [{
+    title: 'Step 1',
+    description: '1.5s bash×2',
+    cells: [
+      {
+        index: 1,
+        kind: 'message',
+        text: 'Checking files',
+        outputDetail: 'Checking files',
+        input: 10,
+        output: 20,
+        think: 5,
+        timeSeconds: 1.5,
+        assistantMetrics: {
+          timingRecorded: true,
+          stepStartTime: 1_000,
+          firstTokenTime: 1_500,
+          completedTime: 2_500,
+          usageProvided: true,
+          outputTokens: 20,
+        },
+      },
+      {
+        index: 2,
+        kind: 'tool',
+        text: 'bash · {"command":"pwd"}',
+        inputDetail: '{"command":"pwd"}',
+        timeSeconds: null,
+      },
+      {
+        index: 3,
+        kind: 'tool',
+        text: 'bash · {"command":"false"}',
+        inputDetail: '{"command":"false"}',
+        outputDetail: 'ToolError: non_zero_exit',
+        result: 'non_zero_exit',
+        isError: true,
+        timeSeconds: 0.2,
+      },
+    ],
+  }],
+}]
+
+const FOLD_PROPS = {
+  collapsedTurns: new Set<number>(),
+  onToggleTurn: () => {},
+  collapsedAssistants: new Set<number>(),
+  onToggleAssistant: () => {},
+}
+
+describe('TrajectoryTable', () => {
+  it('shows assistant timing facts after keyboard selection', () => {
+    render(<TrajectoryTable turns={TURNS} {...FOLD_PROPS} />)
+    fireEvent.keyDown(screen.getByRole('row', { name: /ASSISTANT/ }), { key: 'Enter' })
+    fireEvent.click(screen.getByRole('button', { name: 'Timing' }))
+
+    expect(screen.getByText('500 ms')).toBeTruthy()
+    expect(screen.getByText('1.00 s')).toBeTruthy()
+    expect(screen.getByText('20.0 tok/s')).toBeTruthy()
+  })
+
+  it('breaks output tokens into labeled reasoning and content rows', () => {
+    render(<TrajectoryTable turns={TURNS} {...FOLD_PROPS} />)
+    fireEvent.click(screen.getByRole('row', { name: /ASSISTANT/ }))
+
+    expect(screen.getByText('Tokens')).toBeTruthy()
+    expect(screen.getByText('20 tok')).toBeTruthy()
+    expect(screen.getByText('Reasoning')).toBeTruthy()
+    expect(screen.getByText('5 tok')).toBeTruthy()
+    expect(screen.getByText('Content')).toBeTruthy()
+    expect(screen.getByText('15 tok')).toBeTruthy()
+  })
+
+  it('keeps raw HTML tags in a Markdown-derived context preview', () => {
+    const html = [
+      '<background-task-complete id="trajectory-ui-watch">',
+      'Command: pnpm test',
+      'Exit code: 0',
+      '</background-task-complete>',
+    ].join('\n')
+    const turns: readonly TrajectoryTurnModel[] = [{
+      turn: 1,
+      groups: [{
+        title: 'Message',
+        cells: [{
+          index: 1,
+          kind: 'context',
+          text: '',
+          inputDetail: html,
+          timeSeconds: 0,
+        }],
+      }],
+    }]
+
+    render(<TrajectoryTable turns={turns} {...FOLD_PROPS} />)
+
+    expect(screen.getByText(
+      '<background-task-complete id="trajectory-ui-watch"> Command: pnpm test Exit code: 0 </background-task-complete>',
+    )).toBeTruthy()
+  })
+
+  it('clears the selected row when ledger whitespace is clicked', () => {
+    const onClearSelection = vi.fn()
+    render(
+      <TrajectoryTable
+        turns={TURNS}
+        {...FOLD_PROPS}
+        onClearSelection={onClearSelection}
+      />,
+    )
+    const row = screen.getByRole('row', { name: /ASSISTANT/ })
+    fireEvent.click(row)
+
+    expect(row.getAttribute('aria-selected')).toBe('true')
+    expect(screen.getByRole('complementary', { name: 'Event details' })).toBeTruthy()
+
+    const tablePane = screen.getByRole('table').parentElement
+    expect(tablePane).not.toBeNull()
+    fireEvent.click(tablePane as HTMLElement)
+
+    expect(row.getAttribute('aria-selected')).toBe('false')
+    expect(screen.queryByRole('complementary', { name: 'Event details' })).toBeNull()
+    expect(onClearSelection).toHaveBeenCalledOnce()
+  })
+
+  it('keeps running and failure semantics distinct from record roles', () => {
+    const view = render(<TrajectoryTable turns={TURNS} {...FOLD_PROPS} />)
+    expect(view.container.querySelector('tr[data-kind="tool"][data-running="true"]')).toBeTruthy()
+    expect(view.container.querySelector('tr[data-kind="tool"][data-error="true"]')).toBeTruthy()
+
+    fireEvent.click(screen.getByRole('row', { name: /TOOL, bash \{"command":"pwd"\}/ }))
+    expect(screen.getByText('Pending')).toBeTruthy()
+    fireEvent.click(screen.getByRole('row', { name: /TOOL, bash \{"command":"false"\}/ }))
+    expect(screen.getByText('Failed')).toBeTruthy()
+    fireEvent.click(screen.getByRole('tab', { name: 'Result' }))
+    expect(screen.getByText('ToolError: non_zero_exit')).toBeTruthy()
+  })
+
+  it('renders a single-text JSON tool result as a JSON tree', () => {
+    const turns: readonly TrajectoryTurnModel[] = [{
+      turn: 1,
+      groups: [{
+        title: 'Step 1',
+        cells: [{
+          index: 1,
+          kind: 'tool',
+          text: 'read {"path":"result.json"}',
+          outputDetail: '{"value":1,"nested":{"ok":true}}',
+          outputBlocks: [{
+            type: 'text',
+            content: '{"value":1,"nested":{"ok":true}}',
+          }],
+          timeSeconds: 0.1,
+        }],
+      }],
+    }]
+
+    render(<TrajectoryTable turns={turns} {...FOLD_PROPS} />)
+    fireEvent.click(screen.getByRole('row', { name: /TOOL/ }))
+    fireEvent.click(screen.getByRole('tab', { name: 'Result' }))
+
+    expect(screen.getByRole('tree', { name: 'Result JSON' })).toBeTruthy()
+    expect(screen.getByText('value:')).toBeTruthy()
+  })
+
+  it('keeps the first row and a compact summary when a turn is collapsed', () => {
+    render(
+      <TrajectoryTable
+        turns={TURNS}
+        {...FOLD_PROPS}
+        collapsedTurns={new Set([1])}
+      />,
+    )
+    expect(screen.queryByRole('columnheader')).toBeNull()
+    expect(screen.getByRole('row', { name: /ASSISTANT/ })).toBeTruthy()
+    expect(screen.getByRole('row', { name: /Collapsed turn summary/ })).toBeTruthy()
+  })
+})
diff --git a/packages/client/ui-trajectory/tests/views.spec.tsx b/packages/client/ui-trajectory/tests/views.spec.tsx
index 917beaccc6..c79b7407ab 100644
--- a/packages/client/ui-trajectory/tests/views.spec.tsx
+++ b/packages/client/ui-trajectory/tests/views.spec.tsx
@@ -1,31 +1,34 @@
 // @vitest-environment jsdom
 /**
  * View registration acceptance on the real framework stack: the plugin fiber
- * registers trajectory/waterfall into a real SlotsService view ring, tabs
+ * registers Trajectory into a real SlotsService view ring, tabs
  * switch inside ConversationRoot (renderSlot share driven by the same tab
  * projection apply uses) without collapsing chat, trajectory renders the
- * turn-list chrome (no span stats bar), waterfall keeps in-body stats, and
- * fiber disposal removes both tabs. Span derivation edge cases ride along.
+ * event ledger with its timing overview, and fiber disposal removes the tab.
+ * Timeline projection and inclusive focus edge cases ride along.
  */
 import { Context } from 'cordis'
 import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
 import { cleanup, fireEvent, render, screen } from '@testing-library/react'
-import { createElement, type FC, type ReactNode } from 'react'
+import { createElement, type ComponentProps, type FC, type ReactNode } from 'react'
 import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react'
 import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
 import type { UseSession } from '@deepseek-ai/dsh-client-web-react'
 import { SlotsService } from '@deepseek-ai/dsh-client-runtime/client'
-import type { ConversationSnapshot, SessionId, SessionListState, WorkspaceListState } from '@deepseek-ai/dsh-client-runtime/client'
+import type {
+  ConversationSnapshot, RequestView, SessionHistoryFace, SessionHistoryInspection,
+  SessionHistorySnapshot, SessionId, SessionListState, WorkspaceListState,
+} from '@deepseek-ai/dsh-client-runtime/client'
 import type { ConvViewProps, ViewTab } from '@deepseek-ai/dsh-client-ui-conversation/client'
-// Export discipline: packages/client/AGENTS.md.
 import { ConversationSession, type ConversationSessionProps } from '@deepseek-ai/dsh-client-ui-conversation/src/client/skeleton/ConversationSession.tsx'
 import { createChatStore } from '@deepseek-ai/dsh-client-ui-conversation/src/client/stores.ts'
 import { apply, inject } from '@deepseek-ai/dsh-client-ui-trajectory/client'
-import { deriveSpans, deriveSpanStats, deriveSubSpans } from '@deepseek-ai/dsh-client-ui-trajectory/src/client/spans.ts'
-import { TrajectoryStatsHeader } from '@deepseek-ai/dsh-client-ui-trajectory/src/client/TrajectoryStatsHeader.tsx'
-import { TrajectoryView } from '@deepseek-ai/dsh-client-ui-trajectory/src/client/TrajectoryView.tsx'
-import { WaterfallView } from '@deepseek-ai/dsh-client-ui-trajectory/src/client/WaterfallView.tsx'
 import { apply as nodeApply } from '@deepseek-ai/dsh-client-ui-trajectory'
+import type { TrajectoryTurnModel } from '../src/client/layout.ts'
+import {
+  TrajectoryView, type TrajectoryViewInjected,
+} from '../src/client/TrajectoryView.tsx'
+import { deriveTrajectoryTimeline } from '../src/client/timeline.ts'
 
 const SID = 's1' as SessionId
 afterEach(cleanup)
@@ -40,17 +43,56 @@ beforeEach(() => {
 /** Node fixture: user prologue, two turns, one tool result inside turn 1. */
 const NODES = [
   { kind: 'user', seq: 1, time: 1_000, content: [], source: null },
-  { kind: 'assistant', seq: 2, time: 2_000, turn: 1, step: 1, blocks: [] },
   {
-    kind: 'tool-result', seq: 3, time: 3_000, callId: 'c1', call: null, callTime: null,
+    kind: 'assistant', seq: 2, time: 2_000, turn: 1, step: 1, blocks: [],
+    timing: { stepStartTime: 1_800, firstTokenTime: 1_900, completedTime: 2_000 },
+  },
+  {
+    kind: 'tool-result', seq: 3, time: 3_000, callId: 'c1', call: null, callTime: 2_200,
     content: [], isError: false, callView: null, resultView: null,
   },
-  { kind: 'assistant', seq: 4, time: 4_000, turn: 2, step: 1, blocks: [] },
+  {
+    kind: 'assistant', seq: 4, time: 4_000, turn: 2, step: 1, blocks: [],
+    timing: { stepStartTime: 3_500, firstTokenTime: 3_700, completedTime: 4_000 },
+  },
 ] as unknown as ConversationSnapshot['nodes']
 
+function historySnapshot(
+  nodes: ConversationSnapshot['nodes'],
+  inspection: Partial<SessionHistoryInspection> = {},
+): SessionHistorySnapshot {
+  return {
+    state: 'ready',
+    error: null,
+    hasMore: false,
+    inspection: {
+      eventNodes: nodes,
+      contexts: [{ id: 0, nodes }],
+      requests: [],
+      callSchemas: new Map(),
+      interruptedNodes: [],
+      partial: null,
+      runningCalls: [],
+      codeDispatches: new Map(),
+      ...inspection,
+    },
+  }
+}
+
+function standaloneHistory(
+  snapshot: SessionHistorySnapshot,
+): Pick<ComponentProps<typeof TrajectoryView>, 'useHistory' | 'loadAllHistory'> {
+  const store = createSnapshotStore(snapshot)
+  return {
+    useHistory: bindSnapshotSelector(store),
+    loadAllHistory: () => Promise.resolve(),
+  }
+}
+
 function fakeSession(nodes: ConversationSnapshot['nodes']) {
   const store = createSnapshotStore({
-    nodes, partial: null, runningCalls: [] as ConversationSnapshot['runningCalls'], codeDispatches: new Map(),
+    nodes, pending: [], partial: null,
+    runningCalls: [] as ConversationSnapshot['runningCalls'], codeDispatches: new Map(),
   })
   return { store, useSession: bindSnapshotSelector(store) as unknown as UseSession<ConversationSnapshot> }
 }
@@ -82,9 +124,17 @@ function standaloneProps(nodes: ConversationSnapshot['nodes']): ConvViewProps {
 }
 
 /** Real-stack bench: root Context + real SlotsService ring + the plugin fiber. */
-async function bench() {
+async function bench(snapshot = historySnapshot(NODES)) {
   const ctx = new Context()
   const slots = new SlotsService(ctx)
+  const loadAllHistory = vi.fn((_signal: AbortSignal) => Promise.resolve())
+  const historyStore = createSnapshotStore(snapshot)
+  const history: SessionHistoryFace = {
+    sessionId: SID,
+    getSnapshot: () => historyStore.getSnapshot(),
+    subscribe: listener => historyStore.subscribe(listener),
+    loadAll: loadAllHistory,
+  }
   // The conversation entry's role: declare the ring, then seed the chat entry.
   slots.register({
     name: 'root',
@@ -96,9 +146,10 @@ async function bench() {
   // 'conversation' inject is an ordering edge; the bench declares the ring
   // itself, so a stub satisfies the wait.
   ctx.provide('conversation', {})
+  ctx.provide('sessionHistory', { source: () => history })
   const fiber = ctx.plugin({ inject: [...inject], apply })
   await fiber.await()
-  return { ctx, slots, fiber }
+  return { ctx, slots, fiber, loadAllHistory }
 }
 
 /** Tab projection twin of apply's viewTabs (the render-side consumption path). */
@@ -111,6 +162,8 @@ function tabsOf(slots: SlotsService): ViewTab[] {
 function mount(slots: SlotsService, nodes: ConversationSnapshot['nodes'] = NODES) {
   const sessionSnapshot = createSnapshotStore({
     running: false, removed: false, promptError: null, nodes,
+    pending: [],
+    openState: 'open' as const, hasMore: true, loadingOlder: false,
     partial: null, runningCalls: [] as ConversationSnapshot['runningCalls'], codeDispatches: new Map(),
   })
   const useSession = bindSnapshotSelector(sessionSnapshot) as unknown as UseSession<ConversationSnapshot>
@@ -122,8 +175,21 @@ function mount(slots: SlotsService, nodes: ConversationSnapshot['nodes'] = NODES
     const entry = slots.entries('conversation.view').find(e => e.options.id === opts?.only)
     if (entry === undefined) return null
     const View = entry.component as FC<ConvViewProps>
+    const injectEntry = entry.inject as ((sessionId: SessionId) => object) | undefined
+    const injected = injectEntry === undefined
+      ? {}
+      : injectEntry(SID)
+    const injectedProps = 'hooks' in injected
+      ? {
+        loadAllHistory: (injected as TrajectoryViewInjected).loadAllHistory,
+        useHistory: bindSnapshotSelector(
+          (injected as TrajectoryViewInjected).hooks.history,
+        ),
+      }
+      : injected
     return (
       <View
+        {...injectedProps}
         {...({ sessionId: SID, useSession, useSessions: emptySessions(), useWorkspaces: emptyWorkspaces() } as unknown as ConvViewProps)}
         key={key}
       />
@@ -154,16 +220,15 @@ function mount(slots: SlotsService, nodes: ConversationSnapshot['nodes'] = NODES
 }
 
 describe('plugin registration', () => {
-  it('registers trajectory and waterfall after chat on the ring', async () => {
+  it('registers trajectory after chat on the ring', async () => {
     const b = await bench()
     expect(tabsOf(b.slots)).toEqual([
       { id: 'chat', label: 'Chat' },
       { id: 'trajectory', label: 'Trajectory' },
-      { id: 'waterfall', label: 'Waterfall' },
     ])
   })
 
-  it('fiber disposal removes both tabs and leaves chat standing', async () => {
+  it('fiber disposal removes the tab and leaves chat standing', async () => {
     const b = await bench()
     await b.fiber.dispose()
     expect(tabsOf(b.slots).map(v => v.id)).toEqual(['chat'])
@@ -171,80 +236,275 @@ describe('plugin registration', () => {
 })
 
 describe('tab switching in ConversationRoot', () => {
-  it('renders all three tabs, defaults to chat, and switches to trajectory without stats chrome', async () => {
+  it('renders two tabs, defaults to chat, and switches to the trajectory ledger', async () => {
     const b = await bench()
-    mount(b.slots)
+    const view = mount(b.slots)
     expect(screen.getByTestId('chat-body')).toBeTruthy()
-    expect(screen.getAllByRole('tab').map(t => t.textContent)).toEqual(['Chat', 'Trajectory', 'Waterfall'])
+    expect(screen.getAllByRole('tab').map(t => t.textContent)).toEqual(['Chat', 'Trajectory'])
 
     fireEvent.click(screen.getByRole('tab', { name: 'Trajectory' }))
     expect(screen.queryByText(/turns ·/)).toBeNull()
-    expect(screen.getByText('Turn 1')).toBeTruthy()
-    expect(screen.getByText('Turn 2')).toBeTruthy()
-    expect(screen.getAllByText('Message').length).toBeGreaterThan(0)
-    expect(screen.getAllByText('Step 1').length).toBeGreaterThan(0)
-    expect(screen.getAllByText('Input').length).toBeGreaterThan(0)
+    expect(view.container.querySelectorAll('tr[data-turn-start="true"]')).toHaveLength(2)
+    expect(screen.queryByRole('columnheader')).toBeNull()
+    expect(screen.getByRole('toolbar', { name: 'Trajectory toolbar' })).toBeTruthy()
+    expect(screen.getByRole('region', { name: 'Trajectory timeline' })).toBeTruthy()
+    fireEvent.click(screen.getByRole('button', { name: 'Collapse turns' }))
+    expect(view.container.querySelector('[data-collapsed-summary="turn"]')).toBeTruthy()
+    fireEvent.click(screen.getByRole('button', { name: 'Expand turns' }))
+    expect(screen.getByRole('row', { name: /USER/ })).toBeTruthy()
     expect(screen.queryByTestId('chat-body')).toBeNull()
+    await vi.waitFor(() => {
+      expect(b.loadAllHistory).toHaveBeenCalledOnce()
+    })
+    const signal = b.loadAllHistory.mock.calls[0]?.[0]
+    expect(signal?.aborted).toBe(false)
+    fireEvent.click(screen.getByRole('tab', { name: 'Chat' }))
+    expect(signal?.aborted).toBe(true)
   })
 
-  it('waterfall renders bars and switching back to chat does not collapse it', async () => {
+  it('opens a local record inspector and switches payload tabs without opening chat details', async () => {
     const b = await bench()
     mount(b.slots)
-    fireEvent.click(screen.getByRole('tab', { name: 'Waterfall' }))
-    expect(screen.getByTitle('2 nodes')).toBeTruthy()
-    expect(screen.getByTitle('1 tool calls')).toBeTruthy()
+    fireEvent.click(screen.getByRole('tab', { name: 'Trajectory' }))
 
-    fireEvent.click(screen.getByRole('tab', { name: 'Chat' }))
-    expect(screen.getByTestId('chat-body')).toBeTruthy()
+    fireEvent.keyDown(screen.getByRole('row', { name: /TOOL/ }), { key: 'Enter' })
+    expect(screen.getByRole('complementary', { name: 'Event details' })).toBeTruthy()
+    expect(screen.getByText('Turn 1 · Step 1')).toBeTruthy()
+    expect(screen.getByText('Completed')).toBeTruthy()
+    expect(screen.getByRole('tab', { name: 'Result' })).toBeTruthy()
+
+    fireEvent.click(screen.getByRole('button', { name: 'Close details' }))
+    expect(screen.queryByRole('complementary', { name: 'Event details' })).toBeNull()
   })
 
-  it('empty window: placeholder copy in the body, the stats header renders nothing', async () => {
+  it('dragging the overview focuses overlapping records without filtering the ledger', async () => {
     const b = await bench()
-    mount(b.slots, [])
+    mount(b.slots)
     fireEvent.click(screen.getByRole('tab', { name: 'Trajectory' }))
-    expect(screen.getByText('暂无轨迹数据')).toBeTruthy()
+    const plot = screen.getByLabelText('Timeline overview; drag horizontally to focus events')
+    vi.spyOn(plot, 'getBoundingClientRect').mockReturnValue({
+      x: 0, y: 0, left: 0, top: 0, right: 100, bottom: 72, width: 100, height: 72,
+      toJSON: () => ({}),
+    })
+    fireEvent.pointerDown(plot, { button: 0, clientX: 55, pointerId: 1 })
+    fireEvent.pointerMove(plot, { clientX: 95, pointerId: 1 })
+    fireEvent.pointerUp(plot, { clientX: 95, pointerId: 1 })
+
+    expect(screen.getByRole('row', { name: /USER/ }).getAttribute('data-timeline-focus'))
+      .toBe('outside')
+
+    const tablePane = screen.getByRole('table').parentElement
+    expect(tablePane).not.toBeNull()
+    fireEvent.click(tablePane as HTMLElement)
+    expect(screen.getByRole('row', { name: /USER/ }).getAttribute('data-timeline-focus'))
+      .toBeNull()
+
+    fireEvent.pointerDown(plot, { button: 0, clientX: 55, pointerId: 2 })
+    fireEvent.pointerMove(plot, { clientX: 95, pointerId: 2 })
+    fireEvent.pointerUp(plot, { clientX: 95, pointerId: 2 })
+    expect(screen.getByRole('row', { name: /USER/ }).getAttribute('data-timeline-focus'))
+      .toBe('outside')
+    fireEvent.contextMenu(plot)
+    expect(screen.getByRole('row', { name: /USER/ }).getAttribute('data-timeline-focus'))
+      .toBeNull()
+  })
+
+  it('empty window keeps the toolbar and reports no timing data', async () => {
+    const b = await bench(historySnapshot([]))
+    mount(b.slots)
+    fireEvent.click(screen.getByRole('tab', { name: 'Trajectory' }))
+    expect(screen.getByRole('toolbar', { name: 'Trajectory toolbar' })).toBeTruthy()
+    expect(screen.getByText('No timing data')).toBeTruthy()
+    expect(screen.queryByRole('row')).toBeNull()
     expect(screen.queryByText(/turns ·/)).toBeNull()
   })
 })
 
-describe('span derivation', () => {
-  it('attributes prologue to turn 0 and follows steering turn tags', () => {
-    const nodes = [
-      { kind: 'user', seq: 1 },
-      { kind: 'steering', seq: 2, turn: 5 },
-      { kind: 'user', seq: 3 },
-    ] as unknown as ConversationSnapshot['nodes']
-    const spans = deriveSpans(nodes)
-    expect(spans).toEqual([
-      { turn: 0, steps: 0, calls: 0, nodes: 1 },
-      { turn: 5, steps: 0, calls: 0, nodes: 2 },
-    ])
-    expect(deriveSpanStats(spans)).toEqual({ turns: 2, steps: 0, calls: 0 })
+describe('timeline projection', () => {
+  const turns = [{
+    turn: 1,
+    groups: [{
+      title: 'Step 1',
+      cells: [
+        { index: 1, kind: 'message', text: 'assistant', startedAt: 1_000, timeSeconds: 1 },
+        { index: 2, kind: 'tool', text: 'bash', startedAt: 2_000, timeSeconds: 1 },
+        { index: 3, kind: 'user', text: 'unknown', timeSeconds: 0 },
+      ],
+    }],
+  }] satisfies readonly TrajectoryTurnModel[]
+
+  it('uses equal-width operation slots and stable semantic lanes', () => {
+    expect(deriveTrajectoryTimeline(turns)).toEqual({
+      start: 0,
+      end: 3,
+      spans: [
+        {
+          index: 1, kind: 'message', label: 'assistant', lane: 1, start: 0, end: 1,
+        },
+        { index: 2, kind: 'tool', label: 'bash', lane: 2, start: 1, end: 2 },
+        { index: 3, kind: 'user', label: 'unknown', lane: 0, start: 2, end: 3 },
+      ],
+      turnBoundaries: [{ turn: 1, time: 0 }],
+    })
   })
 
-  it('empty inputs produce zero stats and standalone components render their empty forms', () => {
-    expect(deriveSpanStats(deriveSpans([]))).toEqual({ turns: 0, steps: 0, calls: 0 })
-    const { useSession } = fakeSession([])
-    const { container } = render(createElement(TrajectoryStatsHeader, { useSession: useSession }))
-    expect(container.firstChild).toBeNull()
-    render(createElement(TrajectoryView as FC<ConvViewProps>,
-      standaloneProps([])))
-    expect(screen.getByText('暂无轨迹数据')).toBeTruthy()
+  it('ignores durations and idle gaps while retaining turn boundaries', () => {
+    const separatedTurns = [
+      {
+        turn: 1,
+        groups: [{
+          title: 'Step 1',
+          cells: [
+            { index: 1, kind: 'message', text: 'first', startedAt: 1_000, timeSeconds: 1 },
+            { index: 2, kind: 'tool', text: 'within-turn gap', startedAt: 4_000, timeSeconds: 1 },
+          ],
+        }],
+      },
+      {
+        turn: 2,
+        groups: [{
+          title: 'Step 1',
+          cells: [
+            { index: 3, kind: 'message', text: 'after user idle', startedAt: 40_000, timeSeconds: 1 },
+          ],
+        }],
+      },
+    ] satisfies readonly TrajectoryTurnModel[]
+
+    expect(deriveTrajectoryTimeline(separatedTurns)).toMatchObject({
+      start: 0,
+      end: 3,
+      spans: [
+        { index: 1, start: 0, end: 1 },
+        { index: 2, start: 1, end: 2 },
+        { index: 3, start: 2, end: 3 },
+      ],
+      turnBoundaries: [
+        { turn: 1, time: 0 },
+        { turn: 2, time: 2 },
+      ],
+    })
+  })
+
+  it('empty inputs produce no model and the standalone view reports its empty form', () => {
+    expect(deriveTrajectoryTimeline([])).toBeNull()
+    render(createElement(
+      TrajectoryView,
+      {
+        ...standaloneProps([]),
+        ...standaloneHistory(historySnapshot([])),
+      },
+    ))
+    expect(screen.getByRole('toolbar', { name: 'Trajectory toolbar' })).toBeTruthy()
+    expect(screen.queryByRole('row')).toBeNull()
   })
 })
 
-describe('WaterfallView standalone branches', () => {
-  it('empty window renders the placeholder copy', () => {
-    render(createElement(WaterfallView as FC<ConvViewProps>,
-      standaloneProps([])))
-    expect(screen.getByText('暂无瀑布数据')).toBeTruthy()
+describe('TrajectoryView branches', () => {
+  it('renders only the selected rewind branch while retaining session-global requests', () => {
+    const retained = {
+      kind: 'user',
+      seq: 1,
+      time: 1_000,
+      content: [{ type: 'text', text: 'retained user' }],
+      source: null,
+    } as unknown as ConversationSnapshot['nodes'][number]
+    const abandoned = {
+      kind: 'assistant',
+      seq: 3,
+      time: 3_000,
+      turn: 1,
+      step: 1,
+      blocks: [{ kind: 'text', text: 'abandoned response' }],
+    } as unknown as ConversationSnapshot['nodes'][number]
+    const current = {
+      kind: 'assistant',
+      seq: 5,
+      time: 5_000,
+      turn: 2,
+      step: 1,
+      blocks: [{ kind: 'text', text: 'current response' }],
+    } as unknown as ConversationSnapshot['nodes'][number]
+    const request = (startSeq: number, turn: number): RequestView => ({
+      purpose: 'assistant',
+      startSeq,
+      turn,
+      step: 1,
+      startedAt: startSeq * 1_000,
+      completedAt: startSeq * 1_000 + 100,
+      status: 'complete',
+    })
+    const store = createSnapshotStore(historySnapshot(
+      [retained, abandoned, current],
+      {
+        eventNodes: [retained, abandoned, current],
+        contexts: [
+          { id: 0, nodes: [retained, abandoned] },
+          {
+            id: 1,
+            parentId: 0,
+            origin: 'rewind' as const,
+            originSeq: 4,
+            nodes: [retained, current],
+          },
+        ],
+        requests: [request(2, 1), request(4, 2)],
+        callSchemas: new Map(),
+      },
+    ))
+
+    const view = render(
+      <TrajectoryView
+        {...standaloneProps([])}
+        useHistory={bindSnapshotSelector(store)}
+        loadAllHistory={vi.fn(() => Promise.resolve())}
+      />,
+    )
+
+    expect(screen.queryByText('abandoned response')).toBeNull()
+    expect(screen.getByText('current response')).toBeTruthy()
+    expect(screen.getByRole('row', { name: /Request 2, ASSISTANT/ })).toBeTruthy()
+    expect(view.container.querySelectorAll('[data-request-only="true"]')).toHaveLength(0)
   })
 
-  it('a turn without tool calls renders the node bar only', () => {
-    const nodes = [{ kind: 'user', seq: 1 }] as unknown as ConversationSnapshot['nodes']
-    render(createElement(WaterfallView as FC<ConvViewProps>, standaloneProps(nodes)))
-    expect(screen.getByTitle('1 nodes')).toBeTruthy()
-    expect(screen.queryByTitle(/tool calls/)).toBeNull()
+  it('retains cancellation-frozen assistant and tool nodes outside raw contexts', () => {
+    const retained = {
+      kind: 'user', seq: 1, time: 1_000,
+      content: [{ type: 'text', text: 'stop the task' }], source: null,
+    } as unknown as ConversationSnapshot['nodes'][number]
+    const interruptedAssistant = {
+      kind: 'assistant', seq: 2.1, time: 2_000, turn: 1, step: 1,
+      blocks: [{ kind: 'text', text: 'partial response retained' }],
+      interrupted: true,
+    } as unknown as ConversationSnapshot['nodes'][number]
+    const interruptedTool = {
+      kind: 'tool-result', seq: 2.2, time: 2_100, callId: 'slow-call',
+      call: { name: 'bash', argsRaw: '{"command":"sleep 30"}' }, callTime: 1_900,
+      content: [], isError: true,
+      error: { name: 'Interrupted', code: 'interrupted' },
+      callView: null, resultView: null,
+    } as unknown as ConversationSnapshot['nodes'][number]
+    const store = createSnapshotStore(historySnapshot(
+      [retained],
+      {
+        eventNodes: [retained],
+        contexts: [{ id: 0, nodes: [retained] }],
+        requests: [],
+        callSchemas: new Map(),
+        interruptedNodes: [interruptedAssistant, interruptedTool],
+      },
+    ))
+
+    render(
+      <TrajectoryView
+        {...standaloneProps([])}
+        useHistory={bindSnapshotSelector(store)}
+        loadAllHistory={vi.fn(() => Promise.resolve())}
+      />,
+    )
+
+    expect(screen.getByText('partial response retained')).toBeTruthy()
+    expect(screen.getByRole('row', { name: /TOOL, bash/ })).toBeTruthy()
   })
 })
 
@@ -253,117 +513,3 @@ describe('node half', () => {
     expect(() => { nodeApply() }).not.toThrow()
   })
 })
-
-describe('deriveSubSpans (waterfall lanes)', () => {
-  const dispatchNodes = [
-    { kind: 'assistant', seq: 2, time: 6_000, turn: 3, step: 1, blocks: [] },
-    {
-      kind: 'tool-result', seq: 3, time: 9_000, callId: 'p1',
-      call: { name: 'run_code', argsRaw: '{}' }, callTime: 6_100,
-      content: [], isError: false, callView: null, resultView: null,
-    },
-  ] as unknown as ConversationSnapshot['nodes']
-
-  it('scales settled lanes into the dispatch window with real durations', () => {
-    const codeDispatches = new Map([['p1', [
-      {
-        kind: 'tool-result', seq: 101, time: 7_000, callId: 'p1:code:1',
-        call: { name: 'bash', argsRaw: '{}' }, callTime: 6_200,
-        content: [], isError: false, callView: null, resultView: null,
-      },
-      {
-        kind: 'tool-result', seq: 102, time: 8_200, callId: 'p1:code:2',
-        call: { name: 'read', argsRaw: '{}' }, callTime: 7_000,
-        content: [], isError: false, callView: null, resultView: null,
-      },
-    ]]]) as unknown as ConversationSnapshot['codeDispatches']
-    const lanes = deriveSubSpans(dispatchNodes, codeDispatches)
-    const turn3 = lanes.get(3)
-    expect(turn3).toHaveLength(2)
-    // Window = 6200..8200 (2000ms). bash: 0..0.4; read: 0.4..1.0.
-    expect(turn3?.[0]).toMatchObject({ name: 'bash', durationMs: 800, timing: 'measured', offsetFraction: 0 })
-    expect(turn3?.[0]?.widthFraction).toBeCloseTo(0.4)
-    expect(turn3?.[1]).toMatchObject({ name: 'read', durationMs: 1200 })
-    expect(turn3?.[1]?.offsetFraction).toBeCloseTo(0.4)
-  })
-
-  it('a running lane extends to the window end with a null duration', () => {
-    const codeDispatches = new Map([['p1', [
-      {
-        kind: 'tool-result', seq: 101, time: 8_000, callId: 'p1:code:1',
-        call: { name: 'bash', argsRaw: '{}' }, callTime: 6_200,
-        content: [], isError: false, callView: null, resultView: null,
-      },
-      { callId: 'p1:code:2', name: 'grep', argsRaw: '{}', turn: 0, step: 0, time: 7_000, callView: null },
-    ]]]) as unknown as ConversationSnapshot['codeDispatches']
-    const lanes = deriveSubSpans(dispatchNodes, codeDispatches)
-    const running = lanes.get(3)?.find(lane => lane.name === 'grep')
-    expect(running).toMatchObject({ durationMs: null, timing: 'running' })
-    // Extends from its start to the window end.
-    expect(running!.offsetFraction + running!.widthFraction).toBeCloseTo(1)
-  })
-
-  it('a settle-only entry (null callTime) is unknown timing, never a measured 0 ms', () => {
-    const codeDispatches = new Map([['p1', [
-      {
-        kind: 'tool-result', seq: 101, time: 8_000, callId: 'p1:code:1',
-        call: { name: 'bash', argsRaw: '{}' }, callTime: null,
-        content: [], isError: false, callView: null, resultView: null,
-      },
-    ]]]) as unknown as ConversationSnapshot['codeDispatches']
-    const lane = deriveSubSpans(dispatchNodes, codeDispatches).get(3)?.[0]
-    expect(lane).toMatchObject({ durationMs: null, timing: 'unknown' })
-  })
-
-  it('waterfall renders sub-span lanes under the owning turn row', () => {
-    const codeDispatches = new Map([['p1', [
-      {
-        kind: 'tool-result', seq: 101, time: 8_000, callId: 'p1:code:1',
-        call: { name: 'bash', argsRaw: '{}' }, callTime: 6_200,
-        content: [], isError: false, callView: null, resultView: null,
-      },
-    ]]]) as unknown as ConversationSnapshot['codeDispatches']
-    const store = createSnapshotStore({
-      nodes: dispatchNodes, partial: null,
-      runningCalls: [] as ConversationSnapshot['runningCalls'], codeDispatches,
-    })
-    const props = {
-      sessionId: SID,
-      useSession: bindSnapshotSelector(store) as unknown as UseSession<ConversationSnapshot>,
-      useSessions: emptySessions(),
-      useWorkspaces: emptyWorkspaces(),
-      useProjection: (() => undefined) as never,
-    } as unknown as ConvViewProps
-    const view = render(createElement(WaterfallView as FC<ConvViewProps>, props))
-    const lane = view.container.querySelector('[data-subspan]')
-    expect(lane).not.toBeNull()
-    expect(lane!.textContent).toContain('bash')
-    expect(lane!.querySelector('[title*="1.80s"]')).not.toBeNull()
-    expect(lane!.querySelector('[data-timing="measured"]')).not.toBeNull()
-  })
-
-  it('waterfall labels a settle-only lane as duration unknown', () => {
-    const codeDispatches = new Map([['p1', [
-      {
-        kind: 'tool-result', seq: 101, time: 8_000, callId: 'p1:code:1',
-        call: { name: 'read', argsRaw: '{}' }, callTime: null,
-        content: [], isError: false, callView: null, resultView: null,
-      },
-    ]]]) as unknown as ConversationSnapshot['codeDispatches']
-    const store = createSnapshotStore({
-      nodes: dispatchNodes, partial: null,
-      runningCalls: [] as ConversationSnapshot['runningCalls'], codeDispatches,
-    })
-    const props = {
-      sessionId: SID,
-      useSession: bindSnapshotSelector(store) as unknown as UseSession<ConversationSnapshot>,
-      useSessions: emptySessions(),
-      useWorkspaces: emptyWorkspaces(),
-      useProjection: (() => undefined) as never,
-    } as unknown as ConvViewProps
-    const view = render(createElement(WaterfallView as FC<ConvViewProps>, props))
-    const bar = view.container.querySelector('[data-timing="unknown"]')
-    expect(bar).not.toBeNull()
-    expect(bar!.getAttribute('title')).toContain('duration unknown')
-  })
-})
diff --git a/packages/client/ui-workspace/README.i18n.yaml b/packages/client/ui-workspace/README.i18n.yaml
index 00b639f625..536911a16a 100644
--- a/packages/client/ui-workspace/README.i18n.yaml
+++ b/packages/client/ui-workspace/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/client/ui-workspace/README.md
-README.md: 8acf819121b46512d38b39ff858bb2bf797cfe96
-README.zh.md: e97d93f7e38d00af91b43f0df9fb0e3b17ae8ed5
+README.md: a1b58f4abe0925be3b426d10344777e46caa9ba0
+README.zh.md: a472507bc45549c8feb55a75d294cbd7b3138cc5
diff --git a/packages/client/ui-workspace/README.md b/packages/client/ui-workspace/README.md
index 8acf819121..a1b58f4abe 100644
--- a/packages/client/ui-workspace/README.md
+++ b/packages/client/ui-workspace/README.md
@@ -4,7 +4,7 @@ English | [中文](README.zh.md)
 
 Shared Workspace picker plugin. `WorkspaceBrowser` is registered into the sidebar's `sidebar.workspaces` slot and `WorkspacePicker` into the page-local Session Intent hero's `conversation.hero.workspace` slot, so both surfaces use the same menu and creation flow.
 
-The picker lists real Host Workspace entities through the global `useWorkspaces` hook. Selecting a Workspace invokes the slot owner's `onPick` callback to retarget the frontend Session object. Each registration declares a **directory-flow child hole** (`single` kind: `conversation.hero.workspace.directoryFlow` / `sidebar.workspaces.directoryFlow`) that the composed picker package's client half fills with its picking interaction — the [`-native`](../../host/directory-picker-native/README.md) backend's renderless OS-chooser driver today, an in-app browsing dialog under a `-browse` composition. The flat **Open local folder...** action renders only while the surface's hole is occupied (occupancy read per menu render; an empty hole means the composition has no picking affordance — the seam's documented no-flow default). This package owns the trigger and the adoption: the occupant reports one picked path per open through the hole's owner conversation (`open`/`busy`/`onPicked`/`onCancel`/`onError`), and the owner adopts it through the object layer, selecting the committed Workspace only after its list projection has refreshed; cancellation is silent, and errors land in the retryable folder dialog whose **Choose again** reopens the flow. **Create a new workspace** retains the name dialog and disables names already present in that list, while the Host remains authoritative for concurrent or non-UI callers. The runtime Session and Workspace services own materialization. The Workspace row's Delete action opens a confirmation that states the retention boundary, blocks duplicate submission, and keeps failures open; success removes the group while its Sessions remain under Ungrouped.
+The picker lists real Host Workspace entities through the global `useWorkspaces` hook. Selecting a Workspace invokes the slot owner's `onPick` callback to retarget the frontend Session object. Each registration declares a **directory-flow child hole** (`single` kind: `conversation.hero.workspace.directoryFlow` / `sidebar.workspaces.directoryFlow`) that the composed picker package's client half fills with its picking interaction — the [`-native`](../../host/directory-picker-native/README.md) backend's renderless OS-chooser driver today, an in-app browsing dialog under a `-browse` composition. The flat **Open local folder...** action renders only while the surface's hole is occupied (occupancy read per menu render; an empty hole means the composition has no picking affordance — the seam's documented no-flow default). This package owns the trigger and the adoption: the occupant reports one picked path per open through the hole's owner conversation (`open`/`busy`/`onPicked`/`onCancel`/`onError`), and the owner adopts it through the object layer, selecting the committed Workspace only after its list projection has refreshed; cancellation is silent, and errors land in the retryable folder dialog whose **Choose again** reopens the flow. **Create a new workspace** retains the name dialog and disables names already present in that list, while the Host remains authoritative for concurrent or non-UI callers. The runtime Session and Workspace services own materialization. The Workspace row's Delete action opens a confirmation that states the retention boundary, blocks duplicate submission, and keeps failures open; success removes the group while its Sessions remain under Ungrouped. The Session row's Rename action opens the same browser-owned dialog pattern prefilled with the row's display title: no client-side conflict rule exists (the host normalizes and may reject with `title-invalid`, rendered in the dialog alert), and confirming an unchanged title is deliberately allowed — it pins the current automatic title against regeneration.
 
 Both target slots are declared by other plugins, so `apply` registers through declaration-aware deferral and re-registers after a declaring slot is restored.
 
@@ -18,5 +18,5 @@ None; this package neither assembles nor sends a provider request.
 
 ## Known Limitations and Deferred Work
 
-- **No Session deletion control** — the existing Session menu row remains visual-only; Workspace registration deletion does not delete Sessions.
+- **No Session deletion or fork control** — the Session menu's Fork and Delete rows remain visual-only (Rename is wired); Workspace registration deletion does not delete Sessions.
 - **Native folder selection depends on the local Host carrier** — under the `-native` composition, fixture-only or remote browser deployments cannot open a local operating-system dialog; platform failures are shown in a retryable modal. Remote-capable picking is the `-browse` composition's in-app flow.
diff --git a/packages/client/ui-workspace/README.zh.md b/packages/client/ui-workspace/README.zh.md
index e97d93f7e3..a472507bc4 100644
--- a/packages/client/ui-workspace/README.zh.md
+++ b/packages/client/ui-workspace/README.zh.md
@@ -4,7 +4,7 @@
 
 共享 Workspace 选择器插件。`WorkspaceBrowser` 注册到侧边栏的 `sidebar.workspaces` slot，`WorkspacePicker` 注册到页面局部 Session Intent 主视觉区的 `conversation.hero.workspace` slot，因此两个表层使用同一菜单和创建流程。
 
-该选择器通过全局 `useWorkspaces` hook 列出真实的 Host Workspace 实体。选择 Workspace 会调用 slot owner 的 `onPick` 回调，重新定位前端 Session 对象。每个注册各自声明一个**目录流子洞**（`single` kind：`conversation.hero.workspace.directoryFlow`／`sidebar.workspaces.directoryFlow`），由组合的选择器包 client half 填入其选取交互——今天是 [`-native`](../../host/directory-picker-native/README.md) 后端的无渲染 OS 选择器驱动，`-browse` 组合下则是应用内浏览对话框。平铺显示的 **打开本地文件夹…** 操作仅在本表层的洞被占用时渲染（每次菜单渲染读取占用状态；洞为空意味着该组合没有选目录能力——seam 文档化的无流程默认行为）。本包持有触发与接纳：占用者经洞的 owner 会话（`open`/`busy`/`onPicked`/`onCancel`/`onError`）每次打开上报一个所选路径，owner 通过对象层接纳它，并等待 Workspace 列表投影刷新后才选中已提交的 Workspace；取消操作不会显示提示，错误落入可重试的文件夹对话框，其 **重新选择** 会重新打开流程。**创建新工作区** 操作保留名称对话框，并禁用列表中已有的名称，而 Host 对并发或非 UI 调用方仍具有最终决定权。运行时 Session 与 Workspace 服务负责物化。Workspace 行内的 Delete 操作会打开确认框，说明保留边界、阻止重复提交，并在失败时保持打开；成功后，该分组会被移除，其 Session 则留在 Ungrouped 下。
+该选择器通过全局 `useWorkspaces` hook 列出真实的 Host Workspace 实体。选择 Workspace 会调用 slot owner 的 `onPick` 回调，重新定位前端 Session 对象。每个注册各自声明一个**目录流子洞**（`single` kind：`conversation.hero.workspace.directoryFlow`／`sidebar.workspaces.directoryFlow`），由组合的选择器包 client half 填入其选取交互——今天是 [`-native`](../../host/directory-picker-native/README.md) 后端的无渲染 OS 选择器驱动，`-browse` 组合下则是应用内浏览对话框。平铺显示的 **打开本地文件夹…** 操作仅在本表层的洞被占用时渲染（每次菜单渲染读取占用状态；洞为空意味着该组合没有选目录能力——seam 文档化的无流程默认行为）。本包持有触发与接纳：占用者经洞的 owner 会话（`open`/`busy`/`onPicked`/`onCancel`/`onError`）每次打开上报一个所选路径，owner 通过对象层接纳它，并等待 Workspace 列表投影刷新后才选中已提交的 Workspace；取消操作不会显示提示，错误落入可重试的文件夹对话框，其 **重新选择** 会重新打开流程。**创建新工作区** 操作保留名称对话框，并禁用列表中已有的名称，而 Host 对并发或非 UI 调用方仍具有最终决定权。运行时 Session 与 Workspace 服务负责物化。Workspace 行内的 Delete 操作会打开确认框，说明保留边界、阻止重复提交，并在失败时保持打开；成功后，该分组会被移除，其 Session 则留在 Ungrouped 下。Session 行内的 Rename 操作打开同款浏览器持有的对话框，并以该行的显示标题预填：客户端不设名称冲突规则（host 负责规范化，可能以 `title-invalid` 拒绝，错误渲染在对话框告警区）；确认未修改的标题是有意允许的——这正是把当前自动标题钉住、不再被重新生成覆盖的手势。
 
 两个目标 slot 都由其他插件声明，因此 `apply` 通过声明感知的延迟机制完成注册，并在声明该 slot 的插件恢复后重新注册。
 
@@ -18,5 +18,5 @@
 
 ## 已知限制与暂缓事项
 
-- **没有 Session 删除控件**：现有 Session 菜单行仍仅提供视觉效果；删除 Workspace 注册记录不会删除 Session。
+- **没有 Session 删除与 fork 控件**：Session 菜单的 Fork 与 Delete 行仍仅提供视觉效果（Rename 已接线）；删除 Workspace 注册记录不会删除 Session。
 - **原生文件夹选择依赖本地 Host 载体**：在 `-native` 组合下，仅使用 fixture（测试前置数据）的部署或远程浏览器部署无法打开本地操作系统对话框；模态框会显示平台故障，并允许重试。可远程的选取是 `-browse` 组合的应用内流程。
diff --git a/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx b/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
index 7effff36b5..a1fa0683ac 100644
--- a/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
+++ b/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
@@ -92,12 +92,14 @@ type SessionTreeProps = Pick<
   onRenameRequest: (workspaceId: WorkspaceId, currentTitle: string) => void
   /** Open the browser-owned delete-confirmation dialog for a real Workspace group. */
   onDeleteRequest: (workspaceId: WorkspaceId, currentTitle: string) => void
+  /** Open the browser-owned session rename dialog. */
+  onSessionRename: (sessionId: SessionNode['id'], currentTitle: string) => void
 }
 
 /** The scrolling session tree; unmounting at collapse settle drops the sessions subscription and expansion state. */
 function SessionTree({
   useSessions, startSession, open, workspaces, query,
-  onRenameRequest, onDeleteRequest, insertSessionBefore,
+  onRenameRequest, onDeleteRequest, onSessionRename, insertSessionBefore,
 }: SessionTreeProps) {
   const list = useSessions(s => s)
   const current = list.current
@@ -192,6 +194,7 @@ function SessionTree({
                   currentId={current}
                   now={now}
                   onOpen={open}
+                  onRename={onSessionRename}
                   onToggle={(id) => { setExpandedSessions(l => toggled(l, id)) }}
                   drag={dragProps}
                 />
@@ -206,7 +209,7 @@ function SessionTree({
 }
 
 /** The flat "In one list" body: every session a top-level row, newest-first. */
-function FlatList({ useSessions, open, query }: Pick<SessionTreeProps, 'useSessions' | 'open' | 'query'>) {
+function FlatList({ useSessions, open, onSessionRename, query }: Pick<SessionTreeProps, 'useSessions' | 'open' | 'onSessionRename' | 'query'>) {
   const list = useSessions(s => s)
   const rows = useMemo(() => deriveFlat(list, { query }), [list, query])
   const now = Date.now()
@@ -224,6 +227,7 @@ function FlatList({ useSessions, open, query }: Pick<SessionTreeProps, 'useSessi
             currentId={list.current}
             now={now}
             onOpen={open}
+            onRename={onSessionRename}
             /* v8 ignore next -- required-prop filler: flat rows render no twist, so it never fires. */
             onToggle={() => {}}
             flat
@@ -249,6 +253,7 @@ export function WorkspaceBrowser({
   actions,
   startSession,
   open,
+  renameSession,
   renameWorkspace,
   deleteWorkspace,
   insertSessionBefore,
@@ -309,6 +314,39 @@ export function WorkspaceBrowser({
     })
   }
 
+  // Session rename dialog (same browser-owned pattern as workspace rename;
+  // sessions have no client-side name-conflict rule — the host normalizes).
+  // Unlike workspace rename, an unchanged title is NOT blocked: confirming
+  // the current automatic title is the gesture that pins it.
+  const [sessionRenameTarget, setSessionRenameTarget] = useState<{ sessionId: SessionNode['id']; currentTitle: string } | null>(null)
+  const [sessionRenameDraft, setSessionRenameDraft] = useState('')
+  const [sessionRenaming, setSessionRenaming] = useState(false)
+  const [sessionRenameError, setSessionRenameError] = useState<string | null>(null)
+  const sessionRenameTrimmed = sessionRenameDraft.trim()
+  const sessionRenameBlocked = sessionRenaming || sessionRenameTrimmed === '' || sessionRenameTarget === null
+  const closeSessionRename = () => {
+    if (sessionRenaming) return
+    setSessionRenameTarget(null)
+    setSessionRenameError(null)
+  }
+  const confirmSessionRename = () => {
+    if (sessionRenameBlocked) return
+    setSessionRenaming(true)
+    setSessionRenameError(null)
+    renameSession(sessionRenameTarget.sessionId, sessionRenameTrimmed).then(() => {
+      setSessionRenaming(false)
+      setSessionRenameTarget(null)
+    }).catch((reason: unknown) => {
+      setSessionRenaming(false)
+      setSessionRenameError(reason instanceof Error ? reason.message : String(reason))
+    })
+  }
+  const onSessionRename = (sessionId: SessionNode['id'], currentTitle: string) => {
+    setSessionRenameTarget({ sessionId, currentTitle })
+    setSessionRenameDraft(currentTitle)
+    setSessionRenameError(null)
+  }
+
   // Delete dialog is separate from the row so a successful removal can
   // unmount that row without tearing down the in-flight confirmation state.
   const [deleteTarget, setDeleteTarget] = useState<{ workspaceId: WorkspaceId; title: string } | null>(null)
@@ -424,10 +462,11 @@ export function WorkspaceBrowser({
           itself is wide-only. */}
       <div className={css.listArea}>
         {wide && (groupBy === 'flat'
-          ? <FlatList useSessions={useSessions} open={open} query={query} />
+          ? <FlatList useSessions={useSessions} open={open} onSessionRename={onSessionRename} query={query} />
           : (
             <SessionTree
               useSessions={useSessions}
+              onSessionRename={onSessionRename}
               workspaces={workspaces}
               startSession={startSession}
               open={open}
@@ -479,6 +518,37 @@ export function WorkspaceBrowser({
         )}
         {renameError !== null && <div className={css.renameError} role="alert">{renameError}</div>}
       </Modal>
+
+      <Modal
+        open={sessionRenameTarget !== null}
+        onClose={closeSessionRename}
+        title="Rename session"
+        footer={(
+          <>
+            <Button variant="outline" disabled={sessionRenaming} onClick={closeSessionRename}>Cancel</Button>
+            <Button variant="primary" disabled={sessionRenameBlocked} onClick={confirmSessionRename}>Rename</Button>
+          </>
+        )}
+      >
+        <input
+          className={css.renameInput}
+          value={sessionRenameDraft}
+          aria-label="Session name"
+          autoFocus
+          disabled={sessionRenaming}
+          onFocus={(e) => { e.target.select() }}
+          onChange={(e) => { setSessionRenameDraft(e.target.value); setSessionRenameError(null) }}
+          onCompositionStart={() => { composingRef.current = true }}
+          onCompositionEnd={() => { composingRef.current = false }}
+          onKeyDown={(e) => {
+            if (e.key === 'Enter' && !composingRef.current) {
+              e.preventDefault()
+              confirmSessionRename()
+            }
+          }}
+        />
+        {sessionRenameError !== null && <div className={css.renameError} role="alert">{sessionRenameError}</div>}
+      </Modal>
       <Modal
         open={deleteTarget !== null}
         onClose={closeDelete}
diff --git a/packages/client/ui-workspace/src/client/contract/slots.ts b/packages/client/ui-workspace/src/client/contract/slots.ts
index 63b60d8051..ee0970f73f 100644
--- a/packages/client/ui-workspace/src/client/contract/slots.ts
+++ b/packages/client/ui-workspace/src/client/contract/slots.ts
@@ -93,6 +93,8 @@ export type WorkspaceBrowserInjected = DirectoryPickingInjected & {
   startSession: (workspaceId?: WorkspaceId) => void
   /** Open a real Session. */
   open: (sessionId: SessionId) => void
+  /** Rename a Session (explicit user title; resolves on host acceptance). */
+  renameSession: (sessionId: SessionId, title: string) => Promise<void>
   /** Rename a Host Workspace (rejects on name conflict; resolves on durability). */
   renameWorkspace: (workspaceId: WorkspaceId, title: string) => Promise<void>
   /** Delete only a Host Workspace registration; directory and Session logs remain. */
diff --git a/packages/client/ui-workspace/src/client/index.ts b/packages/client/ui-workspace/src/client/index.ts
index 1cf5a7ae5a..c1f5e61ba4 100644
--- a/packages/client/ui-workspace/src/client/index.ts
+++ b/packages/client/ui-workspace/src/client/index.ts
@@ -51,6 +51,14 @@ export function apply(ctx: ClientContext): void {
     // the runtime's shared action (recent-Workspace projection inside).
     startSession: (workspaceId) => { ctx.workspaces.startSession(workspaceId) },
     open: (sessionId) => { ctx.sessions.open(sessionId) },
+    renameSession: async (sessionId, title) => {
+      // Row → session-face hop: rename is a per-session verb (ISession), not
+      // a list-service verb; the binding resolves any listed session.
+      const session = ctx.sessions.binding(sessionId)?.session
+      if (session === undefined) throw new Error(`unknown session "${sessionId}"`)
+      const result = await session.rename(title)
+      if (!result.ok) throw new Error(result.error.message)
+    },
     renameWorkspace: async (workspaceId, title) => { await ctx.workspaces.rename(workspaceId, title) },
     deleteWorkspace: async (workspaceId) => { await ctx.workspaces.delete(workspaceId) },
     insertSessionBefore: async (workspaceId, sessionId, beforeSessionId) => {
diff --git a/packages/client/ui-workspace/src/client/rows/Rows.tsx b/packages/client/ui-workspace/src/client/rows/Rows.tsx
index 1ee2a84adc..d75fabdd8b 100644
--- a/packages/client/ui-workspace/src/client/rows/Rows.tsx
+++ b/packages/client/ui-workspace/src/client/rows/Rows.tsx
@@ -2,8 +2,8 @@
  * Workspace browser tree row components (figma Cell set 14:3080): pure presentational —
  * all data and callbacks arrive via props. Hover swaps (folder->chevron,
  * time->ellipsis, action buttons) are CSS-only. Row ... menus are visual-only
- * except workspace Rename; the session hover card is suppressed while a menu
- * is open. Workspace Rename/Delete are wired; session actions remain visual-only.
+ * except workspace Rename/Delete and session Rename; the session hover card is
+ * suppressed while a menu is open.
  */
 import { useState } from 'react'
 import clsx from 'clsx'
@@ -159,12 +159,14 @@ function rowHalf(e: { clientY: number; currentTarget: HTMLElement }): 'before' |
   return e.clientY < rect.top + rect.height / 2 ? 'before' : 'after'
 }
 
-export function SessionNodeItem({ node, depth, currentId, now, onOpen, onToggle, drag, flat = false }: {
+export function SessionNodeItem({ node, depth, currentId, now, onOpen, onRename, onToggle, drag, flat = false }: {
   node: SessionNode
   depth: number
   currentId: string | undefined
   now: number
   onOpen: (id: SessionNode['id']) => void
+  /** Open the browser-owned session rename dialog (row menu action). */
+  onRename: (id: SessionNode['id'], currentTitle: string) => void
   onToggle: (id: SessionNode['id']) => void
   /** Present only on draggable rows (workspace-group roots outside search). */
   drag?: RowDragProps | undefined
@@ -232,7 +234,10 @@ export function SessionNodeItem({ node, depth, currentId, now, onOpen, onToggle,
           open={menuOpen}
           onClose={() => { setMenuOpen(false) }}
           items={SESSION_MENU_ITEMS}
-          onSelect={() => { setMenuOpen(false) }} // Visual-only for now.
+          onSelect={(id) => {
+            setMenuOpen(false)
+            if (id === 'rename') onRename(node.id, row.title) // fork/delete stay visual-only.
+          }}
           portal
           closeOnPointerLeave
           anchor={(
@@ -264,6 +269,7 @@ export function SessionNodeItem({ node, depth, currentId, now, onOpen, onToggle,
           currentId={currentId}
           now={now}
           onOpen={onOpen}
+          onRename={onRename}
           onToggle={onToggle}
         />
       ))}
diff --git a/packages/client/ui-workspace/tests/rows.spec.tsx b/packages/client/ui-workspace/tests/rows.spec.tsx
index 15121e929a..bfaa8a36dd 100644
--- a/packages/client/ui-workspace/tests/rows.spec.tsx
+++ b/packages/client/ui-workspace/tests/rows.spec.tsx
@@ -68,7 +68,8 @@ describe('workspace browser rows', () => {
     const onOpen = vi.fn()
     const onToggle = vi.fn()
     const view = render(
-      <SessionNodeItem node={parent} depth={0} currentId={parent.id} now={0} onOpen={onOpen} onToggle={onToggle} />,
+      <SessionNodeItem node={parent} depth={0} currentId={parent.id} now={0} onOpen={onOpen}
+        onRename={vi.fn()} onToggle={onToggle} />,
     )
 
     const parentRow = screen.getByText('Parent').closest('[role="treeitem"]')!
@@ -88,7 +89,8 @@ describe('workspace browser rows', () => {
     view.rerender(
       <SessionNodeItem
         node={{ ...parent, children: [], expanded: false, running: false }}
-        depth={1} currentId={undefined} now={0} onOpen={onOpen} onToggle={onToggle}
+        depth={1} currentId={undefined} now={0} onOpen={onOpen}
+        onRename={vi.fn()} onToggle={onToggle}
       />,
     )
     expect(screen.getByRole('button', { name: 'Expand' })).toBeTruthy()
@@ -135,19 +137,29 @@ describe('workspace browser rows', () => {
     expect(screen.queryByRole('button', { name: /Workspace actions/ })).toBeNull()
   })
 
-  it('session row menu opens without opening the session and closes on selection', () => {
+  it('session row menu opens without opening the session and dispatches rename', () => {
     const onOpen = vi.fn()
+    const onRename = vi.fn()
     const node: SessionNode = {
       id: sid('s1'), title: 'One', children: [], hasChildren: false,
       expanded: false, running: false, updatedAt: 0,
     }
-    render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={onOpen} onToggle={vi.fn()} />)
+    render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={onOpen}
+      onRename={onRename} onToggle={vi.fn()} />)
     fireEvent.click(screen.getByRole('button', { name: 'Session actions for One' }))
     expect(onOpen).not.toHaveBeenCalled()
     expect(screen.getByRole('menuitem', { name: 'Delete session' }).className).toMatch(/danger/)
-    fireEvent.click(screen.getByRole('menuitem', { name: 'Fork session' }))
+    // Rename dispatches with the current display title (dialog prefill).
+    fireEvent.click(screen.getByRole('menuitem', { name: 'Rename' }))
     expect(screen.queryByRole('menu')).toBeNull()
+    expect(onRename).toHaveBeenCalledWith(node.id, 'One')
     expect(onOpen).not.toHaveBeenCalled()
+    // Fork and Delete stay visual-only.
+    fireEvent.click(screen.getByRole('button', { name: 'Session actions for One' }))
+    fireEvent.click(screen.getByRole('menuitem', { name: 'Fork session' }))
+    fireEvent.click(screen.getByRole('button', { name: 'Session actions for One' }))
+    fireEvent.click(screen.getByRole('menuitem', { name: 'Delete session' }))
+    expect(onRename).toHaveBeenCalledOnce()
     // Escape closes without selecting (Menu onClose path).
     fireEvent.click(screen.getByRole('button', { name: 'Session actions for One' }))
     fireEvent.keyDown(document, { key: 'Escape' })
@@ -159,7 +171,8 @@ describe('workspace browser rows', () => {
       id: sid('p'), title: 'Parent', children: [], hasChildren: true,
       expanded: false, running: false, updatedAt: 0,
     }
-    render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()} onToggle={vi.fn()} flat />)
+    render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()}
+      onRename={vi.fn()} onToggle={vi.fn()} flat />)
     expect(screen.queryByRole('button', { name: 'Expand' })).toBeNull()
   })
 
@@ -170,7 +183,8 @@ describe('workspace browser rows', () => {
         id: sid('s1'), title: 'Hovered', children: [], hasChildren: false,
         expanded: false, running: true, updatedAt: 0,
       }
-      render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={60_000} onOpen={vi.fn()} onToggle={vi.fn()} />)
+      render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={60_000} onOpen={vi.fn()}
+        onRename={vi.fn()} onToggle={vi.fn()} />)
       const wrapper = screen.getByRole('treeitem').parentElement as HTMLElement
       fireEvent.pointerEnter(wrapper)
       act(() => { vi.advanceTimersByTime(500) })
@@ -196,7 +210,8 @@ describe('workspace browser rows', () => {
         id: sid('s1'), title: 'Quiet', children: [], hasChildren: false,
         expanded: false, running: false, updatedAt: 0,
       }
-      render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()} onToggle={vi.fn()} />)
+      render(<SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()}
+        onRename={vi.fn()} onToggle={vi.fn()} />)
       fireEvent.pointerEnter(screen.getByRole('treeitem').parentElement as HTMLElement)
       act(() => { vi.advanceTimersByTime(500) })
       expect(screen.getByText('Idle')).toBeTruthy()
@@ -213,7 +228,8 @@ describe('workspace browser rows', () => {
     }
     const inactive = dragProps()
     const { rerender } = render(
-      <SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()} onToggle={vi.fn()} drag={inactive} />,
+      <SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()}
+        onRename={vi.fn()} onToggle={vi.fn()} drag={inactive} />,
     )
     const row = screen.getByRole('treeitem')
     stubRect(row)
@@ -230,7 +246,8 @@ describe('workspace browser rows', () => {
 
     const active = dragProps({ active: true, marker: 'before' })
     rerender(
-      <SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()} onToggle={vi.fn()} drag={active} />,
+      <SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()}
+        onRename={vi.fn()} onToggle={vi.fn()} drag={active} />,
     )
     stubRect(screen.getByRole('treeitem'))
     // Top half hovers/drops 'before'; bottom half 'after' (row mid = 117).
@@ -243,7 +260,8 @@ describe('workspace browser rows', () => {
 
     const after = dragProps({ active: true, marker: 'after' })
     rerender(
-      <SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()} onToggle={vi.fn()} drag={after} />,
+      <SessionNodeItem node={node} depth={0} currentId={undefined} now={0} onOpen={vi.fn()}
+        onRename={vi.fn()} onToggle={vi.fn()} drag={after} />,
     )
     expect(screen.getByRole('treeitem').className).toMatch(/dropAfter/)
   })
diff --git a/packages/client/ui-workspace/tests/workspace-browser.spec.tsx b/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
index 12a5efc264..abe896dffe 100644
--- a/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
+++ b/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
@@ -55,6 +55,7 @@ function mount(overrides: Partial<WorkspaceBrowserProps> = {}) {
     actions: store.actions,
     startSession: vi.fn(),
     open: vi.fn(),
+    renameSession: vi.fn(async () => {}),
     renameWorkspace: vi.fn(async () => {}),
     deleteWorkspace: vi.fn(async () => {}),
     insertSessionBefore: vi.fn(async () => {}),
diff --git a/packages/client/web-react/README.i18n.yaml b/packages/client/web-react/README.i18n.yaml
index 091f2d8ad1..30a8d7cee1 100644
--- a/packages/client/web-react/README.i18n.yaml
+++ b/packages/client/web-react/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/web-react/README.md
 README.md: 7cc80f22bd5527838288d11c819e81b7ec4d17c4
-README.zh.md: ec7143db9a1fe97145a07a0a1fde45c9268c2425
+README.zh.md: 855417ff357b78b05fa54946c6cf84fafaa5180e
diff --git a/packages/client/web-react/README.zh.md b/packages/client/web-react/README.zh.md
index ec7143db9a..855417ff35 100644
--- a/packages/client/web-react/README.zh.md
+++ b/packages/client/web-react/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-slot 终端设计的外壳侧 React 胶水：createSlotRenderer（外壳安装到运行时 SlotsService 的 SlotRenderer 实现）、SessionProvider（框架接线的 render prop，也作为标准 seat 注入到声明会话 scope 子项的配置项）、bindSnapshotSelector（唯一的 hook 构造器：主机与引擎只传递裸 observable source；每个 hook 在此绑定，并按 source 缓存）、useInvoke。链式 slot outlet 在渲染时按链顺序运行已注册 selector，只挂载被选中的配置项，其 select 返回值以 `matched` 加入 props；`renderSlotChain` binding 与 `renderSlot` 一样按配置项缓存。快照 store 引擎与 defineStore 位于 runtime（store 已迁移）；业务插件只依赖 ui-slots 类型，绝不依赖该包。
+slot 终端设计的外壳侧 React 胶水：createSlotRenderer（外壳安装到运行时 SlotsService 的 SlotRenderer 实现）、SessionProvider（由框架接入的 render prop，也作为标准 seat 注入到声明会话 scope 子 slot 的配置项）、bindSnapshotSelector（唯一的钩子构造器：主机与引擎只传递裸 observable source；每个钩子在此绑定，并按 source 缓存）、useInvoke。链式 slot outlet 在渲染时按链顺序运行已注册 selector，只挂载被选中的配置项，其 select 返回值以 `matched` 加入 props；`renderSlotChain` 绑定与 `renderSlot` 一样按配置项缓存。快照 store 引擎与 defineStore 位于运行时（store 已迁移）；业务插件只依赖 ui-slots 类型，绝不依赖该包（package）。
 
 ## 模型体验
 
@@ -14,6 +14,6 @@ slot 终端设计的外壳侧 React 胶水：createSlotRenderer（外壳安装
 
 ## 已知限制与暂缓事项
 
-- **persist 中间件会损坏原始值状态 store**：保存时它会对状态执行对象展开，因此 `SnapshotStore<string>` 往返后会变成字符 map；引擎改为自行实现持久化（见 `attachPersistence`）。
+- **persist 中间件会损坏原始值状态 store**：保存时它会对状态执行对象展开，因此 `SnapshotStore<string>` 往返后会变成字符映射；引擎改为自行实现持久化（见 `attachPersistence`）。
 - **`UseSession` 有意保持宽泛（`object` 快照）**：依赖方向（runtime → web-react，绝不反向）使真实 `ConversationSnapshot` 类型不可访问；会话 slot 消费方在其边界处缩窄一次。
-- **renderSlot 是唯一的 P-I 形式**：没有 Suspense，也没有逐配置项惰性加载；渐进式渲染表层会随其独立项目回归。
+- **renderSlot 是唯一的 P-I 形式**：没有 Suspense，也没有逐配置项惰性加载；渐进式渲染能力将在其独立项目中恢复。
diff --git a/packages/client/web/README.i18n.yaml b/packages/client/web/README.i18n.yaml
index 3a82b79223..9a9733f420 100644
--- a/packages/client/web/README.i18n.yaml
+++ b/packages/client/web/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/client/web/README.md
 README.md: a4325f5dbe8ddd9bbe77086eb16fdb7aed9adc83
-README.zh.md: 9c1eab795145cf0a1b69528bc9822a26ed226023
+README.zh.md: d4cb7a43da4e9c84a401ee0a1ac8c30d4816926e
diff --git a/packages/client/web/README.zh.md b/packages/client/web/README.zh.md
index 9c1eab7951..d4cb7a43da 100644
--- a/packages/client/web/README.zh.md
+++ b/packages/client/web/README.zh.md
@@ -2,19 +2,19 @@
 
 [English](README.md) | 中文
 
-Web 外壳内核：`new AppWebEntry(el, seams?).run()` 通过两阶段启动（web2）挂载整个客户端。第一阶段（模块表层）：构建客户端模块系统（`@deepseek-ai/dsh-client-modules`），以主机推送的配置项图（`window.__DSH_BOOT__`）为基础，并行预抓取 `immediately` 层级；执行组合包只会注册 factory。第二阶段（插件表层）：挂载 vendored cordis Loader，并把模块系统作为其 `internal` seam 注入；为每一行图数据创建一个 loader 配置项，另创建外壳自身的 app-shell 组装配置项（tree.import 会物化各模块）；以 settle 作为 AppRoot 的门禁（loader 完全停稳 + 每个配置项 fiber 都为 ACTIVE → 一次切换显示完整 UI）。组合完全由主机图决定：花名册和 immediately 层级都位于负责组合的应用中；外壳不作任何组合决策。
+Web 外壳内核：`new AppWebEntry(el, seams?).run()` 通过两阶段启动（web2）挂载整个客户端。第一阶段（模块侧）：构建客户端模块系统（`@deepseek-ai/dsh-client-modules`），以主机推送的配置项图（`window.__DSH_BOOT__`）为基础，并行预取 `immediately` 层级；执行组合包只会注册 factory。第二阶段（插件侧）：挂载仓库内置的 Cordis Loader，并把模块系统作为其 `internal` seam 注入；为每一行图数据创建一个 loader 配置项，另创建外壳自身的 app-shell 组装配置项（tree.import 会物化各模块）；以 settle 作为 AppRoot 的门禁（loader 完全停稳 + 每个配置项 fiber 都为 ACTIVE → 一次切换显示完整 UI）。组合完全由主机图决定：花名册和 immediately 层级都位于负责组合的应用中；外壳不作任何组合决策。
 
-外壳自给自足（web2 硬性规则）：内核不对任何插件包执行值导入；启动状态 store 与信号在这里手写（`loader-status.ts`），因此插件失败时（尤其在失败时）加载页面仍能工作。app-shell 组装（`@deepseek-ai/dsh-client-app-shell`，由外壳拥有、背后没有 npm 包的伪配置项）是唯一通过 `registerStatic` 注册的模块；它与任何插件一样，通过 inject 等待 slots/sessions/layout。
+外壳自给自足（web2 硬性规则）：内核不对任何插件包（package）执行值导入；启动状态 store 与信号在这里手写（`loader-status.ts`），因此即使插件失败，加载页面仍能工作，而此时这一点尤其重要。app-shell 组装（`@deepseek-ai/dsh-client-app-shell`，由外壳拥有、背后没有 npm 包的伪配置项）是唯一通过 `registerStatic` 注册的模块；它与任何插件一样，通过 inject 等待 slots/sessions/layout。
 
 `PLATFORM_MODULES`（src/platform.ts）是共享模块表层的唯一真源：种子表 key、tsdown 客户端 external 和 vite alias 集都是它的投影。
 
-可选 `seams` 参数会转发模块系统的 `fetchBundle`／`executeBundle` 传输覆盖（`BootSeams`）；生产调用方省略此参数。它用于测试环境，因为此类环境中的 `<script>` 执行无法到达页面上下文（jsdom）。
+可选 `seams` 参数会转发模块系统的 `fetchBundle`／`executeBundle` 传输覆盖（`BootSeams`）；生产调用方省略此参数。它用于 `<script>` 执行无法到达页面上下文的测试环境（jsdom）。
 
 外壳拥有浏览器标题投影。选中带有持久标题的会话时，它会渲染 `<session title> — <existing HTML title>` 并响应后续标题修订；未选择会话或选中无标题会话时，会保留现有标题；外壳卸载时恢复标题。现有 HTML 标题仍是可配置的产品后缀。
 
 ## 模型体验
 
-无。配置项外壳负责启动浏览器插件树；这里没有任何内容进入模型请求。
+无。入口外壳负责启动浏览器插件树；这里没有任何内容进入模型请求。
 
 #### KV Cache 影响
 
@@ -22,5 +22,5 @@ Web 外壳内核：`new AppWebEntry(el, seams?).run()` 通过两阶段启动（w
 
 ## 已知限制与暂缓事项
 
-- **有意采用一次性渲染**：UI 等待启动 settle；只要一个配置项失败，加载页面就会保留并高声逐项报告，不提供部分可用性（渐进式渲染会随其独立项目回归）。
+- **有意采用一次性渲染**：UI 等待启动 settle；只要一个配置项失败，加载页面就会保留并逐项显示醒目的报告，不提供部分可用性（渐进式渲染将作为独立项目恢复）。
 - **窄窗口验收暂缓**：ui-layout 已实现让步链，但外壳级窄视口演练是 P-II 验收项。
diff --git a/packages/code-runtime/README.i18n.yaml b/packages/code-runtime/README.i18n.yaml
index 002a78c603..a914fa79e3 100644
--- a/packages/code-runtime/README.i18n.yaml
+++ b/packages/code-runtime/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/code-runtime/README.md
 README.md: dbe6b37ffa01d07c6902672a06ebf6f88548ff99
-README.zh.md: fd73e1376afe5ae4390457fa648b9f19591ee71c
+README.zh.md: f97bc091839bf7660a4a8c5ed506037d6b63eb3f
diff --git a/packages/code-runtime/README.zh.md b/packages/code-runtime/README.zh.md
index fd73e1376a..f97bc09183 100644
--- a/packages/code-runtime/README.zh.md
+++ b/packages/code-runtime/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-代码执行能力 seam（参见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：一个抽象运行时接口，用于针对宿主提供的异步绑定执行一段模型编写的程序，并捕获程序打印和返回的内容。消费方是工具注册表的 [Code Mode](../core/tools/README.md)（`tools: { mode: code }`，即 `run_code` 工具与生成的 TypeScript SDK）；设计记录在 [Code Mode Agent Note](../../.agents/notes/implemented/feature/2026-06-15-code-mode.md) 中。这些都是**产品** 包。
+代码执行能力 seam（参见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：一个抽象运行时接口，用于针对宿主提供的异步绑定执行一段模型编写的程序，并捕获程序打印和返回的内容。消费方是工具注册表的 [Code Mode](../core/tools/README.md)（`tools: { mode: code }`，即 `run_code` 工具与生成的 TypeScript SDK）；设计记录在 [Code Mode Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-06-15-code-mode.md) 中。这些都是**产品**包（package）。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `code-runtime/` | 抽象代码执行 seam（接口 + 词汇） | `ctx.codeRuntime` |
-| [`code-runtime-worker/`](code-runtime-worker/README.md) | worker 线程后端：每次运行使用全新 worker，由宿主侧剥离 TypeScript 类型（类型注解仅供参考，绝不执行类型检查）、端口桥接绑定、预算／堆隔离 | 注册 `ctx.codeRuntime` |
+| [`code-runtime-worker/`](code-runtime-worker/README.md) | worker 线程后端：每次运行使用全新 worker，由宿主侧剥离 TypeScript 类型（类型注解仅供参考，绝不执行类型检查）、端口桥接绑定、预算／堆限制 | 注册 `ctx.codeRuntime` |
 
-接口位于 `code-runtime/code-runtime/`，随附后端位于 `code-runtime/code-runtime-worker/`。不同后端可以采用不同执行基底（worker 线程、进程、容器）与源语言；二者都是服务上的只读描述符。后端注册 `ctx.codeRuntime`，无需修改接口或消费方；正是这种拆分，使未来可以直接换入强化后端。
+接口位于 `code-runtime/code-runtime/`，随附的后端位于 `code-runtime/code-runtime-worker/`。不同后端可以采用不同执行基底（worker 线程、进程、容器）与源语言；二者都是服务上的只读描述符。后端注册 `ctx.codeRuntime`，无需修改接口或消费方；正是这种拆分，使未来可以直接换入加固后端。
diff --git a/packages/code-runtime/code-runtime-worker/README.i18n.yaml b/packages/code-runtime/code-runtime-worker/README.i18n.yaml
index 7245b80dcf..9a8eef9301 100644
--- a/packages/code-runtime/code-runtime-worker/README.i18n.yaml
+++ b/packages/code-runtime/code-runtime-worker/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/code-runtime/code-runtime-worker/README.md
 README.md: 83c9a398970831e88cb3ef5d71d3f175da97d1f1
-README.zh.md: 8e6c15eccdd83d3b918249920596dcbeee94f2b2
+README.zh.md: 3a4714fab55b85ffeb237d2a9fd0615ae4789bcf
diff --git a/packages/code-runtime/code-runtime-worker/README.zh.md b/packages/code-runtime/code-runtime-worker/README.zh.md
index 8e6c15eccd..3a4714fab5 100644
--- a/packages/code-runtime/code-runtime-worker/README.zh.md
+++ b/packages/code-runtime/code-runtime-worker/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-这是 [`@deepseek-ai/dsh-code-runtime`](../code-runtime/README.md) seam 的 worker 线程实现：`WorkerCodeRuntime` 会在每次运行中使用一个全新的 Node `worker_threads.Worker`，输入 TypeScript，由宿主侧剥离类型，通过消息端口桥接绑定，输出 `{ value, logs, error? }`。**这是隔离措施，而非安全边界**：其信任立场有意与 bash 等价（参见 [Code Mode Agent Note](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md) 的 Trust posture 章节），但提供 bash 没有的隔离：独立 isolate、空环境、堆上限与强制终止。
+这是 [`@deepseek-ai/dsh-code-runtime`](../code-runtime/README.md) seam 的 worker 线程实现：`WorkerCodeRuntime` 会在每次运行中使用一个全新的 Node `worker_threads.Worker`，输入 TypeScript，由宿主侧剥离类型，通过消息端口桥接绑定，输出 `{ value, logs, error? }`。**这是隔离措施，而非安全边界**：其信任立场有意与 bash 等价（参见 [Code Mode Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md) 的 Trust posture 章节），但提供 bash 没有的隔离：独立 isolate、空环境、堆上限与强制终止。
 
 ## 配置
 
@@ -23,18 +23,18 @@
 - **每次运行使用一个全新 worker，不设池化**：程序所在的世界会随 worker 一同终止，不会留下需要记录的跨运行状态，也无法发生状态泄漏；仅凭会话日志即可重建运行。
 - **在执行上下文中，由宿主侧剥离类型**：程序会包裹在异步函数外壳中，通过 `node:module` 的 `stripTypeScriptTypes` 剥离类型（只支持可擦除语法；`enum`／namespace 会作为程序 `exception` 被拒绝，且不会启动 worker），再按字节位置切回原内容。之后程序作为 `AsyncFunction` 的函数体执行，因此顶层 `await`／`return` 可用。
 - **端口把对端视为不可信**：模型代码能够访问 `parentPort` 并伪造通信，因此任何代码读取入站消息前，系统都会验证其形状并重新构建（`null`、原始值、无效类型和格式错误的载荷会被静默丢弃；伪造的额外字段绝不会被带入）；宿主对每个调用 id 最多响应一次，只将绑定名称解析为自有属性（伪造的 `constructor` 无法沿原型链访问），丢弃结算后的回复，并验证每个绑定 resolve 值与完成值是否为无损 JSON。伪造的 `log`／`done` 消息无法绕过外层上限：宿主会再次验证，并统计每条获准日志以及完成值或诊断。worker 侧命名空间使用 null-prototype 和 `defineProperty`，因此形似 `__proto__` 的绑定名称只是普通键。
-- **绑定 reject 类属于请求数据**：可选命名空间描述符会指定构造器全局变量，以及用于接收失败成员名称的自有属性。worker 会创建并注入该真实类，使 `instanceof` 生效，同时无需硬编码 `tools` 或 `ToolCallError`；全局变量无效或冲突的声明会在启动 worker 前失败。失败路径使用模块捕获的错误与属性定义 intrinsic，以及 null-prototype 描述符，因此模型之后的修改无法把被拒绝的绑定变成 worker 崩溃。
-- **两个独立预算，因为对端不可信**：`computeMs` 统计 worker 实际测得的忙碌时间（轮询 `worker.performance.eventLoopUtilization()`）；热循环无法借助待完成的诱饵 dispatch 隐藏，程序等待慢工具时则不累计。`maxWallMs` 为忙碌时间无法观测的情况兜底（例如等待永远不会 resolve 的 promise）。二者最终都会调用 `worker.terminate()`，连同步热循环也能终止；堆溢出会表现为 worker 的 OOM 退出（`kind: 'worker-exit'`）。`maxWallMs` 在加载时会对照 `MAX_TIMER_DELAY_MS` 做范围校验：`setTimeout` 会把更长的延迟夹到 1 ms，仅有正数校验会放行一个在第一个 tick 就到期的上限。`computeMs` 不需要这道上界，因为它对照的是实测占用率，而不是喂给定时器。
-- **中间绑定值是完整 JSON**：绑定参数与 resolve 值会接受迭代式无损 JSON 验证。程序执行前，worker 会捕获自己 realm 中的普通容器原型身份，以及只用于外部 realm 的原生函数源码检查，因此构造器槽修改和用户编写的仿冒对象都无法改变容器分类。它还会捕获该 JSON 边界使用的每一个结构与计量 intrinsic，以无原型对象创建属性描述符，并绕过可变集合原型管理私有遍历状态；因此，模型对全局对象、原型方法或 `Object.prototype` 上形似描述符字段的修改，都无法改变验证、wire 传输或字节计量。值会展平为有深度上限的前序 wire 值，供 structured clone 使用，并在另一侧迭代式重建。它们没有字节、JavaScript 调用栈或嵌套 structured-clone 深度上限，绝不会进入外层输出账本或模型上下文；上限仍来自提供方／执行器获取限制与进程／worker 内存。
-- **日志主动流入一个外层账本**：console／stdout／stderr 文本按发送顺序穿过端口，因此超时或被终止的程序仍会显示已经打印的内容。worker 会按 JSON 字符串精确计费，并在发送完成值和异常诊断前，根据组合预算的剩余量预检；因此，抛出的百万字节 stack 会在 worker 边界变成固定的 `output-limit` 诊断。绕过补丁 stream 槽的原生写入会到达独立于完成端口的 pipe，因此宿主会针对这些字节和不可信伪造通信再次执行账本统计；在物化结果前，结算过程会持续进行有界 pipe 捕获，直到 worker 完成终止。`maxOutputBytes` 统计外层 `logs` 数组加完成值或失败消息载荷的 JSON 序列化；固定的 `CodeRunResult` 字段名、花括号、有界错误 kind 标签，以及后续呈现空白不计入这份可变载荷账本。未超过上限时会返回精确值；有损完成值属于 `invalid-output`，组合溢出属于 `output-limit`，不会用 inspected string 代替。失败会保留能容纳的已捕获前缀，之后按普通外层 `run_code` 落盘策略处理。
-- **空环境**：worker 使用 `env: {}` 和 `execArgv: []`，既没有环境凭据（比 spawn 命令的清理环境规则更严格），也不会继承 loader 标志。
-- **释放资源时等待完全停稳**：清理会把进行中的运行标记为 `abort`，并在 resolve 前等待每个 worker 退出。
+- **绑定调用被拒绝时使用的异常类属于请求数据**：可选命名空间描述符会指定构造器全局变量，以及用于接收调用失败的成员名称的自有属性。worker 会创建并注入该真实类，使 `instanceof` 生效，同时无需硬编码 `tools` 或 `ToolCallError`；全局变量无效或冲突的声明会在启动 worker 前失败。失败路径使用模块捕获的错误 intrinsic 与属性定义 intrinsic，以及 null-prototype 描述符，因此模型之后的修改无法把被拒绝的绑定变成 worker 崩溃。
+- **两个独立预算，因为对端不可信**：`computeMs` 统计 worker 实际测得的忙碌时间（轮询 `worker.performance.eventLoopUtilization()`）；热循环无法借助待完成的诱饵 dispatch 隐藏，程序等待慢工具时则不累计。`maxWallMs` 为忙碌时间无法观测的情况兜底（例如等待永远不会 resolve 的 promise）。二者最终都会调用 `worker.terminate()`，连同步热循环也能终止；堆溢出会表现为 worker 的 OOM 退出（`kind: 'worker-exit'`）。`maxWallMs` 在加载时会对照 `MAX_TIMER_DELAY_MS` 做范围校验：`setTimeout` 会把更长的延迟限制为 1 ms，仅有正数校验会放行一个在第一个 tick 就到期的上限。`computeMs` 不需要这道上界，因为它对照的是实测占用率，而不是喂给定时器。
+- **中间绑定值是完整 JSON**：绑定参数与 resolve 值会接受迭代式无损 JSON 验证。程序执行前，worker 会捕获自己 realm 中的普通容器原型身份，以及只用于外部 realm 的原生函数源码检查，因此构造器槽修改和用户编写的仿冒对象都无法改变容器分类。它还会捕获该 JSON 边界使用的每一个结构与计量 intrinsic，以无原型对象创建属性描述符，并绕过可变集合原型管理私有遍历状态；因此，模型对全局对象、原型方法或 `Object.prototype` 上形似描述符字段的修改，都无法改变验证、wire 传输或字节计量。值会展平为自身嵌套深度有界的前序 wire 值，供 structured clone 使用，并在另一侧迭代式重建。它们没有字节、JavaScript 调用栈或嵌套 structured-clone 深度上限，绝不会进入外层输出账本或模型上下文；上限仍来自提供方／执行器获取限制与进程／worker 内存。
+- **日志主动流入一个外层账本**：console／stdout／stderr 文本按产生顺序经端口传输，因此超时或被终止的程序仍会显示已经打印的内容。worker 会精确统计 JSON 字符串的字节数，并在发送完成值和异常诊断前，根据组合预算的剩余量预检；因此，抛出的百万字节 stack 会在 worker 边界变成固定的 `output-limit` 诊断。绕过补丁 stream 槽的原生写入会到达独立于完成端口的 pipe，因此宿主会针对这些字节和不可信伪造通信再次执行账本统计；在物化结果前，结算过程会持续进行有界 pipe 捕获，直到 worker 完成终止。`maxOutputBytes` 统计外层 `logs` 数组加完成值或失败消息载荷的 JSON 序列化；固定的 `CodeRunResult` 字段名、花括号、有界错误 kind 标签，以及后续呈现空白不计入这份可变载荷账本。未超过上限时会返回精确值；有损完成值属于 `invalid-output`，组合溢出属于 `output-limit`，不会用 inspected string 代替。失败会保留能容纳的已捕获前缀，之后按普通外层 `run_code` 落盘策略处理。
+- **空环境**：worker 使用 `env: {}` 和 `execArgv: []`，既不会获得环境变量中的凭据（比 spawn 命令的清理环境规则更严格），也不会继承 loader 标志。
+- **dispose（资源释放）时等待完全停稳**：清理会使进行中的运行以 `abort` 失败，并会等待每个 worker 退出后再完成。
 
 ## 未构建与已构建的 worker 入口
 
-源代码模式通过 Node 原生类型剥离加载只包含可擦除语法的 `src/worker.ts`。其传递运行时闭包只包含 Node 内置模块和相对源模块，因此全新 checkout 绝不需要兄弟工作区包尚未构建的 `lib/` 导出。worker 本地 JSON 快照器会与会话自有的规范边界执行一致性测试；消息端口两侧都会展平并重建已验证值，使应用嵌套永远不会进入 structured clone。构建模式会把兄弟文件 `lib/worker.cjs` 作为文件系统路径传入，因为 pkg 的 VFS Worker hook 要求 CommonJS；同一路径也可在普通 Node 下使用。`tests/built-lib.e2e.ts` 固定了 [docs/testing.md](../../../docs/testing.md) 要求的真实加载路径。
+源代码模式通过 Node 原生类型剥离加载只包含可擦除语法的 `src/worker.ts`。其传递运行时闭包只包含 Node 内置模块和相对源模块，因此全新 checkout 绝不需要兄弟工作区包（package）尚未构建的 `lib/` 导出。worker 本地 JSON 快照器会与会话自有的规范边界执行一致性测试；消息端口两侧都会展平并重建已验证值，使应用嵌套永远不会进入 structured clone。构建模式会把兄弟文件 `lib/worker.cjs` 作为文件系统路径传入，因为 pkg 的虚拟文件系统（VFS）Worker hook 要求 CommonJS；同一路径也可在普通 Node 下使用。`tests/built-lib.e2e.ts` 固定了 [docs/testing.md](../../../docs/testing.md) 要求的真实加载路径。
 
-SDK 接口是默认／具名 `WorkerCodeRuntime` 类与 `Config`。可操作的 `./worker` 子路径仅作为打包后的 spawn 入口存在；wire 协议与启动辅助模块是源代码私有的实现细节。
+SDK 对外提供默认及具名导出的 `WorkerCodeRuntime` 类，以及 `Config`。运行所用的 `./worker` 子路径仅作为打包后的 spawn 入口存在；wire 协议与启动辅助模块是源代码私有的实现细节。
 
 ## 模型体验
 
@@ -42,13 +42,13 @@ SDK 接口是默认／具名 `WorkerCodeRuntime` 类与 `Config`。可操作的
 
 #### KV Cache 影响
 
-不会直接失效；由具名消费方负责请求前缀变更。
+不会直接失效；由上述消费方负责请求前缀变更。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- **程序 spawn 的 OS 进程在终止后仍会存活**：`worker.terminate()` 只结束线程，比 bash-local 的进程组终止更弱；在容器后端出现前，孤儿清理属于部署职责。
+- **程序派生的 OS 进程在程序终止后仍会存活**：`worker.terminate()` 只结束线程，比 bash-local 的进程组终止更弱；在容器后端出现前，孤儿进程清理属于部署职责。
 - **类型剥离依赖 Node 的实验性 `stripTypeScriptTypes` API**：依赖的行为由单元测试固定；如其发生变化，amaro／sucrase 是已经点名的直接替代品。
 - **`computeMs` 到期最多可能超过一个轮询间隔**：系统每 25 ms 采样一次忙碌时间（内部常量，有意不做成配置）。
-- **程序获得一个含 5 种方法的 `console` shim**（`log`／`info`／`warn`／`error`／`debug`）：有意不提供 Node 的完整 console 接口。
+- **程序获得一个含 5 个方法的 `console` shim**（`log`／`info`／`warn`／`error`／`debug`）：有意不提供 Node 的完整 console 接口。
 - **中间绑定值没有字节上限**：程序可以用永远不会成为外层输出的值耗尽进程或 worker 内存。
 - **默认 64 MiB 是拒绝边界，不是可恢复存储**：外层落盘只能保存发生 `output-limit` 后返回的有界日志和诊断；在运行时上限之外被拒绝的字节永远不会到达落盘层。
diff --git a/packages/code-runtime/code-runtime/README.i18n.yaml b/packages/code-runtime/code-runtime/README.i18n.yaml
index 57415166ac..4b5476220b 100644
--- a/packages/code-runtime/code-runtime/README.i18n.yaml
+++ b/packages/code-runtime/code-runtime/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/code-runtime/code-runtime/README.md
 README.md: c7a2d519e47d160f5ab123bfc887e7e9f24ec602
-README.zh.md: 10c1b900165e255a3ea57197ae323283822ab7bc
+README.zh.md: 9103e7490e2fccb071cd8a234b224bc17514253e
diff --git a/packages/code-runtime/code-runtime/README.zh.md b/packages/code-runtime/code-runtime/README.zh.md
index 10c1b90016..9103e7490e 100644
--- a/packages/code-runtime/code-runtime/README.zh.md
+++ b/packages/code-runtime/code-runtime/README.zh.md
@@ -4,21 +4,21 @@
 
 这是**代码执行 seam**：抽象的 `CodeRuntime` 服务（`ctx.codeRuntime`）只定义代码运行时做什么，即针对宿主提供的一组异步绑定运行一段模型编写的程序，并报告 `{ value, logs, error? }`，而不规定如何实现。
 
-此包是该能力的接口层（以 bash 三包结构为模板，参见[能力 seam](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：实现通过继承 `CodeRuntime` 并注册服务接入；消费方是工具注册表的 Code Mode，它生成面向模型的 SDK，并桥接工具分发。两者都由 [Code Mode Agent Note](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md) 规定，首个实现是 Node worker 线程后端。运行时不了解工具或会话：调用方只向它提供具名异步函数与程序字符串；所有工具形状的内容都留在消费方。
+此包（package）承担该能力三个组成部分中的接口职责（以 bash 三包结构为模板，参见[能力 seam](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：实现通过继承 `CodeRuntime` 并注册服务接入；消费方是工具注册表的 Code Mode，它生成面向模型的 SDK，并桥接工具分发。这两项职责均由 [Code Mode Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md) 规定，首个实现是 Node worker 线程后端。运行时不了解工具或会话：调用方只向它提供具名异步函数与程序字符串；所有与工具有关的内容都留在消费方。
 
 ## 服务 API（`ctx.codeRuntime`）
 
 | 成员 | 语义 |
 |---|---|
-| `run(request)` | 针对请求的绑定执行一段程序。**每一种程序结果都通过 error 字段完成 resolve**：包括解析／转换失败、抛出异常、无效完成值、输出溢出、预算到期、中止或执行基底死亡（由 `CodeRunFailure` 的正交 `kind` 分类表示）；只有调用方误用 seam 本身时才 reject（例如资源释放后仍提交运行）。程序作为异步函数的函数体运行，因此顶层 `await`／`return` 可用，无损 JSON 完成值会成为 `result.value`。 |
-| `language` | 只读描述符：`run` 期望的源语言（已知值为 `'typescript'`）。仅供参考，不作门禁；生成语言专用呈现的消费方会对该值执行分支，遇到无法呈现的语言时明确失败。 |
+| `run(request)` | 针对请求的绑定执行一段程序。**所有程序失败结果都通过 resolve 结果中的 error 字段报告**：包括解析／转换失败、抛出异常、无效完成值、输出溢出、预算到期、中止或执行基底终止（由 `CodeRunFailure` 的正交 `kind` 分类表示）；只有调用方误用 seam 本身时才 reject（例如 dispose（资源释放）后仍提交运行）。程序作为异步函数的函数体运行，因此顶层 `await`／`return` 可用，无损 JSON 完成值会成为 `result.value`。 |
+| `language` | 只读描述符：`run` 期望的源语言（已知值为 `'typescript'`）。仅供参考，不作门禁；生成语言专用呈现的消费方会根据该值选择分支，遇到无法呈现的语言时明确失败。 |
 | `isolation` | 只读描述符：执行基底（`'worker-thread'`、`'process'`、`'container'`）。供部署与诊断使用，**不构成安全声明**。 |
 
-每个实现都必须遵守以下语义（完整契约见类 JSDoc）：绑定调用会桥接完整的无损 JSON 参数与 resolve 值，seam 层不设字节上限；程序被视为不可信对等方（任意绑定名称都是自有属性，格式错误的通信绝不能使宿主崩溃）；不同运行之间不保留任何状态；资源释放会终止进行中的运行，并且在完成前等待其退出。
+每个实现都必须遵守以下语义（完整契约见类 JSDoc）：绑定调用会桥接完整的无损 JSON 参数与 resolve 值，seam 层不设字节上限；程序被视为敌对对等方（任意绑定名称都会成为自有属性，格式错误的通信绝不能使宿主崩溃）；不同运行之间不保留任何状态；dispose 会终止进行中的运行，并且在完成前等待其退出。
 
 ## 词汇
 
-`CodeRunRequest`（`program`、`bindings`、`signal?`）携带运行时操作所需的全部内容；默认值解析（时间预算与外层输出上限）属于实现的已验证配置，绝不能是隐藏的 `??`，更不能藏在 `run()` 内部。`bindings` 是 `CodeBindingNamespace` 列表（`global` + `functions` + 可选 `errorClass`）；每个命名空间会作为一个由异步可调用函数组成的全局对象公开给程序，这些函数返回 `CodeJsonValue`。后者是 seam 本地、与规范 `JsonValue` 结构等价的类型，使接口包保持独立于会话。`errorClass` 描述符点名真实的程序全局构造器，以及用于接收 reject 成员名称的自有属性；运行时不依赖 `ToolCallError` 等消费方术语。`CodeRunResult` 报告无损 JSON 完成值 `value?`、有序的 `logs: string[]` 和 `error?`（`CodeRunFailure`：`kind` + 可反馈给模型的 `message`）。完整契约见 `src/types.ts`。
+`CodeRunRequest`（`program`、`bindings`、`signal?`）携带运行时操作所需的全部内容；默认值解析（时间预算与外层输出上限）属于实现的已验证配置，绝不能是隐藏的 `??`，更不能藏在 `run()` 内部。`bindings` 是 `CodeBindingNamespace` 列表（`global` + `functions` + 可选 `errorClass`）；每个命名空间会作为一个由异步可调用函数组成的全局对象公开给程序，这些函数返回 `CodeJsonValue`。后者是 seam 本地、与规范 `JsonValue` 结构等价的类型，使接口包保持独立于会话。`errorClass` 描述符点名真实的程序全局构造器，以及用于接收被拒绝成员名称的自有属性；运行时不依赖 `ToolCallError` 等消费方术语。`CodeRunResult` 报告无损 JSON 完成值 `value?`、有序的 `logs: string[]` 和 `error?`（`CodeRunFailure`：`kind` + 可反馈给模型的 `message`）。完整契约见 `src/types.ts`。
 
 ## 模型体验
 
@@ -26,11 +26,11 @@
 
 #### KV Cache 影响
 
-不会直接失效；由具名消费方负责请求前缀变更。
+不会直接失效；由上述消费方负责请求前缀变更。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- **`run()` 是一次性的**：`logs` 只有在 `CodeRunResult` resolve 后才能获得；seam 不提供活跃程序输出的流式日志或进度接口。
+- **`run()` 是一次性的**：`logs` 只有在 `CodeRunResult` resolve 后才能获得；seam 不提供正在运行的程序所产生输出的流式日志或进度接口。
 - **持久 REPL 风格内核已记录为未来工作**：在持久内核后端带来自己的日志方案前，运行之间不保留状态的契约继续有效（参见 [Code Mode Agent Note](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md)）。
-- **目前只提供 worker 线程后端**：`'process'`／`'container'` 是已经声明但没有实现的已知 `isolation` 值；硬安全边界需要等待容器后端。
+- **目前只提供 worker 线程后端**：`'process'`／`'container'` 是已经声明但没有实现的已知 `isolation` 值；强安全边界需要等待容器后端。
 - **中间绑定值没有字节上限**：实现仍受 structured-clone 成本与进程内存约束，而提供方或执行器可能已经应用自己的获取上限。
diff --git a/packages/compact/README.i18n.yaml b/packages/compact/README.i18n.yaml
index fcb73c5293..95109e1c76 100644
--- a/packages/compact/README.i18n.yaml
+++ b/packages/compact/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/compact/README.md
 README.md: 3c3644adce23c12db37241bf797ea614d273a0fb
-README.zh.md: ad3abfd4f6064a44ecb22cd22b6c969247484de3
+README.zh.md: 260a92154ecd33cb127391af5ded399a2bc20038
diff --git a/packages/compact/README.zh.md b/packages/compact/README.zh.md
index ad3abfd4f6..260a92154e 100644
--- a/packages/compact/README.zh.md
+++ b/packages/compact/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-一个压缩（compaction）能力家族（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：抽象接口、摘要后端、不依赖模型的工具结果剪枝配套组件，以及暂缓实现的面向模型消费方。这些全是**产品** 包。
+一个压缩（compaction）能力家族（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：抽象接口、摘要生成后端、不依赖模型的工具结果剪枝配套组件，以及暂缓实现的面向模型消费方。这些全是**产品**包（package）。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
 | `compact/` | 抽象压缩 seam（接口 + `compact/*` 事件 + `CompactionResult`） | `ctx.compact` |
-| `compact-basic/` | 后端：`ctx.tokenMeter` 压力 + token 预算保留 + `llm.stream()` 摘要生成 | （注册 `ctx.compact`） |
+| `compact-basic/` | 后端：`ctx.tokenMeter` 压力 + 按 token 预算保留内容 + `llm.stream()` 摘要生成 | （注册 `ctx.compact`） |
 | `compact-tool-result-prune/` | 可选的不依赖模型的头／中／尾重写，在摘要压缩之前运行 | `ctx.toolResultPrune` |
 | `tool-compact/`（暂缓） | 面向模型的 `/compact` 工具，基于 `ctx.compact` | （注册到 `ctx.tools`） |
 
-接口位于 `compact/compact/`，后端位于 `compact/compact-basic/`，确定性剪枝位于 `compact/compact-tool-result-prune/`。与 bash seam 不同，该接口依赖 `dsh-session` 和 `dsh-llm`，因为它的动词基于 `Session` 定义，输出则使用 `ContentBlock`。这项偏差记录在[压缩能力 seam Agent Note](../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中。token 测量仍是可复用的 LLM 家族服务；模板或模型支持的压缩器可以替换 `compact-basic`，而无需更改计量器、剪枝器或调用方。
+接口位于 `compact/compact/`，后端位于 `compact/compact-basic/`，确定性剪枝位于 `compact/compact-tool-result-prune/`。与 bash seam 不同，该接口依赖 `dsh-session` 和 `dsh-llm`，因为它的操作以 `Session` 为对象，输出则使用 `ContentBlock`。这项偏差记录在[压缩能力 seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中。token 测量仍是可复用的 LLM（大语言模型）家族服务；基于模板或模型的压缩器可以替换 `compact-basic`，而无需更改计量器、剪枝器或调用方。
diff --git a/packages/compact/compact-basic/README.i18n.yaml b/packages/compact/compact-basic/README.i18n.yaml
index 794d20e67a..c7cb49d338 100644
--- a/packages/compact/compact-basic/README.i18n.yaml
+++ b/packages/compact/compact-basic/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/compact/compact-basic/README.md
 README.md: 775355f1ac1a7c79c16f66a5b2489d73df7b960d
-README.zh.md: 2f7ccc7dd00fa5599d3d6bbe66e81d3312d38dd9
+README.zh.md: bfa139596b5ef61c23d29575bdea5534fa82b158
diff --git a/packages/compact/compact-basic/README.zh.md b/packages/compact/compact-basic/README.zh.md
index 2f7ccc7dd0..bfa139596b 100644
--- a/packages/compact/compact-basic/README.zh.md
+++ b/packages/compact/compact-basic/README.zh.md
@@ -4,28 +4,28 @@
 
 **基础压缩（compaction）后端**：`BasicCompactService` 实现 `@deepseek-ai/dsh-compact` seam，使用可复用的 `ctx.tokenMeter` 压力、token 预算保留与摘要。摘要是直接的一次性 `ctx.llm.stream()` 调用，它会回放会话前缀以复用提供方的 KV cache（可在 `llm/stream` 处拦截）。
 
-这是压缩能力的实现层。seam 见 [接口包](../compact/README.md)，设计见 [能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md)。
+这是压缩能力的实现层。seam 见 [接口包（package）](../compact/README.md)，设计见 [能力 seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md)。
 
 ## 拥有的职责
 
 该后端拥有压缩策略：
 
-- **测量**：单例 `ctx.tokenMeter` 会在同一个已消费日志 revision 上为最新规范已记录 envelope 与当前表层计价。因此，步骤边界压力会包含实际系统提示词、工具、路由、assistant 完成、工具结果、缓冲上下文与 steering。
-- **路由策略**：主动压力从拥有最新持久提供方／模型路由的适配器解析容量，再将默认策略与可选的精确目标覆盖缩放为具体 token 预算。模型发现仍只提供建议，不会被咨询。
-- **不依赖模型的剪枝**：在压力或规范溢出符合条件后，可选的 [`ctx.toolResultPrune`](../compact-tool-result-prune/README.md) 服务会在选择范围之前改写超大工具结果。Compact-basic 通过 `ctx.tokenMeter` 重新测量；如果压力已回到安全范围，就跳过摘要，否则摘要已剪枝表层。低于压力的步骤检查绝不剪枝。
-- **保留**：压缩最旧的完整表层单元，同时保留近期尾部，并通过 [`dsh-compact` 边界 helper](../compact/README.md#tool-pairing-boundaries) 保持工具调用／结果 cut 平衡。轮次边界不会保护失控轮次内的旧步骤。开启且不可分的尾部在关闭前会拒绝压缩。当闭合的超大工具单元以文本型结果为可移除主体时，可选 pruner 可以修复它；不可分的非工具单元与不可剪枝的工具剩余部分不在范围内。
+- **测量**：单例 `ctx.tokenMeter` 会在同一个已消费日志 revision 上，计量最新一份规范化已记录 envelope 与当前表层的 token 用量。因此，步骤边界的压力计量会包含实际系统提示词、工具、路由、assistant 完成、工具结果、缓冲上下文与 steering（中途引导）。
+- **路由策略**：主动压力从拥有最新持久提供方／模型路由的适配器解析容量，再将默认策略与可选的精确目标覆盖缩放为具体 token 预算。模型发现仍仅供参考，不参与此处的策略解析。
+- **不依赖模型的剪枝**：在压力或规范溢出符合条件后，可选的 [`ctx.toolResultPrune`](../compact-tool-result-prune/README.md) 服务会在选择范围之前改写超大工具结果。Compact-basic 通过 `ctx.tokenMeter` 重新测量；如果压力已回到安全范围，就跳过摘要，否则对已剪枝的表层进行摘要。低于压力的步骤检查绝不剪枝。
+- **保留**：压缩最旧的完整表层单元，同时保留近期尾部，并通过 [`dsh-compact` 边界 helper](../compact/README.md#tool-pairing-boundaries) 将切分点调整到工具调用／结果配对平衡的位置。轮次边界不会保护失控轮次内的旧步骤。尚未闭合且不可分的尾部会在闭合前拒绝压缩。当闭合的超大工具单元以文本型结果为可移除主体时，可选 pruner 可以修复它；不可分的非工具单元与不可剪枝的工具剩余部分不在范围内。
 - **收敛**：最多按 `compactionRetries` 重试头部检查点压缩；拒绝不能缩小源内容的摘要，如果重试仍无法回到阈值以下，则抛出异常。
-- **摘要**：直接 `llm/stream` 调用使用已配置的提供方／模型对与上限，回退到最新已记录请求目标，然后再回退到 agent（智能体）目标，而不运行仅用于 loop 的 `agent/request` seam。该调用会逐字回放会话自身的系统提示词、工具与已遮蔽区域消息，并将压缩指令作为最后一条 user 消息追加，从而复用提供方的热前缀 cache，而非使它失效。它将 `GenerateOptions.purpose` 设为 `compaction`，适配器可将其作为请求归因转发（DeepSeek 适配器发送 `x-deepseek-harness-compact: 1`），但不会触碰模型可见主体。只有返回文本会进入检查点；会排除可能泄露私有推理或产生遗留调用的 reasoning 与工具调用。
+- **摘要**：直接 `llm/stream` 调用使用已配置的提供方／模型对与上限，回退到最新已记录请求目标，然后再回退到 agent 目标，而不运行仅用于 agent loop 的 `agent/request` seam。该调用会逐字回放会话自身的系统提示词、工具与已遮蔽区域消息，并将压缩指令作为最后一条 user 消息追加，从而复用提供方的热前缀 cache，而非使它失效。它将 `GenerateOptions.purpose` 设为 `compaction`，适配器可将其作为请求归因转发（DeepSeek 适配器发送 `x-deepseek-harness-compact: 1`），但不会触碰模型可见的请求体。只有返回的文本会进入检查点；推理（reasoning）和工具调用都会被排除，以免泄露私有推理或产生遗留调用。
 - **框定**：替换 user 消息使用 `<compacted-summary>` 标签标记已建立的检查点上下文。原始摘要保留在溯源事件上，后续自动周期会合并之前的检查点。
-- **生命周期**：`compactRegion()` 会更改 `agent.session`，并记录开始、摘要、替换与结束。异步摘要后，它会拒绝已改变的表层节点快照，而不相关的仅日志事件可以追加，不会使已选 span 失效。串行 `agent/step` listener 会在派生请求之前检查压力。规范提供方溢出会在失败步骤之后经由 `agent/request-error` 交给本插件；插件在此执行压缩，并且只在表层取得持久进展后才返回重试动作。
-- **溢出恢复**：提供方已确认的溢出不需容量元数据。它会绕过常规压力与保留，执行剪枝，再尝试一次最大平衡头部缩减，并留下最新不可分单元。只要 `surface.replaceGeneration` 前进，就允许重试，包括剪枝在后续摘要工作抛出异常前已落地的情况。如果没有替换、精确目标上限已耗尽、已取消，或遇到未知／非规范错误，则保留原始提供方失败。
-- **失败处理**：不匹配的 `compact/start` 是惰性崩溃标记，因为没有摘要替换落地。区域失败会记录错误结束；除非剪枝已落地，否则表层保持不变。操作性压力失败会发出警告并继续；只当之前没有替换使表层前进时，溢出恢复失败才保留原始提供方错误。取消在任何进展后仍具有最高权威。
+- **生命周期**：`compactRegion()` 会更改 `agent.session`，并记录开始、摘要、替换与结束。异步摘要后，如果表层节点快照已改变，它会拒绝操作，而不相关的仅日志事件可以追加，不会使已选 span 失效。串行 `agent/step` listener 会在派生请求之前检查压力。规范提供方溢出会在失败步骤之后经由 `agent/request-error` 交给本插件；插件在此执行压缩，并且只在表层取得持久进展后才返回重试动作。
+- **溢出恢复**：提供方已确认的溢出不需容量元数据。它会绕过常规压力与保留，执行剪枝，再尝试一次最大平衡头部缩减，并留下最新不可分单元。只要 `surface.replaceGeneration` 前进，就允许重试，包括剪枝在后续摘要工作抛出异常前已落地的情况。如果没有替换、目标特定上限已耗尽、已取消，或遇到未知／非规范错误，则保留原始提供方失败。
+- **失败处理**：未配对的 `compact/start` 是不起作用的崩溃标记，因为没有摘要替换落地。区域失败会记录错误结束；除非剪枝已落地，否则表层保持不变。压力检查中的运行故障会发出警告并继续；只有此前没有替换推进表层时，溢出恢复失败才保留原始提供方错误。即使已经取得进展，取消仍具有最终决定权。
 
-受保护的 `summarize()` 方法是唯一的子类 hook。基于模板或远程摘要器的子类可以覆盖该方法，同时压力、保留、溯源、缩减验证与已遮蔽 token 计量仍位于 `ctx.tokenMeter`。hook 会将摘要块与它使用的调用 envelope 一并返回（`{ summary, provider, model, maxTokens? }`），并记录在 `compact/summary` 上。
+受保护的 `summarize()` 方法是唯一的子类钩子。基于模板或远程摘要器的子类可以覆盖该方法，同时压力、保留、溯源、缩减验证与已遮蔽 token 计量仍由 `ctx.tokenMeter` 负责。钩子会将摘要块与它使用的调用 envelope 一并返回（`{ summary, provider, model, maxTokens? }`），并记录在 `compact/summary` 上。
 
 ## 配置（`BasicCompactConfig`）
 
-所有设置都可选。顶层策略字段是每个已路由模型的默认值；`modelPolicies` 对精确提供方／模型对应用部分覆盖。出现压力时，compact-basic 会请求所属 LLM 适配器提供该路由的上下文容量，并解析绝对预算。无法识别的 key、重复目标、互斥保留形式，以及合并后的 `retainRatio` 不低于 `thresholdRatio`，都会使插件加载失败。不低于缩放后阈值的绝对 `retainTokens` 预算会在第一个可解析目标上失败，因为该比较需要模型容量。
+所有设置都可选。顶层策略字段是每个已路由模型的默认值；`modelPolicies` 对精确提供方／模型对应用部分覆盖。出现压力时，compact-basic 会请求所属 LLM（大语言模型）适配器提供该路由的上下文容量，并解析绝对预算。无法识别的配置键、重复目标、互斥保留形式，以及合并后的 `retainRatio` 不低于 `thresholdRatio`，都会使插件加载失败。不低于缩放后阈值的绝对 `retainTokens` 预算会在首次解析出目标时导致失败，因为该比较需要模型容量。
 
 | Key | 必填 | 含义 |
 |---|---|---|
@@ -34,7 +34,7 @@
 | `retainTokens` | 否 | 逐字保留的近期表层绝对预算；与 `retainRatio` 互斥，并且必须低于已解析阈值。 |
 | `summarizationProvider` | 否（默认 `''`） | 与 `summarizationModel` 一起设置；空对会解析为最新已记录请求目标，再回退到 `AgentOptions` 对。 |
 | `summarizationModel` | 否（默认 `''`） | 与 `summarizationProvider` 一起设置；空对会解析为最新已记录请求目标，再回退到 `AgentOptions` 对。 |
-| `maxTokens` | 否（默认 `8192`） | 摘要调用的提供方生成上限；可包含 reasoning token。 |
+| `maxTokens` | 否（默认 `8192`） | 摘要调用的提供方生成上限；可包含推理 token。 |
 | `compactionRetries` | 否（默认 `1`） | 压力仍高于阈值时，在首次尝试后进行的额外尝试次数。 |
 | `maxOverflowRetries` | 否（默认 `1`） | 规范上下文窗口溢出后的最大重试次数；`0` 只禁用恢复。 |
 | `modelPolicies` | 否（默认 `[]`） | 精确的 `{ provider, model, ...partialPolicy }` 覆盖；匹配使用两个字段，不依赖 `listModels()`。 |
@@ -60,9 +60,9 @@ export function apply(ctx: Context): void {
 }
 ```
 
-加载插件会注册 `ctx.compact`。在该插件之前添加同级 [`dsh-compact-tool-result-prune`](../compact-tool-result-prune/README.md) 以启用可选的不依赖模型 pass。当 `auto: true`（默认）时，它会在 token 压力下自动压缩；消费方（未来的 `/compact` 工具）也可直接调用 `ctx.compact.compactIfNeeded(...)` 或 `ctx.compact.compactRegion(...)`。
+加载插件会注册 `ctx.compact`。在该插件之前添加同级 [`dsh-compact-tool-result-prune`](../compact-tool-result-prune/README.md) 以启用可选的不依赖模型的处理阶段。当 `auto: true`（默认）时，它会在 token 压力下自动压缩；消费方（未来的 `/compact` 工具）也可直接调用 `ctx.compact.compactIfNeeded(...)` 或 `ctx.compact.compactRegion(...)`。
 
-例如，同一个 compact 插件可以安全服务于容量不同的模型，并应用一项目标特定策略：
+例如，同一个压缩插件可以安全服务于容量不同的模型，并应用一项目标特定策略：
 
 ```yaml
 - name: '@deepseek-ai/dsh-compact-basic'
@@ -82,7 +82,7 @@ export function apply(ctx: Context): void {
 
 #### 模型看到的内容
 
-成功步骤越过阈值后，如果已加载可选 pruner，超大工具结果会先被改写。如果仍需摘要，下一个请求会收到下方检查点前导、一个空行、`<compacted-summary>`、取决于数据的摘要以及 `</compacted-summary>`。溢出恢复会根据使表层前进的任何替换重建立即重试。检查点会替换已选较早范围，后面跟随已保留的近期单元。
+成功步骤越过阈值后，如果已加载可选 pruner，超大工具结果会先被改写。如果仍需摘要，下一个请求会收到下方检查点前导、一个空行、`<compacted-summary>`、根据数据生成的摘要以及 `</compacted-summary>`。溢出恢复会根据使表层前进的任何替换重建立即重试。检查点会替换已选较早范围，后面跟随已保留的近期单元。
 
 ##### 会话检查点前导
 
@@ -92,7 +92,7 @@ This is an automatically generated checkpoint condensing an earlier span of the
 
 #### Token 影响
 
-不依赖模型的剪枝可以完全避免辅助调用；否则它会在摘要替换较早范围之前缩减该调用的 transcript。替换会缩减未来输入历史，而非追加第二份副本。摘要会保留到后续压缩将其替换，但不可分的非工具单元仍可能超出预算。
+不依赖模型的剪枝可以完全避免辅助调用；否则它会在摘要替换较早范围之前缩减该调用的 transcript（文本记录）。替换会缩减未来输入历史，而非追加第二份副本。摘要会保留到后续压缩将其替换，但不可分的非工具单元仍可能超出预算。
 
 #### KV Cache 影响
 
@@ -153,9 +153,9 @@ Rules:
 
 ## 已知限制与暂缓事项
 
-- **Meter 准确度遵循固定启发式规则**：可复用提供方用量缺失时，会回退到字符数加结构开销，而非精确 tokenization。
+- **计量准确度取决于固定启发式规则**：可复用提供方用量缺失时，会回退到字符数加结构开销，而非精确的 token 化。
 - **溢出分类由适配器维护**：提供方措辞可能改变；两个 DeepSeek 适配器将当前可识别的上下文限制失败规范化为 `CONTEXT_WINDOW_EXCEEDED`。
 - **部分不可分单元与仅 envelope 溢出仍不在表层压缩范围内**：恢复无法缩减系统／工具／前缀、拆分不可分的非工具节点，或修复不可剪枝剩余部分仍超出窗口的工具单元。可选 pruner 可以缩减原本不可分工具对内的文本型工具结果主体。
-- **`compactRegion` 要求开启轮次**：在完全关闭的会话上手动调用会抛出异常（「no open turn」），而不是执行压缩。
-- **摘要失败会保留最新持久表层**：任何替换前，自动路径会记录警告，并携带完整超预算历史继续。如果剪枝已落地，后续摘要失败会从该持久剪枝表层继续。在 `maxTokens` 处的摘要截断（可能由隐藏 reasoning token 耗尽）遵循同一规则。
-- **摘要调用没有 transcript 快照覆盖**：`dsh-llm-replay` 从 `assistant/chunk` 事件派生调用，因此这次不含 chunk 的直接 `ctx.llm.stream()` 调用无法回放（[seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中明确的暂缓回放基础设施）。
+- **`compactRegion` 要求存在未结束的轮次**：在完全关闭的会话上手动调用会抛出异常（「no open turn」），而不是执行压缩。
+- **摘要失败会保留最新持久表层**：任何替换前，自动路径会记录警告，并携带完整超预算历史继续。如果剪枝已落地，后续摘要失败会从该持久剪枝表层继续。因达到 `maxTokens` 而发生的摘要截断（隐藏推理 token 可能会耗尽该额度）遵循同一规则。
+- **摘要调用没有 transcript 快照覆盖**：`dsh-llm-replay` 从 `assistant/chunk` 事件派生调用，因此这次不含分片的直接 `ctx.llm.stream()` 调用无法回放（[seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中明确的暂缓回放基础设施）。
diff --git a/packages/compact/compact-basic/src/region.ts b/packages/compact/compact-basic/src/region.ts
index 3bd1cd43a9..d5eb7cfc40 100644
--- a/packages/compact/compact-basic/src/region.ts
+++ b/packages/compact/compact-basic/src/region.ts
@@ -126,7 +126,9 @@ export async function compactSurfaceRegion(
     }
     const shadowedTokenCount = selected.reduce((total, node) => total + node.tokens, 0)
     const summarizationInput = buildSummarizationInput(session, shadowedSeqs)
-    const { summary, provider, model, maxTokens } = await dependencies.summarize(summarizationInput, agent, signal)
+    const {
+      summary, rawOutput, provider, model, maxTokens, usage,
+    } = await dependencies.summarize(summarizationInput, agent, signal)
 
     const currentMeasurement = dependencies.meter.measure(session)
     if (!isDeepStrictEqual(currentMeasurement.nodes, lockedMeasurement.nodes)) {
@@ -146,12 +148,14 @@ export async function compactSurfaceRegion(
 
     const summaryEvent = session.append('compact/summary', {
       summary,
+      ...rawOutput === undefined ? {} : { rawOutput },
       shadowedRange: { start, end },
       shadowedSeqs,
       shadowedTokenCount,
       provider,
       model,
       ...maxTokens === undefined ? {} : { maxTokens },
+      ...usage === undefined ? {} : { usage },
     })
     session.append('user/message', checkpointMessage, {
       surfaceOp: { op: 'replace', start, end },
diff --git a/packages/compact/compact-basic/src/summarizer.ts b/packages/compact/compact-basic/src/summarizer.ts
index 4f9d94518b..3202970c7d 100644
--- a/packages/compact/compact-basic/src/summarizer.ts
+++ b/packages/compact/compact-basic/src/summarizer.ts
@@ -6,7 +6,9 @@
 
 import type { Context } from 'cordis'
 import { createUserMessage, BlockAssembler } from '@deepseek-ai/dsh-llm'
-import type { ContentBlock, FinishReason, GenerateOptions, Message, ToolSchema } from '@deepseek-ai/dsh-llm'
+import type {
+  ContentBlock, FinishReason, GenerateOptions, Message, TokenUsage, ToolSchema,
+} from '@deepseek-ai/dsh-llm'
 import type { Agent } from '@deepseek-ai/dsh-agent'
 
 interface SummaryConfig {
@@ -85,9 +87,13 @@ export interface SummarizationInput {
 /** Safe summary content plus the exact auxiliary call envelope recorded in provenance. */
 export interface SummaryResult {
   summary: ContentBlock[]
+  /** Complete provider output before the text-only summary projection. */
+  rawOutput?: ContentBlock[]
   provider: string
   model: string
   maxTokens?: number
+  /** Provider-reported usage for this summarization request. */
+  usage?: TokenUsage
 }
 
 /**
@@ -148,15 +154,18 @@ export async function summarizeWithLlm(
   const error = finishError(assembler.finish)
   if (error !== undefined) throw error
 
-  const summary = textOnly(assembler.blocks())
+  const rawOutput = assembler.blocks()
+  const summary = textOnly(rawOutput)
   if (!summary.some(block => block.text.trim().length > 0)) {
     throw new Error('summarization produced no text summary content')
   }
   return {
     summary,
+    rawOutput,
     provider: options.provider,
     model: options.model,
     maxTokens: config.maxTokens,
+    ...(assembler.usage === undefined ? {} : { usage: assembler.usage }),
   }
 }
 
diff --git a/packages/compact/compact-basic/tests/compact-basic.spec.ts b/packages/compact/compact-basic/tests/compact-basic.spec.ts
index a368a3631d..2b3cdecf17 100644
--- a/packages/compact/compact-basic/tests/compact-basic.spec.ts
+++ b/packages/compact/compact-basic/tests/compact-basic.spec.ts
@@ -19,6 +19,7 @@ import type {
   LlmResolvedModelInfo,
   Message,
   StreamChunk,
+  TokenUsage,
 } from '@deepseek-ai/dsh-llm'
 import { Session, SessionId } from '@deepseek-ai/dsh-session'
 import TokenMeterService from '@deepseek-ai/dsh-token-meter'
@@ -233,6 +234,8 @@ function oversizedToolResult(chars = 3_000, withCompactablePrompt = false): Sess
 
 class TestCompactService extends BasicCompactService {
   summary: ContentBlock[] = [{ type: 'text', text: 'small checkpoint' }]
+  rawOutput: ContentBlock[] | undefined
+  usage: TokenUsage | undefined
   summaryProvider = 'summary-provider'
   summaryModel = 'summary-model'
   error: unknown
@@ -243,15 +246,24 @@ class TestCompactService extends BasicCompactService {
     input: SummarizationInput,
     _agent: Agent,
     signal?: AbortSignal,
-  ): Promise<{ summary: ContentBlock[]; provider: string; model: string; maxTokens?: number }> {
+  ): Promise<{
+    summary: ContentBlock[]
+    rawOutput?: ContentBlock[]
+    provider: string
+    model: string
+    maxTokens?: number
+    usage?: TokenUsage
+  }> {
     this.calls.push({ input, signal })
     this.mutateDuringSummary?.()
     if (this.error !== undefined) throw this.error
     return {
       summary: this.summary,
+      ...this.rawOutput === undefined ? {} : { rawOutput: this.rawOutput },
       provider: this.summaryProvider,
       model: this.summaryModel,
       maxTokens: 123,
+      ...this.usage === undefined ? {} : { usage: this.usage },
     }
   }
 }
@@ -829,6 +841,11 @@ describe('optional model-free tool-result pruning', () => {
 describe('compaction region transaction', () => {
   it('lands a framed, replayable checkpoint with exact pricing provenance', async () => {
     const compact = service()
+    compact.rawOutput = [
+      { type: 'reasoning', text: 'private compact thought' },
+      ...compact.summary,
+    ]
+    compact.usage = { inputTokens: 40, outputTokens: 5 }
     const session = conversation(3)
     const before = [...session.surface.nodes]
     const result = await compact.compactRegion(
@@ -849,6 +866,8 @@ describe('compaction region transaction', () => {
       provider: 'summary-provider',
       model: 'summary-model',
       maxTokens: 123,
+      rawOutput: compact.rawOutput,
+      usage: compact.usage,
     })
     const head = session.deriveMessages()[0]!
     expect(head.content[0]?.type).toBe('text')
@@ -1088,6 +1107,7 @@ describe('compaction region transaction', () => {
 
 class ScriptedAdapter extends LlmAdapter {
   lastOptions: GenerateOptions | undefined
+  usage: TokenUsage | undefined
 
   constructor(
     private readonly blocks: readonly ContentBlock[],
@@ -1108,6 +1128,7 @@ class ScriptedAdapter extends LlmAdapter {
         yield { type: 'block-end', index, block }
       }
     }
+    if (this.usage !== undefined) yield { type: 'usage', usage: this.usage }
     yield { type: 'finish', reason: this.finish }
   }
 }
@@ -1117,7 +1138,14 @@ class ExposedCompactService extends BasicCompactService {
     input: SummarizationInput,
     owner: Agent,
     signal?: AbortSignal,
-  ): Promise<{ summary: ContentBlock[]; provider: string; model: string; maxTokens?: number }> {
+  ): Promise<{
+    summary: ContentBlock[]
+    rawOutput?: ContentBlock[]
+    provider: string
+    model: string
+    maxTokens?: number
+    usage?: TokenUsage
+  }> {
     return this.summarize(input, owner, signal)
   }
 }
@@ -1150,13 +1178,20 @@ describe('default one-shot summarizer', () => {
       maxTokens: 321,
     })
     const session = conversation(1)
+    adapter.usage = { inputTokens: 12, outputTokens: 3 }
     const output = await compact.runSummarize(promptInput('transcript'), agent(session, 'fallback'), SIGNAL)
 
     expect(output).toEqual({
       summary: [{ type: 'text', text: 'public summary' }],
+      rawOutput: [
+        { type: 'reasoning', text: 'private' },
+        { type: 'text', text: 'public summary' },
+        { type: 'tool-call', id: CallId('unexpected'), name: 'x', arguments: '{}' },
+      ],
       provider: MODEL,
       model: MODEL,
       maxTokens: 321,
+      usage: adapter.usage,
     })
     expect(adapter.lastOptions).toMatchObject({
       provider: MODEL,
diff --git a/packages/compact/compact-tool-result-prune/README.i18n.yaml b/packages/compact/compact-tool-result-prune/README.i18n.yaml
index 8da3d60991..a9445fca25 100644
--- a/packages/compact/compact-tool-result-prune/README.i18n.yaml
+++ b/packages/compact/compact-tool-result-prune/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/compact/compact-tool-result-prune/README.md
 README.md: edeba52b189b3cee5530faf7efc04043a326917f
-README.zh.md: fa471c0817cc937162b386cd3d249ce5ecd8c3ff
+README.zh.md: 1b42a9db5d3288c6610e431c57403858a238c07a
diff --git a/packages/compact/compact-tool-result-prune/README.zh.md b/packages/compact/compact-tool-result-prune/README.zh.md
index fa471c0817..1b42a9db5d 100644
--- a/packages/compact/compact-tool-result-prune/README.zh.md
+++ b/packages/compact/compact-tool-result-prune/README.zh.md
@@ -2,31 +2,31 @@
 
 [English](README.md) | 中文
 
-可安全回放、不依赖模型的剪枝服务（`ctx.toolResultPrune`）。它会将超出预算的 `tool/result` 表层节点改写为有界头部、固定省略标记和有界尾部，同时在仅追加会话日志中保留完整原始事件。
+可安全回放、不依赖模型的剪枝服务（`ctx.toolResultPrune`）。它会将超出预算的 `tool/result` 表层节点改写为长度受限的头部、固定省略标记和长度受限的尾部，同时在仅追加会话日志中保留完整原始事件。
 
-这是 [`dsh-compact-basic`](../compact-basic/README.md) 的具体配套服务，不是压缩后端或面向模型的工具。Compact-basic 通过可选的 `ctx.get('toolResultPrune')` 读取它，因此两个包都保持可独立组合。
+这是 [`dsh-compact-basic`](../compact-basic/README.md) 的具体配套服务，不是压缩（compaction）后端或面向模型的工具。Compact-basic 通过可选的 `ctx.get('toolResultPrune')` 读取它，因此这两个包（package）仍可各自独立组合。
 
 ## 服务 API
 
-`pruneSession(session)` 会扫描当前表层的一个稳定快照。每个超出预算的工具结果都会被一个新追加的 `tool/result` 替换，其携带 `{ surfaceOp: { op: 'replace', start: originalSeq, end: originalSeq }, sourceEventSeqs: [originalSeq] }`。替换会展开完整原始数据，只更改 `content`，保留 `turn`、`step`、`callId`、错误字段、`meta` 以及后续添加的数据。原始事件仍可用于持久化、回放和精确日志检查。
+`pruneSession(session)` 会扫描当前表层的一个稳定快照。每个超出预算的工具结果都会被一个新追加的 `tool/result` 替换，其携带 `{ surfaceOp: { op: 'replace', start: originalSeq, end: originalSeq }, sourceEventSeqs: [originalSeq] }`。替换会展开完整原始数据，只更改 `content`，保留 `turn`、`step`、`callId`、错误字段、`meta` 以及以后新增的数据字段。原始事件仍可用于持久化、回放和精确日志检查。
 
-当会话拒绝替换时，该方法会同步抛出异常。本次 pass 中较早提交的替换仍然持久。
+当会话拒绝替换时，该方法会同步抛出异常。本次扫描中先前已提交的替换仍会保留。
 
-`measureContent(blocks)` 会统计 `text` 块中的 Unicode 码点。`pruneContent(blocks)` 会返回有界替换；如果内容已在阈值内，则返回 `null`。非文本块保持原始相对位置；文本切片绝不会拆分 UTF-16 surrogate pair，但可能拆分由多个码点组成的 grapheme cluster。
+`measureContent(blocks)` 会统计 `text` 块中的 Unicode 码点。`pruneContent(blocks)` 会返回长度受限的替换；如果内容已在阈值内，则返回 `null`。非文本块保持原始相对位置；文本切片绝不会拆分 UTF-16 代理项对，但可能拆分由多个码点组成的字素簇。
 
-每个发出的结果在文本码点上都精确包含已配置的头部预算、固定标记和尾部预算，不大于 `thresholdChars`，且严格小于触发输入。因此第二次 pass 不会发出替换。
+每个发出的结果在文本码点上都精确包含已配置的头部预算、固定标记和尾部预算，不大于 `thresholdChars`，且严格小于触发输入。因此第二次扫描不会发出替换。
 
 ## 配置
 
-无法识别的 key 会使插件在构造时失败。已解析配置与输入脱离，并且深度不可变。
+无法识别的配置键会使插件在构造时失败。已解析配置与输入脱离，并且深度不可变。
 
-| Key | 必填 | 含义 |
+| 配置键 | 必填 | 含义 |
 |---|---|---|
 | `thresholdChars` | 否（默认 `8192`） | 合并文本超过此 Unicode 码点数时剪枝。 |
 | `headChars` | 否（默认 `4096`） | 保留的开头 Unicode 码点数。 |
 | `tailChars` | 否（默认 `1024`） | 保留的末尾 Unicode 码点数。 |
 
-所有值都必须是整数；阈值必须为正数，头部／尾部必须为非负数。`headChars + marker + tailChars` 必须能容纳在 `thresholdChars` 内，因此有效配置可以剪枝每个超出预算的结果，不会增长或重复改写。
+所有值都必须是整数；阈值必须为正数，头部／尾部必须为非负数。`headChars + marker + tailChars` 之和不得超过 `thresholdChars`，因此有效配置可以剪枝每个超出预算的结果，不会增长或重复改写。
 
 ## 用法
 
@@ -45,7 +45,7 @@ export function apply(ctx: Context): void {
 
 #### 模型看到的内容
 
-一旦压缩触发器成立，后续请求会看到保留的头部、`\n\n[... tool result middle pruned ...]\n\n` 和保留的尾部，用它们替换已移除文本。富内容块保持顺序。模型不会看到原文的第二份副本。
+一旦满足压缩触发条件，后续请求看到的将是保留的头部、`\n\n[... tool result middle pruned ...]\n\n` 和保留的尾部，而非被移除的文本。非文本块保持原有顺序。模型不会看到原文的第二份副本。
 
 #### Token 影响
 
@@ -59,4 +59,4 @@ export function apply(ctx: Context): void {
 
 - **字符预算不是 token 预算**：不同提供方的 token 密度各异，因此 `ctx.tokenMeter` 仍负责判定剪枝是否缓解了请求压力。
 - **剪枝只基于语法**：它保留开头与结尾，不解释中间哪些行在语义上重要。
-- **Grapheme cluster 可能被拆分**：按码点切片可保护 surrogate pair，但不会执行感知 locale 的 grapheme 分割。
+- **字素簇可能被拆分**：按码点切片可保护代理项对，但不会执行感知区域设置的字素簇分割。
diff --git a/packages/compact/compact/README.i18n.yaml b/packages/compact/compact/README.i18n.yaml
index 921ee62a6d..310abbb287 100644
--- a/packages/compact/compact/README.i18n.yaml
+++ b/packages/compact/compact/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/compact/compact/README.md
 README.md: 17a5420ae9fa23ce4021b4d4927ae5b95962f979
-README.zh.md: d98251649cfdcf6b12e89a192a43b08b0f071f38
+README.zh.md: 56624efad3c4ffacb61c1a2e199e0461afc57242
diff --git a/packages/compact/compact/README.zh.md b/packages/compact/compact/README.zh.md
index d98251649c..56624efad3 100644
--- a/packages/compact/compact/README.zh.md
+++ b/packages/compact/compact/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-**压缩 seam**：抽象 `CompactService`（`ctx.compact`）定义压缩做什么，即判定历史记录是否过大，并将较早范围摘要为单个表层节点，但不规定如何实现。
+**压缩（compaction） seam**：抽象 `CompactService`（`ctx.compact`）定义压缩做什么，即判定历史记录是否过大，并将较早范围摘要为单个表层节点，但不规定如何实现。
 
-本包是压缩能力的接口层，因此各项职责可以独立演进（和替换）：
+这个包（package）是压缩能力的接口层，因此各项职责均可独立演进，也可独立替换：
 
 | 包 | 职责 |
 |---|---|
@@ -12,7 +12,7 @@
 | `@deepseek-ai/dsh-compact-basic` | 后端：`ctx.tokenMeter` 压力 + token 预算保留 + `llm.stream()` 摘要 |
 | `@deepseek-ai/dsh-tool-compact`（暂缓） | 面向模型的 `/compact` 工具，基于 `ctx.compact` 实现 |
 
-与 bash seam 不同，该接口依赖 `@deepseek-ai/dsh-session` 和 `@deepseek-ai/dsh-llm`。契约的动词基于 `Session` 定义，其输出使用 `ContentBlock` 词汇，因此无法在不指名这些包的情况下表达。这项对「接口只依赖 cordis」指引的偏离是有意的，并记录在 [压缩能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中。
+与 bash seam 不同，该接口依赖 `@deepseek-ai/dsh-session` 和 `@deepseek-ai/dsh-llm`。契约的动词基于 `Session` 定义，其输出使用 `ContentBlock` 词汇，因此无法在不指名这些包的情况下表达。这项对「接口只依赖 cordis」指引的偏离是有意的，并记录在 [压缩能力 seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-18-compaction-capability-seam.md) 中。
 
 ## 服务 API（`ctx.compact`）
 
@@ -20,22 +20,22 @@
 
 | 成员 | 语义 |
 |---|---|
-| `compactIfNeeded(agent, trigger, signal)` | 为 `trigger: 'pressure' \| 'context-overflow'` 考虑自动压缩。压力触发可应用后端的阈值与保留尾部策略；已确认溢出可强制进行有效的平衡缩减。返回 `CompactionResult`，无安全范围时则返回 `null`。后端摘要请求是直接 `ctx.llm.stream()` 调用（不是 loop 步骤），因此每次调用在 `llm/stream` 处拦截。 |
-| `compactRegion(start, end, agent, signal?)` | 强制将表层节点 `[start, end]`（包含两端 seq）从 `agent.session` 摘要为单个替换节点，其源为 `COMPACT_CHECKPOINT_SOURCE`。如果压缩已在进行、`start`／`end` 不是表层节点，或 `start` 在表层上位于 `end` 之后，则**抛出异常**。该范围是表层位置 span，不是数值 seq 区间：在之前的 replace 将新鲜高 seq 摘要节点放到已遮蔽范围的位置之后，表层顺序不再跟随 seq 顺序。 |
+| `compactIfNeeded(agent, trigger, signal)` | 根据 `trigger: 'pressure' \| 'context-overflow'` 判断是否需要自动压缩。压力触发可应用后端的阈值与保留尾部策略；已确认溢出可强制进行有效的平衡缩减。返回 `CompactionResult`，无安全范围时则返回 `null`。后端摘要请求是直接的 `ctx.llm.stream()` 调用（不是 agent loop 步骤），因此每次调用都可在 `llm/stream` 处拦截。 |
+| `compactRegion(start, end, agent, signal?)` | 强制将表层节点 `[start, end]`（包含两端 seq）从 `agent.session` 摘要为单个替换节点，其源为 `COMPACT_CHECKPOINT_SOURCE`。如果压缩已在进行、`start`／`end` 不是表层节点，或 `start` 在表层上位于 `end` 之后，则**抛出异常**。该范围是表层位置范围，不是数值 seq 区间：在之前的 replace 将新生成的高 seq 摘要节点放到已遮蔽范围的位置之后，表层顺序不再跟随 seq 顺序。 |
 
-`CompactionResult` 向调用方保留原始摘要与记账事件 seq，同时保留已遮蔽范围与 token 计量；其经漂移检查的形状位于 [压缩数据结构参考](../../../docs/core-data-structures/compaction.md#compactionresult)。
+`CompactionResult` 向调用方保留原始摘要与记录操作过程的事件 seq，同时保留已遮蔽范围与 token 计量；其结构由漂移检查保障，定义见 [压缩数据结构参考](../../../docs/core-data-structures/compaction.md#compactionresult)。
 
-`compactIfNeeded` 要求必填 `signal`；`compactRegion` 的该参数可选。通过 `ctx.llm.stream()` 摘要的后端**必须** 将它转发到调用的 `GenerateOptions.signal`，因此 abort 或 fiber dispose 会停止进行中的摘要，不会留下越过取消时点继续运行的遗留模型调用。可以从所拥有会话的日志（当前开启的轮次）恢复 `compact/*` 事件所属轮次，因此后端从日志中标记该值，而不信任调用方提供的值。
+`compactIfNeeded` 必须传入 `signal`；`compactRegion` 的该参数可选。通过 `ctx.llm.stream()` 摘要的后端**必须** 将它转发到调用的 `GenerateOptions.signal`，因此 abort 或 fiber dispose（资源释放） 会停止进行中的摘要，不会留下越过取消时点继续运行的遗留模型调用。可以从所拥有会话的日志（当前尚未结束的轮次）恢复 `compact/*` 事件所属轮次，因此后端从日志中标记该值，而不信任调用方提供的值。
 
 ## 工具配对边界
 
-该接口导出 `toolPairingBalancedBefore(session, seq)` 与 `toolPairingBalancedAfter(session, seq)`，用于对齐和验证压缩边缘。安全边缘不会被尚未回答的 assistant 工具调用跨越。每个 helper 都会验证事件序列位于当前表层，并使用按表层顺序为每个 cut 缓存的 balance 返回答案。
+该接口导出 `toolPairingBalancedBefore(session, seq)` 与 `toolPairingBalancedAfter(session, seq)`，用于对齐和验证压缩边界。安全边界不会被尚未回答的 assistant 工具调用跨越。每个 helper 都会验证给定事件 seq 位于当前表层，并根据按表层顺序缓存的各切分点配对状态返回结果。
 
-每会话私有 cache 以 `session.surface.replaceGeneration` 与已处理表层配置项数为 key。generation 未变时，fold 只会扩展到尚未处理的尾部配置项；不含新表层配置项的仅日志追加不会读取事件，而 replace generation 会重建当前成员关系与 balance。事件 seq 缺失以及没有之前开启调用的 `tool/result` 会被拒绝为表层状态损坏。
+每个会话的私有 cache 以 `session.surface.replaceGeneration` 和已处理表层条目数为 key。generation 未变时，只需将尚未处理的尾部条目纳入累计结果；仅向日志追加、但未新增表层条目时，不会读取事件。replace generation 变化时则会重建当前成员关系与配对状态。事件 seq 缺失以及 `tool/result` 没有对应的先前未闭合调用，均会被视为表层状态损坏并遭拒绝。
 
 ## 表层契约
 
-`SurfaceEventType` 是封闭联合：只有 `user/message`、`assistant/message`、`tool/result` 和 `steering/message` 可以携带 `surfaceOp`。因此 `compact/*` 事件**不能** 出现在表层上。成功压缩改为：
+`SurfaceEventType` 是封闭联合：只有 `user/message`、`assistant/message`、`tool/result` 和 `steering/message` 可以携带 `surfaceOp`。因此 `compact/*` 事件**不能**出现在表层上。成功压缩改为：
 
 1. 追加 `compact/start`（仅日志）：获取锁；
 2. 摘要该范围；
@@ -43,13 +43,13 @@
 4. 追加单个 `user/message`，其携带 `source: COMPACT_CHECKPOINT_SOURCE` 和包含摘要的 `surfaceOp: { op: 'replace', start, end }`：这是**本操作唯一的表层变更**；
 5. 追加 `compact/end`（仅日志）：释放锁。
 
-表层变更（第 4 步）位于锁括号**内部**：`compact/end` 是最后一个事件，因此表层变更落地前绝不会释放锁。如果在 `compact/start` 与 `compact/end` 之间崩溃，会留下可检测的遗留锁（一个 `compact/start` 没有匹配的 `compact/end`），而不是虚假声称压缩已完成、但表层从未被遮蔽的 `compact/end`。
+表层变更（第 4 步）位于锁的起止范围**内**：`compact/end` 是最后一个事件，因此表层变更落地前绝不会释放锁。如果在 `compact/start` 与 `compact/end` 之间崩溃，会留下可检测的遗留锁（一个 `compact/start` 没有匹配的 `compact/end`），而不是虚假声称压缩已完成、但表层从未被遮蔽的 `compact/end`。
 
 `deriveMessages()` 随后将摘要渲染为 user 角色消息，再跟上已保留节点。已遮蔽事件仍保留在原始日志中，因此回放具有确定性。
 
 ## 阻塞
 
-压缩通过日志记录的锁串行化：`compactRegion` 会拒绝启动，条件是最后一个 `compact/start` 之后没有匹配的 `compact/end`。锁就是日志（不是内存 mutex），因此它能在回放后存活，持久化后端也可以在重新加载时检测遗留 `compact/start`。锁会括住**整个** 操作：摘要、`compact/summary` 溯源记录*以及* `user/message` 表层替换全部发生在 `compact/end` 之前，因此 `session/event` listener 即使在 `compact/end` 时触发，也绝不会看到锁已释放而表层变更仍在等待。基础后端会在摘要后重新验证已选表层：表层变更会导致拒绝，不相关的仅日志追加不会使替换失效。即使摘要抛出异常，也会追加 `compact/end`，因此失败绝不会将锁卡死。
+压缩通过日志记录的锁串行化：`compactRegion` 会拒绝启动，条件是最后一个 `compact/start` 之后没有匹配的 `compact/end`。锁由日志记录（而非内存 mutex），因此回放后仍然有效，持久化后端也可以在重新加载时检测遗留 `compact/start`。锁会覆盖**整个**操作：摘要、`compact/summary` 溯源记录*以及* `user/message` 表层替换全部发生在 `compact/end` 之前，因此 `session/event` listener 即使在 `compact/end` 时触发，也绝不会看到锁已释放而表层变更仍在等待。基础后端会在摘要后重新验证已选表层：表层变更会导致拒绝，不相关的仅日志追加不会使替换失效。即使摘要抛出异常，也会追加 `compact/end`，因此失败绝不会将锁卡死。
 
 ## 事件
 
@@ -65,7 +65,7 @@
 
 #### 模型看到的内容
 
-成功实现会用一个 user 角色摘要检查点替换较早表层范围，即一个 `user/message`，它携带 `surfaceOp: { op: 'replace', start, end }`；原始事件仍会记录，但不再出现在派生模型消息中。seam 本身不执行改写。
+成功的实现会用一个 user 角色摘要检查点替换较早表层范围，即一个 `user/message`，它携带 `surfaceOp: { op: 'replace', start, end }`；原始事件仍会记录，但不再出现在派生模型消息中。seam 本身不执行改写。
 
 #### Token 影响
 
@@ -78,5 +78,5 @@
 ## 已知限制与暂缓事项
 
 - **尚无面向模型的消费方层**：`@deepseek-ai/dsh-tool-compact`（`/compact` 工具）已暂缓；只能通过直接 `ctx.compact` 调用或后端的自动 listener 进行压缩。
-- **部分单元溢出不在契约内**：平衡摘要压缩无法拆分一个不可分单元。当可移除的文本型工具结果体量较大时，可选剪枝配套服务仍可修复闭合工具对；无法压缩大型非工具节点，或不可剪枝剩余部分过大的工具单元。
+- **部分单元溢出不在契约内**：平衡摘要压缩无法拆分一个不可分单元。当闭合工具对中可移除的主要部分是承载文本的工具结果时，可选剪枝配套服务仍可修复该工具对；无法压缩大型非工具节点，或不可剪枝剩余部分过大的工具单元。
 - **单独接近窗口大小的 envelope 不属于表层压缩工作**：压缩缩减派生历史，绝不缩减系统提示词、工具或会话前缀。
diff --git a/packages/compact/compact/src/types.ts b/packages/compact/compact/src/types.ts
index e531430525..b77735b229 100644
--- a/packages/compact/compact/src/types.ts
+++ b/packages/compact/compact/src/types.ts
@@ -7,7 +7,7 @@
  * @module @deepseek-ai/dsh-compact/types
  */
 
-import type { ContentBlock } from '@deepseek-ai/dsh-llm'
+import type { ContentBlock, TokenUsage } from '@deepseek-ai/dsh-llm'
 
 declare module '@deepseek-ai/dsh-session' {
   interface SessionEventMap {
@@ -21,6 +21,8 @@ declare module '@deepseek-ai/dsh-session' {
      */
     'compact/summary': {
       summary: ContentBlock[]
+      /** Complete provider output before the backend's safe summary projection. */
+      rawOutput?: ContentBlock[]
       shadowedRange: { start: number; end: number }
       shadowedSeqs: number[]
       shadowedTokenCount: number
@@ -35,6 +37,8 @@ declare module '@deepseek-ai/dsh-session' {
       model: string
       /** The generation cap the summarize call sent, when one applied. */
       maxTokens?: number
+      /** Provider-reported token usage for the summarization request, when emitted. */
+      usage?: TokenUsage
     }
     /** Marks the end of a compaction — log-only, releases the lock. `error` set if summarization failed. */
     'compact/end': { turn: number; error?: string }
diff --git a/packages/context/README.i18n.yaml b/packages/context/README.i18n.yaml
index 7de339cd9a..7836a2a03c 100644
--- a/packages/context/README.i18n.yaml
+++ b/packages/context/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: a5244dfe99a714605744b57d33f97359d4d6fa4e
-README.zh.md: 2c1bdd6e5b290a771094719daa6e4d7c3bf577db
+#   pnpm run verify-translation-pairing --write packages/context/README.md
+README.md: fce6e21816d261171aaeaa217171580adb7c43f9
+README.zh.md: b8a4d68ca6892b51ed52479a7296513f5edcc292
diff --git a/packages/context/README.md b/packages/context/README.md
index a5244dfe99..fce6e21816 100644
--- a/packages/context/README.md
+++ b/packages/context/README.md
@@ -2,12 +2,13 @@
 
 English | [中文](README.zh.md)
 
-Product plugins that add model-visible request context without defining a tool. `workspace-context` is included by the default `dsh-agent-spine-demo` bundle and can be disabled through bundle config; `time-context` is opt-in, while the standard TUI bundle composes `session-reference` explicitly.
+Product plugins that add model-visible request context without defining a tool. `workspace-context` is included by the default `dsh-agent-spine-demo` bundle and can be disabled through bundle config; `time-context` and `tmux-context` are opt-in, while the standard TUI bundle composes `session-reference` explicitly.
 
 | Package | Role | ctx key |
 |---|---|---|
 | `session-reference/` | Bounded current-surface snapshots of other sessions | `ctx.sessionReferences` |
 | `time-context/` | Durable per-step current time and elapsed-time context | (none) |
+| `tmux-context/` | Durable per-turn context with this agent's tmux pane/window location | (listens on `agent/step`, reads `ctx.bash`) |
 | `workspace-context/` | `AGENTS.md`/`CLAUDE.md` workspace context loader | (listens on `agent/step` + `tools/post-execute`) |
 
 The [`workspace-context` decision record](../../.agents/notes/implemented/feature/2026-06-24-workspace-context.md) explains its per-agent/session isolation and lifecycle split.
diff --git a/packages/context/README.zh.md b/packages/context/README.zh.md
index 2c1bdd6e5b..b8a4d68ca6 100644
--- a/packages/context/README.zh.md
+++ b/packages/context/README.zh.md
@@ -2,12 +2,13 @@
 
 [English](README.md) | 中文
 
-这些产品插件无需定义工具，即可增加模型可见的请求上下文。`workspace-context` 包含在默认的 `dsh-agent-spine-demo` 组合包中，且可通过组合包配置将其禁用；`time-context` 需要选择启用，标准 TUI 组合包则会显式组合 `session-reference`。
+这些产品插件无需定义工具，即可增加模型可见的请求上下文。`workspace-context` 包含在默认的 `dsh-agent-spine-demo` 组合包中，且可通过组合包配置将其禁用；`time-context` 与 `tmux-context` 均需显式启用，标准 TUI 组合包则会显式组合 `session-reference`。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
 | `session-reference/` | 其他会话当前表层的有界快照 | `ctx.sessionReferences` |
-| `time-context/` | 持久的逐步骤当前时间与耗时上下文 | （无） |
+| `time-context/` | 持久化的逐步骤当前时间与已用时上下文 | （无） |
+| `tmux-context/` | 持久化的逐轮次上下文，记录本 agent 所在的 tmux pane/window 位置 | （监听 `agent/step`，读取 `ctx.bash`） |
 | `workspace-context/` | `AGENTS.md`／`CLAUDE.md` 工作区上下文 loader | （监听 `agent/step` + `tools/post-execute`） |
 
-[`workspace-context` 决策记录](../../.agents/notes/implemented/feature/2026-06-24-workspace-context.md)解释了它的逐 agent／会话隔离与生命周期拆分。
+[`workspace-context` 决策记录](../../.agents/notes/implemented/feature/2026-06-24-workspace-context.md)解释了每个 agent（智能体）和会话各自隔离的方式，以及相应的生命周期拆分。
diff --git a/packages/context/session-reference/README.i18n.yaml b/packages/context/session-reference/README.i18n.yaml
index 8ed188f2d7..fff2b196ad 100644
--- a/packages/context/session-reference/README.i18n.yaml
+++ b/packages/context/session-reference/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/context/session-reference/README.md
 README.md: 66df45b18df6d859239c8d3216d6c8b9fa61ab23
-README.zh.md: 7431f1375d50f4734f3efeea9a2597b7296e38c2
+README.zh.md: b59a98b86baf88f53429bc2409ad1776b1638736
diff --git a/packages/context/session-reference/README.zh.md b/packages/context/session-reference/README.zh.md
index 7431f1375d..b59a98b86b 100644
--- a/packages/context/session-reference/README.zh.md
+++ b/packages/context/session-reference/README.zh.md
@@ -6,13 +6,13 @@
 
 ## 公开 API
 
-- `listCandidates(agent, query?, limit?)` 会列出 `agent.id` 之外的会话，按 id、cwd 或日志中最新的标题进行不区分大小写的筛选，再按同 cwd、无 cwd、其他 cwd 记录排序，同时保持每组内的 `listSessions()` 创建顺序。每个已选候选会话都使用该标题作为 mention label；标题不存在或无法读取时回退到会话 id。不搜索消息主体。
-- `prepare(agent, content, references, signal?)` 会保留首次 mention 顺序、对 id 去重，并拒绝自引用或超过已配置不同源上限的情况。它会并行读取所有源，返回与输入脱离的内容，外加零个或一个聚合且带标识的 `UserMessage` 上下文。任何无效引用、读取失败、取消或预算失败都会在宿主调用 `followup()` 或 `steer()` 之前被拒绝。
-- `encodeSessionReferenceUri()` 与 `decodeSessionReferenceUri()` 实现 `dsh-session:<base64url(JSON.stringify(sessionId))>`，因此每个 JavaScript 字符串 id 都能精确往返。`formatSessionReferenceMention()` 发出 `@[label](uri)`，`parseSessionReferenceText()` 将 Markdown mention 或裸规范 URI 替换为可读的 `@label` 文本，并返回结构化引用。显式 Markdown mention 会拒绝每个格式错误的 URI；只当 scheme 后跟非空、符合 base64url 形状的 payload 时，裸文本才被视为引用，匹配但非规范的候选项仍会失败。空 scheme mention 或只含标点符号的 scheme mention 仍是普通讨论文本。
+- `listCandidates(agent, query?, limit?)` 会列出 `agent.id` 之外的会话，按 id、cwd 或以日志为依据的最新标题进行不区分大小写的筛选，再按同 cwd、无 cwd、其他 cwd 记录排序，同时保持每组内的 `listSessions()` 创建顺序。每个已选候选会话都使用该标题作为 mention label；标题不存在或无法读取时回退到会话 id。不搜索消息主体。
+- `prepare(agent, content, references, signal?)` 会保留首次 mention 顺序、对 id 去重，并拒绝自引用或超过已配置不同源上限的情况。它会并行读取所有源，返回与输入脱离的内容，外加零个或一个聚合且带标识的 `UserMessage` 上下文。任何无效引用、读取失败、取消或预算失败，都会使准备操作在宿主调用 `followup()` 或 `steer()` 之前失败。
+- `encodeSessionReferenceUri()` 与 `decodeSessionReferenceUri()` 实现 `dsh-session:<base64url(JSON.stringify(sessionId))>`，因此每个 JavaScript 字符串 id 都能精确往返。`formatSessionReferenceMention()` 发出 `@[label](uri)`，`parseSessionReferenceText()` 将 Markdown mention 或裸规范 URI 替换为可读的 `@label` 文本，并返回结构化引用。解析器会拒绝显式 Markdown mention 中任何格式错误的 URI；只当 scheme 后跟非空、符合 base64url 形状的 payload 时，裸文本才被视为引用，匹配但非规范的候选项仍会失败。空 scheme mention 或只含标点符号的 scheme mention 仍是普通讨论文本。
 
 ## 快照语义
 
-准备阶段会对每个不同源调用一次 `ctx.sessionQuery.readSurface()`，入队后绝不重读。它仅投影折叠后当前表层中的直接 user `user/message`、直接 user `steering/message`、assistant 文本，以及 `user/message` 检查点；这类检查点携带规范 `dsh-compact` 源标记。对于已经包含烘焙前缀上下文的源提示词，投影只读取其对模型隐藏的显示内容，以防止快照递归传播。已遮蔽的压缩前事件、工具、reasoning、上下文、除已标记 compact 检查点外的插件生成 user 消息，以及未完成的 assistant chunk 均会被排除。因此，已压缩源贡献的是最新检查点与之后保留的会话，而非已恢复的遮蔽文本。
+准备阶段会对每个不同源调用一次 `ctx.sessionQuery.readSurface()`，入队后绝不重读。它仅投影折叠后当前表层中的用户直接发出的 `user/message`、用户直接发出的 `steering/message`、assistant 文本，以及 `user/message` 检查点；这类检查点携带规范 `dsh-compact` 源标记。对于已经包含固化前缀上下文的源提示词，投影只读取其对模型隐藏的显示内容，以防止快照递归传播。已遮蔽的压缩（compaction）前事件、工具、推理（reasoning）、上下文、除已标记 compact 检查点外的插件生成 user 消息，以及未完成的 assistant 分片均会被排除。因此，已压缩源只会提供最新检查点及其后保留的会话内容，不会还原已遮蔽的文本。
 
 上下文源为 `{ kind: 'session-reference', version: 1, references }`；每条引用会记录其源 id 与 label、捕获 seq、是否存在 compact、已保留／已省略消息数、已省略 UTF-8 字节数与截断状态。标准 TUI 在不把上下文附加到通用 inbox 记录的情况下保留接纳归属：next-step 接收窗口之外，一次性 `agent/prompt-submit` 包装层只为获准决策添加快照；提示词接纳期间或轮次打开时，`inject()` 与 `steer()` 会并排暂存到同一安全边界。目标日志因此会先记录一条带来源信息的上下文 `user/message`，再记录可读的直接 `user/message` 或 `steering/message`。后续源变更、压缩或删除都无法改变目标回放。
 
@@ -24,7 +24,7 @@
 | `candidateLimit` | `50` | 返回给宿主的默认候选数量。 |
 | `maxReferenceBytes` | `65536` | 一个引用对象的最大序列化 JSON 字节数。 |
 
-保留会对每个源独立应用 `maxReferenceBytes`，保留 compact 检查点与最新消息，再丢弃较旧的非检查点单元，并使用 `dsh-retention` 头部／尾部截断和精确 UTF-8 省略通知。如果某个源的固定序列化字段无法容纳，准备会以 `SESSION_REFERENCE_BUDGET_EXCEEDED` 失败，而不返回部分上下文。
+保留会对每个源独立应用 `maxReferenceBytes`，保留 compact 检查点与最新消息，再丢弃较旧的非检查点单元，并使用 `dsh-retention` 头部／尾部截断和精确 UTF-8 省略通知。如果某个源的固定序列化字段本身就超出限额，准备会以 `SESSION_REFERENCE_BUDGET_EXCEEDED` 失败，而不返回部分上下文。
 
 ## 模型体验
 
@@ -32,11 +32,11 @@
 
 #### 模型看到的内容
 
-模型会看到两条连续的 user 角色消息：先是 `## Referenced sessions` 不受信任快照，再是带可读 `@label` 的当前消息。警告禁止遵循快照中的指令、权限声明或工具请求，除非当前 user 重复这些内容。Label、cwd 值、id 与会话文本作为 JSON 在 `<referenced-sessions>` 标签中序列化；每个数据 `<` 都发出为无损 JSON 转义 `\u003c`，因此源文本无法拼出框定标签。
+模型会看到两条连续的 user 角色消息：先是 `## Referenced sessions` 不受信任快照，再是带可读 `@label` 的当前消息。警告禁止遵循快照中的指令、权限声明或工具请求，除非当前 user 重复这些内容。标签、cwd 值、id 与会话文本会作为 JSON 在 `<referenced-sessions>` 标签中序列化；数据中的每个 `<` 都会以无损 JSON 转义 `\u003c` 的形式发出，因此源文本无法拼出定界标签。
 
 #### Token 影响
 
-每条引用消息都会添加固定警告和最多三个序列化快照，每个快照都受 `maxReferenceBytes` 独立限制。精确快照会保留在目标历史中，直到目标压缩遮蔽或摘要它；源会话变更不会添加更多 token。
+每条包含引用的消息都会添加固定警告和最多三个序列化快照，每个快照都受 `maxReferenceBytes` 独立限制。精确快照会保留在目标历史中，直到目标压缩遮蔽或摘要它；源会话变更不会添加更多 token。
 
 #### KV Cache 影响
 
@@ -44,7 +44,7 @@
 
 ## 已知限制与暂缓事项
 
-- **不支持正文发现**：候选查询会检查折叠后的标题，但不搜索消息主体。非空查询可能通过 session-query 服务有界、可取消的批处理检查每个可见的持久化会话日志；专用标题索引未来可以替换这条发现路径，而不改变 URI、快照或持久化契约。
+- **不支持消息正文检索**：候选查询会检查折叠后的标题，但不搜索消息主体。非空查询可能通过 session-query 服务有界、可取消的批处理检查每个可见的持久化会话日志；专用标题索引未来可以替换这条发现路径，而不改变 URI、快照或持久化契约。
 - **受信任调用方边界**：该服务假设宿主有权读取 `ctx.sessionQuery` 公开的每个会话；它不是面向模型的搜索工具。
 - **只投影文本**：不会在会话间传播非文本 user 与 assistant 块。
 - **没有实时链接**：引用是快照，不是 fork、恢复、订阅或源会话变更。
diff --git a/packages/context/time-context/README.i18n.yaml b/packages/context/time-context/README.i18n.yaml
index f5cd8dfd5d..d54c4ddde0 100644
--- a/packages/context/time-context/README.i18n.yaml
+++ b/packages/context/time-context/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/context/time-context/README.md
 README.md: 9fe818855439466b2a3e349cd54a2f408cf5ec10
-README.zh.md: 337ce3613d7b17134db7cf85a808881017f4e2c3
+README.zh.md: 1133715ebb3348d6e3dbbf8bbef169d6d8c56d8f
diff --git a/packages/context/time-context/README.zh.md b/packages/context/time-context/README.zh.md
index 337ce3613d..1133715ebb 100644
--- a/packages/context/time-context/README.zh.md
+++ b/packages/context/time-context/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-可选的持久上下文，包含模型请求准备期间采样的当前分区时间与已经过时间。`dsh-agent-spine-demo` 与已发布示例不装载它。决策记录：[持久 time-context Agent Note](../../../.agents/notes/implemented/feature/2026-07-16-durable-per-step-time-context.md)。
+可选的持久上下文，包含模型请求准备期间采样的带时区的当前时间与经过时长。`dsh-agent-spine-demo` 与随附示例不挂载该插件。决策记录：[持久 time-context Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-16-durable-per-step-time-context.md)。
 
 ## 配置
 
@@ -16,21 +16,21 @@
 
 省略 `timeZone` 时，插件会在加载时解析一次 Node 进程的系统时区。Node 遵循 `TZ`；如果没有该覆盖，时区由宿主或容器提供。显式 `timeZone` 必须是 IANA 标识符，并在插件加载时验证。
 
-`refreshIntervalMs` 必须是非负安全整数。省略或设为 `0` 时，对每次信号尚未 abort 的合格步骤前尝试执行追加。正数值只会在会话没有早先 time-context 注入、墙上时间向后移动，或自最新注入起已经过至少相应毫秒数时执行追加。
+`refreshIntervalMs` 必须是非负安全整数。省略或设为 `0` 时，对每次信号尚未中止的合格步骤前尝试执行追加。正数值只会在会话没有早先 time-context 注入、挂钟时间倒退，或自最新注入起已经过至少相应毫秒数时执行追加。
 
 ## 时序语义
 
-该插件会前置一个 `agent/step` listener。需要注入时，它会追加一条注入的 `user/message`，通过 `agent.inject()` 完成，时机位于 `step/start` 和普通自动压缩之前，其源为 `{ kind: 'plugin', plugin: 'time-context' }`。被抑制的尝试不追加任何内容。
+该插件会前置一个 `agent/step` 监听器。需要注入时，它会追加一条注入的 `user/message`，通过 `agent.inject()` 完成，时机位于 `step/start` 和普通自动压缩（compaction）之前，其源为 `{ kind: 'plugin', plugin: 'time-context' }`。被抑制的尝试不追加任何内容。
 
-正间隔调度会扫描原始持久会话事件，查找最新的上述源 `user/message`，包括已被压缩遮蔽的 reading。因此，调度可以跨轮次和已恢复进程应用，不需要进程本地 cache 状态。它会降低追加频率与历史增长，但绝不移除现有 reading，且每个会话独立调度。
+正间隔调度会扫描原始持久会话事件，查找最新的上述源 `user/message`，包括已被压缩遮蔽的时间读数。因此，调度可以跨轮次以及进程恢复持续生效，不需要进程本地缓存状态。它会降低追加频率与历史增长，但绝不移除现有时间读数，且每个会话独立调度。
 
-第 1 步从最新的前置模型可见消息起测量，包括开启轮次的提示词。后续步骤从同一轮次中前一个 time-context 事件起测量。两种基线都使用持久会话事件时间戳；墙上时钟向后移动时，已经过时间限制为零。如果第一步缺少基线，或者后续步骤因间隔抑制而没有较早的同轮次 reading，则报告 `unavailable`。
+第 1 步从前一条模型可见消息起测量，包括开启轮次的提示词。后续步骤从同一轮次中前一个 time-context 事件起测量。两种基线都使用持久会话事件时间戳；挂钟时间倒退时，经过时长限制为零。如果第一步缺少基线，或者后续步骤因间隔抑制而没有较早的同轮次时间读数，则报告 `unavailable`。
 
-时间 reading 记录的是一次请求准备尝试，不是已提交步骤或已传输请求。因为 listener 首先运行，后续步骤前 listener 取消或使尝试失败时，该追加可能仍会保留。日志仅追加，该插件不执行回滚。
+时间读数记录的是一次请求准备尝试，不是已提交步骤或已传输请求。因为 listener 首先运行，后续的步骤前监听器取消或使该尝试失败时，该追加可能仍会保留。日志仅追加，该插件不执行回滚。
 
-单独发布的 `./invariant` 配套模块会根据开启轮次、下一个步骤前位置、已经过时间基线与持久事件时间检查每个归因于插件的 reading。其渲染时间戳必须可解析，且不能晚于该事件；采样与追加之间的进程挂起不会使 reading 失效。
+单独发布的 `./invariant` 配套模块会根据当前未结束的轮次、下一个步骤前位置、经过时长基线与持久事件时间检查每个归因于插件的时间读数。其渲染时间戳必须可解析，且不能晚于该事件；采样与追加之间的进程挂起不会使时间读数失效。
 
-时间 reading 会保留在派生会话历史中，直到后续压缩遮蔽它。请求标头不含 time-context 状态。请求重建会在每个 `step/start` 处使用完整持久表层前缀，因此已传输请求无需与 reading 一一对应：失败的准备可能留下额外 reading，而间隔抑制可让请求复用现有历史，无需添加 reading。
+时间读数会保留在派生会话历史中，直到后续压缩遮蔽它。请求标头不含 time-context 状态。请求重建会在每个 `step/start` 处使用完整持久表层前缀，因此已传输请求无需与时间读数一一对应：失败的准备可能留下额外时间读数，而间隔抑制可让请求复用现有历史，无需添加时间读数。
 
 ## 模型体验
 
@@ -38,7 +38,7 @@
 
 #### 模型看到的内容
 
-每次执行注入的准备尝试都会生成一条带源标记的上下文消息，包含下方两行。`<timestamp>` 是带数字偏移与 IANA 时区、形如 ISO 的本地时间戳；持续时间使用紧凑的整秒单位。正间隔可能使某次步骤尝试没有新 reading。
+每次执行注入的准备尝试都会生成一条带源标记的上下文消息，包含下方两行。`<timestamp>` 是带数字偏移与 IANA 时区、形如 ISO 的本地时间戳；持续时间使用紧凑的整秒单位。正间隔可能使某次步骤尝试没有新时间读数。
 
 ##### 第一步
 
@@ -60,11 +60,11 @@ Elapsed since the preceding step context: <duration-or-unavailable>.
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
 - **整秒显示**：时间戳与持续时间省略亚秒精度，尽管持久事件时间保留毫秒。
-- **会话事件基线**：已经过时间从持久追加时间戳起计算，而非客户端传输的原始发送时间戳。
-- **进程本地默认时区**：省略设置时，使用插件加载时捕获的 Node 进程 `TZ`、宿主或容器时区，而非远程 user 的时区；两者不同时，请配置显式 IANA 时区。
-- **压缩之间的历史成本**：省略设置或设为 `0` 会为每次合格准备尝试保留一条 reading，包括后续取消或失败的尝试；正间隔可以降低但无法消除该成本。
+- **会话事件基线**：经过时长从持久追加时间戳起计算，而非客户端传输的原始发送时间戳。
+- **进程本地默认时区**：省略设置时，使用插件加载时捕获的 Node 进程 `TZ`、宿主或容器时区，而非远程用户的时区；两者不同时，请配置显式 IANA 时区。
+- **压缩之间的历史成本**：省略设置或设为 `0` 会为每次合格准备尝试保留一条时间读数，包括后续取消或失败的尝试；正间隔可以降低但无法消除该成本。
diff --git a/packages/context/tmux-context/README.i18n.yaml b/packages/context/tmux-context/README.i18n.yaml
new file mode 100644
index 0000000000..9a6f113cc8
--- /dev/null
+++ b/packages/context/tmux-context/README.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write packages/context/tmux-context/README.md
+README.md: a166a46d20f472cb5d8f045e2456ce3e6de7a2f2
+README.zh.md: 0575d549e352239e7d954870eaf40beea1169cc6
diff --git a/packages/context/tmux-context/README.md b/packages/context/tmux-context/README.md
new file mode 100644
index 0000000000..a166a46d20
--- /dev/null
+++ b/packages/context/tmux-context/README.md
@@ -0,0 +1,68 @@
+# @deepseek-ai/dsh-tmux-context
+
+English | [中文](README.zh.md)
+
+Opt-in durable context naming the tmux session, window, and pane this agent process runs in, plus the window's pane-tree layout. Sampled once per turn during model-request preparation. `dsh-agent-spine-demo` and shipped examples do not mount it. Decision record: [the tmux-context Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-tmux-location-context.md).
+
+## Config
+
+```yaml
+- id: tmux-context
+  name: '@deepseek-ai/dsh-tmux-context'
+  config:
+    refreshIntervalMs: 60000 # optional; omit or set to 0 to inject on every changed turn
+```
+
+`refreshIntervalMs` must be a non-negative safe integer. Omission or `0` injects whenever the tmux state changed since the last injection. A positive value additionally suppresses injections that fall within that many milliseconds of the latest one.
+
+## How it reads tmux
+
+The plugin prepends an `agent/step` listener that runs only on the first step of each turn. When due, it runs one read-only command through the `ctx.bash` executor seam:
+
+```sh
+[ -n "$TMUX_PANE" ] || exit 1
+self_tty=$(ps -o tty= -p <pid> | tr -d ' ')
+pane_tty=$(tmux display-message -t "$TMUX_PANE" -p '#{pane_tty}') || exit 1
+[ "$pane_tty" = "/dev/$self_tty" ] || exit 1
+exec tmux display-message -t "$TMUX_PANE" -p '<format>'
+```
+
+`$TMUX_PANE` alone is insufficient: a terminal launched from a tmux shell (a VS Code integrated terminal, a desktop launcher) **inherits** `$TMUX` and `$TMUX_PANE` from that ancestor, so the variables are present even though the process does not live in that pane. The command therefore also compares the pane's `#{pane_tty}` against this process's own controlling terminal (`ps -o tty=` for its pid): a genuine pane owns this process's tty, while an inherited environment names some other pane's tty. Running through `ctx.bash` applies the deployment's sandbox and policy; the plugin owns no subprocess code. When `ctx.bash` is absent, the process is not in a real tmux pane (`$TMUX_PANE` unset, or the tty does not match ⇒ nonzero exit), or the reading is malformed, the attempt is a no-op, never an error. The location is optional, so an executor rejection — a policy refusal from `resolve()` or an infrastructure failure from `run()` — is contained and logged as a warning rather than failing the turn.
+
+State is pulled on every eligible turn — a moved, renamed, or re-laid-out pane is picked up without any tmux hook or background process. The plugin re-injects only when the rendered tmux state differs from its last injection, so an unchanged location adds nothing.
+
+## Timing semantics
+
+When an injection is due, the plugin appends one injected `user/message` through `agent.inject()` before `step/start`, with source `{ kind: 'plugin', plugin: 'tmux-context' }`. Change suppression and interval scheduling scan the raw durable session events for the latest injection of this source, so the schedule survives compaction and resumed processes without process-local cache state; sessions schedule independently. The reading records a request-preparation attempt, not a committed step; because the listener runs first, its append may remain when a later pre-step listener cancels or fails the attempt (the log is append-only and the plugin performs no rollback).
+
+## Model Experience
+
+### Preparation-time tmux location
+
+#### What the model sees
+
+On each turn whose tmux state changed, one source-tagged context message with the three lines below. `<window-layout>` is tmux's compact pane-tree description; pane and window pixel sizes are intentionally excluded, and the contents of sibling panes are never captured.
+
+##### Changed-turn reading
+
+```markdown
+tmux location (turn <turn>):
+session <session>, window <index> "<name>", pane <index> <pane-id>
+window active=<0|1>, pane active=<0|1>, layout <window-layout>
+```
+
+#### Token effect
+
+Each two-line reading accumulates until compaction shadows it. Unchanged locations and interval suppression add nothing.
+
+#### KV Cache effect
+
+Append-only; newly visible content follows the reusable request prefix and does not invalidate existing KV-cache entries.
+
+## Known Limitations and Deferred Work
+
+- **First step only** — a pane moved or resized mid-turn is reflected on the next turn, not between steps.
+- **Own location only** — the plugin never captures the visible text of sibling panes.
+- **Layout, not size** — pane/window pixel dimensions are omitted; only the layout tree and active flags are reported.
+- **Tab-delimited fields** — a tmux window name containing the literal two-character sequence `\t` would mis-split the reading and be skipped as malformed; ordinary names are unaffected.
+- **tty-based pane detection** — the process is considered "in tmux" only when its controlling terminal matches `$TMUX_PANE`'s `#{pane_tty}`. This deliberately excludes terminals that inherited `$TMUX`/`$TMUX_PANE` from a tmux ancestor (e.g. a VS Code integrated terminal). `ps -o tty=` is POSIX; the check is a no-op wherever it or `#{pane_tty}` is unavailable.
diff --git a/packages/context/tmux-context/README.zh.md b/packages/context/tmux-context/README.zh.md
new file mode 100644
index 0000000000..0575d549e3
--- /dev/null
+++ b/packages/context/tmux-context/README.zh.md
@@ -0,0 +1,68 @@
+# @deepseek-ai/dsh-tmux-context
+
+[English](README.md) | 中文
+
+可选启用的持久上下文，记录本 agent 进程所在的 tmux session、window、pane，以及该 window 的 pane 树布局。在准备模型请求时每轮采样一次。`dsh-agent-spine-demo` 与随附示例均不挂载它。决策记录见：[tmux-context Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-tmux-location-context.md)。
+
+## 配置
+
+```yaml
+- id: tmux-context
+  name: '@deepseek-ai/dsh-tmux-context'
+  config:
+    refreshIntervalMs: 60000 # optional; omit or set to 0 to inject on every changed turn
+```
+
+`refreshIntervalMs` 必须是非负安全整数。省略或 `0` 表示只要 tmux 状态自上次注入以来发生变化就注入。正值会额外抑制距最近一次注入不足该毫秒数的注入。
+
+## 如何读取 tmux
+
+插件前置注册一个 `agent/step` 监听器，仅在每轮的第一个 step 运行。当需要注入时，它通过 `ctx.bash` 执行器 seam 运行一条只读命令：
+
+```sh
+[ -n "$TMUX_PANE" ] || exit 1
+self_tty=$(ps -o tty= -p <pid> | tr -d ' ')
+pane_tty=$(tmux display-message -t "$TMUX_PANE" -p '#{pane_tty}') || exit 1
+[ "$pane_tty" = "/dev/$self_tty" ] || exit 1
+exec tmux display-message -t "$TMUX_PANE" -p '<format>'
+```
+
+仅凭 `$TMUX_PANE` 并不足够：从 tmux shell 启动的终端（VS Code 集成终端、桌面启动器）会从该祖先进程**继承** `$TMUX` 与 `$TMUX_PANE`，因此即使进程并不位于那个 pane 中，这些变量依然存在。为此该命令还会把 pane 的 `#{pane_tty}` 与本进程自己的控制终端（对其 pid 执行 `ps -o tty=`）作比较：真正的 pane 拥有本进程的 tty，而继承而来的环境指向的是另一个 pane 的 tty。通过 `ctx.bash` 运行会应用部署方的沙箱与策略；插件不拥有任何子进程代码。当 `ctx.bash` 缺失、进程不在真实的 tmux pane 内（`$TMUX_PANE` 未设置，或 tty 不匹配 ⇒ 非零退出）或读取结果格式非法时，本次尝试为空操作，绝不报错。由于位置信息是可选的，执行器的拒绝——`resolve()` 的策略拒绝或 `run()` 的基础设施故障——会被兜住并记录为警告，而不会使该轮失败。
+
+状态在每个符合条件的轮次拉取——pane 被移动、改名或重新布局都会被感知，无需任何 tmux hook 或后台进程。插件仅在渲染出的 tmux 状态与上次注入不同时才重新注入，因此位置不变时不会新增任何内容。
+
+## 时序语义
+
+当需要注入时，插件在 `step/start` 之前通过 `agent.inject()` 追加一条注入的 `user/message`，来源为 `{ kind: 'plugin', plugin: 'tmux-context' }`。变化抑制与间隔调度会扫描原始持久会话事件中该来源的最近一次注入，因此调度可跨压缩与恢复的进程存续，无需进程内缓存状态；各会话独立调度。该读数记录的是一次请求准备尝试，而非已提交的 step；由于监听器最先运行，当后续 pre-step 监听器取消或失败时，它的追加可能仍会保留（日志只追加，插件不做回滚）。
+
+## 模型体验
+
+### 准备期 tmux 位置
+
+#### 模型看到的内容
+
+在 tmux 状态发生变化的每一轮，注入一条带来源标记、含以下三行的上下文消息。`<window-layout>` 是 tmux 紧凑的 pane 树描述；pane 与 window 的像素尺寸有意省略，相邻 pane 的内容从不采集。
+
+##### 变化轮次读数
+
+```markdown
+tmux location (turn <turn>):
+session <session>, window <index> "<name>", pane <index> <pane-id>
+window active=<0|1>, pane active=<0|1>, layout <window-layout>
+```
+
+#### Token 影响
+
+每条两行读数会累积，直到压缩将其遮蔽。位置未变化以及间隔抑制不会新增内容。
+
+#### KV 缓存影响
+
+只追加；新增可见内容位于可复用的请求前缀之后，不会使已有 KV 缓存条目失效。
+
+## 已知限制与后续工作
+
+- **仅第一个 step**——轮次中途移动或缩放的 pane 会在下一轮反映，而非在 step 之间。
+- **仅自身位置**——插件从不采集相邻 pane 的可见文本。
+- **只有布局，没有尺寸**——省略 pane/window 像素尺寸；仅报告布局树与活动标志。
+- **制表符分隔字段**——若 tmux window 名称包含字面两字符序列 `\t`，会使读数分割错误并作为非法读数跳过；常规名称不受影响。
+- **基于 tty 的 pane 判定**——只有当进程的控制终端与 `$TMUX_PANE` 的 `#{pane_tty}` 一致时，才视为“位于 tmux 中”。这会有意排除从 tmux 祖先进程继承 `$TMUX`／`$TMUX_PANE` 的终端（如 VS Code 集成终端）。`ps -o tty=` 属于 POSIX；在其或 `#{pane_tty}` 不可用的环境中，该检查即为空操作。
diff --git a/packages/context/tmux-context/package.json b/packages/context/tmux-context/package.json
new file mode 100644
index 0000000000..222cdfa55b
--- /dev/null
+++ b/packages/context/tmux-context/package.json
@@ -0,0 +1,48 @@
+{
+  "name": "@deepseek-ai/dsh-tmux-context",
+  "description": "Opt-in durable per-step context with this agent's tmux pane and window location",
+  "version": "0.0.1",
+  "private": true,
+  "type": "module",
+  "main": "lib/index.js",
+  "types": "lib/types/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./lib/types/index.d.ts",
+      "default": "./lib/index.js"
+    },
+    "./invariant": {
+      "types": "./lib/types/invariant.d.ts",
+      "default": "./lib/invariant.js"
+    },
+    "./src/*": "./src/*",
+    "./package.json": "./package.json"
+  },
+  "files": [
+    "lib/index.js",
+    "lib/invariant.js",
+    "lib/types/**/*.d.ts",
+    "lib/types/**/*.d.ts.map",
+    "src"
+  ],
+  "license": "BSD-3-Clause",
+  "dependencies": {
+    "schemastery": "^3.18.0"
+  },
+  "peerDependencies": {
+    "@deepseek-ai/dsh-agent": "^0.0.1",
+    "@deepseek-ai/dsh-bash": "^0.0.1",
+    "@deepseek-ai/dsh-invariants": "^0.0.1",
+    "@deepseek-ai/dsh-session": "^0.0.1",
+    "cordis": "^4.0.0-rc.7"
+  },
+  "devDependencies": {
+    "@deepseek-ai/dsh-agent": "workspace:^",
+    "@deepseek-ai/dsh-bash": "workspace:^",
+    "@deepseek-ai/dsh-invariants": "workspace:^",
+    "@deepseek-ai/dsh-llm": "workspace:^",
+    "@deepseek-ai/dsh-session": "workspace:^",
+    "@deepseek-ai/dsh-system-prompt": "workspace:^",
+    "cordis": "^4.0.0-rc.7"
+  }
+}
diff --git a/packages/context/tmux-context/src/index.ts b/packages/context/tmux-context/src/index.ts
new file mode 100644
index 0000000000..35f4c4a22f
--- /dev/null
+++ b/packages/context/tmux-context/src/index.ts
@@ -0,0 +1,241 @@
+/**
+ * Opt-in request-preparation tmux-location context. Eligible step attempts
+ * append durable, source-attributed context naming the tmux session, window,
+ * and pane this agent process runs in, plus the window's pane-tree layout.
+ *
+ * The plugin pulls state once per turn, on the first step (`step === 1`), by
+ * running one `tmux display-message` through the `ctx.bash` executor seam. It
+ * confirms this process genuinely runs inside the pane `$TMUX_PANE` names by
+ * matching the pane's `#{pane_tty}` against this process's controlling terminal,
+ * so a terminal that merely inherited `$TMUX`/`$TMUX_PANE` from a tmux ancestor
+ * (e.g. a VS Code integrated terminal) reads as "not in tmux". It re-injects
+ * only when the rendered tmux state changes since the last injection (a moved,
+ * renamed, or re-laid-out pane), with an optional `refreshIntervalMs` floor
+ * between injections. Absent tmux environment, an inherited-only environment,
+ * absent `ctx.bash`, or a failed query is a no-op, never an error: an executor
+ * rejection is contained and logged as a warning so the turn continues.
+ *
+ * @module @deepseek-ai/dsh-tmux-context
+ */
+
+import type { Context, LoggerService } from 'cordis'
+import z from 'schemastery'
+import type { Agent } from '@deepseek-ai/dsh-agent'
+import type { BashExecutor, BashRunResult } from '@deepseek-ai/dsh-bash'
+import { createUserMessage } from '@deepseek-ai/dsh-llm'
+
+/** Cordis plugin name used by loader diagnostics. */
+export const name = 'tmux-context'
+
+/** The agent registry that owns the `agent/step` lifecycle seam. */
+export const inject = ['agents']
+
+/** Per-turn tmux-location scheduling. Invalid values fail plugin load. */
+export interface Config {
+  /** Minimum milliseconds between durable injections in one session. Omit or set to 0 to inject on every eligible change. */
+  refreshIntervalMs?: number
+}
+
+/** Schemastery validation for {@link Config}. */
+export const Config: z<Config> = z.object({
+  refreshIntervalMs: z.number(),
+})
+
+/**
+ * Tab-separated tmux format fields, in query order. Layout (`window_layout`)
+ * is the pane-tree description; pane/window pixel sizes are intentionally
+ * excluded (own location and layout only, per the package scope).
+ */
+const TMUX_FIELDS = [
+  '#{session_name}',
+  '#{window_index}',
+  '#{window_name}',
+  '#{pane_index}',
+  '#{pane_id}',
+  '#{window_active}',
+  '#{pane_active}',
+  '#{window_layout}',
+] as const
+
+/** Structured tmux location parsed from one `display-message` reading. */
+interface TmuxLocation {
+  sessionName: string
+  windowIndex: string
+  windowName: string
+  paneIndex: string
+  paneId: string
+  windowActive: string
+  paneActive: string
+  windowLayout: string
+}
+
+/** Prefix marking the volatile turn/step preamble line of a rendered reading. */
+const READING_PREFIX = 'tmux location (turn '
+
+/**
+ * Field separator between tmux format fields. tmux does not interpret C escapes
+ * in a format, so the literal two-character sequence `\t` is emitted verbatim
+ * and split back out here; this avoids embedding raw whitespace in the command.
+ */
+const FIELD_SEP = '\\t'
+
+/**
+ * Read this process's tmux location through the bash seam, or `undefined` when
+ * this process is not genuinely running inside a tmux pane or the query fails.
+ *
+ * `$TMUX_PANE` alone is insufficient: a terminal launched from a tmux shell
+ * (e.g. VS Code's integrated terminal, a desktop launcher) inherits `$TMUX` and
+ * `$TMUX_PANE` from that ancestor, so the variables are present even though this
+ * process does not live in that pane. The command therefore also compares the
+ * pane's `#{pane_tty}` against this process's own controlling terminal
+ * (`ps -o tty=` for {@link processId}); a genuine pane owns this process's tty,
+ * an inherited environment names some other pane's tty. Fields are emitted only
+ * on a match, so an inherited environment reads as "not in tmux" and injects
+ * nothing.
+ *
+ * The location is optional context, so an executor rejection is a failed query,
+ * not a turn failure: `resolve()` may reject the command on policy grounds and
+ * `run()` only promises to resolve for nonzero exits, timeouts, and aborts, so
+ * both are contained and reported as a warning.
+ *
+ * @param bash - the executor seam used to run the read-only tmux/ps commands.
+ * @param logger - receives a warning when the executor rejects the query.
+ * @param processId - this agent process's pid, whose controlling tty must match the pane.
+ * @param signal - abort signal forwarded to the executor.
+ * @returns the parsed location, or `undefined` when not in a real pane or on any failure.
+ */
+async function queryTmuxLocation(
+  bash: BashExecutor,
+  logger: LoggerService,
+  processId: number,
+  signal: AbortSignal,
+): Promise<TmuxLocation | undefined> {
+  const format = TMUX_FIELDS.join(FIELD_SEP)
+  const command = [
+    '[ -n "$TMUX_PANE" ] || exit 1',
+    `self_tty=$(ps -o tty= -p ${processId} | tr -d ' ')`,
+    '[ -n "$self_tty" ] || exit 1',
+    'pane_tty=$(tmux display-message -t "$TMUX_PANE" -p \'#{pane_tty}\') || exit 1',
+    '[ "$pane_tty" = "/dev/$self_tty" ] || exit 1',
+    `exec tmux display-message -t "$TMUX_PANE" -p '${format}'`,
+  ].join('\n')
+  let result: BashRunResult
+  try {
+    result = await bash.run(bash.resolve({ command, signal }))
+  } catch (error: unknown) {
+    const message = error instanceof Error ? error.message : String(error)
+    logger.warn(`tmux location query failed: ${message}; injecting no location this turn`)
+    return undefined
+  }
+  if (result.exitCode !== 0) return undefined
+  const line = result.stdout.text.split('\n', 1)[0] as string
+  const parts = line.split(FIELD_SEP)
+  if (parts.length !== TMUX_FIELDS.length) return undefined
+  const [
+    sessionName,
+    windowIndex,
+    windowName,
+    paneIndex,
+    paneId,
+    windowActive,
+    paneActive,
+    windowLayout,
+  ] = parts as [string, string, string, string, string, string, string, string]
+  if (paneId.length === 0) return undefined
+  return {
+    sessionName,
+    windowIndex,
+    windowName,
+    paneIndex,
+    paneId,
+    windowActive,
+    paneActive,
+    windowLayout,
+  }
+}
+
+/**
+ * Render the stable tmux state block: the part of a reading compared for
+ * change suppression. It excludes the turn preamble so re-injection is driven
+ * only by tmux state, not by loop position.
+ */
+function renderState(location: TmuxLocation): string {
+  return `session ${location.sessionName}, `
+    + `window ${location.windowIndex} ${JSON.stringify(location.windowName)}, `
+    + `pane ${location.paneIndex} ${location.paneId}\n`
+    + `window active=${location.windowActive}, pane active=${location.paneActive}, `
+    + `layout ${location.windowLayout}`
+}
+
+/** Render the full durable reading, including the volatile turn preamble. */
+function renderReading(location: TmuxLocation, turn: number): string {
+  return `${READING_PREFIX}${turn}):\n${renderState(location)}`
+}
+
+/**
+ * The stable state block of this plugin's latest durable injection, or
+ * `undefined` when the session has none. Scans raw durable events so the
+ * schedule survives compaction and resumed processes without process-local
+ * cache state.
+ */
+function latestInjectedState(agent: Agent): { state: string; time: number } | undefined {
+  for (const event of [...agent.session.events].reverse()) {
+    if (event.type === 'user/message'
+      && event.data.source.kind === 'plugin'
+      && event.data.source.plugin === name) {
+      const [block] = event.data.content
+      if (block?.type !== 'text') return undefined
+      const newline = block.text.indexOf('\n')
+      const state = newline === -1 ? '' : block.text.slice(newline + 1)
+      return { state, time: event.time }
+    }
+  }
+  return undefined
+}
+
+/** Reject refresh intervals that cannot represent an exact elapsed-millisecond threshold. */
+function validateRefreshInterval(refreshIntervalMs: number | undefined): void {
+  if (refreshIntervalMs !== undefined && (
+    !Number.isSafeInteger(refreshIntervalMs)
+    || refreshIntervalMs < 0
+  )) {
+    throw new TypeError(
+      `tmux-context: refreshIntervalMs must be a non-negative safe integer, got ${String(refreshIntervalMs)}`,
+    )
+  }
+}
+
+/**
+ * Register a prepended `agent/step` listener for the lifetime of `ctx`.
+ * @param ctx - plugin context; the listener is disposed with it.
+ * @param config - durable refresh scheduling configuration.
+ * @throws when the refresh interval is invalid.
+ */
+export function apply(ctx: Context, config: Config): void {
+  const refreshIntervalMs = config.refreshIntervalMs
+  validateRefreshInterval(refreshIntervalMs)
+
+  ctx.on('agent/step', async (
+    agent: Agent,
+    turn: number,
+    step: number,
+    signal: AbortSignal,
+  ): Promise<void> => {
+    if (signal.aborted || step !== 1) return
+    const bash = ctx.get('bash')
+    if (bash === undefined) return
+    const previous = latestInjectedState(agent)
+    if (refreshIntervalMs !== undefined && refreshIntervalMs > 0 && previous !== undefined) {
+      const now = Date.now()
+      if (now >= previous.time && now - previous.time < refreshIntervalMs) return
+    }
+    const location = await queryTmuxLocation(bash, ctx.logger, process.pid, signal)
+    if (location === undefined) return
+    const state = renderState(location)
+    if (previous !== undefined && previous.state === state) return
+    agent.inject(createUserMessage({
+      content: [{ type: 'text', text: renderReading(location, turn) }],
+      source: { kind: 'plugin', plugin: name },
+    }))
+  }, { prepend: true })
+}
diff --git a/packages/context/tmux-context/src/invariant.ts b/packages/context/tmux-context/src/invariant.ts
new file mode 100644
index 0000000000..181f1a2289
--- /dev/null
+++ b/packages/context/tmux-context/src/invariant.ts
@@ -0,0 +1,30 @@
+/**
+ * Package-owned invariant companion for `@deepseek-ai/dsh-tmux-context`.
+ * @module @deepseek-ai/dsh-tmux-context/invariant
+ */
+
+/* jscpd:ignore-start */
+import type { Context } from 'cordis'
+import type { InvariantInstaller } from '@deepseek-ai/dsh-invariants'
+
+const PACKAGE_NAME = '@deepseek-ai/dsh-tmux-context'
+
+/** Cordis companion plugin name. */
+export const name = 'tmux-context-invariant'
+/** Service required before the companion can reserve package ownership. */
+export const inject = ['invariants']
+
+/**
+ * No runtime invariant: a reading is a per-turn snapshot of external tmux state, so the session
+ * holds no cross-event relation to check; scheduling and format are owned by pipeline tests.
+ */
+const install: InvariantInstaller = () => {}
+
+/**
+ * Register this package's invariant companion.
+ * @param ctx - Cordis context carrying the invariant service.
+ * @returns the installed registration's disposer after setup succeeds.
+ */
+export const apply = (ctx: Context): Promise<() => void> =>
+  Promise.resolve(ctx.invariants.register(PACKAGE_NAME, install))
+/* jscpd:ignore-end */
diff --git a/packages/context/tmux-context/tests/tmux-context.spec.ts b/packages/context/tmux-context/tests/tmux-context.spec.ts
new file mode 100644
index 0000000000..fa162f8d3f
--- /dev/null
+++ b/packages/context/tmux-context/tests/tmux-context.spec.ts
@@ -0,0 +1,407 @@
+import { afterEach, describe, expect, it, vi } from 'vitest'
+import { Context } from 'cordis'
+import { Session, SessionId } from '@deepseek-ai/dsh-session'
+import AgentRegistry, { agentEvents, type Agent } from '@deepseek-ai/dsh-agent'
+import { createUserMessage } from '@deepseek-ai/dsh-llm'
+import { BashExecutor } from '@deepseek-ai/dsh-bash'
+import type { BashExecRequest, BashExecSpec, BashProcess, BashRunResult } from '@deepseek-ai/dsh-bash'
+import * as tmuxContext from '@deepseek-ai/dsh-tmux-context'
+import type { Config } from '@deepseek-ai/dsh-tmux-context'
+
+const SIGNAL = new AbortController().signal
+
+/** One `#{...}`-joined tmux reading line for the eight queried fields. */
+function tmuxLine(fields: {
+  sessionName?: string
+  windowIndex?: string
+  windowName?: string
+  paneIndex?: string
+  paneId?: string
+  windowActive?: string
+  paneActive?: string
+  windowLayout?: string
+} = {}): string {
+  return [
+    fields.sessionName ?? '0',
+    fields.windowIndex ?? '1',
+    fields.windowName ?? 'node',
+    fields.paneIndex ?? '2',
+    fields.paneId ?? '%90',
+    fields.windowActive ?? '1',
+    fields.paneActive ?? '0',
+    fields.windowLayout ?? 'd517,270x71,0,0{135x71,0,0,87,134x71,136,0[134x35,136,0,90,134x35,136,36,93]}',
+  ].join('\\t')
+}
+
+function runResult(stdout: string, overrides: Partial<BashRunResult> = {}): BashRunResult {
+  return {
+    exitCode: 0,
+    signal: null,
+    timedOut: false,
+    aborted: false,
+    timeoutMs: 60_000,
+    stdout: { text: stdout, truncated: false },
+    stderr: { text: '', truncated: false },
+    ...overrides,
+  }
+}
+
+/** A scriptable fake `ctx.bash` recording the command it was asked to run. */
+class FakeBash extends BashExecutor {
+  commands: string[] = []
+  result: BashRunResult = runResult(`${tmuxLine()}\n`)
+  runError?: Error
+  resolveError?: Error
+
+  override resolve(request: BashExecRequest): BashExecSpec {
+    if (this.resolveError) throw this.resolveError
+    return {
+      command: request.command,
+      workdir: request.workdir ?? '/work',
+      timeoutMs: request.timeoutMs ?? 60_000,
+      stdoutMaxBytes: request.stdoutMaxBytes ?? 64_000,
+      signal: request.signal,
+      sandboxPolicy: request.sandboxPolicy,
+    }
+  }
+  override async run(spec: BashExecSpec): Promise<BashRunResult> {
+    this.commands.push(spec.command)
+    if (this.runError) throw this.runError
+    return this.result
+  }
+  override start(): BashProcess {
+    throw new Error('tmux-context must never start a background task')
+  }
+}
+
+async function mount(config: Config, withBash: true): Promise<{ ctx: Context; bash: FakeBash }>
+async function mount(config?: Config, withBash?: boolean): Promise<{ ctx: Context; bash: FakeBash | undefined }>
+async function mount(
+  config: Config = {},
+  withBash = false,
+): Promise<{ ctx: Context; bash: FakeBash | undefined }> {
+  const ctx = new Context()
+  await ctx.plugin(AgentRegistry)
+  let bash: FakeBash | undefined
+  if (withBash) {
+    await ctx.plugin(FakeBash)
+    bash = ctx.bash as FakeBash
+  }
+  await ctx.plugin(tmuxContext, config)
+  return { ctx, bash }
+}
+
+function sessionAgent(session: Session, id = 'agent'): Agent {
+  return {
+    id: SessionId(id),
+    options: {},
+    session,
+    status: 'running',
+    acceptsNextStep: true,
+    ctx: new Context(),
+    followup: () => {},
+    steer: () => {},
+    inject(input) {
+      session.append('user/message', input, { surfaceOp: 'append' })
+    },
+    send: () => {},
+    cancel() {},
+    whenIdle: () => Promise.resolve(),
+  }
+}
+
+function openMessageTurn(session: Session, turn: number): void {
+  session.append('turn/start', { turn, trigger: { kind: 'message', source: { kind: 'user' } } })
+  session.append('user/message', createUserMessage({
+    content: [{ type: 'text', text: `turn ${turn}` }],
+    source: { kind: 'user' },
+  }), { surfaceOp: 'append' })
+}
+
+function contextTexts(session: Session): string[] {
+  const texts: string[] = []
+  for (const event of session.events) {
+    if (event.type === 'user/message'
+      && event.data.source.kind === 'plugin'
+      && event.data.source.plugin === 'tmux-context') {
+      texts.push(event.data.content.find(block => block.type === 'text')?.text ?? '')
+    }
+  }
+  return texts
+}
+
+async function fire(
+  ctx: Context,
+  agent: Agent,
+  turn: number,
+  step: number,
+  signal: AbortSignal = SIGNAL,
+): Promise<void> {
+  await agentEvents(ctx, agent).serial('agent/step', turn, step, signal)
+}
+
+afterEach(() => {
+  vi.restoreAllMocks()
+  vi.useRealTimers()
+})
+
+describe('tmux-context injection', () => {
+  it('injects the tmux location on the first step of a turn', async () => {
+    const { ctx } = await mount({}, true)
+    const session = new Session(SessionId('first'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toEqual([
+      'tmux location (turn 1):\n'
+      + 'session 0, window 1 "node", pane 2 %90\n'
+      + 'window active=1, pane active=0, '
+      + 'layout d517,270x71,0,0{135x71,0,0,87,134x71,136,0[134x35,136,0,90,134x35,136,36,93]}',
+    ])
+    const event = session.events.at(-1)
+    if (event?.type !== 'user/message') throw new Error('missing tmux context')
+    expect(event.data.source).toEqual({ kind: 'plugin', plugin: 'tmux-context' })
+    expect(event.surfaceOp).toBe('append')
+  })
+
+  it('queries the pane this process runs in and matches its controlling tty', async () => {
+    const { ctx, bash } = await mount({}, true)
+    const session = new Session(SessionId('command'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(bash.commands).toHaveLength(1)
+    const command = bash.commands[0]!
+    expect(command).toContain('[ -n "$TMUX_PANE" ]')
+    expect(command).toContain('tmux display-message -t "$TMUX_PANE" -p')
+    // Guards against an inherited $TMUX_PANE: the pane's tty must equal this
+    // process's controlling tty (resolved for this exact pid).
+    expect(command).toContain(`ps -o tty= -p ${process.pid}`)
+    expect(command).toContain('#{pane_tty}')
+    expect(command).toContain('[ "$pane_tty" = "/dev/$self_tty" ]')
+  })
+
+  it('does not run on later steps of a turn', async () => {
+    const { ctx, bash } = await mount({}, true)
+    const session = new Session(SessionId('later-step'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 2)
+
+    expect(bash.commands).toHaveLength(0)
+    expect(contextTexts(session)).toHaveLength(0)
+  })
+
+  it('re-injects a new turn only when tmux state changed', async () => {
+    const { ctx, bash } = await mount({}, true)
+    const session = new Session(SessionId('change'))
+    const agent = sessionAgent(session)
+
+    openMessageTurn(session, 1)
+    await fire(ctx, agent, 1, 1)
+    session.append('turn/end', { turn: 1, reason: { kind: 'completed' } })
+
+    // Same state on turn 2: suppressed.
+    openMessageTurn(session, 2)
+    await fire(ctx, agent, 2, 1)
+    session.append('turn/end', { turn: 2, reason: { kind: 'completed' } })
+    expect(contextTexts(session)).toHaveLength(1)
+
+    // Moved pane on turn 3: re-injected.
+    bash.result = runResult(`${tmuxLine({ windowName: 'shell', paneId: '%12' })}\n`)
+    openMessageTurn(session, 3)
+    await fire(ctx, agent, 3, 1)
+
+    const texts = contextTexts(session)
+    expect(texts).toHaveLength(2)
+    expect(texts[1]).toContain('tmux location (turn 3):')
+    expect(texts[1]).toContain('window 1 "shell", pane 2 %12')
+  })
+
+  it('honors a positive refresh interval between injections', async () => {
+    vi.useFakeTimers()
+    vi.setSystemTime(1_000)
+    const { ctx, bash } = await mount({ refreshIntervalMs: 10_000 }, true)
+    const session = new Session(SessionId('interval'))
+    const agent = sessionAgent(session)
+
+    openMessageTurn(session, 1)
+    await fire(ctx, agent, 1, 1)
+    session.append('turn/end', { turn: 1, reason: { kind: 'completed' } })
+
+    // Changed state but inside the interval: suppressed, and never queried.
+    bash.result = runResult(`${tmuxLine({ paneId: '%99' })}\n`)
+    vi.setSystemTime(5_000)
+    openMessageTurn(session, 2)
+    await fire(ctx, agent, 2, 1)
+    expect(contextTexts(session)).toHaveLength(1)
+    expect(bash.commands).toHaveLength(1)
+
+    // Past the interval: queried and re-injected.
+    vi.setSystemTime(12_000)
+    openMessageTurn(session, 3)
+    await fire(ctx, agent, 3, 1)
+    expect(contextTexts(session)).toHaveLength(2)
+    expect(bash.commands).toHaveLength(2)
+  })
+})
+
+describe('tmux-context prior-reading resilience', () => {
+  it('treats a prior non-text plugin reading as absent and injects afresh', async () => {
+    const { ctx, bash } = await mount({}, true)
+    const session = new Session(SessionId('prior-non-text'))
+    const agent = sessionAgent(session)
+    openMessageTurn(session, 1)
+    session.append('user/message', createUserMessage({
+      content: [{ type: 'reasoning', text: 'not a location' }],
+      source: { kind: 'plugin', plugin: 'tmux-context' },
+    }), { surfaceOp: 'append' })
+
+    await fire(ctx, agent, 1, 1)
+
+    expect(bash.commands).toHaveLength(1)
+    expect(contextTexts(session).at(-1)).toContain('tmux location (turn 1):')
+  })
+
+  it('treats a prior single-line plugin reading (no newline) as empty state', async () => {
+    const { ctx, bash } = await mount({}, true)
+    const session = new Session(SessionId('prior-single-line'))
+    const agent = sessionAgent(session)
+    openMessageTurn(session, 1)
+    session.append('user/message', createUserMessage({
+      content: [{ type: 'text', text: 'single line, no newline' }],
+      source: { kind: 'plugin', plugin: 'tmux-context' },
+    }), { surfaceOp: 'append' })
+
+    await fire(ctx, agent, 1, 1)
+
+    // Empty prior state never equals the multi-line reading, so it re-injects.
+    expect(bash.commands).toHaveLength(1)
+    expect(contextTexts(session).at(-1)).toContain('tmux location (turn 1):')
+  })
+})
+
+describe('tmux-context no-op paths', () => {
+  it('is a no-op when no bash executor is mounted', async () => {
+    const { ctx } = await mount()
+    const session = new Session(SessionId('no-bash'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+  })
+
+  it('is a no-op when the tmux query exits nonzero (outside tmux, or an inherited env whose tty does not match the pane)', async () => {
+    const { ctx, bash } = await mount({}, true)
+    bash.result = runResult('', { exitCode: 1 })
+    const session = new Session(SessionId('outside-tmux'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+  })
+
+  it('is a no-op when the reading has the wrong field count', async () => {
+    const { ctx, bash } = await mount({}, true)
+    bash.result = runResult('0\\t1\\tnode\n')
+    const session = new Session(SessionId('malformed'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+  })
+
+  it('is a no-op when the pane id is empty', async () => {
+    const { ctx, bash } = await mount({}, true)
+    bash.result = runResult(`${tmuxLine({ paneId: '' })}\n`)
+    const session = new Session(SessionId('empty-pane'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+  })
+
+  it('warns and injects nothing when the executor rejects the run', async () => {
+    const { ctx, bash } = await mount({}, true)
+    bash.runError = new Error('bash executor unavailable')
+    const warn = vi.spyOn(ctx.logger, 'warn')
+    const session = new Session(SessionId('run-rejected'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+    expect(warn).toHaveBeenCalledWith(expect.stringContaining('bash executor unavailable'))
+  })
+
+  it('warns and injects nothing when the executor rejects the command at resolve', async () => {
+    const { ctx, bash } = await mount({}, true)
+    bash.resolveError = new Error('command denied by policy')
+    const warn = vi.spyOn(ctx.logger, 'warn')
+    const session = new Session(SessionId('resolve-rejected'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+    expect(warn).toHaveBeenCalledWith(expect.stringContaining('command denied by policy'))
+  })
+
+  it('reports a non-Error rejection in the warning', async () => {
+    const { ctx, bash } = await mount({}, true)
+    // Non-Error throw: the executor seam is typed, but a bad impl can reject with anything.
+    bash.runError = 'spawn refused' as unknown as Error
+    const warn = vi.spyOn(ctx.logger, 'warn')
+    const session = new Session(SessionId('non-error-rejection'))
+    openMessageTurn(session, 1)
+
+    await fire(ctx, sessionAgent(session), 1, 1)
+
+    expect(contextTexts(session)).toHaveLength(0)
+    expect(warn).toHaveBeenCalledWith(expect.stringContaining('spawn refused'))
+  })
+
+  it('skips an already-aborted step and runs before ordinary agent/step listeners', async () => {
+    const { ctx } = await mount({}, true)
+    const session = new Session(SessionId('ordering'))
+    const agent = sessionAgent(session)
+    openMessageTurn(session, 1)
+    let ordinarySawContext = false
+    ctx.on('agent/step', (subject) => {
+      ordinarySawContext = subject.session.events.some(
+        event => event.type === 'user/message'
+          && event.data.source.kind === 'plugin'
+          && event.data.source.plugin === 'tmux-context',
+      )
+    })
+
+    const abort = new AbortController()
+    abort.abort()
+    await fire(ctx, agent, 1, 1, abort.signal)
+    expect(contextTexts(session)).toHaveLength(0)
+
+    await fire(ctx, agent, 1, 1)
+    expect(ordinarySawContext).toBe(true)
+    expect(contextTexts(session)).toHaveLength(1)
+  })
+})
+
+describe('tmux-context configuration', () => {
+  it('rejects a negative refresh interval at plugin load', async () => {
+    await expect(mount({ refreshIntervalMs: -1 })).rejects.toThrow(
+      /refreshIntervalMs must be a non-negative safe integer/,
+    )
+  })
+
+  it('rejects a non-integer refresh interval at plugin load', async () => {
+    await expect(mount({ refreshIntervalMs: 1.5 })).rejects.toThrow(
+      /refreshIntervalMs must be a non-negative safe integer/,
+    )
+  })
+})
diff --git a/packages/context/tmux-context/tsconfig.json b/packages/context/tmux-context/tsconfig.json
new file mode 100644
index 0000000000..a5983a2c7f
--- /dev/null
+++ b/packages/context/tmux-context/tsconfig.json
@@ -0,0 +1,37 @@
+{
+  "extends": "../../../tsconfig.base.json",
+  "compilerOptions": {
+    "rootDir": "src",
+    "outDir": "lib/types"
+  },
+  "include": ["src"],
+  "references": [
+    {
+      "path": "../../../vendor/cosmokit"
+    },
+    {
+      "path": "../../../vendor/cordis"
+    },
+    {
+      "path": "../../../vendor/schemastery"
+    },
+    {
+      "path": "../../llm/llm"
+    },
+    {
+      "path": "../../bash/bash"
+    },
+    {
+      "path": "../../core/agent"
+    },
+    {
+      "path": "../../core/system-prompt"
+    },
+    {
+      "path": "../../support/invariants"
+    },
+    {
+      "path": "../../core/session"
+    }
+  ]
+}
diff --git a/packages/context/workspace-context/README.i18n.yaml b/packages/context/workspace-context/README.i18n.yaml
index 8191413d37..1cb7ff155c 100644
--- a/packages/context/workspace-context/README.i18n.yaml
+++ b/packages/context/workspace-context/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/context/workspace-context/README.md
 README.md: df75b29dd3e8dbb504aac9e9885c32a809cbf70f
-README.zh.md: 8bd926302f09ecdf453c7832b3a15b0e7fcc1b2a
+README.zh.md: 555e55464071261f5d1752b65ed6884c5452f4ae
diff --git a/packages/context/workspace-context/README.zh.md b/packages/context/workspace-context/README.zh.md
index 8bd926302f..555e554640 100644
--- a/packages/context/workspace-context/README.zh.md
+++ b/packages/context/workspace-context/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-为每个会话加载与 `AGENTS.md` 兼容的工作区指令文件。该插件会将初始 user 全局指令与项目指令链注入持久历史，随后发现嵌套文件，并在成功的文件系统工具调用后报告后续变更或移除。
+为每个会话加载与 `AGENTS.md` 兼容的工作区指令文件。该插件会将初始的用户全局指令与项目指令链注入持久历史，随后发现嵌套文件，并在成功的文件系统工具调用后报告后续变更或移除。
 
 ## 生命周期
 
@@ -10,9 +10,9 @@
 
 该插件还会监听 `tools/post-execute` 中成功的第一方 `read`、`write` 和 `edit` 调用。每次 touch 都会检查新达到的后代 scope 以及之前加载的每个 scope。每个已配置候选名称都是所在目录中的独立 scope：新出现的文件通过结果的 `additionalContexts` 附加；已改变文件追加替换；文件消失或成为同一目录中较早候选文件的重复项时，追加移除通知。原生调用与 Code Mode 子分派共享该路径：`run_code` 将每个嵌套上下文延迟到外层结果，因此 loop 仍会在工具调用／结果相邻关系完成后追加更新。这种发现跟随结构化文件系统活动，而不是 shell `cd`，因为每次本地 bash 调用都启动新 shell，解析任意 shell 语法也不可靠。
 
-指令读取使用可选 `ctx.fs` 提供方。该插件不会静态注入 `fs`，因此没有提供方的产品树仍可启动，指令加载在提供方出现前不执行任何操作。它会解析每个候选文件并获取结果状态，因此会跟随最终组件 symlink 到其目标：指向常规文件的链接会加载目标内容，缺失路径或非文件目标（包括指向目录的链接）则已确认不存在。resolve 或 stat 异常会改为将该候选文件的 scope 标记为暂时不可用。步骤取消与动态工具取消会传播到解析、元数据探测与流式读取。文件加载后的提供方失败会视为暂时不可用，而非文件已删除的证据。
+指令读取使用可选 `ctx.fs` 提供方。该插件不会静态注入 `fs`，因此没有提供方的产品树仍可启动，指令加载在提供方出现前不执行任何操作。它会解析每个候选文件并对解析结果执行 stat，因此会跟随路径最后一段的 symlink 到其目标：指向常规文件的链接会加载目标内容，缺失路径或非文件目标（包括指向目录的链接）则已确认不存在。resolve 或 stat 异常会改为将该候选文件的 scope 标记为暂时不可用。前缀取消与动态工具取消会传播到解析、元数据探测与流式读取。文件加载后的提供方失败会视为暂时不可用，而非文件已删除的证据。
 
-## 提示词形状
+## 提示词结构
 
 基线指令是持久的 user 角色消息，使用熟悉的 system-reminder 模式框定：
 
@@ -42,15 +42,15 @@ These instructions apply to work under `packages/app`. Use them as guidance when
 </system-reminder>
 ```
 
-同一文件的编辑以 `Updated instructions from: <path>` 开头，并说明使用新内容替代之前加载的内容。候选文件消失或成为同一目录中较早候选文件的重复项时，消息是 `Instructions removed: <path>`，后跟 `The previously loaded instructions from this file no longer apply.`。指令文件中的字面 `</system-reminder>` 文本会转义，因此文件内容无法关闭插件拥有的 frame。
+同一文件的编辑以 `Updated instructions from: <path>` 开头，并说明使用新内容替代之前加载的内容。候选文件消失或成为同一目录中较早候选文件的重复项时，消息是 `Instructions removed: <path>`，后跟 `The previously loaded instructions from this file no longer apply.`。指令文件中的字面 `</system-reminder>` 文本会转义，因此文件内容无法关闭插件控制的框架。
 
-该插件拥有完整 `<system-reminder>` framing，每个注入的 `user/message` 都会在没有核心包装的情况下逐字达到模型。
+该插件控制完整的 `<system-reminder>` 框架，每个注入的 `user/message` 都不经核心包装便原样传给模型。
 
 ## 状态与刷新
 
 模型可见文本不含隐藏状态标记。每个基线或动态上下文事件改为携带带类型的 `workspace-instructions` 来源，其中包含 `{ action, scope, path, digest? }` 变更列表；完整的启动或恢复基线还会携带 `baseline: true`。每次相关工具 touch 时，插件会从可见会话事件重建已加载状态，并叠加一个短暂内存 pending 窗口，用于不可变顶层 `tools/result` 上存在但 loop 尚未追加的上下文。匹配的持久 `user/message` 会确认 pending 转换。如果所属 `step/end` 在匹配上下文进入日志之前到达，插件会清除 pending 转换及其版本快速路径，使下一次成功 touch 可以重新加载。嵌套 Code Mode 结果会在外层执行 token 下暂存 pending 变更，用于抑制同次运行中的重复项；外层结果会回滚该状态，再只重新提交经过外层策略的上下文。
 
-路径与 SHA-1 内容 digest 都未变时，不会重复注入。每会话、每 scope 提供方 cache 只存储 `{ path, version, digest, trimmedDigest }`：当提供方的不透明 `FsVersion` 与有效可见状态都匹配时，对账会跳过内容读取；版本改变会在任何模型可见更新之前触发有界读取与 SHA-1 确认。`trimmedDigest` 是针对去除空白后内容的 SHA-1，也是每目录重复 key，因此较早候选文件与某个未更改文件的内容收敛后，后者仍可被移除。恢复可行，因为 SHA-1 状态持久化在带类型的来源中，而空的内存版本 cache 只会导致一次确认读取。压缩会在 scope 的上下文事件离开可见表层后重新启用它，即使缓存版本未变。移除是 tombstone，因此候选文件之后重新出现时会重新加载。只有在字节预算内实际渲染的模型可见变更才会进入来源、pending 状态和版本 cache；已省略变更仍可在后续 touch 处理，而相同 digest 的版本刷新只更新提供方 cache。
+路径与 SHA-1 内容 digest 都未变时，不会重复注入。每会话、每 scope 提供方 cache 只存储 `{ path, version, digest, trimmedDigest }`：当提供方的不透明 `FsVersion` 与有效可见状态都匹配时，对账会跳过内容读取；版本改变会在任何模型可见更新之前触发有界读取与 SHA-1 确认。`trimmedDigest` 是针对去除空白后内容的 SHA-1，也是每目录重复 key，因此较早候选文件与某个未更改文件的内容收敛后，后者仍可被移除。恢复可行，因为 SHA-1 状态持久化在带类型的来源中，而空的内存版本 cache 只会导致一次确认读取。压缩（compaction）会在 scope 的上下文事件离开可见表层后重新启用它，即使缓存版本未变。移除是 tombstone，因此候选文件之后重新出现时会重新加载。只有在字节预算内实际渲染的模型可见变更才会进入来源、pending 状态和版本 cache；已省略变更仍可在后续 touch 处理，而相同 digest 的版本刷新只更新提供方 cache。
 
 初始基线事件自身不会被改写。其带类型的变更仅在该事件仍位于可见会话表层时才是权威状态；下一次成功的文件系统 touch 会在压缩后重新添加未变的基线 scope，或追加其替换或移除。内存中的 scope 标记和提供方版本 cache 只负责选择探测对象并加速探测。插件热重挂只有在其带类型的事件仍然可见时才保留基线，同时会重建当前 scope 与版本跟踪状态；否则会注入当前基线。恢复的 loop 始终重新组合当前基线，并在第一个请求前对账仍可见的动态 scope。没有文件 watcher，因此磁盘变更会在下一次成功 `read`、`write` 或 `edit` touch 时可见，也会在恢复 loop 准备基线时可见。
 
@@ -67,15 +67,15 @@ export interface Config {
 }
 ```
 
-`maxBytes` 必填，因此每个部署都必须显式选择提示词预算。`maxSourceBytes` 在渲染前限制每个源指令文件，默认为 1 MiB。`projectRootMarkers` 默认为 `['.git']`，`instructionFileCandidates` 默认为 `['AGENTS.md', 'CLAUDE.md']`。每个项目目录中的所有现有候选文件都会加载，在去除周围空白后与较早候选文件内容匹配的文件会被丢弃。因此，使用默认设置时，内容相同的 `AGENTS.md` 与 `CLAUDE.md` 只渲染一次（作为 `AGENTS.md`），真正不同的同级文件则同时应用。`localInstructionFileCandidates` 默认为 `['AGENTS.local.md', 'CLAUDE.local.md']`，会与同一目录的基础文件一起加载其现有 overlay（渲染在它们之后），并应用同一个每目录去重；空列表会禁用 overlay。两个列表的候选配置项都必须是同目录文件名，因此会忽略空配置项、`.`／`..` 以及包含 `/` 或 `\` 的配置项。
+`maxBytes` 必填，因此每个部署都必须显式选择提示词预算。`maxSourceBytes` 在渲染前限制每个源指令文件，默认为 1 MiB。`projectRootMarkers` 默认为 `['.git']`，`instructionFileCandidates` 默认为 `['AGENTS.md', 'CLAUDE.md']`。每个项目目录中的所有现有候选文件都会加载，在去除周围空白后与较早候选文件内容匹配的文件会被丢弃。因此，使用默认设置时，内容相同的 `AGENTS.md` 与 `CLAUDE.md` 只渲染一次（作为 `AGENTS.md`），真正不同的同级文件则同时应用。`localInstructionFileCandidates` 默认为 `['AGENTS.local.md', 'CLAUDE.local.md']`，会与同一目录的基础文件一起加载其现有 overlay（渲染在它们之后），并应用同一个每目录去重；空列表会禁用 overlay。两个列表中的候选项都必须是同一目录下的文件名，因此会忽略空项、`.`／`..` 以及包含 `/` 或 `\` 的项。
 
-user 全局文件始终是 `$DSH_HOME/AGENTS.md`，没有本地 overlay；两个候选列表只控制项目 scope。`$DSH_HOME` 默认为 `~/.dsh`，已配置的 `~`、`~/...` 与 Windows 风格 `~\...` 前缀会基于操作系统 home 目录展开。非正数或非有限渲染预算会同时禁用基线与动态加载；已配置 `maxSourceBytes` 必须是正整数。
+用户全局文件始终是 `$DSH_HOME/AGENTS.md`，没有本地 overlay；两个候选列表只控制项目 scope。`$DSH_HOME` 默认为 `~/.dsh`，已配置的 `~`、`~/...` 与 Windows 风格 `~\...` 前缀会基于操作系统 home 目录展开。非正数或非有限渲染预算会同时禁用基线与动态加载；已配置 `maxSourceBytes` 必须是正整数。
 
 ## 预算与有界读取
 
 渲染会优先保留最具体的指令文件。它会先丢弃完整的较宽泛文件，再截断最具体文件，并发出可见 `Workspace instruction budget ...` 通知，其中指名已省略与已截断路径。渲染后字节数绝不超过 `maxBytes`。
 
-即使提供方元数据省略大小，或文件在元数据探测后增长，指令内容仍会通过 `streamText()` 在 `maxSourceBytes` 下读取。超大文件会被忽略；在动态对账期间，它会暂时不可用，而不是被移除。该插件不保留进程级 cache，绝不缓存指令文本。其会话本地 scope cache 只将提供方版本用作快速失效信号；失效后，对有界读取计算的 SHA-1 仍是存储在结构化消息来源中的跨提供方内容身份。
+即使提供方元数据省略大小，或文件在元数据探测后增长，指令内容仍会通过 `streamText()` 在 `maxSourceBytes` 下读取。超大文件会被忽略；在动态对账期间，它会暂时不可用，而不是被移除。该插件不保留进程级 cache，绝不缓存指令文本。其会话本地 scope cache 只将提供方版本用作快速失效信号；失效后，对有界读取计算的 SHA-1 仍是存储在结构化消息来源中的跨提供方内容标识。
 
 ## 模型体验
 
@@ -83,7 +83,7 @@ user 全局文件始终是 `$DSH_HOME/AGENTS.md`，没有本地 overlay；两个
 
 #### 模型看到的内容
 
-在每个 loop 实例的第一个请求中，模型会收到一条持久 user 角色消息，其中按从宽泛到具体的顺序包含有界 user 全局指令与项目指令链。
+在每个 loop 实例的第一个请求中，模型会收到一条持久 user 角色消息，其中按从宽泛到具体的顺序包含有界用户全局指令与项目指令链。
 
 ##### 基线指令模板
 
@@ -133,7 +133,7 @@ These instructions apply to work under `packages/app`. Use them as guidance when
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 已改变或移除的指令上下文
 
@@ -157,13 +157,13 @@ The previously loaded instructions from this file no longer apply.
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
 - **发现跟随结构化 fs 工具，而非 shell 导航**：更改目录的 `bash` 命令不会触发嵌套指令发现，因为 shell 语法与每次调用 shell 状态不是可靠的文件系统 seam。
 - **刷新由 touch 驱动**：没有 watcher；外部编辑会在下一次成功的第一方 `read`、`write` 或 `edit` 时可见，也会在恢复 loop 准备基线时可见。
-- **候选语义有意保持简单**：不解释小写名称、`.claude/rules/` 与 `@path` import；项目 scope 默认加载 `AGENTS.local.md`／`CLAUDE.local.md` overlay，但 user 全局 `$DSH_HOME` scope 没有本地 overlay，其他自定义名称需要显式候选配置。
+- **候选语义有意保持简单**：不解释小写名称、`.claude/rules/` 与 `@path` import；项目 scope 默认加载 `AGENTS.local.md`／`CLAUDE.local.md` overlay，但用户全局 `$DSH_HOME` scope 没有本地 overlay，其他自定义名称需要显式候选配置。
 - **每目录去重基于内容**：只有在去除首尾空白后字节完全一致时，才折叠同级候选文件。`CLAUDE.md` 若 symlink 到同级 `AGENTS.md`，会解析为相同内容，并像任何重复项一样折叠；从 `AGENTS.md` 漂移的独立实体副本则会与它一起完整加载。
-- **Symlink 指令文件会跨越信任边界跟随**：最终组件是 symlink 的候选文件会被解析并加载其目标，因此克隆仓库可以将树外文件内容呈现为较低权限的工作区指引（它绝不会覆盖 system、developer 或直接 user 指令）。加载不受信任仓库时，请用文件系统策略门禁或 OS 沙箱限制 `ctx.fs`。
-- **指令内容受限但不会摘要**：超出预算的宽泛文件会被省略，最具体文件可能被截断；该插件绝不请求模型压缩指令文本。
+- **Symlink 指令文件会跨越信任边界跟随**：最终组件是 symlink 的候选文件会被解析并加载其目标，因此克隆仓库可以将树外文件内容呈现为较低优先级的工作区指引（它绝不会覆盖 system、developer 或用户直接下达的指令）。加载不受信任仓库时，请用文件系统策略门禁或 OS 沙箱限制 `ctx.fs`。
+- **指令内容受限但不会被摘要**：超出预算的宽泛文件会被省略，最具体文件可能被截断；该插件绝不请求模型压缩指令文本。
diff --git a/packages/cordis/README.i18n.yaml b/packages/cordis/README.i18n.yaml
index fbc8586428..aaa96435d3 100644
--- a/packages/cordis/README.i18n.yaml
+++ b/packages/cordis/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/cordis/README.md
 README.md: a47b9ba20789bb6b9a36b1af9b3942b90e61b365
-README.zh.md: cc91e68f0dfa9fb343c332eba2579c2a077beb64
+README.zh.md: 3ee1ddb1db28352cd05b3e79e88228035bc39ac4
diff --git a/packages/cordis/README.zh.md b/packages/cordis/README.zh.md
index cc91e68f0d..3ee1ddb1db 100644
--- a/packages/cordis/README.zh.md
+++ b/packages/cordis/README.zh.md
@@ -2,8 +2,8 @@
 
 [English](README.md) | 中文
 
-面向模型、作用于 agent（智能体）所在实时 Cordis 运行时的工具：检查当前 DSH 进程，并挂载或卸载仅存于内存的临时 Plugin。设计归档见[工具集 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
+这些面向模型的工具作用于 agent（智能体）自身所在的实时 Cordis 运行时，可检查已加载的插件和服务接口、挂载模型编写的插件，并将其 dispose（资源释放）。设计说明见[工具集 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
 
 | 包（package） | 角色 | ctx 键 |
 |---|---|---|
-| [`tool-cordis/`](tool-cordis/README.md) | `cordis_inspect`／`cordis_mount`／`cordis_unmount` 工具：读取当前进程运行时，并在一个自有分组 fiber 下管理临时 Plugin | 注册到 `ctx.tools` |
+| [`tool-cordis/`](tool-cordis/README.md) | `cordis_inspect`／`cordis_mount`／`cordis_unmount` 工具：读取当前进程运行时，并在一个自有分组 fiber 下管理内存中的临时插件 | 注册到 `ctx.tools` |
diff --git a/packages/cordis/tool-cordis/README.i18n.yaml b/packages/cordis/tool-cordis/README.i18n.yaml
index 64d5ac7711..8be2474092 100644
--- a/packages/cordis/tool-cordis/README.i18n.yaml
+++ b/packages/cordis/tool-cordis/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/cordis/tool-cordis/README.md
 README.md: 5b58e665dae95aea0d0ad094238fef5d3dc0fb97
-README.zh.md: 11e5be11dca84247b19888fefa7d70e5d74da174
+README.zh.md: 237b72244be2336a82c8c48cb341d7f9796d08f4
diff --git a/packages/cordis/tool-cordis/README.zh.md b/packages/cordis/tool-cordis/README.zh.md
index 11e5be11dc..237b72244b 100644
--- a/packages/cordis/tool-cordis/README.zh.md
+++ b/packages/cordis/tool-cordis/README.zh.md
@@ -2,19 +2,19 @@
 
 [English](README.md) | 中文
 
-自引用 Cordis 工具集：三个面向模型的工具，操作当前 DSH 进程中的存活运行时。设计归属（沙箱语义、临时 Plugin 生命周期与组合、生成的 API 目录、既定决策）见[工具集 Agent Note](../../../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
+自引用 Cordis 工具集：三个面向模型的工具，操作当前 DSH 进程中的实时运行时。沙箱语义、临时插件生命周期与组合、生成的 API 目录及既定决策详见[工具集 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-08-self-referential-cordis-toolset.md)。
 
 ## 功能
 
-- `cordis_inspect`：当前进程运行时的只读报告，包括服务、全部存活 Plugin fiber、已注册工具、`cordis_mount` 临时 Plugin 子集，以及目录支持的 `api`／`events` 参考。精确的 `name` 配合 `what: "api"` 或 `what: "events"` 可缩窄报告，并附上原始源代码 JSDoc。
-- `cordis_mount`：立即求值模型编写的 JavaScript 且不保存到任何位置；代码必须返回一个以 `dyn-<n>` 跟踪、仅存于内存的临时 Plugin。
-- `cordis_unmount`：卸载一个 `dyn-<n>` 临时 Plugin，并只在其自有效果完全停稳后返回；它不能删除 Loader、配置或已安装的 Plugin。
+- `cordis_inspect`：当前进程运行时的只读报告，包括服务、全部存活插件 fiber、已注册工具、`cordis_mount` 临时插件子集，以及目录支持的 `api`／`events` 参考。精确的 `name` 配合 `what: "api"` 或 `what: "events"` 可缩窄报告，并附上原始源代码 JSDoc。
+- `cordis_mount`：立即求值模型编写的 JavaScript 且不保存到任何位置；代码必须返回一个仅存于内存、以 `dyn-<n>` 为标识进行跟踪的临时插件。
+- `cordis_unmount`：卸载一个 `dyn-<n>` 临时插件，并只在其拥有的 effect 完全停稳后返回；它不能移除 Loader 插件、已配置插件或已安装插件。
 
-精确的面向模型 schema 见[生成的工具目录](../../../docs/tool-catalog.md)。
+面向模型的确切 schema 见[生成的工具目录](../../../docs/tool-catalog.md)。
 
-规范成功值分别为检查字符串、挂载 `{ id, pluginName, state, provides, waitingFor }`，以及卸载 `{ id, pluginName }`。原生 renderer 会说明临时 Plugin 正在运行还是等待中，并说明它可用至被卸载或 DSH 重启；卸载结果确认它已移除。
+规范成功结果分别为检查字符串、挂载 `{ id, pluginName, state, provides, waitingFor }`，以及卸载 `{ id, pluginName }`。原生渲染会说明临时插件正在运行还是等待中，并说明它可用至被卸载或 DSH 重启；卸载结果确认它已移除。
 
-临时 Plugin 只存在于共享 DSH 进程内存中。它可跨后续 turn 保持活跃，也可能影响同一进程中的其他 session，但会在 `cordis_unmount`、工具集卸载或 DSH 重启后消失。它不会创建 Plugin 文件、安装 package、修改 `cordis.yml` 或个人／项目配置、跨重启存续，也不能自动转为正式 Plugin。若要保留实验结果，应让 Agent 通过常规开发流程实现普通的本地、项目或仓库 Plugin。
+临时插件只存在于共享 DSH 进程内存中。它可跨后续轮次保持活跃，也可能影响同一进程中的其他会话，但会在 `cordis_unmount`、工具集卸载或 DSH 重启后消失。它不会创建插件文件、安装任何包（package）、修改 `cordis.yml` 或个人／项目配置、跨重启存续，也不能自动转为正式插件。若要保留实验结果，应让 agent（智能体）通过常规开发流程实现普通的本地、项目或仓库插件。
 
 ## 信任立场
 
@@ -24,17 +24,17 @@
 
 | 字段 | 默认值 | 含义 |
 |---|---|---|
-| `vmTimeoutMs` | `5000` | 临时 Plugin 代码求值中同步部分的边界；async 主体可逃出该边界 |
+| `vmTimeoutMs` | `5000` | 临时插件代码求值中同步部分的边界；async 主体可逃出该边界 |
 
 ## 生成的 API 目录
 
-`src/api-catalog.ts` 由 `scripts/gen-cordis-api.ts` 生成，使用与 [docs/cordis-catalog](../../../docs/cordis-catalog/services.md) 相同的 AST 遍历，并由 `pnpm run verify-cordis-api`（位于 `doc-sync` 中）实施新鲜度门禁，绝不可手工编辑。`cordis_inspect` 在调用时把该目录与存活服务 store 取交集。宽泛的 `api`／`events` 报告只渲染摘要与签名；精确 `name` 会选择保留的方法／事件 JSDoc，未知或未运行的服务目标会高声失败。
+`src/api-catalog.ts` 由 `scripts/gen-cordis-api.ts` 生成，使用与 [docs/cordis-catalog](../../../docs/cordis-catalog/services.md) 相同的 AST 遍历，并由 `pnpm run verify-cordis-api`（位于 `doc-sync` 中）实施新鲜度门禁，绝不可手工编辑。`cordis_inspect` 在调用时把该目录与存活服务 store 取交集。宽泛的 `api`／`events` 报告只渲染摘要与签名；精确 `name` 会选择保留的方法／事件 JSDoc，未知或未运行的服务目标会明确报错。
 
 ## 渲染
 
-三个工具都渲染 `generic` 卡片（`read`／`execute`／`delete`）；`cordis_mount` 以 `rawInput` 携带临时 Plugin 代码。presenter 是 args 的纯函数；结果保留默认文本渲染。
+三个工具都渲染 `generic` 卡片（`read`／`execute`／`delete`）；`cordis_mount` 以 `rawInput` 携带临时插件代码。presenter 是 args 的纯函数；结果保留默认文本渲染。
 
-## 导出形状
+## 导出形式
 
 Namespace 插件：命名导出 `name`／`inject`／`Config`／`apply`，无默认导出（[docs/postmortem/0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)）。
 
@@ -58,29 +58,29 @@ Namespace 插件：命名导出 `name`／`inject`／`Config`／`apply`，无默
 
 #### 模型看到的内容
 
-检查会精确地用 `## <section>` 加换行及数据相关主体来拼接选中区段，各区段之间留一个空行；`what: "temporary"` 使用 `## Temporary Plugins` 标题。每个临时 Plugin 行都会报告 running／pending 状态、提供与等待的服务，以及持续至卸载或 DSH 重启的生命周期；空状态说明 `cordis_mount` Plugin 会在重启时消失。宽泛的 API／事件报告省略 JSDoc；`name` 配合 `what: "api"` 或 `what: "events"` 返回一个精确目标及其原始 JSDoc。挂载返回 `Temporary Plugin <id> is running (...)` 或 `Temporary Plugin <id> is pending (...)`；卸载返回 `Temporary Plugin <id> was unmounted and removed.`。提交的程序保留在 assistant 工具调用历史中。
+检查会精确地用 `## <section>` 加换行及取决于数据的正文来拼接选中区段，各区段之间留一个空行；`what: "temporary"` 使用 `## Temporary Plugins` 标题。每个临时插件行都会报告 running／pending 状态，以及其提供和等待的服务，以及持续至卸载或 DSH 重启的生命周期；空状态说明 `cordis_mount` 插件会在重启时消失。宽泛的 API／事件报告省略 JSDoc；`name` 配合 `what: "api"` 或 `what: "events"` 返回一个精确目标及其原始 JSDoc。挂载返回 `Temporary Plugin <id> is running (...)` 或 `Temporary Plugin <id> is pending (...)`；卸载返回 `Temporary Plugin <id> was unmounted and removed.`。提交的程序保留在 assistant 工具调用历史中。
 
 #### Token 影响
 
-检查输出与挂载代码取决于数据，并在压缩前重复发送；生命周期确认文本很短。
+检查输出与挂载代码取决于数据，并在压缩（compaction）前重复发送；生命周期确认文本很短。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### cordis_mount 后的后续请求
 
 #### 模型看到的内容
 
-临时 Plugin 可以注册工具、提示词贡献或监听器，改变其目标 scope 的后续请求；`cordis_unmount` 会在完全停稳后移除这些贡献。
+临时插件可以注册工具、提示词贡献或监听器，改变其目标 scope 的后续请求；`cordis_unmount` 会在完全停稳后移除这些贡献。
 
 #### Token 影响
 
-间接 token 影响等于临时 Plugin 的贡献，且只在其进程内生命周期内持续。
+间接 token 影响等于临时插件的贡献，且只在其进程内生命周期内持续。
 
 #### KV Cache 影响
 
-挂载或卸载提示词／工具贡献会改变后续请求前缀，并可能使从第一个变化的贡献起的复用失效；临时 Plugin 集合不变时，前缀保持稳定。
+挂载或卸载提示词／工具贡献会改变后续请求前缀，并可能使从第一个变化的贡献起的复用失效；临时插件集合不变时，前缀保持稳定。
 
 ## 已知限制与暂缓事项
 
diff --git a/packages/cordis/tool-cordis/src/api-catalog.ts b/packages/cordis/tool-cordis/src/api-catalog.ts
index 7d9b087b8f..a19d49f860 100644
--- a/packages/cordis/tool-cordis/src/api-catalog.ts
+++ b/packages/cordis/tool-cordis/src/api-catalog.ts
@@ -734,6 +734,10 @@ export const SERVICE_API: readonly ServiceApiEntry[] = [
         signature: 'get(session: Session): SessionTitleSnapshot | undefined',
         jsDoc: '/**\n * Read the latest folded title from one live or replayed session.\n * @param session - session whose log is the title source of truth.\n * @returns latest title snapshot, or `undefined` before eligible input.\n */',
       },
+      {
+        signature: 'rename(session: Session, title: string): SessionTitleSnapshot',
+        jsDoc: '/**\n * Accept an explicit user title. Appends a `session/title` event with the\n * `user` source, which pins the title: in-flight automatic generation is\n * superseded and later user messages schedule none (an explicit\n * {@link SessionTitleService.refresh} remains the deliberate unpin).\n * @param session - exact live session to rename.\n * @param title - raw user input; normalized before acceptance.\n * @returns the accepted title snapshot.\n * @throws {SessionTitleInvalidError} when the title normalizes to empty.\n * @throws {Error} when the session is not live or the service is disposed.\n */',
+      },
       {
         signature: 'async refresh(session: Session, signal?: AbortSignal): Promise<SessionTitleSnapshot | undefined>',
         jsDoc: '/**\n * Explicitly retry the registered provider, or materialize the built-in\n * fallback when no provider is registered.\n * @param session - exact live session to refresh.\n * @param signal - optional caller cancellation.\n * @returns latest accepted title, or `undefined` when no eligible text exists.\n */',
@@ -2325,7 +2329,7 @@ export const TYPE_API: readonly TypeApiEntry[] = [
   },
   {
     name: 'SessionTitleSource',
-    declaration: 'export type SessionTitleSource = {\n    readonly kind: \'fallback\';\n} | {\n    readonly kind: \'provider\';\n    readonly provider: SessionTitleProviderId;\n    readonly model?: SessionTitleModelProvenance;\n};',
+    declaration: 'export type SessionTitleSource = {\n    readonly kind: \'fallback\';\n} | {\n    readonly kind: \'provider\';\n    readonly provider: SessionTitleProviderId;\n    readonly model?: SessionTitleModelProvenance;\n} | {\n    readonly kind: \'user\';\n};',
   },
   {
     name: 'SessionTitleUserMessage',
diff --git a/packages/core/README.i18n.yaml b/packages/core/README.i18n.yaml
index b5efad0494..58d1534fd0 100644
--- a/packages/core/README.i18n.yaml
+++ b/packages/core/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/core/README.md
 README.md: 63ca0f7711c2c9deb193d5a9a574ff909a04602e
-README.zh.md: a7310ff238eaa141ec53d1f2b5705397dbc04699
+README.zh.md: a217868f6c25ddce690ca32fc09b1d68b8ab48d2
diff --git a/packages/core/README.zh.md b/packages/core/README.zh.md
index a7310ff238..a217868f6c 100644
--- a/packages/core/README.zh.md
+++ b/packages/core/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-会话日志、系统提示词组装、工具注册表、agent 词汇，以及构成 harness 默认控制主干的具体循环。这些是 **产品** 包（package），插件和消费方以其稳定表层为基础构建。
+会话日志、系统提示词组装、工具注册表、agent（智能体）词汇，以及构成 harness 默认控制主干的具体循环。这些是 **产品** 包（package），插件和消费方以其稳定接口为基础构建。
 
 | 包 | 角色 | ctx 键 |
 |---|---|---|
@@ -17,4 +17,4 @@
 
 `agent-loop` 是 `agent` seam 的唯一具体实现，位于此处是因为它就是 harness 的默认产品循环。它在 `ctx.agents.withInitiator()` 中运行每个驱动器。扩展插件依赖 `agent`，即使需要发起调用的 Agent 也是如此；它们绝不直接依赖 `agent-loop`，因此循环保持可替换。
 
-将这条主干接成可运行 agent 的默认组合位于 [`examples/agent-spine-demo`](../examples/agent-spine-demo/README.md)：一个 bundle（组合包）插件，加载控制主干及所选默认能力（`timer` + `llm` + 会话 + 后备会话标题 + 系统提示词 + 工具 + agent + 不变式 + 本地[技能系列](../skill/README.md) + `tool-bash` + workspace 上下文 + `agent-loop`），并将 `agent-loop` 的 `agents` 列表作为自身配置转发。它位于 `examples/`，即开箱可运行的演示／参考组合包，而不是 `core/`：`core/` 交付可替换的主干组件，演示组合包则选定其中一种具体组合并添加前端入口。
+将这条主干接成可运行 agent 的默认组合位于 [`examples/agent-spine-demo`](../examples/agent-spine-demo/README.md)：一个 bundle（组合包）插件，加载控制主干及所选默认能力（`timer` + `llm` + 会话 + 后备会话标题 + 系统提示词 + 工具 + agent + 不变式 + 本地 [skill（技能）系列](../skill/README.md) + `tool-bash` + 工作区上下文 + `agent-loop`），并将 `agent-loop` 的 `agents` 列表作为自身配置转发。它位于 `examples/`，即开箱可运行的演示／参考组合包，而不是 `core/`：`core/` 交付可替换的主干组件，演示组合包则选定其中一种具体组合并添加一个对外交互入口。
diff --git a/packages/core/agent-loop/README.i18n.yaml b/packages/core/agent-loop/README.i18n.yaml
index 75edc1357f..339c91a2ff 100644
--- a/packages/core/agent-loop/README.i18n.yaml
+++ b/packages/core/agent-loop/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/core/agent-loop/README.md
-README.md: 6bb8b12af69f54c2a75cd672e4d3802887808c76
-README.zh.md: f9eb8aa3cdead427a88492e35c00eab80ba12f91
+README.md: 16d70cc06498fec1221b7872f988a0126f69f39f
+README.zh.md: ce68595072766ebbf1e4cbd9f7c262cee36c5eff
diff --git a/packages/core/agent-loop/README.md b/packages/core/agent-loop/README.md
index 6bb8b12af6..16d70cc064 100644
--- a/packages/core/agent-loop/README.md
+++ b/packages/core/agent-loop/README.md
@@ -67,7 +67,7 @@ After `agent/request` returns a provider/model call config, the loop asks `ctx.l
 
 Plugin failure ends the current turn, not the loop. Only final adapter dispatch/iteration failures and terminal in-band error or aborted finishes enter `agent/request-error`; middleware, result processing, tools, and other extension failures close directly. Recovery receives the exact live error, immutable provider facts, immutable prior failures, the immutable retry policy of the adapter registration that served the request, and the turn signal after the failed step closes; the policy is absent if no final adapter served it. A handling listener returns `{ kind: 'retry' }`; the loop closes the failed turn with its error and opens one numbered retry turn without an intervening idle notification. Success clears the consecutive history, and an unhandled failure is terminal. AgentLoop owns one cancellation signal for the current admission or turn. An effective `cancel(cause)` clears pending work unless `keepInbox` is set and cooperatively aborts that signal; idle cancellation is a no-op. Durable `turn/end` records `aborted` for `user` and `parent`, while disposal records `disposed`; undispatched model tool calls receive synthetic `tool/call` and `ABORTED_BEFORE_DISPATCH` result pairs. The cancellation cause changes reporting, not how result context finalized after cancellation is handled. Disposal waits for signal-ignoring work before registry removal. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns the lifecycle and race contract.
 
-Within a step, exclusive calls form barriers; parallel-safe calls use a bounded rolling pool and are reclassified before start. Only dispatch/body overlaps. Policy, durable results, and result context remain model-ordered. Abort stops new calls, drains started results, and retains their finalized result context without distinguishing the cancellation cause.
+Within a step, exclusive calls form barriers; parallel-safe calls use a bounded rolling pool and are reclassified before start. Only dispatch/body overlaps. Policy, durable results, and result context remain model-ordered. Abort stops new calls, drains started results, and retains their finalized result context without distinguishing the cancellation cause. An internal scheduler failure stops new dispatches, waits for already-started dispatches, and reaches the turn error boundary without fabricating tool results.
 
 ### What belongs to plugins
 
diff --git a/packages/core/agent-loop/README.zh.md b/packages/core/agent-loop/README.zh.md
index f9eb8aa3cd..ce68595072 100644
--- a/packages/core/agent-loop/README.zh.md
+++ b/packages/core/agent-loop/README.zh.md
@@ -12,11 +12,11 @@
 
 创建与恢复属于同一个受回滚保护的事务：构造私有会话、实体 agent 和带作用域的上下文；等待可选 setup；进入两个注册表；依次宣告 `session/created` 和 `agent/created`；发出 `agent/session-start`；此后才启动驱动器。Setup 接收完整的带作用域 `Context`，作为受信任的同进程组合代码，并且不得驱动尚未发布的 agent。普通的类型化身份与选项输入遵循只读契约以借用方式传入；seed 事件与会话元数据会跨越持久会话边界，因此系统会验证并快照它们。可选的 `AbortSignal` 只取消加载／setup／发布，并在返回的 handle 可见前分离。
 
-调用方 fiber 与 AgentLoop 提供方共同拥有 agent。`AgentFactory.createAgent(ownerCtx, options)` 与 `resume(ownerCtx, options)` 显式接收调用方所有权，而工厂为 `sessions`/`llm`/`tools`/`systemPrompt` 保留自身的依赖上下文；这样，调用方可以只注入 `agents`，而不会缩减新 agent 的服务接口。调用方卸载、handle 释放或提供方卸载都会汇合到同一个记忆化的完全停稳边界。提供方关闭会同时等待资源 teardown，以及已经观测到停用的公开 create/resume 包装层，因此依赖消失后，任何 continuation 都无法继续发布。
+调用方 fiber 与 AgentLoop 提供方共同拥有 agent。`AgentFactory.createAgent(ownerCtx, options)` 与 `resume(ownerCtx, options)` 显式接收调用方所有权，而工厂为 `sessions`/`llm`/`tools`/`systemPrompt` 保留自身的依赖上下文；这样，调用方可以只注入 `agents`，而不会缩减新 agent 的服务接口。调用方卸载、handle dispose（资源释放）或提供方卸载都会汇合到同一个记忆化的完全停稳边界。提供方关闭会同时等待资源 teardown，以及已经观测到停用的公开 create/resume 包装层，因此依赖消失后，任何 continuation 都无法继续发布。
 
 每个 agent 与其会话共享一个由调用方选择的 `SessionId`，并假设它在全局唯一；意外的 UUID 冲突不属于受支持模型。两个使用同一 id 的并发操作都可以进行准备，但最终的 `enter()` 调用会裁决发布，所有失败方都会回滚各自的私有资源。每次 detach 都绑定到确切进入的对象，因此陈旧 disposer 无法移除之后出现的同 id 替代项。在同步创建通知期间请求的 detach 会等待该次分发退栈，从而保留 created/disposed 配对。Teardown 顺序为停止并 drain → 撤销作用域 → detach agent → detach 会话；私有作用域清理完成后，该 id 即可复用。普通、不可 veto 的 `agent/*` 通知通过 `agentEvents(ctx, agent)` 发出；逐步骤组装通过 `assembleContextFor(agent)` 完成。
 
-- `ctx.agentLoop.create(id: SessionId, options?: AgentOptions, meta?: { cwd?: string }): Agent`：在确切共享的 agent／会话 id 下同步创建，不运行 setup，并随调用 fiber 释放。声明式配置把 `agents[].id` 视为稳定 label，通常会先生成 `${label}-session-<uuid>`，再调用此边界。应用也可以提供稳定且确切的 `sessionId`：首次使用时创建；重新挂载且持久化内容已存在时，则恢复已经实体化的历史。`resumeSessionId` 要求并加载现有的持久化 id，且与 `sessionId` 互斥。这样，默认的全新重启不会冲突，也无需保留第二个实时路由身份。
+- `ctx.agentLoop.create(id: SessionId, options?: AgentOptions, meta?: { cwd?: string }): Agent`：在确切共享的 agent／会话 id 下同步创建，不运行 setup，并随调用 fiber dispose。声明式配置把 `agents[].id` 视为稳定 label，通常会先生成 `${label}-session-<uuid>`，再调用此边界。应用也可以提供稳定且确切的 `sessionId`：首次使用时创建；重新挂载且持久化内容已存在时，则恢复已经实体化的历史。`resumeSessionId` 要求并加载现有的持久化 id，且与 `sessionId` 互斥。这样，默认的全新重启不会冲突，也无需保留第二个实时路由身份。
 
 `AgentLoop` 还实现 `AgentFactory` seam，并通过 `ctx.agents.setFactory(this)` 注册自身，因此插件会通过接口 `ctx.agents` 创建／恢复 agent：
 
@@ -49,7 +49,7 @@ interface Config {
 }
 ```
 
-通过配置创建的 agent 会自动启动。模型调用同时需要 `provider` 和 `model`；`agent/request` 可以在分发前补齐缺失的这一对值。可选的正整数 `maxTokens` 会为每次对话请求提供初始输出上限，并记录在请求 header 中。`maxParallelToolCalls` 限制每个 agent 针对并行安全调用使用的滚动池，默认值为 `10`。`cwd` 仅应用于全新会话，而 `resumeSessionId` 保留持久化元数据。通过配置创建的 agent 使用部署 persona；编程式 setup 可以按 agent 遮蔽它。该插件提供逐 agent 的 `provider`、`model` 和 `cwd` 提示词变量；harness 身份与部署 persona 属于 `dsh-system-prompt`。
+通过配置创建的 agent 会自动启动。模型调用同时需要 `provider` 和 `model`；`agent/request` 可以在分发前补齐缺失的这一对值。可选的正数 `maxTokens` 会为每次对话请求提供初始输出上限，并记录在请求 header 中。`maxParallelToolCalls` 限制每个 agent 针对并行安全调用使用的滚动池，默认值为 `10`。`cwd` 仅应用于全新会话，而 `resumeSessionId` 保留持久化元数据。通过配置创建的 agent 使用部署 persona；编程式 setup 可以按 agent 遮蔽它。该插件为每个 agent 提供 `provider`、`model` 和 `cwd` 提示词变量；harness 身份与部署 persona 属于 `dsh-system-prompt`。
 
 ### 包内部实体驱动器
 
@@ -61,19 +61,19 @@ interface Config {
 
 驱动器在其整个生命周期内拥有一个 agent，并在 `ctx.agents.withInitiator(agent, ...)` 内运行。包私有的编排入口点会恢复确切的 Agent，一次性派生 `agent.session`，并让操作局部的辅助函数捕获它，而不是通过浅层接口继续传递实体驱动器或每次操作的 `Session`。如果显式 `Session` 正是辅助函数的实际接口，该辅助函数会保留它；创建、持久化加载、未发布 setup、服务、worker、进程、持久化和 wire 协议则继续保留各自的显式身份。[agent 服务](../agent/README.md#initiating-agent-scope)规定传播、teardown 和分离工作规则。
 
-每次提供方调用成功结束时，都会恰好追加一个 `assistant/message` 完成锚点，包括无内容调用和以 `max-tokens` 结束的调用。该锚点原样记录组装后的内容，保留确切的 chunk 溯源（流没有 chunk 时为 `[]`），并在用量可用时包含用量；空内容不会进入派生消息历史。
+每次提供方调用成功结束时，都会恰好追加一个 `assistant/message` 完成锚点，包括无内容调用和以 `max-tokens` 结束的调用。该锚点原样记录组装后的内容，保留确切的分片溯源（流没有分片时为 `[]`），并在用量可用时包含用量；空内容不会进入派生消息历史。
 
 在 `agent/request` 返回提供方／模型调用配置后，循环会调用 `ctx.llm.prepareCall()`，在活跃轮次信号的控制下校验由适配器持有的推理（reasoning）强度，并填入其配置默认值。准备完成的调用会在这次异步解析、`request/header` 日志记录和最终分派期间保留同一项确切的适配器注册，因此 HMR（热模块替换）不会把某个适配器的能力解析结果与另一适配器的请求混用。生效配置会在分派前写入日志，因此监听器可以在步骤之间更改推理强度，而不会产生未记录的请求变化。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 监听器可以接管并短路该请求；最终分派仍会以 `NO_ADAPTER` 拒绝未得到处理的路由。新循环实例仅在初始提供方／模型路由与日志路由完全一致时恢复上次的推理强度；路由变化会丢弃由前一模型持有的不透明 ID，并单独解析新模型。
 
 插件失败会结束当前轮次，而不是结束循环。只有最终适配器分发／迭代失败以及带内的终止错误或中止结束才进入 `agent/request-error`；中间件、结果处理、工具及其他扩展失败会直接关闭轮次。失败步骤关闭后，恢复逻辑会接收确切的实时错误、不可变的提供方事实、不可变的先前失败、为请求提供服务的适配器注册所对应的不可变重试策略，以及轮次信号；如果没有最终适配器为其提供服务，则该策略缺失。处理失败的监听器返回 `{ kind: 'retry' }`；循环用其错误关闭失败轮次，并在不插入空闲通知的情况下开启一个编号重试轮次。成功会清除连续失败历史；未被处理的失败是终态。AgentLoop 为当前接纳或轮次拥有一个取消信号。有效的 `cancel(cause)` 在未设置 `keepInbox` 时清除待处理工作，并以协作方式中止该信号；空闲取消是空操作。持久 `turn/end` 为 `user` 和 `parent` 记录 `aborted`，dispose（资源释放）则记录 `disposed`；未分发的模型工具调用会收到合成的 `tool/call` 与 `ABORTED_BEFORE_DISPATCH` 结果对。取消原因只改变报告方式，不改变对取消后已定案结果上下文的处理。dispose 会等待忽略信号的工作完成，然后才从注册表移除。[显式取消决策](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)规定生命周期与竞态契约。
 
-在步骤内，独占调用形成屏障；并行安全调用使用有界滚动池，并在启动前重新分类。只有分发／主体会重叠。策略、持久结果和结果上下文仍保持模型顺序。中止会停止新调用，drain 已启动的结果，并保留其已定案的结果上下文，不区分取消原因。
+在步骤内，独占调用形成屏障；并行安全调用使用有界滚动池，并在启动前重新分类。只有分发／主体会重叠。策略、持久结果和结果上下文仍保持模型顺序。中止会停止新调用，drain 已启动的结果，并保留其已定案的结果上下文，不区分取消原因。内部调度器故障会停止新的分发，等待已启动的分发，然后在不虚构工具结果的情况下到达轮次错误边界。
 
 ### 插件负责的内容
 
 超出「调用模型、运行工具、重复」的所有内容，都属于监听事件分类体系的插件：
 - 钩子与策略：相关的 `agent/*` 检查点，加上受守卫保护的 `tools/pre-execute` → `tools/execute` → `tools/post-execute` → 定义拥有的 `finalizeContent` → `tools/result` 流水线；确切事件签名与 mode 位于生成的[事件目录](../../../docs/cordis-catalog/events.md)
-- 压缩（compaction）：在 `agent/step` 上观测压力；在 `agent/request-error` 上修复规范溢出
+- 压缩（compaction）：在 `agent/step` 上观测压力；在 `agent/request-error` 上进行规范的溢出修复
 - 模型请求恢复：`dsh-llm-retry` 在 `agent/request-error` 上记录并等待按确切提供方配置的 normal 或无界退避，发出不进入表层的 `llm/retry` 状态，然后返回重试动作
 - 沙箱、权限、计划模式：使用 `tools/pre-execute` 提供可扩展的拒绝／询问，使用 `tools.guard()` 提供单调拥有方策略，使用 `tools/post-execute` 处理结果决定，并使用 `tools/result` 进行最终观测
 - subagent：在循环外部实现为 `ctx.subagents` 提供方；进程内提供方使用 `ctx.agents.create()` 和拥有的 `AgentHandle` 进行 teardown，而通用的 [`ctx.tasks`](../../tasks/tasks/) 与 [`dsh-tool-subagent`](../../subagent/tool-subagent/) 负责后台收集。
@@ -84,7 +84,7 @@ interface Config {
 
 ### 完整对话请求
 
-#### 模型所见
+#### 模型看到的内容
 
 每个步骤中，循环会发送针对该 agent 呈现的系统提示词、可见工具 schema 和会话派生消息。它提供 `provider`、`model` 与 `cwd` 变量值，但不添加固定文案。
 
@@ -98,7 +98,7 @@ interface Config {
 
 ### 保留的消息历史
 
-#### 模型所见
+#### 模型看到的内容
 
 已接纳的 user 消息、assistant 消息、工具调用与结果、注入上下文和 steering 都会记录，并在后续步骤中发送。原始流分片、生命周期边界和其他仅写入日志的事件会被排除。
 
@@ -108,11 +108,11 @@ interface Config {
 
 #### KV Cache 影响
 
-普通历史增长仅追加，并保留可复用条目。接口替换或压缩会从第一个被遮蔽的历史 token 起使复用失效。
+普通历史增长仅追加，并保留可复用条目。表层替换或压缩会从第一个被遮蔽的历史 token 起使复用失效。
 
 ### 取消后未分发的调用
 
-#### 模型所见
+#### 模型看到的内容
 
 如果后续请求回放一个中止的步骤，取消所阻止分发的每个工具调用都有错误码 `ABORTED_BEFORE_DISPATCH`，结果文本为 `Error: tool call aborted before dispatch`。
 
@@ -124,7 +124,7 @@ interface Config {
 
 仅追加；每个合成结果都位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **分类是一元的**：安全性取决于比较同级调用或资源的调用必须保持独占（参见[设计原理](../../../.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md)）。
 - **配置 label 默认每次新建**：省略 `sessionId` 会在每次启动时创建全新的 `${id}-session-<uuid>`；确切的恢复或创建行为要求显式提供稳定的 `sessionId`，而 `resumeSessionId` 要求已有持久化历史。
diff --git a/packages/core/agent-loop/src/tool-calls.ts b/packages/core/agent-loop/src/tool-calls.ts
index b7ffdaf041..cd57a28f44 100644
--- a/packages/core/agent-loop/src/tool-calls.ts
+++ b/packages/core/agent-loop/src/tool-calls.ts
@@ -2,10 +2,12 @@
  * Schedules one assistant step's tool calls. Exclusive calls form barriers;
  * parallel calls use a bounded rolling pool and are reclassified before start.
  * Dispatch may overlap, while policy, results, and result context remain
- * model-ordered. Abort stops replenishment and drains started calls.
+ * model-ordered. Abort or an internal scheduler failure stops replenishment
+ * and drains started calls.
  *
- * Each advertised call records a balanced `tool/call`/`tool/result` pair. Calls
- * skipped after abort receive synthetic error results so replay stays valid.
+ * Abort records synthetic error results for skipped calls so replay stays
+ * valid. A terminal scheduler failure preserves already-recorded `tool/call`
+ * events without fabricating results.
  * @module dsh-agent-loop/tool-calls
  */
 
@@ -37,10 +39,13 @@ interface GroupOutcome {
 
 /**
  * Schedule one assistant step's tool calls by their live concurrency mode.
- * Started calls receive ordered results. Abort drains them, records synthetic
- * results for unstarted calls, and returns with the signal still aborted after
- * accepting started-call context through the caller-supplied acceptor (the
- * machine stages it on its outbox for the next step boundary).
+ * Ordinary completion and abort commit started-call results in order. Abort
+ * drains them, records synthetic results for unstarted calls, and returns with
+ * the signal still aborted after accepting started-call context through the
+ * caller-supplied acceptor (the machine stages it on its outbox for the next
+ * step boundary). An internal scheduler failure stops new dispatches, drains
+ * already-started dispatches, and rejects with the first failure without
+ * fabricating tool results.
  * The committed step's AgentLoop driver boundary supplies the initiating Agent
  * that becomes each explicit {@link ToolExecutionInput.agent}.
  *
@@ -110,7 +115,8 @@ function parseArguments(raw: string): unknown {
  * drain and remains for the caller's next barrier. Results and contexts commit
  * in model order. Abort stops starts, drains and commits started calls, accepts
  * their contexts into the owning batch, records results for skipped calls, and
- * returns an aborted outcome.
+ * returns an aborted outcome. Scheduler failure drains dispatches without
+ * committing synthetic recovery results.
  */
 async function runGroup(
   ctx: Context,
@@ -131,6 +137,10 @@ async function runGroup(
   let started = 0
   let aborted: boolean = signal.aborted
   let concluded = false
+  let schedulerFailure: { error: unknown } | undefined
+  const throwSchedulerFailure = (): void => {
+    if (schedulerFailure !== undefined) throw schedulerFailure.error
+  }
 
   // `committed` advances only across contiguous model-order slots.
   const commitReady = async (): Promise<void> => {
@@ -157,12 +167,19 @@ async function runGroup(
     callSeqs[index] = appendToolCall(session, turn, step, call.block)
     started++
     const prepared = await ctx.tools[TOOL_REGISTRY_SCHEDULER].prepare(call.exec)
+    throwSchedulerFailure()
     switch (prepared.kind) {
       case 'dispatch': {
-        const promise = ctx.tools[TOOL_REGISTRY_SCHEDULER].dispatch(prepared.exec).then((outcome) => {
-          slots[index] = { exec: prepared.exec, result: outcome.result, needsPost: outcome.kind === 'post-result' }
-          return index
-        })
+        const promise = ctx.tools[TOOL_REGISTRY_SCHEDULER].dispatch(prepared.exec).then(
+          (outcome) => {
+            slots[index] = { exec: prepared.exec, result: outcome.result, needsPost: outcome.kind === 'post-result' }
+            return index
+          },
+          (error: unknown) => {
+            schedulerFailure ??= { error }
+            return index
+          },
+        )
         inFlight.set(index, promise)
         break
       }
@@ -187,24 +204,34 @@ async function runGroup(
         && ctx.tools.executionMode(nextCall.exec).kind !== 'parallel') break
       await startCall(nextToStart)
       nextToStart++
+      throwSchedulerFailure()
       await commitReady()
+      throwSchedulerFailure()
       // Abort may arrive while pre-execute awaits.
       if (signal.aborted) aborted = true
     }
   }
 
-  // Ordered pre-execute may await; only dispatch/body overlaps.
-  // TODO: Drain every started call before rethrowing a scheduler error; tool
-  // bodies must not outlive the failed turn.
-  await fillPool()
-  while (inFlight.size > 0) {
-    const settledIndex = await Promise.race(inFlight.values())
-    inFlight.delete(settledIndex)
-    await commitReady()
-    // Abort may arrive while a tool or ordered commit awaits.
-
-    if (signal.aborted) aborted = true
+  // Ordered pre-execute may await; only dispatch/body overlaps. A scheduler
+  // failure stops new dispatches and reaches the turn boundary after every
+  // already-started dispatch settles.
+  try {
     await fillPool()
+    while (inFlight.size > 0) {
+      const settledIndex = await Promise.race(inFlight.values())
+      inFlight.delete(settledIndex)
+      throwSchedulerFailure()
+      await commitReady()
+      throwSchedulerFailure()
+      // Abort may arrive while a tool or ordered commit awaits.
+
+      if (signal.aborted) aborted = true
+      await fillPool()
+    }
+  } catch (error: unknown) {
+    schedulerFailure ??= { error }
+    await Promise.allSettled(inFlight.values())
+    throw schedulerFailure.error
   }
 
   if (aborted) {
diff --git a/packages/core/agent-loop/tests/tool-calls.spec.ts b/packages/core/agent-loop/tests/tool-calls.spec.ts
index a0266fd3f8..dd086d8e7f 100644
--- a/packages/core/agent-loop/tests/tool-calls.spec.ts
+++ b/packages/core/agent-loop/tests/tool-calls.spec.ts
@@ -9,7 +9,7 @@ import { createUserMessage, CallId, StreamChunk  } from '@deepseek-ai/dsh-llm'
 import SessionStore, { SessionEvent, SessionId } from '@deepseek-ai/dsh-session'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
 import LlmService from '@deepseek-ai/dsh-llm'
-import ToolRegistry, { defineContentToolFixture, TOOL_ABORTED_BEFORE_DISPATCH, type PostToolDecision, type PreToolDecision } from '@deepseek-ai/dsh-tools'
+import ToolRegistry, { defineContentToolFixture, TOOL_ABORTED_BEFORE_DISPATCH, TOOL_REGISTRY_SCHEDULER, type PostToolDecision, type PreToolDecision } from '@deepseek-ai/dsh-tools'
 import AgentRegistry, { type Agent } from '@deepseek-ai/dsh-agent'
 import AgentLoop, { DEFAULT_MAX_PARALLEL_TOOL_CALLS } from '@deepseek-ai/dsh-agent-loop'
 import { MockAdapter, textResponse } from './mock-adapter.ts'
@@ -613,3 +613,66 @@ describe('tool-call scheduler: abort handling', () => {
       })
   })
 })
+
+describe('tool-call scheduler: failure quiescence', () => {
+  it('stops new dispatches and drains started bodies before surfacing the first failure', async () => {
+    const adapter = new MockAdapter([
+      multiCall([
+        { id: 'c1', name: 'p', args: { id: '1' } },
+        { id: 'c2', name: 'p', args: { id: '2' } },
+        { id: 'c3', name: 'p', args: { id: '3' } },
+      ]),
+    ])
+    const ctx = await harness(adapter, 3)
+    const gated = gatedParallelTool('p')
+    ctx.tools.register(gated.tool)
+    // The registry contains expected failures as results; replace its internal
+    // view only to inject the invariant violation this boundary must contain.
+    const scheduler = ctx.tools[TOOL_REGISTRY_SCHEDULER]
+    const prepare = scheduler.prepare.bind(scheduler)
+    const dispatch = scheduler.dispatch.bind(scheduler)
+    const prepareGate = Promise.withResolvers<undefined>()
+    let thirdPrepareEntered = false
+    scheduler.prepare = async (exec) => {
+      const prepared = await prepare(exec)
+      if (exec.callId === CallId('c3')) {
+        thirdPrepareEntered = true
+        await prepareGate.promise
+      }
+      return prepared
+    }
+    const schedulerError = new Error('scheduler exploded')
+    const drainedError = new Error('sibling failed while draining')
+    let rejectFirst: ((error: Error) => void) | undefined
+    scheduler.dispatch = exec => exec.callId === CallId('c1')
+      ? new Promise((_resolve, reject) => { rejectFirst = reject })
+      : dispatch(exec).then(() => { throw drainedError })
+    const agent = ctx.agentLoop.create(SessionId('scheduler-failure'), { provider: 'mock', model: 'mock' })
+    const errors: unknown[] = []
+    ctx.on('agent/error', (subject, _turn, _step, error) => {
+      if (subject === agent) errors.push(error)
+    })
+    let idle = false
+    const idlePromise = waitForIdle(ctx, agent).then(() => { idle = true })
+
+    agent.followup(createUserMessage({ content: [{ type: 'text', text: 'go' }], source: { kind: 'user' } }))
+    await until(() => gated.started.includes('2') && thirdPrepareEntered && rejectFirst !== undefined)
+    rejectFirst?.(schedulerError)
+    await new Promise<void>(resolve => setImmediate(resolve))
+    prepareGate.resolve(undefined)
+    await new Promise<void>(resolve => setImmediate(resolve))
+
+    const startedBeforeDrain = [...gated.started]
+    const idleBeforeDrain = idle
+    const errorsBeforeDrain = [...errors]
+    for (const id of gated.pending()) gated.release(id)
+    await idlePromise
+
+    expect(startedBeforeDrain).toEqual(['2'])
+    expect(idleBeforeDrain).toBe(false)
+    expect(errorsBeforeDrain).toEqual([])
+    expect(gated.pending()).toEqual([])
+    expect(errors).toEqual([schedulerError])
+    expect(errors[0]).toBe(schedulerError)
+  })
+})
diff --git a/packages/core/agent/README.i18n.yaml b/packages/core/agent/README.i18n.yaml
index 59dcdac249..c361cf499b 100644
--- a/packages/core/agent/README.i18n.yaml
+++ b/packages/core/agent/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/core/agent/README.md
 README.md: 9ca79f28506b133a555bd7d1e984386c715fd9d6
-README.zh.md: 165f71f1b395bdf0c229e2c4b1a30e89347b6be1
+README.zh.md: a32f75431a6d7ced64ae2f2171b6aa924d23de3e
diff --git a/packages/core/agent/README.zh.md b/packages/core/agent/README.zh.md
index 165f71f1b3..a32f75431a 100644
--- a/packages/core/agent/README.zh.md
+++ b/packages/core/agent/README.zh.md
@@ -4,7 +4,7 @@
 
 Agent 接口、注册表、进程本地发起方作用域，以及 `agent/*` 事件词汇。每个插件（UI、钩子、编排器）都面向此处定义的 `Agent` handle 编程；它不依赖循环，因此循环可以替换。
 
-可选配套包 `@deepseek-ai/dsh-agent/invariant` 会向 `ctx.invariants` 注册此包的 agent 状态转换检查。根 agent 服务不会隐式加载诊断。
+可选配套包（package）`@deepseek-ai/dsh-agent/invariant` 会向 `ctx.invariants` 注册此包的 agent（智能体）状态转换检查。根 agent 服务不会隐式加载诊断。
 
 ## 服务：`AgentRegistry`（ctx 键：`agents`）
 
@@ -12,11 +12,11 @@ Agent 接口、注册表、进程本地发起方作用域，以及 `agent/*` 事
 
 ### 公开 API
 
-带作用域的注册表层：`Agent.ctx` 是 agent 的作用域上下文（`dsh-scope`，键 = 该 agent）。通过它注册工具／段／变量／监听器，只对该 agent 生效，并在释放时全部撤销。`agentEvents(ctx, agent)` 是普通 agent 主体操作的融合分发器（一次完成载体 + 注入主体）；其通知 mode 会调用每个监听器，并同时收容同步抛出和返回 Promise 的拒绝。注册表生命周期对复用一个稳定路由载体。`assembleContextFor(agent)` 构建按 agent 的组装上下文（同时包含 `agent` + `scope`）。`installAgentLlmTarget(agentCtx, target)` 在提示词组装期间快照可变的提供方／模型／推理（reasoning）强度选择，将路由应用到提示词变量，并将完整目标应用到一个步骤的请求路由；如果没有选定推理强度，则会清除继承的推理强度，使该目标使用适配器／提供方默认值。`CreateAgentOptions.setup(agentCtx)` 和 `ResumeAgentOptions.setup(agentCtx)` 在新建或恢复的 agent 尚未发布时，组合其带作用域的世界。Setup 是受信任、仅用于组合的同进程代码：只有创建完成后才能驱动 agent。
+带作用域的注册接口：`Agent.ctx` 是 agent 的作用域上下文（`dsh-scope`，键 = 该 agent）。通过它注册工具／段／变量／监听器，只对该 agent 生效，并在 dispose（资源释放）时全部撤销。`agentEvents(ctx, agent)` 是普通 agent 主体操作的融合分发器（一次完成载体 + 注入主体）；其通知 mode 会调用每个监听器，并同时收容同步抛出和返回 Promise 的拒绝。注册表生命周期对复用一个稳定路由载体。`assembleContextFor(agent)` 构建按 agent 的组装上下文（同时包含 `agent` + `scope`）。`installAgentLlmTarget(agentCtx, target)` 在提示词组装期间快照可变的提供方／模型／推理（reasoning）强度选择，将路由应用到提示词变量，并将完整目标应用到一个步骤的请求路由；如果没有选定推理强度，则会清除继承的推理强度，使该目标使用适配器／提供方默认值。`CreateAgentOptions.setup(agentCtx)` 和 `ResumeAgentOptions.setup(agentCtx)` 在新建或恢复的 agent 尚未发布时，组合其带作用域的世界。Setup 是受信任、仅用于组合的同进程代码：只有创建完成后才能驱动 agent。
 
-`AgentOptions` 提供初始的提供方／模型路由，以及可选的正整数 `maxTokens` 输出上限。实体循环会把该上限记录到请求 header，并应用到每次对话模型请求；调用方省略时由提供方默认值控制。
+`AgentOptions` 提供初始的提供方／模型路由，以及可选的正数 `maxTokens` 输出上限。实体循环会把该上限记录到请求 header，并应用到每次对话模型请求；调用方省略时由提供方默认值控制。
 
-- `ctx.agents.register(agent: Agent): () => void`：记录一个 **已经构造完成** 的 agent。随调用 fiber 释放。
+- `ctx.agents.register(agent: Agent): () => void`：记录一个 **已经构造完成** 的 agent。随调用 fiber dispose。
 - 高级有序生命周期：`enter(agent, owner): () => void` 强制 `agent.id === agent.session.id`，执行权威 ID 冲突检查，并在不通知的情况下插入；`owner` 显式记录实时创建方 agent 关系（根 agent 为 `undefined`），与持久会话谱系无关。`announce(agent)` 恰好发出一次 `agent/created`。创建监听器同步请求的 detach 会延后到该次分发结束；每次 detach 都会检查捕获的条目对象，因此陈旧能力无法删除后续使用同一 ID 的替代项。异步工厂使用这一拆分；普通插件使用 `register()`。
 - `ctx.agents.get(id: SessionId): Agent | undefined`
 - `ctx.agents.isOwnedBy(id: SessionId, owner: Agent): boolean`：该确切实时条目是否通过父 agent 的作用域上下文创建；运行时所有权与持久会话谱系无关。
@@ -32,41 +32,41 @@ Agent 接口、注册表、进程本地发起方作用域，以及 `agent/*` 事
 - `ctx.agents.withInitiator(agent, operation)`：使用一个确切 Agent 运行，并保留操作的确切同步值或 Promise。
 - `ctx.agents.withoutInitiator(operation)`：对无关的进程本地工作隐藏继承的发起方。
 
-该作用域携带 `Agent` 本身，并且只在进程内有效。环境中的身份既不是存活证明，也不是授权；在服务、worker、进程、持久化和 wire 边界，显式 Agent 字段仍是权威来源。Teardown 会拒绝新边界，允许注入的依赖方和返回 Promise 的边界 drain，然后禁用底层 `AsyncLocalStorage`；未返回的工作仍归将其分离的子系统所有。如果某个边界继承的异步链开始卸载一个拥有它的 Cordis fiber，该嵌套边界链会从 drain 中释放，使卸载不会等待自身；其 continuation 会在 teardown 后观察到已释放的服务。详细边界与 teardown 契约由[发起方作用域决策](../../../.agents/notes/implemented/architecture/2026-07-15-agent-initiator-scope.md)拥有。
+该作用域携带 `Agent` 本身，并且只在进程内有效。环境中的身份既不是存活证明，也不是授权；在服务、worker、进程、持久化和 wire 边界，显式 Agent 字段仍是权威来源。Teardown 会拒绝新边界，允许注入的依赖方和返回 Promise 的边界 drain，然后禁用底层 `AsyncLocalStorage`；未返回的工作仍归将其分离的子系统所有。如果某个边界继承的异步链开始卸载一个拥有它的 Cordis fiber，该嵌套边界链会从 drain 中释放，使卸载不会等待自身；其 continuation 会在 teardown 后观察到已 dispose 的服务。详细边界与 teardown 契约由[发起方作用域决策](../../../.agents/notes/implemented/architecture/2026-07-15-agent-initiator-scope.md)拥有。
 
 #### 工厂 seam（创建）
 
 Agent *创建* 由实现 `AgentFactory` 的插件（`dsh-agent-loop`）提供，并通过 `setFactory` 注册。这样，创建功能留在 `dsh-agent` 接口上，消费方（UI、ACP 桥接层）可以面向 `ctx.agents` 编程，而不依赖具体循环包。注册表会把已经 traced 的 Service 规范化为具体目标，并通过调用方上下文重新 trace 每次调用；这既避免嵌套 Cordis shadow，也会把显式、绑定调用方的 `ownerCtx` 传给普通工厂。
 
-- `ctx.agents.setFactory(factory: AgentFactory): () => void`：注册创建工厂（循环在构造时调用）。第二个工厂会导致抛出；释放时清空槽位。
+- `ctx.agents.setFactory(factory: AgentFactory): () => void`：注册创建工厂（循环在构造时调用）。第二个工厂会导致抛出；dispose 时清空槽位。
 - `ctx.agents.create(options: CreateAgentOptions): Promise<AgentHandle>`：创建会话和 agent，在不发布的情况下等待可选 setup，然后通过最终的 `SessionStore.enter()` 与 `AgentRegistry.enter()` 检查发布。不支持并发创建同一 ID：多个操作可以进行准备，但只有一个能进入；每个失败方都会回滚其私有作用域／会话／驱动器。可选且只用于创建的 `signal` 会取消未发布的 setup，并在返回 handle 前分离；之后的取消使用 `handle.dispose()` 或 `agent.cancel()`。发布包含在回滚范围内，回滚期间每条已交付创建边都会成对处理。未注册工厂时拒绝。
 - `ctx.agents.resume(options: ResumeAgentOptions): Promise<AgentHandle>`：加载持久化会话（[会话持久化](../../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md)），创建新的未发布 agent 作用域，等待可选 setup，并使用相同的最终进入发布序列。其可选 `signal` 同样只用于创建。未注册工厂或未配置会话持久化时拒绝。
 
-`AgentHandle = { agent: Agent; dispose(): Promise<void> }`。Disposer 是一项 **消费方能力**；仅持有裸注册表条目的观察方不能 teardown agent。调用方 fiber 和已注册工厂提供方是结构化共同拥有者：调用方卸载会强制结构化所有权，而工厂卸载必须停止旧实例，因为它们的作用域依赖表层属于该提供方。任意拥有者调用 `dispose()` 都会到达同一个记忆化静默边界：它停止循环，等待循环退出，注销 agent，从存储中移除其会话，最后撤销其作用域世界。`ctx.agents.get(id)` 仍返回裸 `Agent`；ACP 桥接层与进程内 subagent 后端持有消费方 handle，而配置创建的 agent 已由循环 fiber 拥有。
+`AgentHandle = { agent: Agent; dispose(): Promise<void> }`。Disposer 是一项 **消费方能力**；仅持有裸注册表条目的观察方不能 teardown agent。调用方 fiber 和已注册工厂提供方是结构化共同拥有者：调用方卸载会强制结构化所有权，而工厂卸载必须停止旧实例，因为它们的作用域依赖范围属于该提供方。任意拥有者调用 `dispose()` 都会到达同一个记忆化完全停稳边界：它停止循环，等待循环退出，注销 agent，从存储中移除其会话，最后撤销其作用域世界。`ctx.agents.get(id)` 仍返回裸 `Agent`；ACP 桥接层与进程内 subagent 后端持有消费方 handle，而配置创建的 agent 已由循环 fiber 拥有。
 
 ### 实时事件
 
 `dsh-agent` 声明实时 `agent/*` 协调词汇，使插件不必依赖具体循环。确切签名、分发 mode、作用域筛选规则与 payload 契约位于生成的 [Cordis 事件目录](../../../docs/cordis-catalog/events.md)；[架构轮次流](../../../docs/architecture.md#turn-flow) 展示它们与持久会话事件的相对顺序。
 
-生命周期边有两个重要的本地注意事项。`agent/created` 在作用域 setup 之后、会话与 agent 注册表条目都存在之后运行。Setup 是受信任、仅用于组合的代码；紧随其后且不可 veto 的 `agent/session-start` 通知是第一个受支持的启动注入点。`agent/disposed` 始终表示确切 agent 已离开注册表。AgentLoop 在其驱动器静默后发出该事件，而有序 teardown 此时可能仍在分离会话并撤销作用域；直接注册的自定义 agent 自行拥有任何更强的驱动器顺序契约。
+生命周期边有两个重要的本地注意事项。`agent/created` 在作用域 setup 之后、会话与 agent 注册表条目都存在之后运行。Setup 是受信任、仅用于组合的代码；紧随其后且不可 veto 的 `agent/session-start` 通知是第一个受支持的启动注入点。`agent/disposed` 始终表示确切 agent 已离开注册表。AgentLoop 在其驱动器完全停稳后发出该事件，而有序 teardown 此时可能仍在分离会话并撤销作用域；直接注册的自定义 agent 自行拥有任何更强的驱动器顺序契约。
 
-大多数拦截点都是协作式 waterfall。轮次作用域的异步 seam 接收一个显式 `AbortSignal`，其中 `signal` 紧邻 waterfall 最终的 `next`；监听器可以配合，但不得将它保留为控制另一轮次的权限。`agent/step` 是派生请求前的串行检查点，而 `agent/request-error` 是失败模型请求的恢复 waterfall：失败步骤关闭后，它接收确切错误、规范化失败事实和信号。拥有恢复权的监听器返回 `{ kind: 'retry' }` 且不调用 `next()`；循环会关闭失败轮次，并打开一个编号重试轮次。`agent/turn-stopping` 在本可完成的轮次关闭前运行。普通排队提示词保持原样。有效的广义取消会先发出只观测的 `agent/cancel-requested` 及其解析后的类型化原因，再清空队列并中止；通知失败会被收容，不能 veto 停止。信号生命周期由[显式取消决策](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)拥有；作用域分发与终止结算由 [agent 作用域 runtime 设计 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-12-agent-scope-runtime-design.md#three-execution-boundaries-are-deliberately-one-way)拥有。
+大多数拦截点都是协作式 waterfall（瀑布式事件）。轮次作用域的异步 seam 接收一个显式 `AbortSignal`，其中 `signal` 紧邻 waterfall 最终的 `next`；监听器可以配合，但不得将它保留为控制另一轮次的权限。`agent/step` 是派生请求前的串行检查点，而 `agent/request-error` 是失败模型请求的恢复 waterfall：失败步骤关闭后，它接收确切错误、规范化失败事实和信号。拥有恢复权的监听器返回 `{ kind: 'retry' }` 且不调用 `next()`；循环会关闭失败轮次，并打开一个编号重试轮次。`agent/turn-stopping` 在本可完成的轮次关闭前运行。普通排队提示词保持原样。有效的广义取消会先发出只观测的 `agent/cancel-requested` 及其解析后的类型化原因，再清空队列并中止；通知失败会被收容，不能 veto 停止。信号生命周期由[显式取消决策](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)拥有；作用域分发与终止结算由 [agent 作用域 runtime 设计 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-12-agent-scope-runtime-design.md#three-execution-boundaries-are-deliberately-one-way)拥有。
 
 `PromptDecision.additionalContexts` 是由带标识且冻结的 `UserMessage` 值组成的数组，因此每个上下文都保留自己的标识和来源。获准的提示词与每个附加上下文都会在轮次运行前成为各自独立、面向模型的 `user/message` 事件。包装下游允许决策的监听器会保留其 `content` 与 `additionalContexts`，除非有意替换任一字段；替换获准内容时仍会保留提示词的标识。
 
-轮次和步骤边界以及模型 token 流是持久 `session/event` 事实，而不是镜像的 `agent/*` 通知。消费方从会话 feed 读取 `turn/*`、`step/*` 和 `assistant/chunk`；工具策略与结果观测属于 [`dsh-tools`](../tools/README.md) 记录的完整流水线。
+轮次和步骤边界以及模型 token 流是持久 `session/event` 事实，而不是镜像的 `agent/*` 通知。消费方从会话事件流读取 `turn/*`、`step/*` 和 `assistant/chunk`；工具策略与结果观测属于 [`dsh-tools`](../tools/README.md) 记录的完整流水线。
 
 ### Agent 接口（`types.ts`）
 
 每个插件面向的 handle：
 
-- `agent.send(message, options)`：覆盖（`target` × `wakeup`）矩阵的唯一投递原语。`message` 是已有标识且已冻结的 `UserMessage`；调用方通常会在开始路由前使用 `createUserMessage()` 创建它。`SendOptions` 只持有 `target` 与 `wakeup` 策略。agent 会原样发布或排队完整值，不会生成或替换其标识。该消息的 `agent/inbox/enqueue`/`dequeue`/`discard` 事件会携带完整消息，调用方可据此把排队项与其生命周期关联；入队与出队事件还会携带解析出的 `queued | steering` 路由归类，使重复出现的消息标识能在正确的 FIFO 中完成结算。`target: 'next-turn'` 排队一条独立 FIFO 项，获准后成为其轮次中唯一的普通提示词。`target: 'next-step'` 且 `wakeup: true` 提交 steering（中途引导），而 `target: 'next-step'` 且 `wakeup: false` 注入持久上下文，不运行模型。轮次原理由 [one-send-one-turn Agent Note](../../../.agents/notes/implemented/simplification/2026-07-17-one-send-one-turn.md)拥有。
+- `agent.send(message, options)`：覆盖（`target` × `wakeup`）矩阵的唯一投递原语。`message` 是已有标识且已冻结的 `UserMessage`；调用方通常会在开始路由前使用 `createUserMessage()` 创建它。`SendOptions` 只持有 `target` 与 `wakeup` 策略。agent 会原样发布或排队完整值，不会生成或替换其标识。该消息的 `agent/inbox/enqueue`/`dequeue`/`discard` 事件会携带完整消息，调用方可据此把排队项与其生命周期关联；入队与出队事件还会携带解析出的 `queued | steering` 路由归类，使重复出现的消息标识能在正确的 FIFO 中完成结算。`target: 'next-turn'` 在 FIFO 中排入一个独立项，获准后成为其轮次中唯一的普通提示词。`target: 'next-step'` 且 `wakeup: true` 提交 steering（中途引导），而 `target: 'next-step'` 且 `wakeup: false` 注入持久上下文，不运行模型。轮次原理由 [one-send-one-turn Agent Note](../../../.agents/notes/implemented/simplification/2026-07-17-one-send-one-turn.md)拥有。
 - `agent.followup(input)`：`send()` 的 `next-turn`／wakeup 预设：排队一个普通后续轮次并唤醒驱动器。
 - `agent.steer(input)`：`next-step`／wakeup 预设：提示词接纳期间或轮次打开时，为下一个安全边界暂存 steering，且不分发 `agent/prompt-submit`；该接收窗口之外则委托给会唤醒的后续轮次。接纳失败会保留暂存的 steering，以供重试或之后获准的提示词使用，而取消或 dispose 可能丢弃它。
 - `agent.inject(input)`：`next-step`／不唤醒预设：追加面向模型的上下文而不运行模型；下一次请求会看到一条逐字的 user role 消息，其来源由必填的 `input.source` 携带。提示词接纳期间或轮次打开时，注入会在 outbox 中等待下一个安全边界。该接收窗口之外，它会立即追加而不开启轮次；如果接纳结束却未开启轮次，仅含上下文的接纳批次会采用这一回退，而与 steering 一同暂存的上下文则会随其继续待处理。持久化独立地响应 `session/event`。注入不发出 `agent/inbox/*` 事件。
 - `agent.acceptsNextStep`：当前发送 `next-step` 时，是否会加入提示词接纳或已打开的轮次。当调用方必须在 steering 与新接纳的提示词之间选择时，应使用这一更窄的路由判定；`status === 'running'` 还涵盖接纳收尾与轮次结算阶段。
 - `agent.cancel(cause, options?)`：取消活动轮次，并在未设置 `options.keepInbox` 时取消全部待处理工作。调用方必须显式选择 `user | parent` 原因；活动持有者会在中止前把其判别字段复制为已分离、冻结的信号原因。有效调用会在清除排队与 steering 工作前，随原因发出 `agent/cancel-requested`；丢弃项在 `agent/inbox/discard` 上报告，观察方可以同步状态，但不能 veto 取消。`keepInbox: true` 会中止轮次，但保留排队与 steering 项（不丢弃，且不删除尚未开始的工作）。同进程类型化 seam 不会为无类型调用方添加运行时校验或兼容回退。重复取消活动轮次时，首个信号生效；空闲取消是安全空操作，不发通知。ACP 映射到 `user`，进程内父传播映射到 `parent`。原因只存在于运行时；持久 `turn/end` 保持粗粒度的 `aborted`。
-- `agent.whenIdle()`：agent 从 `running` 结算后达到静默时解析（idle ⇒ 立即；disposed ⇒ 等待循环退出）。这是非拥有者的静默观测钩子：观察工作结算，但不 teardown agent。Teardown 独立存在；生命周期拥有者通过 `AgentHandle.dispose()` 停止并注销，并直接等待循环退出。
+- `agent.whenIdle()`：agent 从 `running` 结算后达到完全停稳时解析（idle ⇒ 立即；disposed ⇒ 等待循环退出）。这是非拥有者的完全停稳观测钩子：观察工作结算，但不 teardown agent。Teardown 独立存在；生命周期拥有者通过 `AgentHandle.dispose()` 停止并注销，并直接等待循环退出。
 - `agent.session`、`agent.status`、`agent.options`、`agent.id`
 
 `running` 描述驱动器范围的 drain 区间，而不是轮次仍打开的证明；它可以覆盖轮次关闭、持久性检查点和连续的排队轮次。
@@ -81,13 +81,13 @@ Agent *创建* 由实现 `AgentFactory` 的插件（`dsh-agent-loop`）提供，
 
 ### 用户、steering 与注入消息
 
-#### 模型所见
+#### 模型看到的内容
 
 `send`、`steer` 与 `inject` 会向所属会话提供输入。`agent/prompt-submit`、`agent/step` 和其他已声明事件让插件能够阻止提示词或添加持久请求材料；此接口本身不贡献固定文案。
 
 #### Token 影响
 
-已接受内容成为保留历史或重复会话前缀；被阻止内容不贡献请求 token。大小取决于调用方与插件。
+已接受内容成为保留历史，或成为每次请求都会重复的会话前缀；被阻止内容不贡献请求 token。大小取决于调用方与插件。
 
 #### KV Cache 影响
 
@@ -95,21 +95,21 @@ Agent *创建* 由实现 `AgentFactory` 的插件（`dsh-agent-loop`）提供，
 
 ### Agent 作用域的请求组合
 
-#### 模型所见
+#### 模型看到的内容
 
 通过 `agent.ctx` 进行的注册可以遮蔽提示词段或工具，也可以在未发布 setup 期间安装仅适用于该 agent 的拦截器。
 
 #### Token 影响
 
-此包自身不增加 token；带作用域贡献只影响该 agent，并在释放时消失。
+此包自身不增加 token；带作用域贡献只影响该 agent，并在 dispose 时消失。
 
 #### KV Cache 影响
 
 只要 agent 的作用域注册不变，前缀就保持稳定。改变提示词段、工具定义或请求监听器的 setup 或 reload，可能从第一个受影响的请求 token 起使复用失效。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
-- **发起方作用域只存在于进程内**：worker、子进程、HTTP、持久队列和重启会显式物化所需身份。
+- **发起方作用域只存在于进程内**：worker、子进程、HTTP、持久队列和重启必须显式传递所需身份。
 - **环境身份可能比存活状态更久**：消费方在生命周期敏感工作前，仍要检查 `agent.status`、取消状态和所属能力契约。
 - **委派以外的 agent 间通道**：共享状态、流式子输出和后台／轮询语义仍在当前同步 `ctx.subagents` seam 之外。
 - **`agent/session-start` 不能为启动设置门禁**：它仍是同步且不可 veto 的通知；必须在发布前完成的异步组合属于工厂的 `setup(agentCtx)` 事务。
diff --git a/packages/core/scope/README.i18n.yaml b/packages/core/scope/README.i18n.yaml
index 734e6a52a7..f4050f251f 100644
--- a/packages/core/scope/README.i18n.yaml
+++ b/packages/core/scope/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/core/scope/README.md
 README.md: 4f32573779a15e8c34b4936bfe75549dfc86d9f6
-README.zh.md: 96c7d740bd3945558413e5a2763ddaaea5b85f2c
+README.zh.md: b060d8f8e44a28e717ee4724249b797941388798
diff --git a/packages/core/scope/README.zh.md b/packages/core/scope/README.zh.md
index 96c7d740bd..b060d8f8e4 100644
--- a/packages/core/scope/README.zh.md
+++ b/packages/core/scope/README.zh.md
@@ -2,14 +2,14 @@
 
 [English](README.md) | 中文
 
-带作用域的注册原语。`createScope(ctx, key)` 创建一个带标签的 Cordis 上下文，其底层 fiber 拥有通过该上下文进行的每项注册。`scopeOf(ctx)` 读取标签；`scopeTarget(base, key)` 将带作用域的事件路由到键相同的监听器，同时让无作用域监听器保持全局可见。Agent loop 为每个实时 agent 创建一个作用域，但该机制与键的具体含义无关，因此底层包无需依赖 agent 即可使用。
+带作用域的注册原语。`createScope(ctx, key)` 创建一个带标签的 Cordis 上下文，其底层 fiber 拥有通过该上下文进行的每项注册。`scopeOf(ctx)` 读取标签；`scopeTarget(base, key)` 将带作用域的事件路由到键相同的监听器，同时让无作用域监听器保持全局可见。agent loop（智能体循环）为每个实时 agent 创建一个作用域，但该机制与键的具体含义无关，因此底层包（package）无需依赖 agent 即可使用。
 
 ## 公开 API
 
-- `createScope(ctx: Context, key: ScopeKey): Scope`：在 `ctx` 的 fiber 下创建作用域。可以同步使用（effect 收集受 uid 门禁约束；服务解析会沿创建该作用域的插件依赖表层继续查找）。同进程、带类型的键受信任；处于非活动状态的创建上下文仍会通过 Cordis 失败（`INACTIVE_EFFECT`）。
+- `createScope(ctx: Context, key: ScopeKey): Scope`：在 `ctx` 的 fiber 下创建作用域。可以同步使用（effect 收集受 uid 门禁约束；服务解析会沿创建该作用域的插件依赖范围继续查找）。同进程、带类型的键受信任；处于非活动状态的创建上下文仍会通过 Cordis 失败（`INACTIVE_EFFECT`）。
 - `Scope.ctx`：带标签的上下文。通过它进行的注册既具备作用域可见性，也服从作用域生命周期。派生上下文（一次 `extend`、挂载于其下的 fiber）继承标签；嵌套作用域会遮蔽外层标签（最近的标签生效）。
 - `Scope.rawDispose`：底层 fiber 的原样 Cordis disposer。组合式（generator）effect 会 yield 此函数，从而把作用域 teardown 嵌套在该 yield 位置（Cordis 按函数标识去重嵌套 effect；yield 一个包装函数会使作用域 teardown 成为并行的同级操作）。
-- `Scope.dispose(): Promise<void>`：通过作用域进行的每项注册所共用的幂等静默边界。竞态调用或重复调用会等待同一次 teardown；即使 `rawDispose` 先调用了底层单次 Cordis disposer 也是如此。
+- `Scope.dispose(): Promise<void>`：通过作用域进行的每项注册所共用的幂等完全停稳边界。竞态调用或重复调用会等待同一次 teardown；即使 `rawDispose` 先调用了底层单次 Cordis disposer 也是如此。
 - `scopeOf(ctx: Context): ScopeKey | undefined`：上下文或其任意派生上下文携带的标签；`undefined` 表示上下文全局。
 - `scopeTarget(base: T, key: ScopeKey | undefined): Scoped<T>`：为按作用域筛选的事件构造不透明分发 `thisArg`。它把 `base` 现有的 `Context.filter` 与作用域谓词组合起来（无标签监听器 ⇒ 放行；有标签监听器 ⇒ 仅当标签 === key 时放行；`key === undefined` ⇒ 仅放行无标签监听器）。载体只包含路由状态；真实主体由事件参数携带。带 `{ global: true }` 的监听器绕过筛选（Cordis 语义）。
 - `Scoped<T>`：编译期不透明载体 brand。按作用域筛选的事件要求它作为 `this` 类型，因此使用裸主体分发会产生编译错误。类型参数记录主体类型，但不公开其属性。
@@ -19,18 +19,18 @@
 - `NamedEntries<V>`：按插入顺序排列的具名存储，调用方拥有重复项诊断、查找，以及一个非空表世代内的实时迭代。表清空后，现有迭代器与后续插入项脱离；`insert()` 返回幂等的精确条目撤销函数。
 - `AnonymousEntries<V>`：按插入顺序排列的匿名存储；唯一内部键使相同值仍作为独立注册存在。它使用相同的清空世代迭代器边界；`append()` 返回幂等的精确条目撤销函数。
 
-可选配套包 `@deepseek-ai/dsh-scope/invariant` 拥有该运行时断言。它使用生成的 `scoped-events.generated.ts` 解析器映射，要求每个已声明的带作用域事件都携带载体；当 payload 公开路由主体时，还要求主体与载体键标识相同。基于 Program 的生成器根据事件声明和真实的 `scopeTarget(base, key)` 调用生成该映射。
+可选配套包 `@deepseek-ai/dsh-scope/invariant` 拥有该运行时断言。它使用生成的 `scoped-events.generated.ts` 解析器映射，要求每个已声明的带作用域事件都携带载体；当 payload 公开路由主体时，还要求路由主体与载体键严格相等。基于 Program 的生成器根据事件声明和真实的 `scopeTarget(base, key)` 调用生成该映射。
 
 ## 设计契约
 
-注册上下文同时决定可见性和所有权，防止注册在一个作用域中可见、却随另一个作用域释放。作用域用于路由受信任的同进程插件；它们不是沙箱或权限边界。原理与明确排除的安全目标见 [agent 作用域 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-08-agent-scope-contexts.md#security-and-authority-are-explicit-non-goals)。
+注册上下文同时决定可见性和所有权，防止注册在一个作用域中可见、却随另一个作用域 dispose（资源释放）。作用域用于路由受信任的同进程插件；它们不是沙箱或权限边界。原理与明确排除的安全目标见 [agent 作用域 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-08-agent-scope-contexts.md#security-and-authority-are-explicit-non-goals)。
 
-感知作用域的服务会定义具体 `ScopeLayer`，聚合各自不同的表与领域辅助函数。`ScopedLayers.effect()` 接受一个返回同步撤销函数的同步动作，在可选通知前安装该撤销函数，并且只有在完整聚合为空时才回收精确作用域层。`notify` 默认为 `true`；所提供的回调拥有决定观测方失败是抛出还是受控的职责。`EntryValues` 保持内部可见；存储类从包根而非 `/store` 子路径导入；共享存储不定义注册表专属的筛选或迭代策略。详见[共享作用域层存储 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-12-scoped-layers-store.md)。
+感知作用域的服务会定义具体 `ScopeLayer`，聚合各自不同的表与领域辅助函数。`ScopedLayers.effect()` 接受一个返回同步撤销函数的同步动作，在可选通知前安装该撤销函数，并且只有在完整聚合为空时才回收精确作用域层。`notify` 默认为 `true`；由所提供的回调决定观测方失败是向外抛出还是在内部处理。`EntryValues` 保持内部可见；存储类从包根而非 `/store` 子路径导入；共享存储不定义注册表专属的筛选或迭代策略。详见[共享作用域层存储 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-12-scoped-layers-store.md)。
 
-交出带作用域的上下文，也会交出创建该上下文的插件所能解析的服务表层（解析会沿创建者 fiber 的依赖链，而非持有者的依赖链行进），因此应由具备这些带作用域注册所需依赖的插件来创建它。
+交出带作用域的上下文，也会交出创建该上下文的插件的服务解析范围（解析会沿创建者 fiber 的依赖链，而非持有者的依赖链行进），因此应由具备这些带作用域注册所需依赖的插件来创建它。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - **只有感知作用域的表层才会隔离状态**：注册表必须按 `scopeOf()` 归档，事件必须通过 `scopeTarget()` 分发；仅仅通过带作用域的上下文调用任意 Cordis 服务，并不会改变该服务仍为上下文全局这一事实。
 - **一个上下文只携带一个最近的作用域键**：嵌套作用域会遮蔽父作用域的标签，而不会形成层级策略集或多成员策略集。
-- **服务可达性来自作用域创建者**：交出 `Scope.ctx` 也会交出创建插件注入的服务表层，因此持有者无法再收窄一个较宽的创建者表层。
+- **服务可达性来自作用域创建者**：交出 `Scope.ctx` 也会交出创建插件注入的服务表层，因此，若作用域创建者提供的服务范围较宽，持有者之后也无法将其收窄。
diff --git a/packages/core/system-prompt/README.i18n.yaml b/packages/core/system-prompt/README.i18n.yaml
index 9d82f19bfd..1d1430b228 100644
--- a/packages/core/system-prompt/README.i18n.yaml
+++ b/packages/core/system-prompt/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/core/system-prompt/README.md
 README.md: 79badba0b84b27c01f25e9c31b5df78c556411ea
-README.zh.md: 1d983e44721dbc637efc10824065aa3b88087e1d
+README.zh.md: fb3ed08bc9ea546033acac3b577980f500ce243d
diff --git a/packages/core/system-prompt/README.zh.md b/packages/core/system-prompt/README.zh.md
index 1d983e4472..fb3ed08bc9 100644
--- a/packages/core/system-prompt/README.zh.md
+++ b/packages/core/system-prompt/README.zh.md
@@ -2,22 +2,22 @@
 
 [English](README.md) | 中文
 
-系统提示词组装注册表。插件贡献有序段、工具 schema 和具名变量。循环在每个步骤组装一次，并将结果渲染为完整模型提示词。此插件拥有静态 harness 身份和全局部署 persona；agent 作用域的 persona 会遮蔽全局默认值。
+系统提示词组装注册表。插件贡献有序段、工具 schema 和具名变量。循环在每个步骤组装一次，并将结果渲染为完整的模型提示词。此插件拥有静态 harness 身份和全局部署 persona；agent（智能体）作用域的 persona 会遮蔽全局默认值。
 
 ## 配置
 
 | 键 | 默认值 | 含义 |
 |---|---|---|
-| `persona` | `''` | 全局部署 persona 默认值：唯一由配置创作的提示词片段，渲染为顺序为 0 的 `deployment:persona` 段，除非 agent 作用域的贡献将其遮蔽。它是模板，完整的 `{{…}}` 组会严格按已注册变量解释（已交付循环注册 `{{model}}`/`{{cwd}}`），目前没有表达字面量花括号的转义语法。为空 ⇒ 渲染时删除该段。 |
-| `toolOrder` | 无 | 显式的面向模型工具顺序：一个 `ToolSchema.name` 列表，包含一个 `'<unlisted-tools>'` 其余项（`TOOL_ORDER_REST`）。已列工具占据列出的位置；未列工具按名称字典序落在其余项位置。缺席 ⇒ 直接按名称字典序排列。在 `system-prompt/assemble` waterfall 之前应用于已收集工具；与段的 `order` 排序一样，它会规范化注册表贡献的内容（注册顺序是插件加载工件），而修改列表的 waterfall 监听器拥有其输出的确定性。配置错误会明确失败：列表没有恰好一个其余项或存在重复项，会在加载时抛出；已列名称没有对应已注册工具，会使每次 `assemble()` 被拒绝；工具提供方返回保留的其余项名称也会被拒绝。在已交付循环下，轮次会在任何模型请求前失败。为何采用中心列表而非每插件权重，见[显式面向模型工具顺序](../../../.agents/notes/implemented/feature/2026-07-06-explicit-tool-order.md)。 |
+| `persona` | `''` | 全局部署 persona 默认值：唯一由配置创作的提示词片段，渲染为顺序为 0 的 `deployment:persona` 段，除非 agent 作用域的贡献将其遮蔽。它是模板，完整的 `{{…}}` 组会严格按已注册变量解释（随附循环注册 `{{model}}`/`{{cwd}}`），目前没有表达字面量花括号的转义语法。为空 ⇒ 渲染时删除该段。 |
+| `toolOrder` | 无 | 显式的面向模型工具顺序：一个 `ToolSchema.name` 列表，包含一个 `'<unlisted-tools>'` 其余项（`TOOL_ORDER_REST`）。已列工具占据列出的位置；未列工具按名称字典序落在其余项位置。缺席 ⇒ 直接按名称字典序排列。在 `system-prompt/assemble` waterfall（瀑布式事件）之前应用于已收集工具；与段的 `order` 排序一样，它会规范化注册表贡献的内容（注册顺序是插件加载产物），而修改列表的 waterfall 监听器拥有其输出的确定性。配置错误会明确失败：列表没有恰好一个其余项或存在重复项，会在加载时抛出；已列名称没有对应已注册工具，会使每次 `assemble()` 被拒绝；工具提供方返回保留的其余项名称也会被拒绝。在随附循环下，轮次会在任何模型请求前失败。为何采用中心列表而非每插件权重，见[显式面向模型工具顺序](../../../.agents/notes/implemented/feature/2026-07-06-explicit-tool-order.md)。 |
 
 ## 服务：`SystemPrompt`（ctx 键：`systemPrompt`）
 
 ### 公开 API
 
-- `ctx.systemPrompt.section(section: PromptSection): () => void`：贡献一个段。层由调用上下文的作用域决定：`agent.ctx` 只为该 agent 贡献，并在该处遮蔽同名全局段。同一层中的重复名称和非有限顺序会抛出。随调用 fiber 释放。
-- `ctx.systemPrompt.tools(provider: (context: AssembleContext) => ToolProviderResult): () => void`：贡献工具 schema；每次组装时使用该次组装的上下文求值。`ToolProviderResult` = `{ schemas, knownNames? }`：`schemas` 是限制后的可见集合；`knownNames` 是限制前由 `toolOrder` 使用的全集。提供方不得返回名为 `TOOL_ORDER_REST` 的 schema。带作用域提供方只在其作用域的组装中查询。随调用 fiber 释放。
-- `ctx.systemPrompt.variable(name: string, provider: (context) => string | undefined): () => void`：贡献提示词变量，在段文本中以 `{{name}}` 引用。带作用域变量会为该 agent 遮蔽同名全局变量。同层重复或无法引用的名称会抛出；`undefined` 表示「本次组装没有值」。随调用 fiber 释放。
+- `ctx.systemPrompt.section(section: PromptSection): () => void`：贡献一个段。层由调用上下文的作用域决定：`agent.ctx` 只为该 agent 贡献，并在该处遮蔽同名全局段。同一层中的重复名称和非有限顺序会抛出。随调用 fiber 一并 dispose（资源释放）。
+- `ctx.systemPrompt.tools(provider: (context: AssembleContext) => ToolProviderResult): () => void`：贡献工具 schema；每次组装时使用该次组装的上下文求值。`ToolProviderResult` = `{ schemas, knownNames? }`：`schemas` 是限制后的可见集合；`knownNames` 是限制前由 `toolOrder` 使用的全集。提供方不得返回名为 `TOOL_ORDER_REST` 的 schema。带作用域提供方只在其作用域的组装中查询。随调用 fiber 一并 dispose。
+- `ctx.systemPrompt.variable(name: string, provider: (context) => string | undefined): () => void`：贡献提示词变量，在段文本中以 `{{name}}` 引用。带作用域变量会为该 agent 遮蔽同名全局变量。同层重复或无法引用的名称会抛出；`undefined` 表示「本次组装没有值」。随调用 fiber 一并 dispose。
 - `ctx.systemPrompt.assemble(context?: AssembleContext): Promise<PromptAssembly>`：为一个调用方组装提示词：将全局层与 `context.scope` 的层合并，并在变换 seam 前分离工具 schema。它经过按作用域筛选的 `system-prompt/assemble` waterfall，并返回其权威结果。可选的 `context.signal` 显式控制本次组装请求；提供方与监听器可以配合该信号，但不得将它保留给另一轮次。当已配置的 `toolOrder` 指名提供方 `knownNames` 全集以外的工具，或提供方返回保留的其余项名称时，调用会被拒绝。
 
 ### 实时事件
@@ -35,18 +35,18 @@
 
 ### 扩展点
 
-- 段提供方：工具包拥有跨调用引导（`tool:bash`、`tool:read` 等）；此插件拥有 `harness:identity` 与 `deployment:persona`。
-- 变量提供方：agent loop 注册 `model` 与 `cwd`；任何插件都可以注册自己拥有的事实（未来的 `date`、git 状态等）。
+- 段提供方：工具包（package）拥有跨调用引导（`tool:bash`、`tool:read` 等）；此插件拥有 `harness:identity` 与 `deployment:persona`。
+- 变量提供方：agent loop（智能体循环）注册 `model` 与 `cwd`；任何插件都可以注册自己拥有的事实（未来的 `date`、git 状态等）。
 - 工具 schema 提供方：`ToolRegistry` 自动将自身注册为工具提供方。
 - [`system-prompt/assemble` waterfall](#live-events)：按调用方协作式修改或替换组装结果。
 
-设计原理：[提示词变量 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.md)。
+设计原理：[提示词变量 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.md)。
 
 ## 模型体验
 
 ### 系统提示词
 
-#### 模型所见
+#### 模型看到的内容
 
 每次组装都从下方 harness 身份开始，然后在严格变量插值后追加已配置 persona 与有序插件段。空段会消失；带作用域的段和变量可以为一个 agent 遮蔽全局项。最终 `system-prompt/assemble` waterfall 结果是权威来源，因此专家监听器的变更决定交付的提示词与工具 schema。
 
@@ -66,21 +66,21 @@ You are an AI agent powered by the DeepSeek Harness SDK.
 
 ### 工具 schema
 
-#### 模型所见
+#### 模型看到的内容
 
 对于已交付工具，模型会收到[生成工具 schema](../../../docs/tool-catalog.md#tool-package-map) 中对每个 agent 可见的子集；限制与组装拦截完成后，按配置或字典序排列。扩展可以通过同一注册表贡献其他定义。段与 schema 提供方是独立的组装输入，因此工具限制不会移除独立注册的引导。
 
 #### Token 影响
 
-Schema token 在每次请求中重复。限制工具会为该 agent 移除其全部 schema 成本，但不会移除独立提示词段；重排序会改变 cache 形状，但不改变语义内容。
+Schema token 在每次请求中重复。限制工具会为该 agent 移除其全部 schema 成本，但不会移除独立提示词段；重排序会改变缓存形状，但不改变语义内容。
 
 #### KV Cache 影响
 
 只要可见 schema 集合、渲染与顺序不变，前缀就保持稳定。注册、限制或重排序可能从第一个变化的 schema token 起使复用失效。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
-- **部署创作的提示词文本只来自配置／组合**：此插件拥有全局 persona 默认值；创建方插件可以注册 agent 作用域的遮蔽项；其他段来自拥有相应事实的插件。不存在终端用户提示词编辑 API。
+- **部署方编写的提示词文本只来自配置／组合**：此插件拥有全局 persona 默认值；创建方插件可以注册 agent 作用域的遮蔽项；其他段来自拥有相应事实的插件。不存在终端用户提示词编辑 API。
 - **没有表示字面量 `{{…}}` 花括号的转义语法**：每个完整组都会按已注册变量插值；只有实际提示词需要转义时才会实现。
 - **`toolOrder` 配置错误在提示词组装（首轮）时出现，而不是启动时**：只有形状违规会在配置加载时抛出。
-- **共享同一 `order` 值的段按注册顺序打破平局**：这是插件加载工件；确定性依赖不同顺序区间的约定，与已规范化的工具顺序不同。
+- **共享同一 `order` 值的段按注册顺序打破平局**：这是插件加载产物；确定性依赖不同顺序区间的约定，与已规范化的工具顺序不同。
diff --git a/packages/core/tools/README.i18n.yaml b/packages/core/tools/README.i18n.yaml
index db07e76181..429e76ed6a 100644
--- a/packages/core/tools/README.i18n.yaml
+++ b/packages/core/tools/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/core/tools/README.md
 README.md: e5adb153e77d7a2d8c4068b016194ab6abb6473e
-README.zh.md: 0177c2a6a4c2db121d90c83044bb1e3de7d0099f
+README.zh.md: c67a2f2ee4ac2a9d587c6efbf2b5c60d14fc58c2
diff --git a/packages/core/tools/README.zh.md b/packages/core/tools/README.zh.md
index 0177c2a6a4..c67a2f2ee4 100644
--- a/packages/core/tools/README.zh.md
+++ b/packages/core/tools/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-工具注册表与执行流水线。工具插件注册各自的 schema 和执行器；agent loop（智能体循环）依次让每次调用经过 `tools/pre-execute`（可扩展的允许／拒绝门禁）→ 单调注册守卫 → `tools/execute`（供超时／重试／指标插件使用的环绕分发包装层）→ `tools/post-execute`（检查／替换结果、附加上下文）→ 由定义拥有的 `finalizeContent` 边界 → 仅观测的 `tools/result` 通知。注册表还负责决定如何向模型呈现其工具：`mode` 配置可以选择原生 Function Calling（函数调用）、[Code Mode](#code-mode)，或同时选择两者。
+工具注册表与执行流水线。工具插件注册各自的 schema 和执行器；agent loop（智能体循环）依次让每次调用经过 `tools/pre-execute`（可扩展的允许／拒绝门禁）→ 已注册的单调守卫 → `tools/execute`（供超时／重试／指标插件使用的环绕分发包装层）→ `tools/post-execute`（检查／替换结果、附加上下文）→ 由定义拥有的 `finalizeContent` 边界 → 仅观测的 `tools/result` 通知。注册表还负责决定如何向模型呈现其工具：`mode` 配置可以选择原生 Function Calling（函数调用）、[Code Mode](#code-mode)，或同时选择两者。
 
 ## 服务：`ToolRegistry`（ctx 键：`tools`）
 
@@ -17,25 +17,25 @@ tools:
 
 ### 公开 API
 
-- `ctx.tools.register(definition: ToolDefinition): () => void`：注册一个受信任、带类型的同进程定义，其中必须包含规范的 `output` 声明。所在层由调用上下文的作用域决定：普通插件上下文会全局注册；agent 的 `agent.ctx` 只为该 agent 注册，并在此处遮蔽同名全局工具。同一层内名称重复会抛出；非原生模式还会拒绝保留的 `run_code` 传输名称。缺失或不受支持的输出声明，以及非正数或非有限的 `timeoutMs`，都会使注册失败。可选的同步 `finalizeContent` 回调会在调用开始时创建快照；在所有流水线结果规范化之后，它只能替换最终面向模型的内容，包括实体化其他结果字段时发现的错误。随调用 fiber 释放。
+- `ctx.tools.register(definition: ToolDefinition): () => void`：注册一个受信任、带类型的同进程定义，其中必须包含规范的 `output` 声明。所在层由调用上下文的作用域决定：普通插件上下文会全局注册；agent 的 `agent.ctx` 只为该 agent 注册，并在此处遮蔽同名全局工具。同一层内名称重复会抛出；非原生模式还会拒绝保留的 `run_code` 传输名称。缺失或不受支持的输出声明，以及非正数或非有限的 `timeoutMs`，都会使注册失败。可选的同步 `finalizeContent` 回调会在调用开始时创建快照；在所有流水线结果规范化之后，它只能替换最终面向模型的内容，包括实体化其他结果字段时发现的错误。随调用 fiber dispose（释放资源）。
 - `ctx.tools.restrict(filter)`：对全局工具应用 agent 作用域的允许／拒绝掩码；从普通上下文调用会抛出。筛选器在注册时创建快照；多个掩码取交集，随后再合并作用域本地工具。拒绝掩码会接纳后来出现且未点名的全局工具，而允许掩码会排除后来出现的名称。未知、本地或保留名称以及空筛选器都会被拒绝。这是实时可见性组合，不是权限边界；参见[作用域安全非目标](../../../.agents/notes/implemented/architecture/2026-07-08-agent-scope-contexts.md#security-and-authority-are-explicit-non-goals)。
 - `ctx.tools.get(name: string, scope?: ScopeKey): ToolDefinition | undefined`：按某个作用域所见的结果解析（应用遮蔽；被限制掉的全局工具视为不存在）。呈现器会传入发起调用的 agent，使卡片与实际执行内容一致。
 - `ctx.tools.schemas(scope?: ScopeKey): ToolSchema[]`：返回该作用域可见的所有 schema（不含 `execute` 函数）。已交付工具的 schema 收录在 [docs/tool-catalog.md](../../../docs/tool-catalog.md) 中；该目录通过启动每个工具插件并采集此方法的结果生成（参见[工具 schema 目录 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/process/2026-07-02-tool-schema-catalog.md)）。
-- `ctx.tools.guard(guard: ToolGuard): () => void`：在 `tools/pre-execute` 之后注册单调同步执行守卫：返回理由会拒绝调用，返回 `undefined` 则保持原决定。普通上下文守卫全局生效；`agent.ctx` 守卫只对该 agent 生效。后续 waterfall（瀑布式事件）监听器无法将守卫的拒绝重新变为允许。随调用 fiber 释放。
+- `ctx.tools.guard(guard: ToolGuard): () => void`：在 `tools/pre-execute` 之后注册单调同步执行守卫：返回理由会拒绝调用，返回 `undefined` 则保持原决定。普通上下文守卫全局生效；`agent.ctx` 守卫只对该 agent 生效。后续 waterfall（瀑布式事件）监听器无法将守卫的拒绝重新变为允许。随调用 fiber dispose。
 - `ctx.tools.execute(exec)`：以无损方式快照并冻结参数，分配不透明 token，运行完整的策略／分发／结果流水线，然后在最终观测前独立快照权威结果。无效参数会进入同一结果路径，但不会到达策略或工具主体。环绕包装层只能替换 `signal`；注册表会在调用主体前立即重新融合调用方的原始信号。
 - `ctx.tools.executionMode(exec)`：返回 `parallel` 的唯一条件是可见定义的 `isConcurrencySafe(exec.arguments)` 分类器恰好返回 `true`；未知、隐藏、未声明、无效或抛出异常的分类结果均为独占。
 
 ### 注入的服务
 
-`SystemPrompt`：注册表通过 `ctx.systemPrompt.tools()` 自动将工具 schema 送入系统提示词组装。审批 seam 则按需使用（`ctx.get('approval')`，无静态注入）：未部署该 seam 时仍会将询问退化为拒绝，而无论是否存在该 seam，注册表都会保持活动。
+`SystemPrompt`：注册表通过 `ctx.systemPrompt.tools()` 自动将工具 schema 送入系统提示词组装。审批 seam 则在可用时使用（`ctx.get('approval')`，无静态注入）：未部署该 seam 时仍会将询问退化为拒绝，而无论是否存在该 seam，注册表都会保持活动。
 
 ### 取消
 
-取消采用协作方式，并等待完全停稳。每次类型化调用都提供由调用方拥有的 `AbortSignal`；工具主体通过必填的只读 `exec.signal` 接收它，只有 `tools/execute` 包装层可以临时替换这个必填信号。注册表会在替换期间保留调用方取消，并且绝不会在已启动的同进程 Promise 尚未结算时提前返回。调用主体前发生的取消为 `ABORTED_BEFORE_DISPATCH`；调用后的取消只能把成功结果替换为 `ABORTED`。拒绝、包装层失败、工具失败、后置策略失败或超时拥有的 `TOOL_TIMEOUT` 仍保留更具体的结果。入口处已中止的调用会实体化并冻结参数，随后跳过所有策略和分发阶段，只发布一个结果。每个异步工具都必须观测或转发该信号，并且只能在自身拥有的工作停止后结算。[工具取消 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-19-cooperative-tool-cancellation.md) 规定完整契约和强制终止边界。
+取消采用协作方式，并等待完全停稳。每次类型化调用都提供由调用方拥有的 `AbortSignal`；工具主体通过必填的只读 `exec.signal` 接收它，只有 `tools/execute` 包装层可以临时替换这个必填信号。注册表会在替换期间保留调用方取消，并且绝不会在已启动的同进程 Promise 尚未结算时提前返回。调用主体前发生的取消为 `ABORTED_BEFORE_DISPATCH`；调用后的取消只能把成功结果替换为 `ABORTED`。拒绝、包装层失败、工具失败、后置策略失败或由超时机制产生的 `TOOL_TIMEOUT` 仍保留更具体的结果。入口处已中止的调用会实体化并冻结参数，随后跳过所有策略和分发阶段，只发布一个结果。每个异步工具都必须观测或转发该信号，并且只能在自身拥有的工作停止后结算。[工具取消 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-19-cooperative-tool-cancellation.md) 规定完整契约和强制终止边界。
 
 ### 实时事件
 
-实时注册表流水线先经过 3 个可变换的 waterfall，再经过由定义拥有的内容终结器，最后到达仅观测的 `tools/result` 边界；注册表变更有意作为不过滤的共享状态通知。确切签名、分发 mode、作用域筛选和故障收容契约位于生成的 [Cordis 事件目录](../../../docs/cordis-catalog/events.md)，完整顺序则在生成的[工具执行流水线](../../../docs/tool-execution-pipeline.md)中可视化。`tools/result` 是实时事件；名称相近的 `tool/result` 是 agent loop 随后追加的持久会话事件。
+实时注册表流水线先经过 3 个可变换的 waterfall，再经过由定义拥有的内容终结器，最后到达仅观测的 `tools/result` 边界；注册表变更有意作为不过滤的共享状态通知。确切签名、分发模式、作用域筛选和故障收容契约位于生成的 [Cordis 事件目录](../../../docs/cordis-catalog/events.md)，完整顺序则在生成的[工具执行流水线](../../../docs/tool-execution-pipeline.md)中可视化。`tools/result` 是实时事件；名称相近的 `tool/result` 是 agent loop 随后追加的持久会话事件。
 
 ### 关键类型
 
@@ -44,7 +44,7 @@ tools:
 - `ToolExecutionToken`：注册表分配的全新带品牌 `Symbol`。它只支持通过相等性进行关联，绝不会跨越模型、日志或 worker 边界。
 - `ToolExecution`：只读流水线视图：不可变的 `{ token, callId, name, arguments, signal, agent?, parent? }`；注册表会另行保留并重新融合调用方的原始信号。`ToolDispatchExecution` 是仅供 `tools/execute` 使用的视图，其必填信号可变，因此包装层可以替换并还原它，但不能删除它。嵌套调用的 `parent` 是 `ToolExecutionToken`，而不是执行对象。
 - `ToolRunContext`：传给工具主体的执行上下文，在 `ToolExecution` 基础上增加 `deferContext(context)`。组合工具借此把嵌套分发产生的上下文传递到外层结果，即使工具后来抛出或取消胜出也不例外；该方法绝不会立即注入上下文。
-- `ToolExecutionResult`：可辨识的执行局部结果。成功形态为 `{ isError:false, value:JsonValue, content, meta?, additionalContexts? }`；失败形态为 `{ isError:true, error:{ message, info? }, content, meta?, additionalContexts? }`，且不含值。调用身份保留在不可变的 `ToolExecution` 上。注册表会在呈现前快照、验证并冻结规范值，随后在最终观测前实体化持久呈现字段。`ToolFailure.info` 携带内部的 `{ name, code }`，用于表示 `HarnessError`；`additionalContexts` 为循环在结果后的 FIFO 保留每个延迟或后置执行且带标识的 `UserMessage`。
+- `ToolExecutionResult`：可辨识的执行局部结果。成功形态为 `{ isError:false, value:JsonValue, content, meta?, additionalContexts? }`；失败形态为 `{ isError:true, error:{ message, info? }, content, meta?, additionalContexts? }`，且不含值。调用身份保留在不可变的 `ToolExecution` 上。注册表会在呈现前快照、验证并冻结规范值，随后在最终观测前实体化持久呈现字段。`ToolFailure.info` 携带内部的 `{ name, code }`，用于表示 `HarnessError`；`additionalContexts` 会保留每个通过延迟或 post-execute 加入且带标识的 `UserMessage`，供循环在结果后按 FIFO 顺序处理。
 - `PreToolDecision`：`{kind:'allow'}` | `{kind:'deny', reason}` | `{kind:'ask', reason?}`。该类型有意不提供输入改写；`ask` 在挂载 [`ctx.approval`](../../ui/user-approval/README.md) 时由它处理，否则退化为拒绝。
 - `PostToolDecision`：接受决定可以替换 `content` 或 `value`（不能同时替换），并可附加 `additionalContexts`；阻止决定会把反馈变成无值失败。替换内容会保留规范值和元数据。替换值会重新验证，并重新呈现内容／元数据。接受决定会先保留工具延迟的上下文，再附加决定上下文；阻止决定会丢弃工具延迟的上下文，只公开阻止决定显式提供的上下文。
 - `ToolGuard`：`(execution) => string | undefined`；返回的字符串是最终单调拒绝理由，在可重排的前置执行 waterfall 之后、分发之前求值。
@@ -54,14 +54,14 @@ tools:
 
 - 工具插件调用 `ctx.tools.register()`：schema 会自动流入组装结果。
 - `tools/pre-execute` 是可重排的允许／拒绝／询问门禁；`ctx.tools.guard()` 在其后添加单调的拥有方策略。
-- `tools/execute` 为超时、重试或指标环绕已经规范化的规范分发。包装层只能替换操作信号；包装层创作的成功结果会根据已解析工具的输出声明进行规范化。规范结果的来源属于一个不可变分发 token，因此，来自其他调用或工具的缓存结果会根据当前声明重新验证。
+- `tools/execute` 会环绕包装规范化后的规范分发，以支持超时、重试或指标采集。包装层只能替换操作信号；包装层生成的成功结果会根据已解析工具的输出声明进行规范化。规范结果的来源属于一个不可变分发 token，因此，来自其他调用或工具的缓存结果会根据当前声明重新验证。
 - `tools/post-execute` 可以替换呈现内容、替换规范值、通过反馈阻止，或附加有序上下文。随后，定义可选的 `finalizeContent` 会在普通结果和外层流水线失败中维护其最终、仅涉及内容的不变式；`tools/result` 观测不可变的最终结果。内容替换不是保密边界：当编程消费方不得接收某个值时，应阻止或替换该值。
 - 确切签名与顺序位于生成的[事件目录](../../../docs/cordis-catalog/events.md)和[流水线](../../../docs/tool-execution-pipeline.md)中。
 - MCP 服务器：每个服务器使用一个插件；发现工具后，使用服务器的 schema 调用 `ctx.tools.register()`。
 
 ### 类型化工具参数 schema
 
-第一方插件作者可以使用本包导出的 `defineTool()` 辅助函数定义类型化工具参数 schema：
+第一方插件作者可以使用本包（package）导出的 `defineTool()` 辅助函数定义类型化工具参数 schema：
 
 ```ts
 import { readFile } from 'node:fs/promises'
@@ -114,11 +114,11 @@ ctx.tools.register(defineTool({
 
 ### Code Mode
 
-在 `code` 或 `both` 模式下，注册表为当前作用域公开保留的 `run_code` 传输和确定性的 TypeScript SDK；只有程序的外层日志与返回值会重新进入模型上下文。SDK 为每个可见工具声明精确的 `ToolArgsMap` 和 `ToolOutputMap` 条目，每个绑定都会解析为该工具的规范 JSON 值。每个无损 JSON 绑定调用都会在原生调度契约下重新进入完整工具流水线（并发安全的调用最多可重叠 `maxParallelSubCalls` 个；独占调用单独运行并构成排序屏障），并在日志中与外层调用建立关联。拒绝及其他失败结果会以程序实际可见的 `ToolCallError` 进行 reject，且只携带 `toolName` 和 `message`；Native 内容和内部错误码留在 Code 契约之外。普通副作用不会回滚，子调用的 `additionalContexts` 会通过父结果延迟，以保持调用／结果相邻。运行结算会中止并 drain 尚未完成的绑定；运行时失败以 `CodeRunFailedError` 形式出现。参见 [Code Mode 基础](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md)、[类型化返回契约](../../../.agents/notes/implemented/feature/2026-07-20-code-mode-typed-tool-returns.md)和[代码运行时 seam](../../code-runtime/README.md)。可以运行 `pnpm run demo:code-mode` 试用。
+在 `code` 或 `both` 模式下，注册表为当前作用域公开保留的 `run_code` 传输和确定性的 TypeScript SDK；只有程序的外层日志与返回值会重新进入模型上下文。SDK 为每个可见工具声明精确的 `ToolArgsMap` 和 `ToolOutputMap` 条目，每个绑定都会解析为该工具的规范 JSON 值。每个无损 JSON 绑定调用都会在原生调度契约下重新进入完整工具流水线（并发安全的调用最多可重叠 `maxParallelSubCalls` 个；独占调用单独运行并构成排序屏障），并在日志中与外层调用建立关联。拒绝及其他失败结果会以程序实际可见的 `ToolCallError` 形式拒绝，且只携带 `toolName` 和 `message`；Native 内容和内部错误码留在 Code 契约之外。普通副作用不会回滚，子调用的 `additionalContexts` 会通过父结果延迟，以保持调用／结果相邻。运行结算会中止并排空尚未完成的绑定；运行时失败以 `CodeRunFailedError` 形式出现。参见 [Code Mode 基础](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md)、[类型化返回契约](../../../.agents/notes/implemented/feature/2026-07-20-code-mode-typed-tool-returns.md)和[代码运行时 seam](../../code-runtime/README.md)。可以运行 `pnpm run demo:code-mode` 试用。
 
 - **SDK 段**（`tools:sdk`，顺序 150）：一个惰性提示词段，每次组装时都会重新生成 `JsonValue`、精确的 `ToolArgsMap` / `ToolOutputMap`、`ToolName`、`ToolCallError` 声明、面向调用作用域可见最终能力的映射 `tools` 命名空间（特殊名称使用带引号的键），以及固定用法说明。其输出具有确定性：工具按字典序排列；工具集合不变时，文本逐字节相同（有利于前缀 cache）。导出的代码生成器 `jsonSchemaToTs` 会处理统一 schema 的每种构造，并将不受支持的原始构造降级为 `unknown`，绝不会在提示词组装期间抛出。
 - **分发桥接层**（`run_code` 的 execute）：每个绑定调用都会在分发前快照为无损 JSON（`undefined`、`BigInt`、循环、稀疏数组、`-0` 和特殊对象会使该次调用被拒绝），经由每次运行独有、复用原生并发契约的池调度——调用严格按提交顺序启动，连续的 `isConcurrencySafe` 调用最多可重叠经校验的 `maxParallelSubCalls` 配置个（默认 10；设为 `1` 即恢复串行分发），被分类为独占的调用先排空池、单独运行并阻挡其后的调用——以外层执行的不透明 token 作为 `parent`，并经过完整的 pre-execute → guards → execute → post-execute → result 流水线。成功会返回策略处理后的最终规范值；失败以一条消息到达 worker，并成为 `ToolCallError(toolName, message)`。每个已启动的子调用在进入流水线时记录一条 `tool/code-dispatch-start` 事件（确定性 id `<parent>:code:<n>`，按提交顺序编号），并以一条携带完整模型可见 `content`/`isError` 结果的 `tool/code-dispatch` 事件完结（采用 `tool/result` 词汇，因此 UI 会沿原生路径呈现子调用——这对事件的 `time` 字段承载每个子调用的计时）；因 run 结算而被放弃的排队调用两者都不记录。`deriveMessages()` 既不公开这两个事件，也不持久化规范值。token 关联让以提交为语义的观察器能够把内部成功延迟到最终 `run_code` 结果，而无需公开实时外层执行；普通工具副作用不会回滚。每个子调用的 `additionalContexts` 条目都会按分发顺序通过外层 `ToolRunContext` 延迟；循环只在父级 `run_code` 结果之后追加这些上下文，从而保持相邻关系，并且即使程序后来失败，也会保留各自的来源／元数据。
-- **结算纪律**：桥接层拥有一次运行作用域的中止；该中止会跟随传入的外层信号，并在运行因任何原因结算时触发，因此预算耗尽会中止正在运行的子工具，而不会将其遗留。桥接层随后会在返回之前 drain 队列，使每个 `tool/code-dispatch` 都落在仍打开的轮次内。失败的运行会抛出 `CodeRunFailedError`（`code: 'CODE_RUN_FAILED'`，message = 失败类型 + 已捕获日志），流水线会将其转换为模型可据以自我修正的结构化 `isError`。
+- **结算纪律**：桥接层拥有一个运行作用域的中止机制；该中止会跟随传入的外层信号，并在运行因任何原因结算时触发，因此预算耗尽会中止正在运行的子工具，而不会将其遗留。桥接层随后会在返回之前排空队列，使每个 `tool/code-dispatch` 都落在仍打开的轮次内。失败的运行会抛出 `CodeRunFailedError`（`code: 'CODE_RUN_FAILED'`，message = 失败类型 + 已捕获日志），流水线会将其转换为模型可据以自我修正的结构化 `isError`。
 - **结果边界**：中间绑定值会完整跨越 worker 边界，且没有逐绑定字节上限。`run_code` 返回规范的 `{ logs: string[], result?: JsonValue }`；字符串原样呈现，其他所有存在的 JSON 根都通过栈安全的美化 JSON 遍历呈现，总缩进最多为 10 个字符（更深的子树保持紧凑），`null` 保持显式，而缺少 `result` 表示程序返回 `undefined`。worker 可配置的 `maxOutputBytes`（默认 64 MiB）只应用于组合序列化后的外层日志数组、完成值或失败消息载荷；固定的结果 envelope 语法和呈现空白不计入该账本。无效和超限的完成会明确失败，只有此外层结果可以使用普通 spill。
 
 ### 并行执行
@@ -129,7 +129,7 @@ agent loop 将连续的 `parallel` 调用归入有界滚动池，并把每个 `e
 
 ### 普通工具 schema
 
-#### 模型所见
+#### 模型看到的内容
 
 在普通模式下，模型会看到每个可见定义的确切名称、描述和 JSON schema；已交付定义记录在生成的[工具包映射和 schema 章节](../../../docs/tool-catalog.md#tool-package-map)中。agent 作用域的限制、遮蔽和扩展注册会改变该 agent 的最终工具集合。
 
@@ -139,11 +139,11 @@ agent loop 将连续的 `parallel` 调用归入有界滚动池，并把每个 `e
 
 #### KV Cache 影响
 
-只要可见定义及其顺序不变，前缀就保持稳定。注册、释放或作用域限制可能从第一个改变的 schema token 起使复用失效。
+只要可见定义及其顺序不变，前缀就保持稳定。注册、dispose 或作用域限制可能从第一个改变的 schema token 起使复用失效。
 
 ### Code Mode schema 与系统提示词
 
-#### 模型所见
+#### 模型看到的内容
 
 Code Mode 会公开生成的 [`run_code` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tools)、下方 SDK 说明，以及生成的精确 `declare const tools` 块。`both` 会同时公开普通 schema 与此 Code Mode 接口。
 
@@ -172,24 +172,24 @@ The available tools:
 
 ### 工具调用历史与结果
 
-#### 模型所见
+#### 模型看到的内容
 
 循环会保留模型发出的参数和注册表的最终内容。任何抛出或被拒绝的调用都会恰好变为 `Error: <message>`。Code Mode 只返回外层程序打印的行和呈现后的返回值；两者都为空时返回 `(run_code completed with no output)`；失败时返回 `Error: code run failed (<kind>): <message>`，并根据是否存在已捕获内容，在其后附加 `Captured output:` 与捕获的行。内部分发事件只保留在日志中；后置执行监听器可以在结果之后追加带来源归属的上下文。
 
 #### Token 影响
 
-参数、结果和附加上下文取决于数据，并会重复发送直至压缩。隐藏工具的限制还会在模型可以调用这些工具之前移除其 schema。
+参数、结果和附加上下文取决于数据，并会重复发送直至压缩（compaction）。隐藏工具的限制还会在模型可以调用这些工具之前移除其 schema。
 
 #### KV Cache 影响
 
 仅追加；新的可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **并发策略不是事件 seam**：`executionMode()` 直接读取已解析的工具定义；插件只能在自身拥有的定义上声明分类器。
 - **`tools/pre-execute` 有意不允许改写 `exec.arguments`**：否则日志记录和呈现的参数会与实际运行内容失去同步；改写设计记录在[拟议的 Agent Note](../../../.agents/notes/proposed/feature/2026-06-30-pre-tool-input-rewrite.md)中。
 - **调用方定义的 subagent 与工作流结构化输出仍要求对象根**：这是消费方层面的守卫；共享 schema 词汇和工具输出支持任意 JSON 根。
 - **定义上的 `timeoutMs` 仅为声明**：注册表绝不会强制执行截止时间；要强制执行，必须使用 `@deepseek-ai/dsh-timeout-policy` 包装层。
 - **Code Mode 只支持 TypeScript，且呈现模式在服务内统一**：`mode: code`/`both` 会拒绝组装提示词，除非 `ctx.codeRuntime.language === 'typescript'`；作用域限制／遮蔽仍会选择每个 agent 的可见绑定，但不能让一个工具仅使用 Native，而另一个仅使用 Code。
-- **Code Mode 中间值只存在于执行局部，且没有字节上限**：这些规范的类型化值无法从会话回放重建，并可能耗尽进程或 worker 内存；只有外层 `run_code` 输出受 worker 可配置的硬上限约束。每个子调用的持久日志副本则**有**上限：`tools/code-dispatch-log` waterfall 允许 spill 策略把过大的 `tool/code-dispatch` 内容替换为预览加定位符（[原理](../../../.agents/notes/implemented/feature/2026-07-26-code-dispatch-log-spill.md)）。
+- **Code Mode 中间值只存在于执行局部，且没有字节上限**：这些规范的类型化值无法从会话回放重建，并可能耗尽进程或 worker 内存；只有外层 `run_code` 输出受 worker 可配置的硬上限约束。每个子调用的持久日志副本则确实有上限：`tools/code-dispatch-log` waterfall 允许 spill 策略把过大的 `tool/code-dispatch` 内容替换为预览加定位符（[原理](../../../.agents/notes/implemented/feature/2026-07-26-code-dispatch-log-spill.md)）。
 - **每次运行都会获得全新的 `run_code` 状态**：MVP 不采用持久 REPL 风格内核（跨调用状态不会出现在日志中）；参见 [Code Mode Agent Note](../../../.agents/notes/implemented/feature/2026-06-15-code-mode.md)。
diff --git a/packages/examples/README.i18n.yaml b/packages/examples/README.i18n.yaml
index 615d4194c6..ceb4a2df7a 100644
--- a/packages/examples/README.i18n.yaml
+++ b/packages/examples/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/examples/README.md
 README.md: c229cef22087ac290bf862d6b3e31fdb533858c4
-README.zh.md: 5472fe7be76b015a2f3b06b2738b728a927acdd4
+README.zh.md: 208b9a138506785ea1dd2d83ddfbba29e4b7968e
diff --git a/packages/examples/README.zh.md b/packages/examples/README.zh.md
index 5472fe7be7..208b9a1385 100644
--- a/packages/examples/README.zh.md
+++ b/packages/examples/README.zh.md
@@ -2,17 +2,17 @@
 
 [English](README.md) | 中文
 
-预先组合的插件 bundle（组合包），供轻量叶节点 `cordis.yml` 加载，无需手工组装主干和前端入口。这些是 **演示／参考** 包；npm 名称的 `-demo` 后缀把每个包标为非产品表层，直接查看包名即可辨认。仓库根目录 [`examples/`](../../examples/AGENTS.md) 下的可运行叶节点与 [Python SDK runtime](../../python/sdk-runtime/README.md) 是消费方；每个叶节点都只包含可替换后端和一个组合包入口。
+预先组合的插件 bundle（组合包），供轻量叶节点 `cordis.yml` 加载，无需手工组装主干和前端入口。这些是 **演示／参考** 包（package）；npm 名称的 `-demo` 后缀把每个包标为非产品表层，直接查看包名即可辨认。仓库根目录 [`examples/`](../../examples/AGENTS.md) 下的可运行叶节点与 [Python SDK 运行时](../../python/sdk-runtime/README.md) 是消费方；每个消费方都只包含可替换后端和一个组合包入口。
 
 | 包 | npm 名称 | 角色 |
 |---|---|---|
-| `agent-spine-demo/` | `@deepseek-ai/dsh-agent-spine-demo` | 不含执行器和 UI 的 agent 主干，打包为一个组合包插件，带后备会话标题和选用的持久目标栈 |
-| `tui-demo/` | `@deepseek-ai/dsh-tui-demo` | 全屏终端应用组合包：主干 + 持久目标 + `/goal` 命令 + JSONL 持久化 + `dsh-tui` + 预创建的 `main` agent；没有 bin，由 [`dsh`](../../apps/cli/README.md) CLI 启动 |
+| `agent-spine-demo/` | `@deepseek-ai/dsh-agent-spine-demo` | 不含执行器和 UI 的 agent（智能体）主干，打包为一个组合包插件，带后备会话标题和可选择启用的持久化目标栈 |
+| `tui-demo/` | `@deepseek-ai/dsh-tui-demo` | 全屏终端应用组合包：主干 + 持久化目标 + `/goal` 命令 + JSONL 持久化 + `dsh-tui` + 预创建的 `main` agent；没有 bin，由 [`dsh`](../../apps/cli/README.md) CLI（命令行界面）启动 |
 | `cli-demo/` | `@deepseek-ai/dsh-cli-demo` | 无头单次应用：主干 + JSONL 持久化 + 预创建的 `main` agent，提供文本和 DSH 原生 JSON 输出 |
-| `acp-demo/` | `@deepseek-ai/dsh-acp-demo` | ACP 自动化服务器应用：主干 + 持久目标 + JSONL 持久化 + [`acp`](../acp/acp/README.md) 桥接层（无 stdout logger），带启动 `bin` |
-| `jsonrpc-demo/` | `@deepseek-ai/dsh-jsonrpc-demo` | 只有 bin 的 runtime，用于启动外部 `cordis.yml`，供 stdio JSON-RPC SDK 客户端使用 |
+| `acp-demo/` | `@deepseek-ai/dsh-acp-demo` | ACP（Agent Client Protocol）自动化服务器应用：主干 + 持久化目标 + JSONL 持久化 + [`acp`](../acp/acp/README.md) 桥接层（无 stdout logger），带启动 `bin` |
+| `jsonrpc-demo/` | `@deepseek-ai/dsh-jsonrpc-demo` | 只有 bin 的运行时，用于启动外部 `cordis.yml`，供 stdio JSON-RPC SDK 客户端使用 |
 
-`agent-spine-demo` 是共享组合包；`tui-demo`、`cli-demo` 和 `acp-demo` 分别将它与全屏终端、无头单次和 ACP 自动化前端入口组合。`cli-demo` 与 `acp-demo` 拥有各自的启动 bin；`tui-demo` 只交付组合包插件，产品 [`dsh`](../../apps/cli/README.md) CLI 是它的终端前端入口。`jsonrpc-demo` 自身不挂载任何组合，而是启动部署的 `cordis.yml` 所指名的任意插件树；Python SDK runtime 会启动它。
+`agent-spine-demo` 是共享组合包；`tui-demo`、`cli-demo` 和 `acp-demo` 分别将它与全屏终端、无头单次和 ACP 自动化前端入口组合。`cli-demo` 与 `acp-demo` 拥有各自的启动 bin；`tui-demo` 只交付组合包插件，产品 [`dsh`](../../apps/cli/README.md) CLI 是它的终端前端入口。`jsonrpc-demo` 自身不挂载任何组合，而是启动部署的 `cordis.yml` 所指名的任意插件树；Python SDK 运行时会启动它。
 
 这些 **不是** 产品 API。它们打包的主干组件位于 [`core/`](../core/README.md)，人类／SDK 通道和启动粘合代码位于 [`ui/`](../ui/README.md)，自动化传输位于 [`acp/`](../acp/README.md)，可替换后端位于各自能力组；演示组合包只选定其中一种具体组合。可以自由替换或 fork。
 
@@ -20,4 +20,4 @@
 
 ## jsonrpc bin／exe 名称是历史遗留
 
-`jsonrpc-demo` 已像同级包一样重命名，但其 bin 仍为 `dsh-jsonrpc-agent`，单文件可执行程序仍为 `dsh-jsonrpc-agent-pkg`（在 [Python 分发](../../python/sdk-runtime/README.md)各处被引用）。这些名称属于 SDK 的 runtime 启动表层；只有 SDK 统一该启动流程时才会协调它们，而不会在此次移动中处理。
+`jsonrpc-demo` 已像同级包一样重命名，但其 bin 仍为 `dsh-jsonrpc-agent`，单文件可执行程序仍为 `dsh-jsonrpc-agent-pkg`（在 [Python 分发](../../python/sdk-runtime/README.md)各处被引用）。这些名称属于 SDK 的运行时启动表层；只有 SDK 统一该启动流程时才会协调它们，而不会在此次移动中处理。
diff --git a/packages/examples/acp-demo/README.i18n.yaml b/packages/examples/acp-demo/README.i18n.yaml
index 7e17f48b20..01a4e4cd62 100644
--- a/packages/examples/acp-demo/README.i18n.yaml
+++ b/packages/examples/acp-demo/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/examples/acp-demo/README.md
 README.md: 395ab230146568989c4e6d1361218efb72d857e7
-README.zh.md: 1dfd2d99f4ab80953df77feba19649b934775d68
+README.zh.md: bfba1f83e506e60ab117f11f85851b3d3a7ea16a
diff --git a/packages/examples/acp-demo/README.zh.md b/packages/examples/acp-demo/README.zh.md
index 1dfd2d99f4..bfba1f83e5 100644
--- a/packages/examples/acp-demo/README.zh.md
+++ b/packages/examples/acp-demo/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-ACP 自动化服务器应用：默认 agent 主干、客户端通过 [`@deepseek-ai/dsh-acp`](../../acp/acp/README.md) 创建的 agent、JSONL 持久化，以及由一个 JSON-RPC stdio bin 提供的语义检查点。程序化客户端创建新会话；此包不挂载人类 UI。
+ACP（Agent Client Protocol）自动化服务器应用：默认 agent（智能体）主干、客户端通过 [`@deepseek-ai/dsh-acp`](../../acp/acp/README.md) 创建的 agent、JSONL 持久化，以及由一个 JSON-RPC stdio bin 提供的语义检查点。程序化客户端创建新会话；此包（package）不挂载人工交互 UI。
 
 ## 组合
 
@@ -14,7 +14,7 @@ ACP 自动化服务器应用：默认 agent 主干、客户端通过 [`@deepseek
 | `@deepseek-ai/dsh-session-query-sqlite` | 派生的精确／FTS 会话查询服务；先于 ACP 传输打开，使叶节点消费方在首次模型请求前就绪。 |
 | `@deepseek-ai/dsh-acp` | 通过 stdin／stdout 提供的纯自动化 ACP 传输。 |
 
-应用不安装命令、用户交互、会话导航、配置选择器或 stdout logger。它通过一个有序 effect 拥有这些插件，因此查询服务会在 ACP 接受工作前就绪，而 ACP 会话会在检查点与持久化分离前静默。叶节点配置负责提供 LLM、执行器、沙箱、批准、文件系统和面向模型的工具插件。
+应用不安装命令、用户交互、会话导航、配置选择器或 stdout logger。它通过一个有序 effect 拥有这些插件，因此查询服务会在 ACP 接受工作前就绪，而 ACP 会话会在检查点与持久化插件卸载前完全停稳。叶节点配置负责提供 LLM（大语言模型）、执行器、沙箱、审批、文件系统和面向模型的工具插件。
 
 ## 配置
 
@@ -22,26 +22,26 @@ ACP 自动化服务器应用：默认 agent 主干、客户端通过 [`@deepseek
 |---|---|---|
 | `provider` | 必填 | 每个由 ACP 创建的 agent 所用的提供方路由。 |
 | `model` | 必填 | 每个由 ACP 创建的 agent 所用的模型。 |
-| `maxParallelToolCalls` | agent-loop 默认值 | 正整数工具调用并发上限；`1` 表示串行。 |
+| `maxParallelToolCalls` | agent loop 默认值 | 正整数工具调用并发上限；`1` 表示串行。 |
 | `persona` | 无 | 供 `dsh-system-prompt` 使用的部署 persona 模板。 |
 | `toolOrder` | 字典序 | 供 `dsh-system-prompt` 使用的显式面向模型工具顺序。 |
 | `tools` | `{ mode: 'native' }` | Native、Code Mode 或组合式模型工具传输。 |
-| `dshHome` | `$DSH_HOME` 或 `~/.dsh` | bash 与本地 skill 发现共享的 harness 主目录。 |
+| `dshHome` | `$DSH_HOME` 或 `~/.dsh` | bash 与本地 skill（技能）发现共享的 harness 主目录。 |
 | `sessionTitle` | 主干示例限制 | 持久后备标题限制；标题仍不会进入 ACP wire。 |
 | `persistenceRoot` | `./.sessions` | JSONL 后端根目录，以及派生 `session-query.db` 索引的父目录。 |
-| `packChunks` | `true` | 在存储中打包连续的增量 chunk 事件。 |
+| `packChunks` | `true` | 在存储中打包连续的增量分片事件。 |
 | `persistenceCompression` | `zstd` | 带校验和的 Zstandard 帧，或原始 `none`。 |
-| `workspaceContext` | 必填 | Workspace 指令字节预算／配置，或 `false`。 |
-| `skills` | 拥有者默认值 | Skill 注册表、本地提供方和面向模型的 skill 工具。 |
+| `workspaceContext` | 必填 | 工作区指令字节预算／配置，或 `false`。 |
+| `skills` | 拥有者默认值 | skill 注册表、本地提供方和面向模型的 skill 工具。 |
 | `toolBash` | 拥有者默认值 | 面向模型的 bash 工具配置。 |
 | `toolTasks` | 拥有者默认值 | 通用后台任务控制配置，或 `false`。 |
-| `goals` | 拥有者默认值 | 持久的同会话目标领域与模型工具，或 `false`。 |
+| `goals` | 拥有者默认值 | 持久化的同会话目标领域与模型工具，或 `false`。 |
 
-已交付的 [`examples/acp-agent/cordis.yml`](../../../examples/acp-agent/cordis.yml) 添加 DeepSeek 适配器、沙箱化 bash 与文件系统提供方、一次性批准策略、压缩、subagent、工作流、钩子，以及面向模型的工具。应用提供派生会话查询索引，而面向模型的查询消费方仍由叶节点显式选用。快照 overlay 只替换非确定性提供方或策略值。
+已交付的 [`examples/acp-agent/cordis.yml`](../../../examples/acp-agent/cordis.yml) 添加 DeepSeek 适配器、沙箱化 bash 与文件系统提供方、一次性审批策略、压缩（compaction）、subagent、工作流、钩子，以及面向模型的工具。应用提供派生会话查询索引，而面向模型的查询消费方仍由叶节点显式选用。快照 overlay 只替换非确定性提供方或策略值。
 
 ## Bin
 
-`dsh-acp-demo [--config path-to-cordis.yml]`（短形式 `-c`；默认为 `./cordis.yml`）会加载 gitignore 排除的 `.env`，回放 mode 除外；`DSH_SNAPSHOT=replay` 选择同级 `cordis.snapshot.yml`；stdin EOF 会在退出前释放上下文并刷新会话。Loader 已安装的可选 peer `node-addon-require-builtin` 使纯 Node 下构建后的 bin 可以解析裸插件说明符。诊断使用 stderr，因为 stdout 是 ACP wire。
+`dsh-acp-demo [--config path-to-cordis.yml]`（短形式 `-c`；默认为 `./cordis.yml`）会加载 gitignore 排除的 `.env`，回放模式除外；`DSH_SNAPSHOT=replay` 选择同级 `cordis.snapshot.yml`；stdin EOF 会在退出前释放上下文并刷新会话。loader 已安装的可选对等依赖（peer dependency）`node-addon-require-builtin` 使纯 Node 下构建后的 bin 可以解析裸插件说明符。诊断使用 stderr，因为 stdout 是 ACP wire。
 
 ## 模型体验
 
@@ -49,10 +49,10 @@ ACP 自动化服务器应用：默认 agent 主干、客户端通过 [`@deepseek
 
 #### KV Cache 影响
 
-每个会话只追加；应用本身不添加请求前缀内容。
+每个会话仅追加；应用本身不添加请求前缀内容。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - **JSONL 持久化固定不变**：使用其他后端需要另一种组合。
 - **同级插件可能破坏 stdout**：应用无法阻止另一个条目写入非协议字节。
-- **只支持新建自动化会话**：恢复和人类交互属于其他前端入口。
+- **只支持新建自动化会话**：恢复和人工交互属于其他前端入口。
diff --git a/packages/examples/agent-spine-demo/README.i18n.yaml b/packages/examples/agent-spine-demo/README.i18n.yaml
index a7c71523b2..e08ce5d9ac 100644
--- a/packages/examples/agent-spine-demo/README.i18n.yaml
+++ b/packages/examples/agent-spine-demo/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/examples/agent-spine-demo/README.md
 README.md: 359e7153be2f480ba3fea4b06782acdc9f89ebb9
-README.zh.md: 57fec3f32f5bbc8f3d82ff8971d36d376d722753
+README.zh.md: acd8c06940b03e90e368314cd725846a2b92b656
diff --git a/packages/examples/agent-spine-demo/README.zh.md b/packages/examples/agent-spine-demo/README.zh.md
index 57fec3f32f..acd8c06940 100644
--- a/packages/examples/agent-spine-demo/README.zh.md
+++ b/packages/examples/agent-spine-demo/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-将 **默认的不含执行器、不含 UI 的 agent 主干** 作为一个 Cordis 组合包插件。它加载每个 harness agent 所需的固定服务集合，包括本地 skill 提供方，并将循环的 `agents` 列表作为自身配置转发。因此，应用包只需添加前端入口和可替换后端，就能组合出可工作的 agent。
+将 **默认的不含执行器、不含 UI 的 agent（智能体）主干** 作为一个 Cordis 组合包插件。它加载每个 harness agent 所需的固定服务集合，包括本地 skill（技能）提供方，并将循环的 `agents` 列表作为自身配置转发。因此，应用包（package）只需添加前端入口和可替换后端，就能组合出可工作的 agent。
 
 阅读此包可了解完整插件树及其组合顺序。
 
@@ -43,11 +43,11 @@
 
 主干包含每个前端入口都共有的全部组件。可替换组件和与前端入口耦合的组件留在外部，由加载组合包的一方选择：
 
-- **LLM 适配器**：组合包交付抽象 `llm` 服务；叶节点在 `ctx.llm` 上注册具体适配器（`llm-deepseek`、`llm-pi-ai`、`llm-replay`）。
-- **模型支持的会话标题提供方**：组合包挂载带可覆盖示例限制的后备服务（5 个词、40 个后备字节、80 个可接受标题字节）；叶节点可以恰好选用一个首消息或全消息 LLM 提供方。
+- **LLM（大语言模型）适配器**：组合包交付抽象 `llm` 服务；叶节点在 `ctx.llm` 上注册具体适配器（`llm-deepseek`、`llm-pi-ai`、`llm-replay`）。
+- **基于模型的会话标题提供方**：组合包挂载带可覆盖示例限制的后备服务（5 个词、40 个后备字节、80 个可接受标题字节）；叶节点可以恰好选用一个首消息或全消息 LLM 提供方。
 - **bash 执行器**：组合包交付 `tool-bash`（消费方 schema）；叶节点提供 `ctx.bash`（`bash-local` 或沙箱化实现）。
 - **非本地 skill 提供方**：组合包交付 skill 注册表、本地文件系统提供方和 `skill` 工具；部署可以把嵌入式目录或远程目录等其他提供方作为同级插件添加。
-- **前端入口与各应用基础设施**：终端 TUI 或 ACP 自动化传输，以及 `hmr`。应用包（[`dsh-tui-demo`](../tui-demo/README.md)、[`dsh-acp-demo`](../acp-demo/README.md)）拥有这些选择。`timer` 位于主干中，因为它是共有组件且不写 stdout；前端入口拥有 stdout，因此留在组合包外。
+- **前端入口与各应用基础设施**：终端 TUI 或 ACP（Agent Client Protocol）自动化传输，以及 `hmr`。应用包（[`dsh-tui-demo`](../tui-demo/README.md)、[`dsh-acp-demo`](../acp-demo/README.md)）拥有这些选择。`timer` 位于主干中，因为它是共有组件且不写 stdout；前端入口拥有 stdout，因此留在组合包外。
 
 这把[接口／实现／消费方 seam](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md) 提升到组合层：组合包拥有共享主干，叶节点拥有后端，应用包拥有前端入口。
 
@@ -59,25 +59,25 @@ import type { Config } from '@deepseek-ai/dsh-agent-spine-demo'
 // workspaceContext requires { maxBytes } or false; the other owner schemas supply defaults.
 ```
 
-组合包将每个字段转发给拥有它的子节点：`agents` 与 `maxParallelToolCalls` 交给 `agent-loop`（`agents` 默认为 `[]`，上限在该处默认），因此每个应用提供自己的预创建 agent；TUI 和无头应用预创建 `main`，ACP 应用则在 `session/new` 按需创建 agent；`persona` 与 `toolOrder` 交给 `dsh-system-prompt`；`tools` 交给工具注册表以配置呈现 mode；`sessionTitle` 交给后备标题服务；`skills.registry`、`skills.local` 与 `skills.tool` 分别交给 skill 注册表、本地提供方和面向模型的消费方；必填的 `workspaceContext` 选择交给 `dsh-workspace-context`（`{ maxBytes }` 启用加载，`false` 禁用）；`invariants` 交给不变式服务；`toolBash`/`toolTasks` 交给组合包拥有的两个面向模型工具插件。组合包始终挂载 `dsh-llm-retry`，而每个叶节点适配器拥有自己的嵌套 `retryPolicy`。省略 `sessionTitle` 时采用显式示例策略：5 个词、40 个后备字节、80 个可接受标题字节。`goals` 对象会选用持久领域、模型工具和同会话驱动器，并将 `goals.domain` 与 `goals.tool` 转发给各自拥有者；省略或设为 `false` 会让整个栈缺席，使无头调用方继续以一轮结算。设置 `skills.enabled: false` 会同时省略本地提供方和面向模型的 skill 工具；设置 `toolTasks: false` 会保留供前台生产方使用的任务服务，但不公开 `task_output`/`task_list`/`task_kill`。它对 `dshHome` 只解析一次，解析通过 [`@deepseek-ai/dsh-paths`](../../util/paths/README.md) 完成，并将所得绝对值转发给 tool-bash 的托管环境和已启用的本地 skill 发现。顶层 `dshHome` 缺席时采用 `skills.local.dshHome`；两者同时提供但解析后的路径不同会明确失败。`toolBash.enableRunInBackground` 只控制 bash 生产方；独立加载的生产方保留各自配置。Workspace 指令先于 skill 目录注册，因此其会话前缀消息先渲染。应用包使用 `pickSpineConfig()`，只复制这些由组合包拥有的字段。
+组合包将每个字段转发给拥有它的子节点：`agents` 与 `maxParallelToolCalls` 交给 `agent-loop`（`agents` 默认为 `[]`，上限在该处默认），因此每个应用提供自己的预创建 agent；TUI 和无头应用预创建 `main`，ACP 应用则在 `session/new` 按需创建 agent；`persona` 与 `toolOrder` 交给 `dsh-system-prompt`；`tools` 交给工具注册表以配置呈现模式；`sessionTitle` 交给后备标题服务；`skills.registry`、`skills.local` 与 `skills.tool` 分别交给 skill 注册表、本地提供方和面向模型的消费方；必填的 `workspaceContext` 选择交给 `dsh-workspace-context`（`{ maxBytes }` 启用加载，`false` 禁用）；`invariants` 交给不变式服务；`toolBash`/`toolTasks` 交给组合包拥有的两个面向模型工具插件。组合包始终挂载 `dsh-llm-retry`，而每个叶节点适配器拥有自己的嵌套 `retryPolicy`。省略 `sessionTitle` 时采用显式示例策略：5 个词、40 个后备字节、80 个可接受标题字节。`goals` 对象会选用持久化领域、模型工具和同会话 Goal Round 驱动器，并将 `goals.domain` 与 `goals.tool` 转发给各自拥有者；省略或设为 `false` 会让整个栈缺席，使无头调用方继续以单轮次结算。设置 `skills.enabled: false` 会同时省略本地提供方和面向模型的 skill 工具；设置 `toolTasks: false` 会保留供前台生产方使用的任务服务，但不公开 `task_output`/`task_list`/`task_kill`。它对 `dshHome` 只解析一次，解析通过 [`@deepseek-ai/dsh-paths`](../../util/paths/README.md) 完成，并将所得绝对值转发给 tool-bash 的托管环境和已启用的本地 skill 发现。顶层 `dshHome` 缺席时采用 `skills.local.dshHome`；两者同时提供但解析后的路径不同会明确失败。`toolBash.enableRunInBackground` 只控制 bash 生产方；独立加载的生产方保留各自配置。工作区指令先于 skill 目录注册，因此其会话前缀消息先渲染。应用包使用 `pickSpineConfig()`，只复制这些由组合包拥有的字段。
 
 例如，`{ invariants: { enabled: true, package_allowlist: ['^@deepseek-ai/dsh-'], package_blocklist: ['agent-loop$'] } }` 会让包拥有的配套插件保持挂载，但抑制被阻止的拥有者。Blocklist 匹配优先于 allowlist 匹配；正则表达式与生命周期规则见 [`dsh-invariants`](../../support/invariants/README.md)。
 
 ## 为何使用代码组合包，而非共享 YAML include
 
-YAML include 可以去重配置，却无法拥有 bin 或提供前端入口默认值。ACP 应用包默认接出协议纯净的 stdout，但叶节点仍可添加不安全的 logger。组合包子节点把服务注册到根 isolate-keyed store，因此注入这些服务的叶节点同级插件无需依赖加载顺序即可看到它们。
+YAML include 可以去重配置，却无法拥有 bin 或提供前端入口默认值。ACP 应用包默认接出协议纯净的 stdout，但叶节点仍可添加不安全的 logger。组合包子节点把服务注册到根 isolate-keyed store，因此叶节点的同级插件无需依赖加载顺序即可通过注入看到它们。
 
-重试策略可能在新的编号步骤中重复失败的请求。重试状态、提供方错误和失败的部分 chunk 不进入模型历史；每次提供方尝试仍可能产生计费；always mode 没有尝试次数上限；前端入口从所有已记录步骤推导用量；重建的请求保留先前前缀，以便复用提供方 cache。
+重试策略可能在新的编号步骤中重复失败的请求。重试状态、提供方错误和失败的部分分片不进入模型历史；每次提供方尝试仍可能产生计费；always 模式没有尝试次数上限；前端入口从所有已记录步骤推导用量；重建的请求保留先前前缀，以便复用提供方缓存。
 
 ## 模型体验
 
-模型通过 `dsh-system-prompt`、`dsh-tool-skill`、`dsh-tool-bash`、`dsh-tools` 和 `dsh-llm-retry` 间接获得体验；还会通过 `dsh-tool-goal` 与目标轮次提示词获得体验，前提是启用 `goals`。组合包自身不添加面向模型的包装内容。
+模型通过 `dsh-system-prompt`、`dsh-tool-skill`、`dsh-tool-bash`、`dsh-tools` 和 `dsh-llm-retry` 间接获得体验；还会通过 `dsh-tool-goal` 与 Goal Round 提示词获得体验，前提是启用 `goals`。组合包自身不添加面向模型的包装内容。
 
 #### KV Cache 影响
 
-不会直接失效；具名消费方拥有请求前缀的任何变更。
+不会直接失效；上述消费方负责请求前缀的任何变更。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
-- **大部分主干集合固定在代码中**：`apply()` 始终挂载核心服务与 `tool-bash`；配置可以省略组合包内的目标、skill 与任务控制工具，但要替换循环或删除其他主干成员，就必须组合另一个 bundle。
+- **大部分主干集合固定在代码中**：`apply()` 始终挂载核心服务与 `tool-bash`；配置可以省略组合包内的目标、skill 与任务控制工具，但要替换循环或删除其他主干成员，就必须组合另一个组合包。
 - **不变式 seam 与配套插件仍是固定成员**：`invariants.enabled: false` 或包筛选器会抑制检查，但不会移除服务或配套插件注册；Session 始终启用的校验与冻结是另一套机制。
diff --git a/packages/examples/cli-demo/README.i18n.yaml b/packages/examples/cli-demo/README.i18n.yaml
index 80715af77d..f73cad16df 100644
--- a/packages/examples/cli-demo/README.i18n.yaml
+++ b/packages/examples/cli-demo/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/examples/cli-demo/README.md
 README.md: b8f2bde962738a1a23f0e57218ab0f90e8e0b705
-README.zh.md: 322ba3fb3b253d867832534bd33f65df3a5b8d37
+README.zh.md: 0e03375ced4e087d44eed7ff33666abf1f2cec10
diff --git a/packages/examples/cli-demo/README.zh.md b/packages/examples/cli-demo/README.zh.md
index 322ba3fb3b..0e03375ced 100644
--- a/packages/examples/cli-demo/README.zh.md
+++ b/packages/examples/cli-demo/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-无头单次应用及 bin，用于在没有交互式 UI 或编辑器客户端的情况下运行一项 agent 任务。它组合 [`@deepseek-ai/dsh-agent-spine-demo`](../agent-spine-demo/README.md)、JSONL 持久化，以及恰好一个新建顶层 agent。Bin 提交任务，等待其持久轮次结束，渲染所选输出，释放至静默，然后退出。
+无头单次应用及 bin，用于在没有交互式 UI 或编辑器客户端的情况下运行一项 agent（智能体）任务。它组合 [`@deepseek-ai/dsh-agent-spine-demo`](../agent-spine-demo/README.md)、JSONL 持久化，以及恰好一个新建顶层 agent。bin 提交任务，等待其已持久化的轮次结束状态，渲染所选输出，执行 dispose（资源释放）直至完全停稳，然后退出。
 
-该包不挂载 console logger、交互式 UI、用户交互服务或 `ask_user_question` 工具。Stdout 专用于所选输出格式；诊断使用 stderr。
+该包（package）不挂载 console logger、交互式 UI、用户交互服务或 `ask_user_question` 工具。Stdout 专用于所选输出格式；诊断使用 stderr。
 
 ## 配置
 
@@ -12,26 +12,26 @@
 |---|---|---|
 | `provider` | 必填 | 已配置 agent 的提供方路由 |
 | `model` | 必填 | 已配置 agent 的模型 |
-| `maxParallelToolCalls` | agent-loop 默认值 | 正整数并发工具调用上限；`1` 表示串行 |
+| `maxParallelToolCalls` | agent loop 默认值 | 正整数并发工具调用上限；`1` 表示串行 |
 | `persona` | 无 | `dsh-system-prompt` 中的部署 persona |
 | `toolOrder` | 字典序 | `dsh-system-prompt` 中显式的面向模型工具顺序 |
 | `tools` | `{ mode: 'native' }` | 通过 `dsh-agent-spine-demo` 提供的工具注册表呈现配置 |
-| `dshHome` | `$DSH_HOME` 或 `~/.dsh` | 向模型 bash 公开并用于本地 skill 发现的 Harness 主目录 |
+| `dshHome` | `$DSH_HOME` 或 `~/.dsh` | 向模型 bash 公开并用于本地 skill（技能）发现的 harness 主目录 |
 | `sessionTitle` | 主干示例限制 | 通过 `dsh-agent-spine-demo` 提供的后备标题词数／字节限制 |
-| `skills` | 拥有者默认值 | Skill 注册表、本地提供方和面向模型的 skill 工具 |
-| `toolBash` | 拥有者默认值 | 面向模型的 bash 配置，包括此生产方对后台任务的选用 |
+| `skills` | 拥有者默认值 | skill 注册表、本地提供方和面向模型的 skill 工具 |
+| `toolBash` | 拥有者默认值 | 面向模型的 bash 配置，包括此生产方对后台任务的显式启用 |
 | `toolTasks` | 拥有者默认值 | 通用 `task_output` 等待边界 |
 | `persistenceRoot` | `./.sessions` | JSONL 会话根目录 |
-| `persistenceCompression` | `'zstd'` | JSONL 工件编码（`'zstd'` 或原始 `'none'`） |
-| `workspaceContext` | 必填 | Workspace 指令字节预算，或以 `false` 禁用加载 |
+| `persistenceCompression` | `'zstd'` | JSONL 产物编码（`'zstd'` 或原始 `'none'`） |
+| `workspaceContext` | 必填 | 工作区指令字节预算，或以 `false` 禁用加载 |
 
-## CLI 契约
+## CLI（命令行界面）契约
 
 ```sh
 dsh-cli-demo [--config path] [--output-format text|json|stream-json] <task>
 ```
 
-`--config` 默认为 `./cordis.yml`；`--output-format` 默认为 `text`。必须恰好提供一个非空位置任务，因此含空格的任务需要加引号。`--help` 在不启动的情况下打印用法。不存在 `-p` 或 `--print` 标志。
+`--config` 默认为 `./cordis.yml`；`--output-format` 默认为 `text`。必须恰好提供一个非空的任务位置参数，因此含空格的任务需要加引号。`--help` 在不启动的情况下打印用法。不存在 `-p` 或 `--print` 标志。
 
 根 headless-agent 示例提供其叶节点：
 
@@ -39,7 +39,7 @@ dsh-cli-demo [--config path] [--output-format text|json|stream-json] <task>
 pnpm run demo:headless "inspect the failing test and fix it"
 ```
 
-Loader 配置通过仓库安装的可选原生辅助程序解析裸包说明符，因此根命令不需要特殊 Node 标志。
+loader 配置通过仓库安装的可选原生辅助程序解析裸包说明符，因此根命令不需要特殊 Node 标志。
 
 ### 输出格式
 
@@ -47,32 +47,32 @@ Loader 配置通过仓库安装的可选原生辅助程序解析裸包说明符
 - `json` 写入一条 DSH 原生结果记录：`{ type: "result", success, sessionId, turn, result, reason, usage? }`。`usage` 对任务轮次中的每个模型步骤恰好求和一次，包括产生用量但没有提交 assistant 消息的已计费失败重试。
 - `stream-json` 将顶层会话任务轮次中的每个规范事件写成 `{ type: "session_event", sessionId, event }`，然后写入同一结果记录。子 agent 活动只通过父工具事件与结果出现。
 
-只有 `reason.kind === "completed"` 会成功退出。其他持久轮次结尾仍会输出部分文本或结果记录，向 stderr 添加诊断，并以非零状态退出。参数和启动失败会让 stdout 保持为空。SIGINT 与 SIGTERM 会取消活动工作，等待释放，并分别以 130 和 143 退出。
+只有 `reason.kind === "completed"` 会成功退出。其他已持久化的轮次结束状态仍会输出部分文本或结果记录，向 stderr 添加诊断，并以非零状态退出。参数和启动失败会让 stdout 保持为空。SIGINT 与 SIGTERM 会取消正在进行的工作，等待 dispose 完成，并分别以 130 和 143 退出。
 
 任务轮次会在最终输出前显式刷新。进程退出后，会话日志仍保留在 `persistenceRoot` 下。
 
 ## 操作安全
 
-Headless-agent 叶节点提供本地 bash、文件系统、skill、subagent、工作流和 todo 能力。因此任务可以修改启动 workspace、运行命令、生成子 agent，并消耗提供方 token。请从目标项目目录运行 CLI，检查叶节点的能力与沙箱配置，不要把非交互式执行当作批准边界。
+headless-agent 叶节点提供本地 bash、文件系统、skill、subagent、工作流和 todo 能力。因此任务可以修改启动工作区、运行命令、spawn 子 agent，并消耗提供方 token。请从目标项目目录运行 CLI，检查叶节点的能力与沙箱配置，不要把非交互式执行当作批准边界。
 
 ## 模型体验
 
 ### 单次任务轮次
 
-#### 模型所见
+#### 模型看到的内容
 
-位置任务会成为一条用户消息。通过 `dsh-agent-spine-demo`，顶层 agent 还会收到已配置的 workspace 指令与 persona、skill 目录、可见工具 schema，以及同一轮次后续步骤所需的保留工具结果。
+任务位置参数会成为一条用户消息。通过 `dsh-agent-spine-demo`，顶层 agent 还会收到已配置的工作区指令与 persona、skill 目录、可见工具 schema，以及同一轮次后续步骤所需的保留工具结果。
 
 #### Token 影响
 
-每个模型步骤中的任务、提示词段、工具 schema、assistant 输出和工具结果都会消耗 token。JSON 事件流和最终渲染不增加模型 token；委派的子工作有自己的模型用量，不计入父结果的 `usage` 总量。
+每个模型步骤中的任务、提示词段、工具 schema、assistant 输出和工具结果都会消耗 token。JSON 事件流式输出和最终渲染不增加模型 token；委派的子工作有自己的模型用量，不计入父结果的 `usage` 总量。
 
 #### KV Cache 影响
 
-只要单次 agent 的提示词、schema、模型路由和会话前缀保持不变，工具轮次历史就只追加。改变该组合会建立不同的请求前缀；JSON 输出 mode 不影响 cache。
+只要单次 agent 的提示词、schema、模型路由和会话前缀保持不变，工具轮次历史就仅追加。改变该组合会建立不同的请求前缀；JSON 输出模式不影响缓存。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
-- **每个进程只创建一个新的顶层会话**：其 workspace cwd 是启动目录；此应用不支持恢复、第二条提示词、stdin 上下文或并发顶层会话。
-- **没有交互式问题或批准提供方**：需要人类回答的工具无法完成，除非其他叶节点按显式策略组合一个非交互式提供方。
-- **流只包含顶层会话**：子会话不会平铺到流中，聚合用量只涵盖父任务轮次记录的模型步骤。
+- **每个进程只创建一个新的顶层会话**：其工作区 cwd 是启动目录；此应用不支持恢复、第二条提示词、stdin 上下文或并发顶层会话。
+- **没有交互式问题或批准提供方**：需要人工回答的工具无法完成，除非其他叶节点按显式策略组合一个非交互式提供方。
+- **流式输出仅限顶层会话**：子会话不会平铺到流中，聚合用量只涵盖父任务轮次记录的模型步骤。
diff --git a/packages/examples/jsonrpc-demo/README.i18n.yaml b/packages/examples/jsonrpc-demo/README.i18n.yaml
index 3b3bbdf8b3..4808b7defa 100644
--- a/packages/examples/jsonrpc-demo/README.i18n.yaml
+++ b/packages/examples/jsonrpc-demo/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/examples/jsonrpc-demo/README.md
 README.md: 75e9e3943982c08e53afdbb73d1e9085b2e332bc
-README.zh.md: 0e0f8ba33ab4cfbf57e28219a553a5d042b1cbed
+README.zh.md: 95a626285c2f531ff6c8cceed297b42626aed98c
diff --git a/packages/examples/jsonrpc-demo/README.zh.md b/packages/examples/jsonrpc-demo/README.zh.md
index 0e0f8ba33a..95a626285c 100644
--- a/packages/examples/jsonrpc-demo/README.zh.md
+++ b/packages/examples/jsonrpc-demo/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-只包含 bin 的应用，启动外部 `cordis.yml`；其 [`jsonrpc`](../../ui/jsonrpc/README.md) 入口通过按换行分隔的 stdio 为 SDK 客户端提供服务。配置负责组合主干、后端和服务插件。`lib/bin.js` 也是[单文件可执行 runtime](../../../.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md) 的入口。
+只包含 bin 的应用，启动外部 `cordis.yml`；其 [`jsonrpc`](../../ui/jsonrpc/README.md) 入口通过按换行分隔的 stdio 为 SDK 客户端提供服务。配置负责组合主干、后端和服务插件。`lib/bin.js` 也是[单文件可执行运行时](../../../.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md) 的入口。
 
 ## 配置发现
 
@@ -12,7 +12,7 @@
 
 ## 退出生命周期
 
-stdin EOF 和 `SIGTERM` 会将根上下文释放至静默并以 0 退出；`SIGINT` 完成同样的释放后以 130 退出。EOF 可能按[分发 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md) 所述截断正在处理的轮次。`jsonrpc` 插件拥有先响应再退出的协议关闭流程；两条路径均幂等，可以安全竞态。
+stdin EOF 和 `SIGTERM` 会 dispose（释放资源）根上下文，等待完全停稳后以 0 退出；`SIGINT` 完成同样的 dispose 后以 130 退出。EOF 可能按[分发 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md) 所述截断正在处理的轮次。`jsonrpc` 插件拥有先响应再退出的协议关闭流程；两条路径均幂等，即使发生竞态也安全。
 
 ## stdout 是协议
 
@@ -20,14 +20,14 @@ stdout 只承载 JSON-RPC 帧。bin 和启动守卫在 stderr 上输出诊断，
 
 ## 模型体验
 
-模型通过外部 `cordis.yml` 加载的插件间接获得体验；每个插件拥有自身面向模型的提示词、schema、消息和结果，此 bin 不添加任何内容。
+模型体验由外部 `cordis.yml` 加载的插件间接提供；这些插件负责所有面向模型的提示词、schema、消息和结果，此 bin 不添加任何内容。
 
 #### KV Cache 影响
 
-不会直接失效；具名消费方拥有请求前缀的任何变更。
+不会直接失效；由上述消费方负责请求前缀的任何变更。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - **bin 无法证明配置提供 JSON-RPC 服务**：不含 `dsh-jsonrpc` 条目的有效配置也能成功启动，但不会提供任何服务。
-- **不存在内置或默认配置**：每次启动都必须提供 `DSH_CORDIS_CONFIG` 或位置路径；部署拥有完整插件树和 stdout 纪律。
+- **不存在内置或默认配置**：每次启动都必须提供 `DSH_CORDIS_CONFIG` 或位置路径；部署方负责完整的插件树和 stdout 纪律。
 - **stdin EOF 会截断正在处理的工作**：客户端消失时立即释放根上下文；需要有序完成的调用方应使用协议级 `shutdown` 请求。
diff --git a/packages/examples/tui-demo/README.i18n.yaml b/packages/examples/tui-demo/README.i18n.yaml
index bf1e760913..a9815be3a3 100644
--- a/packages/examples/tui-demo/README.i18n.yaml
+++ b/packages/examples/tui-demo/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/examples/tui-demo/README.md
 README.md: 058ebe87af5f041bd19fbfb205a97753ccacf6b9
-README.zh.md: 254bee76dff0d400a7b133013e8322898599f73e
+README.zh.md: 43681e9c77cbd77459ec0539b69e0ab73274d2c1
diff --git a/packages/examples/tui-demo/README.zh.md b/packages/examples/tui-demo/README.zh.md
index 254bee76df..43681e9c77 100644
--- a/packages/examples/tui-demo/README.zh.md
+++ b/packages/examples/tui-demo/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-全屏终端应用组合包：一个 Cordis 插件，组合 [`@deepseek-ai/dsh-agent-spine-demo`](../agent-spine-demo/README.md)、持久的同会话目标、人类命令注册表与 `/goal` 生产方、JSONL 持久化、键盘支持的用户交互、预创建的 `main` agent，以及 [`@deepseek-ai/dsh-tui`](../../ui/tui/README.md)。一份 `cordis.yml` 将它作为单个条目挂载；[`dsh`](../../../apps/cli/README.md) CLI 是启动此类配置的前端入口。
+全屏终端应用组合包：一个 Cordis 插件，组合 [`@deepseek-ai/dsh-agent-spine-demo`](../agent-spine-demo/README.md)、持久化的同会话目标、人类命令注册表与 `/goal` 生产方、JSONL 持久化、键盘支持的用户交互、预创建的 `main` agent（智能体），以及 [`@deepseek-ai/dsh-tui`](../../ui/tui/README.md)。一份 `cordis.yml` 将它作为一个 Cordis 配置项挂载；[`dsh`](../../../apps/cli/README.md) CLI（命令行界面）是启动此类配置的入口。
 
 管道、脚本和其他非交互式运行应使用 [`@deepseek-ai/dsh-cli-demo`](../cli-demo/README.md)。此组合包需要一对 TTY，不提供面向行的回退。
 
@@ -11,16 +11,16 @@
 | 插件 | 设置在此处的原因 |
 |---|---|
 | `@deepseek-ai/dsh-agent-spine-demo` | 共享服务、面向模型的工具，以及一个已配置的 `main` agent |
-| `@deepseek-ai/dsh-commands` | 供 TUI 和命令插件消费的纯人类命令发现与分发 |
-| `@deepseek-ai/dsh-command-goal` | 直接在主干的持久目标栈上提供 `/goal` 状态与变更 |
+| `@deepseek-ai/dsh-commands` | 供 TUI 和命令插件消费、仅面向人类的命令发现与分发 |
+| `@deepseek-ai/dsh-command-goal` | 直接在主干的持久化目标栈上提供 `/goal` 状态与变更 |
 | `@deepseek-ai/dsh-session-persistence-jsonl` | 位于 `persistenceRoot` 下的持久会话日志 |
 | `@deepseek-ai/dsh-session-checkpoint-policy` | 模型请求和顶层工具 effect 前的语义持久性屏障，以及已完成步骤的检查点 |
 | `@deepseek-ai/dsh-session-query-sqlite` + `@deepseek-ai/dsh-session-reference` | TUI 消费的组合式精确／FTS 会话查询与有界 `@session` 快照；面向模型的查询工具仍由叶节点选用 |
 | `@deepseek-ai/dsh-user-interaction` | 与提供方无关的人类问题服务 |
-| `@deepseek-ai/dsh-tui` | 全屏记录、编辑器、工具卡片、计划与问题 overlay |
+| `@deepseek-ai/dsh-tui` | 全屏 transcript（文本记录）、编辑器、工具卡片、计划与问题 overlay |
 | `@deepseek-ai/dsh-tool-ask-user` | 面向模型的 `ask_user_question` 工具 |
 
-可替换的 LLM、bash、文件系统和其他能力提供方仍留在叶节点配置中。`@cordisjs/plugin-hmr` 也仍是仅叶节点使用的开发条目，因为它需要 Loader 内部表层。
+可替换的 LLM（大语言模型）、bash、文件系统和其他能力提供方仍留在叶节点配置中。`@cordisjs/plugin-hmr` 也仍是仅叶节点使用的开发条目，因为它需要 Loader 内部实现。
 
 ## 配置
 
@@ -31,27 +31,27 @@
 | `maxParallelToolCalls` | agent-loop 默认值 | 组合包内循环的并发上限 |
 | `persona` | 无 | 系统提示词 persona 模板 |
 | `toolOrder` | 字典序 | 显式的面向模型工具顺序 |
-| `tools` | 拥有者默认值 | 工具呈现 mode |
-| `dshHome` | 拥有者默认值 | bash 与 skill 使用的 Harness 主目录 |
+| `tools` | 拥有者默认值 | 工具呈现模式 |
+| `dshHome` | 拥有者默认值 | bash 与 skill（技能）使用的 harness 主目录 |
 | `sessionTitle` | 主干示例限制 | 后备标题词数／字节限制 |
-| `skills` | 拥有者默认值 | Skill 注册表、本地提供方和工具配置 |
+| `skills` | 拥有者默认值 | skill 注册表、本地提供方和工具配置 |
 | `toolBash` | 拥有者默认值 | 面向模型的 bash 工具配置 |
 | `toolTasks` | 拥有者默认值 | 后台任务控制工具配置，或 `false` |
 | `goals` | 拥有者默认值 | 持久目标领域与模型工具配置；`false` 会移除目标栈与 `/goal` 生产方 |
 | `workspaceContext` | 必填 | Workspace 指令配置，或 `false` |
 | `persistenceRoot` | `./.sessions` | JSONL 持久化根目录，以及派生 `session-query.db` 索引的父目录 |
-| `persistenceCompression` | `'zstd'` | JSONL 工件编码（`'zstd'` 或原始 `'none'`） |
+| `persistenceCompression` | `'zstd'` | JSONL 产物编码（`'zstd'` 或原始 `'none'`） |
 | `sessionReferences` | 服务默认值 | 路由到 `dsh-session-reference` 的跨会话候选项与快照限制 |
 | `welcome` | `ready.` | TUI 副标题 |
 | `resumeCommand` | 无 | 退出和无宿主回退的命令模板；选择器本身使用会话查询与宿主移交 |
-| `ui` | 拥有者默认值 | 推理、颜色、卡片高度等 TUI 呈现设置 |
+| `ui` | 拥有者默认值 | 推理（reasoning）、颜色、卡片高度等 TUI 呈现设置 |
 | `resumeSessionId` | 无 | 要恢复的确切持久化会话 |
 
-新运行会创建 `main-session-<uuid>` 会话 id，并将它同时传给 TUI 与已配置的 agent。恢复运行会将两个组件都绑定到 `resumeSessionId`。TUI 先于主干挂载，因此它可以渲染匹配的配置启动失败，而不会留下空白终端。应用为 `/resume` 组合持久化和会话查询；嵌入宿主还可以提供 `tuiResumeHost`，以原地移交进程。
+新运行会创建 `main-session-<uuid>` 会话 id，并将它同时传给 TUI 与已配置的 agent。恢复运行会将两个组件都绑定到 `resumeSessionId`。TUI 先于主干挂载，因此它可以渲染匹配的配置启动失败，而不会留下空白终端。应用为 `/resume` 组合持久化和会话查询；嵌入宿主还可以提供 `tuiResumeHost`，用于原地移交进程。
 
-## 前端入口
+## 入口
 
-此包不交付 bin。[`dsh`](../../../apps/cli/README.md) CLI 是终端前端入口：裸 `dsh` 启动已交付的 `examples/tui-agent/cordis.yml`（它挂载此组合包），而 `dsh --config <path-to-cordis.yml>` 启动另一个挂载此组合包的叶节点配置。它加载 cwd 下可选的 `.env`，驱动 Cordis Loader，并等待完整插件树。仓库安装了 Loader 的可选原生辅助程序，因此裸包说明符可以在纯 Node 下解析。
+此包（package）不交付 bin。[`dsh`](../../../apps/cli/README.md) CLI 是终端入口：裸 `dsh` 启动已交付的 `examples/tui-agent/cordis.yml`（它挂载此组合包），而 `dsh --config <path-to-cordis.yml>` 启动另一个挂载此组合包的叶节点配置。它加载 cwd 下可选的 `.env`，驱动 Cordis Loader，并等待完整插件树。仓库安装了 Loader 的可选原生辅助程序，因此裸包说明符可以在纯 Node 下解析。
 
 ## 叶节点示例
 
@@ -78,21 +78,21 @@
 
 ### 交互式终端轮次
 
-#### 模型所见
+#### 模型看到的内容
 
-每次非空、非命令的编辑器提交都会成为用户消息；运行中轮次内的提交成为 steering。斜杠命令输入和输出仍只面向人类，而已接受的 `/goal` 变更会追加领域拥有的模型可见状态。共享主干提供已配置的 persona、workspace 指令、skill 目录、目标控制和可见工具 schema。TUI 渲染本身对模型不可见。
+每次非空、非命令的编辑器提交都会成为用户消息；运行中轮次内的提交成为 steering（中途引导）。斜杠命令输入和输出仍只面向人类，而已接受的 `/goal` 变更会追加领域拥有的模型可见状态。共享主干提供已配置的 persona、workspace 指令、skill 目录、目标控制和可见工具 schema。TUI 渲染本身对模型不可见。
 
 #### Token 影响
 
-用户、assistant 与工具历史按常规会话和压缩规则增长。Header、卡片、计划、Markdown 样式和快捷键不增加 token。
+用户、assistant 与工具历史按常规会话和压缩（compaction）规则增长。Header、卡片、计划、Markdown 样式和快捷键不增加 token。
 
 #### KV Cache 影响
 
-只要组合后的提示词、schema、路由和保留历史前缀保持稳定，就只追加。组合变更与压缩可能从第一个变化的 token 起使复用失效。
+只要组合后的提示词、schema、路由和保留历史前缀保持稳定，就保持仅追加。组合方式变更与 compaction 可能从第一个变化的 token 起使复用失效。
 
 ### 人类问题答案
 
-#### 模型所见
+#### 模型看到的内容
 
 `ask_user_question` 会保留工具调用，以及 `dsh-tool-ask-user` 定义的精简答案或稳定中断错误。问题 overlay 只在终端显示。
 
@@ -102,11 +102,11 @@
 
 #### KV Cache 影响
 
-只追加；答案跟在可复用请求前缀之后。
+仅追加；答案跟在可复用请求前缀之后。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - **只支持 TTY**：stdin 与 stdout 都必须是终端；自动化使用 `dsh-cli-demo`。
-- **一个已配置的终端会话**：记录与编辑器绑定到一个确切会话 id。
+- **一个已配置的终端会话**：transcript 与编辑器绑定到一个确切会话 id。
 - **应用集群固定不变**：JSONL 持久化与 ask-user 工具内置；不同策略需要另一种组合。
 - **批准机制独立存在**：此应用回答 `ctx.userInteraction`，而不是 `ctx.approval`；权限提示需要批准服务和回答方。
diff --git a/packages/fs/README.i18n.yaml b/packages/fs/README.i18n.yaml
index 1ead814be1..d856f1f5e6 100644
--- a/packages/fs/README.i18n.yaml
+++ b/packages/fs/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/README.md
 README.md: 4d954455ea920be4882530bcfe90b48a364c29b5
-README.zh.md: e818210abaded987edbb8bf38c6d1b43d40ad9c7
+README.zh.md: ed154cbbaf7c7f45c42c17799957254fee918153
diff --git a/packages/fs/README.zh.md b/packages/fs/README.zh.md
index e818210aba..ed154cbbaf 100644
--- a/packages/fs/README.zh.md
+++ b/packages/fs/README.zh.md
@@ -2,19 +2,19 @@
 
 [English](README.md) | 中文
 
-文件系统栈包括：提供方 seam（文本 I/O 与带可选版本防护的原子变更）、本地实现、政策门禁插件（已观察状态、编辑前读取、版本防护的写入/编辑）、面向模型的文件工具与执行器，以及基于 bash 的发现工具。全部都是**产品** 包。
+文件系统栈包括：提供方 seam（文本 I/O 与带可选版本防护的原子变更）、本地实现、策略门禁插件（已观察状态、编辑前读取、版本防护的写入/编辑）、面向模型的文件工具与执行器，以及基于 bash 的发现工具。全部都是**产品**包（package）。
 
 | 包 | 角色 | ctx 键 |
 |---|---|---|
-| `fs/` | 提供方 seam：文本 I/O 与原子变更原语（可选版本防护）；拥有 `fs/*` 政策事件 | `ctx.fs` |
+| `fs/` | 提供方 seam：文本 I/O 与原子变更原语（可选版本防护）；拥有 `fs/*` 策略事件 | `ctx.fs` |
 | `fs-local/` | 本地文件系统 `FileSystem` 实现 | （注册 `ctx.fs`） |
-| `fs-sandbox/` | 强制沙箱的 `FileSystem`：扩展 `fs-local`，并按每次调用的模式与工作区根政策约束写入/编辑（只读模式拒绝，工作区写入模式限制在会话工作区与临时根目录内）；读取直接通过 | （注册 `ctx.fs`） |
-| `fs-policy/` | 政策门禁插件：通过 `fs/*` 事件门禁提供已观察状态、编辑前读取和版本防护的写入/编辑 | （无服务，仅有 `fs/*` 监听器） |
-| `tool-fs/` | 面向模型的 `read`/`write`/`edit` 工具以及执行器（通过 `ctx.fs` 读取，拥有读取窗口逻辑，分派 `fs/*`）；为会话 cwd 相对路径保留文件系统语义，并在已挂载的 `ctx.fs` 实施约束时声明沙箱升级字段 | （注册到 `ctx.tools`） |
+| `fs-sandbox/` | 强制沙箱的 `FileSystem`：扩展 `fs-local`，并按每次调用的模式与工作区根目录策略约束写入/编辑（只读模式拒绝，工作区写入模式限制在会话工作区与临时根目录内）；读取则直接放行 | （注册 `ctx.fs`） |
+| `fs-policy/` | 策略门禁插件：通过 `fs/*` 事件门禁提供已观察状态、编辑前读取和版本防护的写入/编辑 | （无服务，仅有 `fs/*` 监听器） |
+| `tool-fs/` | 面向模型的 `read`/`write`/`edit` 工具以及执行器（通过 `ctx.fs` 读取，拥有读取窗口逻辑，分派 `fs/*`）；为会话 cwd 相对路径保留文件系统语义，并在已挂载的 `ctx.fs` 实施约束时声明沙箱升权字段 | （注册到 `ctx.tools`） |
 | `tool-fs-search/` | 面向模型的 `glob`/`grep` 发现工具；当 `rg` 位于 bash 执行器 `PATH` 上时注册，通过 `ctx.bash` 运行固定 ripgrep 命令，而不是使用 `ctx.fs` 提供方方法 | （注册到 `ctx.tools`） |
 
-接口位于 `fs/fs/`。沙箱化、远程或限定项目作用域的文件系统后端可以替换 `fs-local`，而无需更改 seam、政策门禁或面向模型的工具 schema；`fs-sandbox` 是第一个这样的替代实现（基于共享沙箱模式的进程内路径围栏；见[跨能力族 fs 沙箱 Agent Note](../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)）。政策（`fs-policy/`）是一个只通过 `fs/*` 事件门禁参与的插件，不是工具注入的服务；因此移除它会平稳失去政策，留下不受约束的裸提供方，而不会破坏工具。加载 `tool-fs/` 的部署也应加载该插件。模式围栏与编辑前读取门禁彼此正交，可以组合。发现（`tool-fs-search/`）有意不扩展提供方 seam：搜索是在 bash 执行器上运行 `rg`、由进程支持的工作流，因此文件系统后端无需承担通用搜索契约；只有当执行器能找到 `rg` 时，其工具才会注册。如果 bash 工作目录与 `read` 根目录是同一工作区，结果就能继续读取，这也是其 README 所述的共置部署。
+接口位于 `fs/fs/`。沙箱化、远程或限定项目作用域的文件系统后端可以替换 `fs-local`，而无需更改 seam、策略门禁或面向模型的工具 schema；`fs-sandbox` 是第一个这样的替代实现（基于共享沙箱模式的进程内路径围栏；见[跨能力族 fs 沙箱 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)）。策略（`fs-policy/`）是一个只通过 `fs/*` 事件门禁参与的插件，不是工具注入的服务；因此移除它只会使策略失效，留下不受约束的裸提供方，而不会破坏工具。加载 `tool-fs/` 的部署也应加载该插件。模式围栏与编辑前读取门禁彼此正交，可以组合。发现（`tool-fs-search/`）有意不扩展提供方 seam：搜索是在 bash 执行器上运行 `rg`、基于进程的工作流，因此文件系统后端无需承担通用搜索契约；只有当执行器能找到 `rg` 时，其工具才会注册。如果 bash 工作目录与 `read` 根目录是同一工作区，其结果便可供后续读取，这也是其 README 所述的共置部署。
 
 ## 文件 I/O 不设超时
 
-`read`/`write`/`edit` **不** 接受 `timeoutMs`，提供方 seam 也不启动 deadline。这与 bash 和 web（两者使用 [`@deepseek-ai/dsh-timeout`](../util/timeout/README.md)）及基于 bash 的 `glob`/`grep` 不同（它们声明的 `timeoutMs` 由 `@deepseek-ai/dsh-timeout-policy` 强制执行）：这些工作由进程支持，deadline 可以实际终止工作。本地系统调用至多只能尽力中止：超时无法强制正在进行的 `fsync`/`rename` 停止，因此这里的 deadline 会成为无法兑现承诺的配置项。在此添加 deadline 还会在「显式优于隐式」明确禁止的地方引入隐式默认值。两个参考 agent（Claude Code、Codex）出于同一原因都不为文件 I/O 计时；取消仍通过工具执行信号传播，在系统调用边界尽力中止。
+`read`/`write`/`edit` **不** 接受 `timeoutMs`，提供方 seam 也不设置截止期限。这与 bash 和 web（两者使用 [`@deepseek-ai/dsh-timeout`](../util/timeout/README.md)）及基于 bash 的 `glob`/`grep` 不同（它们声明的 `timeoutMs` 由 `@deepseek-ai/dsh-timeout-policy` 强制执行）：这些工作基于进程运行，截止期限可以实际终止工作。本地系统调用至多只能尽力中止：超时无法强制正在进行的 `fsync`/`rename` 停止，因此这里的截止期限会成为无法兑现承诺的配置项。在此添加截止期限还会在「显式优于隐式」明确禁止的地方引入隐式默认值。两个参考 agent（Claude Code、Codex）出于同一原因都不为文件 I/O 计时；取消仍通过工具执行信号传播，在系统调用边界尽力中止。
diff --git a/packages/fs/fs-local/README.i18n.yaml b/packages/fs/fs-local/README.i18n.yaml
index 7533e614c9..5a567ea309 100644
--- a/packages/fs/fs-local/README.i18n.yaml
+++ b/packages/fs/fs-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/fs-local/README.md
 README.md: 6d344fa3fef7f6bda6c0daa50184661156a925a7
-README.zh.md: 90813831768c09676cfd1b0f053e13e87c1c2fe8
+README.zh.md: 3d488004aa638931bcbf3a7b660ad687220101da
diff --git a/packages/fs/fs-local/README.zh.md b/packages/fs/fs-local/README.zh.md
index 9081383176..3d488004aa 100644
--- a/packages/fs/fs-local/README.zh.md
+++ b/packages/fs/fs-local/README.zh.md
@@ -14,24 +14,24 @@ await ctx.plugin(LocalFileSystem, { cwd: process.cwd() })
 
 ## 行为
 
-- **`resolve(path, opts?)`**：相对 `path` 在调用方提供 `opts.cwd` 时以该值为基准解析（面向模型的工具会传入调用 agent（智能体）的会话 cwd；见[每会话 cwd Agent Note](../../../.agents/notes/implemented/architecture/2026-07-02-fs-per-session-cwd.md)），否则以 `config.cwd` 为基准（默认 `process.cwd()`）；绝对 `path` 会忽略两者。`opts.signal` 会在本地解析前后检查，远程同级后端则可以用它中止往返。`targetKey` 是文件的 `realpath`，因此经符号链接到达同一文件的两个输入路径会共享一个身份，写入/编辑落在链接目标上，同时保留链接。尚不存在的路径在父目录存在时使用 realpath 后的父目录加 basename；只有父目录无法解析时才回退到绝对路径。`displayPath` 是绝对但未经解析的路径。
-- **`stat` / `lstat`**：返回目标元数据；目标不存在时返回 `undefined`。`stat` 为已解析目标报告 `FsInfo`（`version` 是由 bigint `dev:ino:size:mtimeNs:ctimeNs` 派生的不透明 token，`type` 为 `file`/`directory`/`other`，`size` 以字节计）；路径形态的 `lstat` 不跟随最后一个符号链接，报告 `FsPathInfo`，因此可以返回 `symlink`。两者都会在异步元数据探测前后检查取消，因此飞行中的中止会报告 `FS_ABORTED`，而非陈旧的不存在结果。
-- **`readText` / `streamText`**：只支持 UTF-8。`readText` 读取整个文件；`streamText` 按分片流式读取（跨分片解码），因此超大文件无需整体保存在内存中。两者都会拒绝无效 UTF-8、包含 NUL 字节的二进制样本（`FS_NOT_TEXT`）以及非普通文件目标。`read` 工具（`@deepseek-ai/dsh-tool-fs`）按大小决定调用哪个方法，并拥有行窗口逻辑。
+- **`resolve(path, opts?)`**：相对 `path` 在调用方提供 `opts.cwd` 时以该值为基准解析（面向模型的工具会传入调用 agent（智能体）的会话 cwd；见[每会话 cwd Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-02-fs-per-session-cwd.md)），否则以 `config.cwd` 为基准（默认 `process.cwd()`）；绝对 `path` 会忽略两者。`opts.signal` 会在本地解析前后检查，远程同级后端则可以用它中止往返。`targetKey` 是文件的 `realpath`，因此经符号链接到达同一文件的两个输入路径会共享一个身份，写入/编辑落在链接目标上，同时保留链接。尚不存在的路径在父目录存在时使用 realpath 后的父目录加 basename；只有父目录无法解析时才回退到绝对路径。`displayPath` 是绝对但未经解析的路径。
+- **`stat` / `lstat`**：返回目标元数据；目标不存在时返回 `undefined`。`stat` 为已解析目标报告 `FsInfo`（`version` 是由 bigint `dev:ino:size:mtimeNs:ctimeNs` 派生的不透明 token，`type` 为 `file`/`directory`/`other`，`size` 以字节计）；路径形态的 `lstat` 不跟随最后一个符号链接，报告 `FsPathInfo`，因此可以返回 `symlink`。两者都会在异步元数据探测前后检查取消，因此异步探测进行期间发生的中止会报告 `FS_ABORTED`，而非已失效的「不存在」结果。
+- **`readText` / `streamText`**：只支持 UTF-8。`readText` 读取整个文件；`streamText` 按分片流式读取（跨分片解码），因此超大文件无需整体保存在内存中。两者都会拒绝无效 UTF-8、包含 NUL 字节的二进制样本（`FS_NOT_TEXT`）以及非普通文件目标。`read` 工具（`@deepseek-ai/dsh-tool-fs`）按大小决定调用哪个方法，并负责行窗口逻辑。
 - **`listDir`**：按稳定的 `name.localeCompare()` 顺序列出一层目录。每个条目携带子项 basename、类型、解析后的子目标（`displayPath` 位于所列目录下，`targetKey` 是 realpath 身份）和低成本 stat 元数据（`version`，普通文件另有 `size`）。它绝不会打开或解码文件内容。缺失目标报告 `FS_NOT_FOUND`，文件/特殊文件目标报告 `FS_NOT_DIRECTORY`，已中止调用报告 `FS_ABORTED`，权限失败报告 `FS_PERMISSION_DENIED`，其他列出或子项元数据 I/O 失败报告 `FS_IO_ERROR`。损坏/消失的子项以无元数据的 `other` 返回，但解析子项时出现权限/I/O 失败会让整个列表以结构化 `FsError` 失败。
-- **`writeText`**：原子写入。它会向排他打开的临时文件（`wx`、`0o600`）写入；该文件位于目标旁随机命名的私有暂存目录（`0o700`）内。完成写入和 fsync 后，以 rename 覆盖目标。现有文件的 mode 会保留，新文件默认为 `0o600`；Windows 上的新文件继承目标目录的 DACL，而替换会在写入前把目标 DACL 复制到空临时文件，并通过 `ReplaceFileW` 发布，使原访问政策得以保留（见 [Windows DACL 保留 Agent Note](../../../.agents/notes/implemented/bug-fix/2026-07-19-windows-atomic-write-dacl-preservation.md)）。`expected` 防护是可选的：省略时无条件创建或覆盖；`createIfAbsent` 创建缺失目标并拒绝现有目标（`FS_NOT_OBSERVED`）；`replaceIfVersion` 只在观察到的版本上替换（目标缺失或版本不匹配均为 `FS_STALE_VERSION`）。
-- **`editText`**：在同一原语之上依次执行原子的字面量读取、修改和写入，并通过变更锁按目标串行化。`expected` 防护是可选的：提供时，会在字面量匹配之前校验版本（陈旧编辑报告 `FS_STALE_VERSION`，绝不会针对较新内容报告 `FS_EDIT_NOT_FOUND`/`FS_AMBIGUOUS_EDIT`）；省略时，无条件编辑当前内容。无论哪种情况，目标缺失都报告 `FS_STALE_VERSION`。匹配时规范化为 LF，随后恢复文件主要的 CRLF/LF 风格；空 `oldString` / 零匹配报告 `FS_EDIT_NOT_FOUND`，未设置 `replace_all` 的多个匹配则报告 `FS_AMBIGUOUS_EDIT`。
+- **`writeText`**：原子写入。它会向排他打开的临时文件（`wx`、`0o600`）写入；该文件位于目标旁随机命名的私有暂存目录（`0o700`）内。完成写入和 fsync 后，以 rename 覆盖目标。现有文件的 mode 会保留，新文件默认为 `0o600`；Windows 上的新文件继承目标目录的 DACL，而替换会在写入前把目标 DACL 复制到空临时文件，并通过 `ReplaceFileW` 发布，使原访问策略得以保留（见 [Windows DACL 保留 Agent Note](../../../.agents/notes/implemented/bug-fix/2026-07-19-windows-atomic-write-dacl-preservation.md)）。`expected` 防护是可选（OPTIONAL）的：省略时无条件创建或覆盖；`createIfAbsent` 创建缺失目标并拒绝现有目标（`FS_NOT_OBSERVED`）；`replaceIfVersion` 只在观察到的版本上替换（目标缺失或版本不匹配均为 `FS_STALE_VERSION`）。
+- **`editText`**：在同一原语之上执行原子式的字面量读取-修改-写入，并通过变更锁按目标串行化。`expected` 防护是可选（OPTIONAL）的：提供时，会在字面量匹配之前校验版本（陈旧编辑报告 `FS_STALE_VERSION`，绝不会针对较新内容报告 `FS_EDIT_NOT_FOUND`/`FS_AMBIGUOUS_EDIT`）；省略时，无条件编辑当前内容。无论哪种情况，目标缺失都报告 `FS_STALE_VERSION`。匹配时规范化为 LF，随后恢复文件主要的 CRLF/LF 风格；空 `oldString` / 零匹配报告 `FS_EDIT_NOT_FOUND`，未设置 `replace_all` 的多个匹配则报告 `FS_AMBIGUOUS_EDIT`。
 
-包根 SDK 接口包含默认/具名 `LocalFileSystem` 类和 `Config`。原始 I/O 位于 `src/fsio.ts`（不依赖 Cordis，单独进行单元测试）；`src/index.ts` 是轻量服务接线。
+包（package）根目录的 SDK 接口包含默认/具名 `LocalFileSystem` 类和 `Config`。原始 I/O 位于 `src/fsio.ts`（不依赖 Cordis，单独进行单元测试）；`src/index.ts` 是轻量服务接线。
 
 ## 模型体验
 
-通过 [`dsh-tool-fs`](../tool-fs/README.md) 间接产生影响；该消费方把本提供方带行窗口的 UTF-8 内容、变更确认和精确提供方消息渲染为有上限且保留的结果，而版本、原子写入机制和目录元数据保持内部可见。
+通过 [`dsh-tool-fs`](../tool-fs/README.md) 间接产生影响；该消费方在有上限的保留结果中渲染本提供方带行窗口的 UTF-8 内容、变更确认和精确提供方消息，而版本、原子写入机制和目录元数据仍属内部实现。
 
 #### KV Cache 影响
 
 不会直接使缓存失效；具名消费方负责请求前缀的任何变化。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **`config.cwd` 不是沙箱**：它是解析默认值，而非约束；绝对路径和 `..` 可以逃逸。请使用更严格的 `ctx.fs` 后端或 `tools/execute` waterfall（瀑布式事件）上的权限插件实施约束（见[能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-17-filesystem-capability-seam.md#consequences)）。
 - **覆盖会把整个旧文件读入内存**：只用于 UI diff；在大小阈值之上限制这次预读取的工作延期处理（`TODO(overwrite-diff-bound)`）。
diff --git a/packages/fs/fs-policy/README.i18n.yaml b/packages/fs/fs-policy/README.i18n.yaml
index 548f9b530c..6690227dbc 100644
--- a/packages/fs/fs-policy/README.i18n.yaml
+++ b/packages/fs/fs-policy/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/fs-policy/README.md
 README.md: dc4e9377793570c80b8d71ec84196bebe7fe583a
-README.zh.md: 956eb132f8ba42cbdf585d21d86ec7098ecb3446
+README.zh.md: aa0cb25899f5906ac9f531583ba48d01ad6095b4
diff --git a/packages/fs/fs-policy/README.zh.md b/packages/fs/fs-policy/README.zh.md
index 956eb132f8..aa0cb25899 100644
--- a/packages/fs/fs-policy/README.zh.md
+++ b/packages/fs/fs-policy/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-**fs-policy 插件**：它在 `ctx.fs` 提供方 seam（[`@deepseek-ai/dsh-fs`](../fs)）之上增加已观察状态、编辑前读取和版本防护的写入/编辑；它通过 `fs/*` 事件门禁参与，**不是** 通过方法服务。该插件**不** 注册 `ctx.fsPolicy` 服务，也没有公开的 `read`/`write`/`edit`/`resolve` 方法。它是文件系统栈的政策层：不是可替换 seam，而是不应位于 `FileSystem` 提供方基类上的政策。
+**fs-policy 插件**：它在 `ctx.fs` 提供方 seam（[`@deepseek-ai/dsh-fs`](../fs)）之上增加已观察状态、编辑前读取和版本防护的写入/编辑；它通过 `fs/*` 事件门禁参与，**不是**通过方法服务。该插件**不**注册 `ctx.fsPolicy` 服务，也没有公开的 `read`/`write`/`edit`/`resolve` 方法。它是文件系统栈的策略层：不是可替换 seam，而是不应位于 `FileSystem` 提供方基类上的策略。
 
 ```ts
 import type { Context } from 'cordis'
@@ -20,10 +20,10 @@ await ctx.plugin(FsPolicy)
 
 ## 四层拆分
 
-| 层 | 包 | 角色 |
+| 层 | 包（package） | 角色 |
 |---|---|---|
 | 工具/执行器 | `@deepseek-ai/dsh-tool-fs` | 面向模型的 schema、读取窗口和文本渲染；通过 `ctx.fs` 读取/写入/编辑，并分派 `fs/*` 事件 |
-| 政策 | `@deepseek-ai/dsh-fs-policy`（本包） | 已观察状态、编辑前读取和版本防护的写入/编辑，通过 `fs/*` 事件门禁贡献（无服务） |
+| 策略 | `@deepseek-ai/dsh-fs-policy`（本包） | 通过 `fs/*` 事件门禁提供已观察状态、编辑前读取和版本防护的写入/编辑（无服务） |
 | 提供方 seam | `@deepseek-ai/dsh-fs` | `ctx.fs`：文本 I/O 与原子变更原语（可选版本防护）；拥有 `fs/*` 事件词汇 |
 | 提供方 | `@deepseek-ai/dsh-fs-local` | `ctx.fs` 的本地实现 |
 
@@ -39,7 +39,7 @@ await ctx.plugin(FsPolicy)
 
 ## 已观察状态是先前观察记录；新鲜度由提供方 CAS 保证
 
-已观察状态是一张从弱引用所有者映射到目标版本的表，每次读取或变更成功后都会更新；记录存在本身就是先前观察凭据。插件不执行文件系统 I/O：它把观察到的版本提供给提供方的原子变更防护。窗口读取会观察整个文件的版本，因此只有文件保持不变时才允许后续的定向编辑。插件 dispose（资源释放）时会丢弃状态，并且不会跨会话持久化。
+已观察状态是一张以所有者为弱键、记录各目标版本的映射表，每次读取或变更成功后都会更新；记录存在本身就是先前观察凭据。插件不执行文件系统 I/O：它把观察到的版本提供给提供方的原子变更防护。窗口读取会观察整个文件的版本，因此只有文件保持不变时才允许后续的定向编辑。插件 dispose（资源释放）时会丢弃状态，并且不会跨会话持久化。
 
 ## 单槽、先到者胜
 
@@ -47,7 +47,7 @@ await ctx.plugin(FsPolicy)
 
 ## 不与方法耦合
 
-由于插件只通过事件影响外部世界，移除它不会在服务注入边界破坏 `@deepseek-ai/dsh-tool-fs`：工具会直接落到裸 `ctx.fs` 提供方（无条件写入/编辑，无已观察状态）。重新加载则会再次叠加政策。相比必需的方法服务，这种可平稳增删的性质正是事件门禁的全部目的。
+由于插件只通过事件影响外部世界，移除它不会在服务注入边界破坏 `@deepseek-ai/dsh-tool-fs`：工具会直接落到裸 `ctx.fs` 提供方（无条件写入/编辑，无已观察状态）。重新加载则会再次叠加策略。相比必需的方法服务，这种可平稳增删的性质正是事件门禁的全部目的。
 
 ## 模型体验
 
@@ -65,9 +65,9 @@ await ctx.plugin(FsPolicy)
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **已观察状态无法跨会话恢复**：`WeakMap` 记录的持久化延期处理，因此恢复的会话必须重新读取文件，才能执行防护写入/编辑。
-- **没有 agent 会话的参与者绝无法满足政策**：它们的编辑会抛出 `FS_NOT_OBSERVED`，写入总会解析为 `createIfAbsent`，因此非 agent（智能体）调用方无法通过门禁覆盖现有文件。
+- **已观察状态无法在会话恢复后保留**：`WeakMap` 记录的持久化工作延期处理，因此恢复的会话必须重新读取文件，才能执行防护写入/编辑。
+- **没有 agent（智能体）会话的参与者绝无法满足策略**：它们的编辑会抛出 `FS_NOT_OBSERVED`，写入总会解析为 `createIfAbsent`，因此非 agent 调用方无法通过门禁覆盖现有文件。
 - **直接 `ctx.fs` 读取不会发出 `fs/observed`**：在 `read` 工具之外读取的文件仍未观察；后续防护编辑会以 `FS_NOT_OBSERVED` 拒绝，直到工具读取该文件。
-- **授权依据是版本新鲜度，而非视图完整性**：任何窗口读取都会授权对未变文件执行全文件覆盖，这有意弱于完整视图规则（见 [seam 拆分 Agent Note](../../../.agents/notes/implemented/simplification/2026-06-26-fsspec-style-fs-seam.md)）。
+- **授权依据是版本新鲜度，而非视图完整性**：任何窗口读取都会授权对未变文件执行全文件覆盖，这有意弱于完整视图规则（见 [seam 拆分 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/simplification/2026-06-26-fsspec-style-fs-seam.md)）。
diff --git a/packages/fs/fs-sandbox/README.i18n.yaml b/packages/fs/fs-sandbox/README.i18n.yaml
index daa23d9bfe..2b12b6c282 100644
--- a/packages/fs/fs-sandbox/README.i18n.yaml
+++ b/packages/fs/fs-sandbox/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/fs-sandbox/README.md
 README.md: 790444a4184b9bcccd3a0798cf0c09cb6f1b166e
-README.zh.md: ad0adacda7cfde41f7dcb1d0603da3bad115e8fa
+README.zh.md: d54bdcbe65673b6892ebd1d539dd066f66d68cb6
diff --git a/packages/fs/fs-sandbox/README.zh.md b/packages/fs/fs-sandbox/README.zh.md
index ad0adacda7..d54bdcbe65 100644
--- a/packages/fs/fs-sandbox/README.zh.md
+++ b/packages/fs/fs-sandbox/README.zh.md
@@ -4,21 +4,21 @@
 
 `SandboxedFileSystem` 扩展 [`LocalFileSystem`](../fs-local/README.md) 并注册为 `ctx.fs`。它逐字继承全部文本存储机制（解析、stat、读取／流式读取、列出、原子写入、按读取、匹配、写入顺序执行的编辑临界区），只为 `writeText`/`editText` 增加按调用的模式围栏。读取始终直接通过：所有模式都允许读取。
 
-只需加载它来替代 `dsh-fs-local`，并同时加载 [`ctx.sandboxPolicy`](../../sandbox/sandbox-policy/README.md)，即可完成替换；面向模型的工具（`dsh-tool-fs`）无需改动。工具层把调用会话的模式和 cwd 解析为与 bash 相同的按调用政策，因此两个能力族绝不会约束到不同根目录。
+只需加载它来替代 `dsh-fs-local`，并同时加载 [`ctx.sandboxPolicy`](../../sandbox/sandbox-policy/README.md)，即可完成替换；面向模型的工具（`dsh-tool-fs`）无需改动。工具层把调用会话的模式和 cwd 解析为与 bash 相同的按调用策略，因此两个能力族绝不会约束到不同根目录。
 
 ## 围栏
 
-按调用政策携带有效模式（会话覆盖值或升级授权）和调用会话不可变的 cwd 根目录；只有没有会话的调用才回退到部署政策：
+按调用策略携带有效模式（会话覆盖值或升级授权）和调用会话不可变的 cwd 根目录；只有没有会话的调用才回退到部署策略：
 
 - `read-only`：以结构化 `FS_SANDBOX_DENIED` 拒绝所有变更；
 - `workspace-write`：只有目标规范化后位于可写根目录下，才允许变更。可写根包括工作区根目录和平台临时区域（`/tmp`、`os.tmpdir()`），与 Seatbelt profile 授权的集合相同；该集合由唯一的 [`writableRoots`](../../sandbox/README.md) 函数派生，使 fs 围栏与 bash runner 不会漂移。规范拼写使用词法快速路径；基于身份的祖先回退可以识别 Windows 长名称和 8.3 名称等别名等价根目录，而不会把无关前缀视为包含关系。委托前会立即重新规范化目标，因此工具解析后被替换的祖先符号链接也会被发现；
 - `danger-full-access`：不加围栏直接委托。
 
-## 威胁模型：政策围栏，而非内核边界
+## 威胁模型：策略围栏，而非内核边界
 
 围栏是在可信代码中检查模型控制的路径。操作本身属于 seam（open、rename），只有目标路径不可信，因此「规范化后检查包含关系」就是该接口的完整答案。这与 `code-runtime` 的立场相同：提供约束，但不是安全边界。不可信代码的内核级隔离仍由 `ctx.bash` 负责（[`dsh-bash-sandbox`](../../bash/bash-sandbox/README.md)）。剩余 TOCTOU（在包含关系复查与系统调用之间替换祖先符号链接）会通过写入前立即重新规范化来缩小，并为该威胁模型所接受；内核严密边界需要 `openat2` 一类原语，其可移植性成本在此不值得。
 
-拒绝是结构化 `FsError`（`FS_SANDBOX_DENIED`，携带有效模式），不通过 stderr 文本推断（不同于 bash 的内核拒绝），因为进程内围栏准确知道自己拒绝了什么。面向模型的 `[sandbox: file access denied under <mode> mode]` 标记以及唯一一次获批的更宽权限重试位于工具层（`dsh-tool-fs`），与 bash 完全相同。见[跨能力族 fs 沙箱 Agent Note](../../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)。
+拒绝是结构化 `FsError`（`FS_SANDBOX_DENIED`，携带有效模式），不通过 stderr 文本推断（不同于 bash 的内核拒绝），因为进程内围栏准确知道自己拒绝了什么。面向模型的 `[sandbox: file access denied under <mode> mode]` 标记以及唯一一次获批的更宽权限重试位于工具层（`dsh-tool-fs`），与 bash 完全相同。见[跨能力族 fs 沙箱 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)。
 
 ## 模型体验
 
@@ -26,10 +26,10 @@
 
 #### KV Cache 影响
 
-不会直接使缓存失效；具名消费方负责请求前缀的任何变化。
+不会直接使缓存失效；上述消费方负责请求前缀的任何变化。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **政策围栏，而非内核边界**：该检查是可信代码处理模型控制的路径，因此解析到系统调用之间残留的 TOCTOU 会被原位重新规范化缩小，但不会消除；对抗性宿主进程不在范围内。不可信代码的内核级隔离仍属于 `ctx.bash`。
+- **策略围栏，而非内核边界**：该检查是可信代码处理模型控制的路径，因此解析到系统调用之间残留的 TOCTOU 会被原位重新规范化缩小，但不会消除；对抗性宿主进程不在范围内。不可信代码的内核级隔离仍属于 `ctx.bash`。
 - **围栏与 runner 的一致性来自派生，而非断言**：可写集合来自 `writableRoots`，该函数与 Seatbelt profile 共享，并由一致性测试固定；不通过该函数更改可写集合的 runner profile 会发生漂移。
-- **要求 `ctx.sandboxPolicy`**：工具使用它解析每个会话政策，后端用它处理无 agent 调用的回退；未组合该服务时，后端不会实施约束。
+- **要求 `ctx.sandboxPolicy`**：工具使用它解析每个会话策略，后端用它处理无 agent 调用的回退；未组合该服务时，后端不会实施约束。
diff --git a/packages/fs/fs/README.i18n.yaml b/packages/fs/fs/README.i18n.yaml
index 4b59d669c2..a4547d8b81 100644
--- a/packages/fs/fs/README.i18n.yaml
+++ b/packages/fs/fs/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/fs/README.md
 README.md: 9e6c954abad124fb2b30ebc01368a55746752013
-README.zh.md: 262d689be916c2983b203072c713a924b35fc3af
+README.zh.md: ff97513490d4a2855564042ba0ebf9403f28289c
diff --git a/packages/fs/fs/README.zh.md b/packages/fs/fs/README.zh.md
index 262d689be9..ff97513490 100644
--- a/packages/fs/fs/README.zh.md
+++ b/packages/fs/fs/README.zh.md
@@ -2,18 +2,18 @@
 
 [English](README.md) | 中文
 
-**文件系统提供方 seam**：抽象 `FileSystem` 服务（`ctx.fs`），定义后端提供的存储原语，包括路径解析、stat 元数据、不跟随链接的路径元数据、读取/流式读取文本、列出目录、原子写入和应用字面量编辑，但不规定实现方式。两个变更操作都**可选** 接收版本防护，因此 `ctx.fs` 本身就是完整且不受约束的文本存储 seam。本包还拥有由工具分派、政策插件监听的 `fs/*` 政策事件词汇。
+**文件系统提供方 seam**：抽象 `FileSystem` 服务（`ctx.fs`），定义后端提供的存储原语，包括路径解析、stat 元数据、不跟随链接的路径元数据、读取/流式读取文本、列出目录、原子写入和应用字面量编辑，但不规定实现方式。两个变更操作都**可选**接收版本防护，因此 `ctx.fs` 本身就是完整且不受约束的文本存储 seam。本包（package）还拥有由工具分派、策略插件监听的 `fs/*` 策略事件词汇。
 
-本包是四层文件系统栈中的提供方 seam 层；该拆分使每个关注点可以独立演进和替换（见[能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)、[文件系统能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-17-filesystem-capability-seam.md)、[拆分文件系统 seam Agent Note](../../../.agents/notes/implemented/simplification/2026-06-26-fsspec-style-fs-seam.md)和[文件上下文事件门禁 Agent Note](../../../.agents/notes/implemented/architecture/2026-06-26-file-context-as-event-gate.md)）：
+本包是四层文件系统栈中的提供方 seam 层；该拆分使每个关注点可以独立演进和替换（见[能力 seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)、[文件系统能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-17-filesystem-capability-seam.md)、[拆分文件系统 seam Agent Note](../../../.agents/notes/implemented/simplification/2026-06-26-fsspec-style-fs-seam.md)和[文件上下文事件门禁 Agent Note](../../../.agents/notes/implemented/architecture/2026-06-26-file-context-as-event-gate.md)）：
 
 | 层 | 包 | 角色 |
 |---|---|---|
 | 工具/执行器 | `@deepseek-ai/dsh-tool-fs` | 面向模型的 `read`/`write`/`edit` schema、读取窗口和文本渲染；通过 `ctx.fs` 读取/写入/编辑，并分派 `fs/*` 事件 |
-| 政策 | `@deepseek-ai/dsh-fs-policy` | 已观察状态、编辑前读取和版本防护的写入/编辑，通过 `fs/*` 事件门禁贡献（无服务） |
+| 策略 | `@deepseek-ai/dsh-fs-policy` | 已观察状态、编辑前读取和版本防护的写入/编辑，通过 `fs/*` 事件门禁贡献（无服务） |
 | 提供方 seam | `@deepseek-ai/dsh-fs`（本包） | `ctx.fs`：文本 I/O 与原子变更原语（可选版本防护）；拥有 `fs/*` 事件词汇 |
 | 提供方 | `@deepseek-ai/dsh-fs-local` | 宿主文件系统实现 |
 
-未来的沙箱化、虚拟或远程后端只需实现该接口，政策层和工具层无需改变。
+未来的沙箱化、虚拟或远程后端只需实现该接口，策略层和工具层无需改变。
 
 ## 服务 API（`ctx.fs`）
 
@@ -23,24 +23,24 @@
 |---|---|
 | `resolve(path, opts?)` | 把路径解析为稳定的 `FsTarget`（不透明 `targetKey`、`displayPath`）。`opts.cwd` 是相对 `path` 解析所依据的基准（调用方提供其会话工作区；绝对路径忽略该值；省略时使用后端默认值），`opts.signal` 则中止后端往返。该方法是异步的，因为远程后端可能需要 I/O。经不同路径到达的同一文件必须产生相同 `targetKey`。 |
 | `stat(target, signal?)` | 返回 `FsInfo` 元数据（`version`、`type`、可选 `size`）；目标不存在时返回 `undefined`。绝不返回内容。 |
-| `lstat(path, opts?, signal?)` | 当最后一个路径组件是符号链接时，不跟随该组件，返回 `FsPathInfo` 元数据。该方法采用路径形态，使消费方能在 `resolve` 跟随仓库所有的符号链接进入目标前拒绝它。 |
+| `lstat(path, opts?, signal?)` | 当最后一个路径组件是符号链接时，不跟随该组件，返回 `FsPathInfo` 元数据。该方法采用路径形态，使消费方能在 `resolve` 跟随仓库自有的符号链接进入目标前拒绝它。 |
 | `readText(target, signal?)` | 把整个普通文本文件读取为一个解码后的字符串。负责普通文件检查、UTF-8 解码和二进制/NUL 拒绝（`FS_NOT_TEXT`）。 |
 | `streamText(target, signal?)` | 为大文件按解码后的分片流式读取相同文本（跨分片 UTF-8 解码仍由此处负责）。 |
-| `listDir(target, signal?)` | 按稳定名称顺序列出直接子项。返回条目名称、条目类型、解析后的子目标和低成本元数据（若可用则包括 `version`/文件 `size`）；绝不读取文件内容。缺失目标抛出 `FS_NOT_FOUND`，非目录抛出 `FS_NOT_DIRECTORY`，权限失败抛出 `FS_PERMISSION_DENIED`，其他后端 I/O 失败抛出 `FS_IO_ERROR`。损坏/消失的子项可以作为无元数据的 `other` 返回；子项权限/I/O 失败会使用相同结构化代码使整个列表失败。 |
+| `listDir(target, signal?)` | 按稳定名称顺序列出直接子项。返回条目名称、条目类型、解析后的子目标和低成本元数据（若可用则包括 `version`/文件 `size`）；绝不读取文件内容。缺失目标抛出 `FS_NOT_FOUND`，非目录抛出 `FS_NOT_DIRECTORY`，权限失败抛出 `FS_PERMISSION_DENIED`，其他后端 I/O 失败抛出 `FS_IO_ERROR`。损坏/消失的子项可以作为无元数据的 `other` 返回；子项权限/I/O 失败会使用相同结构化代码使整个列出操作失败。 |
 | `writeText(target, content, expected?, signal?)` | 原子创建/替换。`expected` 是可选的：省略 ⇒ 无条件创建或覆盖；提供 `FsWriteIntent`（`createIfAbsent`/`replaceIfVersion`）⇒ 添加防护。 |
 | `editText(target, edit, expected?, signal?)` | 字面量编辑。`expected` 是可选的：省略 ⇒ 无条件编辑当前内容；提供 `{ version }` ⇒ 添加防护，并在匹配之前校验。无论哪种情况，目标缺失都报告 `FS_STALE_VERSION`。应用和写入以原子方式完成，使用同一个变更临界区。 |
 
 无论是否有版本防护，变更都在后端的每目标锁内运行，因此无条件写入/编辑仍是原子的；「无条件」只移除*版本*前置条件，不移除原子性。
 
-## `fs/*` 政策事件
+## `fs/*` 策略事件
 
-本包声明三个事件（见已生成的[事件目录](../../../docs/cordis-catalog/events.md)），使发出方（`@deepseek-ai/dsh-tool-fs`）和政策监听器（`@deepseek-ai/dsh-fs-policy`）共享词汇，而无需让发出方依赖政策插件。`fs/write-intent` 和 `fs/edit-intent` 是单槽决策 waterfall（监听器完整决策，绝不调用 `next()`）；`fs/observed` 是发后即忘的记录事件。它们只携带 `dsh-fs` 词汇和一个不透明 `object` 参与者，不含面向模型的概念或 agent（智能体）/会话所有者结构。
+本包声明三个事件（见已生成的[事件目录](../../../docs/cordis-catalog/events.md)），使发出方（`@deepseek-ai/dsh-tool-fs`）和策略监听器（`@deepseek-ai/dsh-fs-policy`）共享词汇，而无需让发出方依赖策略插件。`fs/write-intent` 和 `fs/edit-intent` 是单槽决策 waterfall（瀑布式事件）（监听器完整决策，绝不调用 `next()`）；`fs/observed` 是发后即忘的记录事件。它们只携带 `dsh-fs` 词汇和一个不透明 `object` 参与者，不含面向模型的概念或 agent（智能体）/会话所有者结构。
 
-## 提供方 seam，不是政策层
+## 提供方 seam，不是策略层
 
-`ctx.fs` 有意接近 fsspec 风格的存储原语，比字节级 `cat`/`open` 高半层，因为它会解码文本并拒绝二进制，使政策层绝不接触原始字节。它负责 UTF-8 解码、二进制拒绝、原子写入和字面量编辑临界区。它**不** 负责行窗口、编号行、渲染 footer 或已观察状态。已观察状态、编辑前读取和版本防护的写入/编辑属于插件（`@deepseek-ai/dsh-fs-policy`）通过提供可选防护而添加的政策，并非提供方行为，因此沙箱化/远程后端不会继承任何面向模型的观察政策。
+`ctx.fs` 有意接近 fsspec 风格的存储原语，比字节级 `cat`/`open` 高半层，因为它会解码文本并拒绝二进制，使策略层绝不接触原始字节。它负责 UTF-8 解码、二进制拒绝、原子写入和字面量编辑临界区。它**不**负责行窗口、编号行、渲染 footer 或已观察状态。已观察状态、编辑前读取和版本防护的写入/编辑属于插件（`@deepseek-ai/dsh-fs-policy`）通过提供可选防护而添加的策略，并非提供方行为，因此沙箱化/远程后端不会继承任何面向模型的观察策略。
 
-`editText` 留在该 seam 上，不由政策层通过读取加写入组合，因为版本防护、字面量匹配和原子重写必须处于同一临界区内，才能正确归因错误并实现一方胜出/一方陈旧的并发；远程后端也可以将其实现为原生比较并编辑操作。
+`editText` 留在该 seam 上，不由策略层通过读取加写入组合，因为版本防护、字面量匹配和原子重写必须处于同一临界区内，才能正确归因错误并实现一方胜出/一方陈旧的并发；远程后端也可以将其实现为原生比较并编辑操作。
 
 ## 词汇
 
@@ -52,9 +52,9 @@
 
 #### KV Cache 影响
 
-不会直接使缓存失效；具名消费方负责请求前缀的任何变化。
+不会直接使缓存失效；上述消费方负责请求前缀的任何变化。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **契约只支持文本**：后端以 `FS_NOT_TEXT` 拒绝二进制/非 UTF-8 内容；二进制安全操作是[工具 schema Agent Note](../../../.agents/notes/implemented/feature/2026-06-17-filesystem-tool-schemas.md)有意延期的工作。
 - **只有八个原语**：没有删除、重命名/移动、复制或监视；`listDir` 只支持一层，递归、glob、分页和搜索不在范围内，见[目录列出 Agent Note](../../../.agents/notes/archived/architecture/2026-07-03-filesystem-directory-listing-seam.md)。
diff --git a/packages/fs/tool-fs-search/README.i18n.yaml b/packages/fs/tool-fs-search/README.i18n.yaml
index 07aaa3c9dc..4d305c9c4c 100644
--- a/packages/fs/tool-fs-search/README.i18n.yaml
+++ b/packages/fs/tool-fs-search/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/tool-fs-search/README.md
 README.md: 88a80fb51d7161e6940a3460b7f506575592f9cb
-README.zh.md: 87be92bb5a8e06dfc275aa6a1fcf97274a761025
+README.zh.md: 9767c545ddc87d323c9fefa43a3145d1a2df5914
diff --git a/packages/fs/tool-fs-search/README.zh.md b/packages/fs/tool-fs-search/README.zh.md
index 87be92bb5a..9767c545dd 100644
--- a/packages/fs/tool-fs-search/README.zh.md
+++ b/packages/fs/tool-fs-search/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-**面向模型的文件系统发现工具**（`glob`、`grep`）由 **bash 执行器 seam** 支持，而不是由 `ctx.fs` 提供方方法支持。加载时，本包探测 `command -v rg`，探测通过 `ctx.bash` 进行；如果执行器无法在其 `PATH` 上找到 ripgrep，就记录警告，并且不注册工具或提示词段。每次调用都会组装固定的 ripgrep 命令（所有模型控制的值都经过同一个包私有 shell 引用辅助函数），通过 `ctx.bash.resolve(request)` → `ctx.bash.run(spec)` 作为普通前台工具调用运行，解析原始 `rg` 输出，并返回相对于工作目录的规范值。本包注入 `tools`、`systemPrompt` 和 `bash`，有意**不** 注入 `fs`；格式化结果 spill 为可选功能，因此机会性读取 `ctx.spillStore`，调用方式为 `ctx.get()`。
+**面向模型的文件系统发现工具**（`glob`、`grep`）由 **bash 执行器 seam** 支持，而不是由 `ctx.fs` 提供方方法支持。加载时，本包（package）探测 `command -v rg`，探测通过 `ctx.bash` 进行；如果执行器无法在其 `PATH` 上找到 ripgrep，就记录警告，并且不注册工具或提示词段。每次调用都会组装固定的 ripgrep 命令（所有模型控制的值都经过同一个包私有 shell 引用辅助函数），通过 `ctx.bash.resolve(request)` → `ctx.bash.run(spec)` 作为普通前台工具调用运行，解析原始 `rg` 输出，并返回相对于工作目录的规范值。本包注入 `tools`、`systemPrompt` 和 `bash`，有意**不**注入 `fs`；格式化结果 spill 为可选功能，因此机会性读取 `ctx.spillStore`，调用方式为 `ctx.get()`。
 
 ```ts ignore-check
 // Default deployment: a bash executor whose PATH includes rg, then the discovery tools.
@@ -16,7 +16,7 @@ await ctx.plugin(LocalSpillStore)                           // @deepseek-ai/dsh-
 
 ## 部署要求：rg 与共置的 bash/文件系统
 
-已挂载的 bash 执行器必须能在插件加载时解析 `rg`，其来源是执行器的 `PATH`；否则面向模型的工具 schema 中不会出现 `glob` 和 `grep`。返回路径会相对于解析后的 bash 工作目录显示（调用 agent（智能体）存在会话 cwd 时使用该值，否则使用执行器配置的默认值）；只有 bash 工作目录与文件系统根目录是同一工作区时，才能用 `read` 继续读取。v1 只记录这项共置要求，不执行运行时跨服务校验；远程或虚拟文件系统搜索需等待共享工作区契约或特定提供方的搜索后端。
+已挂载的 bash 执行器必须能在插件加载时解析 `rg`，其来源是执行器的 `PATH`；否则面向模型的工具 schema 中不会出现 `glob` 和 `grep`。返回路径会相对于解析后的 bash 工作目录显示（调用方 agent（智能体）有会话 cwd 时使用该 cwd，否则使用执行器配置的默认值）；只有 bash 工作目录与文件系统根目录是同一工作区时，才能用 `read` 继续读取。v1 只记录这项共置要求，不执行运行时跨服务校验；远程或虚拟文件系统搜索需等待共享工作区契约或特定提供方的搜索后端。
 
 ## 配置
 
@@ -34,14 +34,14 @@ await ctx.plugin(LocalSpillStore)                           // @deepseek-ai/dsh-
 
 | 工具 | 参数 | 行为 |
 |---|---|---|
-| `glob` | `pattern`、`path?` | 运行 `rg --files --glob <pattern> --sort=modified --no-ignore --hidden`，并排除 VCS 元数据（`.git`、`.svn`、`.hg`、`.bzr`、`.jj`、`.sl`）。`path` 是可选的**目录** 搜索根；省略时使用解析后的 bash 工作目录。每行返回一个路径，按修改时间排序。 |
-| `grep` | `pattern`、`path?`、`include?` | 按行解析 `rg --json`，避免按冒号拆分的歧义。`pattern` 是 ripgrep 正则表达式；`path` 是可选的**文件或目录** 目标；`include` 是一个正向 glob 过滤器，前置拒绝逗号分隔列表或否定值（`!…`），但允许 `*.{ts,tsx}` 等花括号交替。返回按文件分组、形如 `Line N: <preview>` 的匹配。 |
+| `glob` | `pattern`、`path?` | 运行 `rg --files --glob <pattern> --sort=modified --no-ignore --hidden`，并排除 VCS 元数据（`.git`、`.svn`、`.hg`、`.bzr`、`.jj`、`.sl`）。`path` 是可选的**目录**搜索根；省略时使用解析后的 bash 工作目录。每行返回一个路径，按修改时间排序。 |
+| `grep` | `pattern`、`path?`、`include?` | 按行解析 `rg --json`，避免按冒号拆分的歧义。`pattern` 是 ripgrep 正则表达式；`path` 是可选的**文件或目录**目标；`include` 是一个正向 glob 过滤器，前置拒绝逗号分隔列表或否定值（`!…`），但允许 `*.{ts,tsx}` 等花括号交替。返回按文件分组、形如 `Line N: <preview>` 的匹配。 |
 
 常规预算不进入面向模型的 schema（没有 `head_limit`/`offset`/`case_insensitive`/输出模式）：模型需要周边上下文时，用 `read` 读取匹配文件；需要后续结果时，遵循返回的 spill locator 检索提示。
 
 ## 两类预算、两类产物
 
-原始 `rg` stdout 是内部传输细节。每次搜索从 bash seam 请求 `stdoutMaxBytes: rawOutputMaxBytes`，且只解析完整保留的 stdout；如果执行器仍返回 `stdout.truncated`，搜索会以 `SEARCH_RAW_OUTPUT_OVERFLOW` 失败，并要求模型缩小查询。成功的 `glob` 在 `{ paths }` 中保留所有已取得路径；`grep` 保留所有已取得的 `{ path, lineNumber, line }`，并将其存入 `{ matches }`。内联条目和每行预览上限只应用于 Native 渲染器。直接接口调用的逻辑结果超过内联上限时，后置政策会尽力通过 `ctx.spillStore.saveText()` 保存完整格式化预览，并只把呈现替换为头部页面加 locator。嵌套 Code 分派会跳过 spill，因为其完整规范值不会进入模型上下文。spill 缺失/失败时保留内联页面，并报告完整结果无法保存，绝不会成为 `isError`。
+原始 `rg` stdout 是内部传输细节。每次搜索从 bash seam 请求 `stdoutMaxBytes: rawOutputMaxBytes`，且只解析完整保留的 stdout；如果执行器仍返回 `stdout.truncated`，搜索会以 `SEARCH_RAW_OUTPUT_OVERFLOW` 失败，并要求模型缩小查询。成功的 `glob` 在 `{ paths }` 中保留所有已取得路径；`grep` 保留所有已取得的 `{ path, lineNumber, line }`，并将其存入 `{ matches }`。内联条目和每行预览上限只应用于原生渲染器。直接接口调用的逻辑结果超过内联上限时，后置策略会尽力通过 `ctx.spillStore.saveText()` 保存完整格式化预览，并只把呈现替换为头部页面加 locator。嵌套 Code 分派会跳过 spill，因为其完整规范值不会进入模型上下文。spill 缺失/失败时保留内联页面，并报告完整结果无法保存，绝不会成为 `isError`。
 
 ## 错误
 
@@ -117,7 +117,7 @@ Use the grep tool — not shell grep or rg — to search file contents. Use read
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **搜索和文件访问没有共享工作区证明**：只有 bash 工作目录和文件系统根目录表示同一工作区时，返回路径才能继续读取；本包不执行运行时跨服务校验。
 - **Ripgrep 是部署依赖**：缺失 `rg` 可执行文件时，本包不注册工具或指导；可执行文件不兼容或注册后消失时，调用以 `SEARCH_FAILED` 失败。远程或虚拟文件系统需要共置执行器或其他搜索消费方。
diff --git a/packages/fs/tool-fs/README.i18n.yaml b/packages/fs/tool-fs/README.i18n.yaml
index 13f1ecd649..a3ebe97bc4 100644
--- a/packages/fs/tool-fs/README.i18n.yaml
+++ b/packages/fs/tool-fs/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/fs/tool-fs/README.md
 README.md: 4ff9b043525e8e7a0b59e3d91410951d88bb9a69
-README.zh.md: f94a903c9c37f7d45b7f8cebabe21082388bd041
+README.zh.md: ce93e10072d74ce268273aa472bfbb3f34f46259
diff --git a/packages/fs/tool-fs/README.zh.md b/packages/fs/tool-fs/README.zh.md
index f94a903c9c..ce93e10072 100644
--- a/packages/fs/tool-fs/README.zh.md
+++ b/packages/fs/tool-fs/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-**面向模型的文件系统工具**（`read`、`write`、`edit`）及其**执行器**。这是文件系统栈的消费方层：拥有工具名称、JSON schema、参数校验、提示词段、**读取窗口逻辑** 和结果格式化。它**直接** 通过 `ctx.fs` 提供方 seam（[`@deepseek-ai/dsh-fs`](../fs)）读取/写入/编辑：注入 `fs`（以及 `tools`/`systemPrompt`），**不** 注入政策服务。新鲜度/观察政策由独立插件（[`@deepseek-ai/dsh-fs-policy`](../fs-policy)）通过 `fs/*` 事件门禁贡献；工具不与其方法耦合。
+**面向模型的文件系统工具**（`read`、`write`、`edit`）及其**执行器**。这是文件系统栈的消费方层：拥有工具名称、JSON Schema、参数校验、提示词段、**读取窗口逻辑**和结果格式化。它**直接**通过 `ctx.fs` 提供方 seam（[`@deepseek-ai/dsh-fs`](../fs)）读取/写入/编辑：注入 `fs`（以及 `tools`/`systemPrompt`），**不**注入策略服务。新鲜度/观察策略由独立插件（[`@deepseek-ai/dsh-fs-policy`](../fs-policy)）通过 `fs/*` 事件门禁贡献；工具不与其方法耦合。
 
 ```ts ignore-check
 // Default deployment: a ctx.fs provider, the policy plugin, then the tools.
@@ -11,7 +11,7 @@ await ctx.plugin(FsPolicy)                             // @deepseek-ai/dsh-fs-po
 await ctx.plugin(ToolFs)                                  // this package — registers read/write/edit
 ```
 
-`@deepseek-ai/dsh-fs-policy` 是**可选的**：省略时，工具直接使用裸提供方（无条件写入/覆盖/编辑，无已观察状态）。加载这些工具的部署也应加载该插件，从而提供编辑前读取行为。
+`@deepseek-ai/dsh-fs-policy` 是**可选的**：省略时，工具直接使用裸提供方（无条件写入/覆盖/编辑，无已观察状态）。加载这些工具的部署也应加载该插件，从而提供写入/编辑前读取行为。
 
 ## 配置
 
@@ -24,21 +24,21 @@ await ctx.plugin(ToolFs)                                  // this package — re
 | `readMaxBytes` | `51200` | 一次 `read` 调用所选行的字节上限；溢出时以「已达上限」footer 结束窗口。 |
 | `readStreamMinSize` | `10485760` | 大于等于该大小或大小未知的文件采用流式读取，而不是整体加载到内存。 |
 
-## 工具（schema 见[文件系统工具 schema Agent Note](../../../.agents/notes/implemented/feature/2026-06-17-filesystem-tool-schemas.md)）
+## 工具（schema 见[文件系统工具 schema Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-17-filesystem-tool-schemas.md)）
 
 | 工具 | 参数 | 行为 |
 |---|---|---|
 | `read` | `file_path`、`offset?`、`limit?` | 带行号的 UTF-8 内容和分页 footer。`offset` 从 1 开始；`limit` 默认为配置的 `readLimit`（2000），上限也为该值。 |
-| `write` | `file_path`、`content` | 创建文件或完整替换文件。有政策插件时：覆盖现有文件要求先在未变版本上执行 `read`；创建新文件不需要。没有插件时：无条件执行。 |
-| `edit` | `file_path`、非空 `old_string`、`new_string`、`replace_all?` | 字面量替换；除非 `replace_all` 为 true，否则要求唯一匹配。有政策插件时：要求先执行 `read`（任何窗口），且文件此后未变。没有插件时：无条件执行。 |
+| `write` | `file_path`、`content` | 创建文件或完整替换文件。有策略插件时：覆盖现有文件要求先在未变版本上执行 `read`；创建新文件不需要。没有插件时：无条件执行。 |
+| `edit` | `file_path`、非空 `old_string`、`new_string`、`replace_all?` | 字面量替换；除非 `replace_all` 为 true，否则要求唯一匹配。有策略插件时：要求先执行 `read`（任何窗口），且文件此后未变。没有插件时：无条件执行。 |
 
 字段名使用 snake_case，与 Claude Code 和现有 harness 工具 schema 一致。
 
-规范成功值分别为：`read` → `{ path, offset, lines: [{ number, text }], totalLines }`，`write` → `{ path, operation: 'create' | 'update', before: string | null, after }`，`edit` → `{ path, before, after }`。Native 渲染器会保留下方带行号的读取结果和变更确认。写入/编辑从这些值派生可回放的 diff 卡片元数据；值本身仅用于执行，不会添加到 `tool/result`。
+规范成功值分别为：`read` → `{ path, offset, lines: [{ number, text }], totalLines }`，`write` → `{ path, operation: 'create' | 'update', before: string | null, after }`，`edit` → `{ path, before, after }`。原生渲染器会保留下方带行号的读取结果和变更确认。写入/编辑从这些值派生可回放的 diff 卡片元数据；这些值本身仅限于本次执行，不会添加到 `tool/result`。
 
-## 工具就是执行器；政策是事件门禁
+## 工具就是执行器；策略是事件门禁
 
-工具**不** 注入政策服务，也不检查任何缓存。每个工具通过 `ctx.fs.resolve(path, { cwd, signal })` 解析路径；它会传入调用 agent（智能体）的会话 cwd（`exec.agent.session.header.cwd`），使相对路径以会话工作区为基准解析并与 `dsh-tool-bash` 一致，同时把工具取消转发到解析过程（见[每会话 cwd Agent Note](../../../.agents/notes/implemented/architecture/2026-07-02-fs-per-session-cwd.md)）。随后执行：
+工具**不**注入策略服务，也不检查任何缓存。每个工具通过 `ctx.fs.resolve(path, { cwd, signal })` 解析路径；它会传入调用 agent 的会话 cwd（`exec.agent.session.header.cwd`），使相对路径以会话工作区为基准解析并与 `dsh-tool-bash` 一致，同时把工具取消转发到解析过程（见[每会话 cwd Agent Note](../../../.agents/notes/implemented/architecture/2026-07-02-fs-per-session-cwd.md)）。随后执行：
 
 - **read**：一次 `ctx.fs.stat`（用于类型、大小路由和版本），随后调用 `readText`/`streamText`，构建行窗口，再发出 `fs/observed`，使用普通 `ctx.emit`。（1 次 stat。）
 - **write**：调用 `ctx.waterfall('fs/write-intent', target, exec, () => undefined)` 取得可选防护，然后调用 `ctx.fs.writeText(target, content, intent)`，再发出 `fs/observed`。（0 次 stat。）
@@ -52,7 +52,7 @@ await ctx.plugin(ToolFs)                                  // this package — re
 
 `read` 允许并发调度，因为其唯一变更是同步版本记录器。稍后的 `write` 或 `edit` 会在目标锁内重新检查版本，因此记录器竞态会以拒绝方式关闭；两个变更工具仍保持互斥。见[并行工具调用 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md)。
 
-包根目录只导出 Cordis 插件契约（`name`、`inject`、`Config` 和 `apply`）。读取渲染（行窗口与输出格式化）位于 `src/read-render.ts`（不依赖 Cordis，单独进行单元测试）；`src/read.ts`/`write.ts`/`edit.ts` 是工具执行器，`src/index.ts` 负责组合。
+包（package）根目录只导出 Cordis 插件契约（`name`、`inject`、`Config` 和 `apply`）。读取渲染（行窗口与输出格式化）位于 `src/read-render.ts`（不依赖 Cordis，单独进行单元测试）；`src/read.ts`/`write.ts`/`edit.ts` 是工具执行器，`src/index.ts` 负责组合。
 
 ## 模型体验
 
@@ -134,7 +134,7 @@ Use the edit tool for targeted changes to existing UTF-8 text files. It replaces
 
 #### 模型看到的内容
 
-失败会规范化为 `Error: <message>`。本包稳定的校验和读取消息是 `file_path must be a non-empty string`、`limit must be less than or equal to <max>`、`old_string must be a non-empty string`、`old_string and new_string must differ`、`cannot read "<path>": not found`、`cannot read "<path>": not a regular file` 和 `offset <offset> is out of range for "<path>" (<total> lines)`；提供方和政策模板在各自包的 README 中逐字列出。
+失败会规范化为 `Error: <message>`。本包稳定的校验和读取消息是 `file_path must be a non-empty string`、`limit must be less than or equal to <max>`、`old_string must be a non-empty string`、`old_string and new_string must differ`、`cannot read "<path>": not found`、`cannot read "<path>": not a regular file` 和 `offset <offset> is out of range for "<path>" (<total> lines)`；提供方和策略模板在各自包的 README 中逐字列出。
 
 #### Token 影响
 
@@ -144,8 +144,8 @@ Use the edit tool for targeted changes to existing UTF-8 text files. It replaces
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **未交付面向模型的目录列出工具**：`ctx.fs.listDir` 服务于 skill（技能）发现等提供方代码，同级 [`dsh-tool-fs-search`](../tool-fs-search/) 包则提供基于 bash 的 `glob` 与 `grep`，而不是扩展文件系统 seam。
+- **未交付面向模型的目录列表工具**：`ctx.fs.listDir` 服务于 skill（技能）发现等提供方代码，同级 [`dsh-tool-fs-search`](../tool-fs-search/) 包则提供基于 bash 的 `glob` 与 `grep`，而不是扩展文件系统 seam。
 - **`read` 只处理 UTF-8 文本文件**：二进制安全读取和 PDF/图像/多模态内容均延期处理；目录目标为 `FS_NOT_REGULAR_FILE`。
 - **没有超时接口**：`read`/`write`/`edit` 不接受超时参数，也不声明 `timeout-policy` 预算；取消只通过 `exec.signal` 传递（见有意采用的 [fs 能力族立场](../README.md)）。
diff --git a/packages/goal/README.i18n.yaml b/packages/goal/README.i18n.yaml
index e72c1d21db..1a4406e288 100644
--- a/packages/goal/README.i18n.yaml
+++ b/packages/goal/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/goal/README.md
 README.md: f43dfd8258eabe8342207c0b1b9d6acc9e215e9f
-README.zh.md: f663939bb19d8bc295a96fc0a673c56251e9a2a9
+README.zh.md: 70b38caac723d757f44109e4ec75e3c31e7c34b8
diff --git a/packages/goal/README.zh.md b/packages/goal/README.zh.md
index f663939bb1..70b38caac7 100644
--- a/packages/goal/README.zh.md
+++ b/packages/goal/README.zh.md
@@ -7,8 +7,8 @@ goal 家族负责持久目标状态，与消费该状态的面向模型工具和
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `goal/` | 事件溯源的目标生命周期、回放折叠、比较并设置变更，以及进程本地激活 | `ctx.goals` |
-| `goal-session/` | 同会话 goal round 的准入、结果映射与生命周期竞态隔离 | 无 |
+| `goal-session/` | 同会话 Goal Round 的准入、结果映射与生命周期竞态隔离 | 无 |
 | `tool-goal/` | 面向模型的读取／创建／更新工具，并在执行时检查权限 | 无 |
 | `command-goal/` | 面向用户的 `/goal` 状态，以及通过命令平面执行的生命周期控制 | 无 |
 
-目标状态属于所属会话日志。消费方依赖 `dsh-goal`，而不是具体的 agent loop；续行行为由基于公开 agent seam 的独立插件负责。
+目标状态是其所属会话日志的一部分。消费方依赖 `dsh-goal`，而不是具体的 agent loop（智能体循环）；续行行为由基于公开 agent seam 的独立插件负责。
diff --git a/packages/goal/command-goal/README.i18n.yaml b/packages/goal/command-goal/README.i18n.yaml
index 3bc8846656..5082633d83 100644
--- a/packages/goal/command-goal/README.i18n.yaml
+++ b/packages/goal/command-goal/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/goal/command-goal/README.md
 README.md: 8e1a5b417467c8701ea935e25acfece11c5a70d4
-README.zh.md: fc7229e1fa14160b8247ae0c33a5fbc3cb062254
+README.zh.md: 66f237ac7ea8ef5d158917998b0dfd49189289f1
diff --git a/packages/goal/command-goal/README.zh.md b/packages/goal/command-goal/README.zh.md
index fc7229e1fa..66f237ac7e 100644
--- a/packages/goal/command-goal/README.zh.md
+++ b/packages/goal/command-goal/README.zh.md
@@ -2,22 +2,22 @@
 
 [English](README.md) | 中文
 
-面向用户的 `/goal` 控制，基于 [`ctx.goals`](../goal/README.md) 实现。该插件通过 [`ctx.commands`](../../ui/commands/README.md) 注册一个全局命令，因此每个已组合的命令适配器都能发现它；随附 TUI 无需模型轮次即可执行。[用户 goal 命令 Agent Note](../../../.agents/notes/implemented/feature/2026-07-19-human-goal-command.md)负责用户体验与组合决策。
+面向用户的 `/goal` 控制，基于 [`ctx.goals`](../goal/README.md) 实现。该插件通过 [`ctx.commands`](../../ui/commands/README.md) 注册一个全局命令，因此每个已组合的命令适配器都能发现它；随附 TUI 无需模型轮次即可执行。[用户 goal 命令 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-human-goal-command.md) 负责用户体验与组合决策。
 
 ## 命令契约
 
 | 输入 | 结果 |
 |---|---|
-| `/goal` | 显示当前目标、持久 phase、round 计数／上限、进程本地激活状态与有效的下一步命令；被阻塞的 goal 还会显示策略代码和说明，没有 goal 时则显示用法。 |
-| `/goal <objective>` | 创建并激活 goal，或用全新身份替换已完成 goal。未完成 goal 绝不会在没有显式 clear 的情况下被替换。 |
-| `/goal edit <objective>` | 编辑当前目标，不改变其 phase 或激活状态。编辑已完成 goal 会创建新的 active goal。 |
-| `/goal pause` | 暂停 active goal，并撤销续行激活。 |
-| `/goal resume` | 恢复已停止 goal，或在会话 resume／fork 后重新激活 active goal；仍受剩余 round 上限约束。 |
+| `/goal` | 显示当前目标、持久 phase、Round 计数／上限、进程本地续行启用状态与有效的下一步命令；被阻塞的 goal 还会显示策略代码和说明，没有 goal 时则显示用法。 |
+| `/goal <objective>` | 创建 goal 并启用续行，或用全新身份替换已完成 goal。未完成 goal 绝不会在没有显式 clear 的情况下被替换。 |
+| `/goal edit <objective>` | 编辑当前目标，不改变其 phase 或续行启用状态。编辑已完成 goal 会创建新的 active goal。 |
+| `/goal pause` | 暂停 active goal，并停用续行。 |
+| `/goal resume` | 恢复已停止 goal，或在会话 resume／fork 后为 active goal 重新启用续行；仍受剩余 Round 上限约束。 |
 | `/goal clear` | 清除当前指针，同时保留其持久历史和 tombstone。 |
 
-只有控制词占据完整输入时才不区分大小写。其他任何非空后缀都属于目标，因此 `/goal pause after verification` 会创建该字面目标。goal 领域会修剪并验证目标。由于通用命令平面没有模态编辑器或确认原语，`edit` 会内联接收替换内容；若替换内容不完整，则直接返回错误，提示用户执行 edit 或 clear。
+只有控制词占据完整输入时才不区分大小写。其他任何非空后缀都属于目标，因此 `/goal pause after verification` 会创建该字面目标。goal 领域会去除目标首尾空白并进行验证。由于通用命令平面没有模态编辑器或确认原语，`edit` 会内联接收替换内容；若试图替换未完成的 goal，则直接返回错误，提示用户执行 edit 或 clear。
 
-可预期的领域拒绝会变成稳定的直接命令错误，不公开品牌化 id 或 revision。意外实现失败仍会 reject 分发，使适配器能将其报告为命令失败。通用命令文本和输出只属于活跃 UI 状态；每项已接受变更都由 `dsh-goal` 持久化并提供给模型，而不是由此插件完成。
+可预期的领域拒绝会变成稳定的直接命令错误，不公开带品牌类型的 id 或 revision。意外实现失败仍会 reject 分发，使适配器能将其报告为命令失败。通用命令文本和输出仍属于实时 UI 状态；每项已接受变更都由 `dsh-goal` 持久化并提供给模型，而不是由此插件完成。
 
 ## 组合
 
@@ -32,7 +32,7 @@
   name: '@deepseek-ai/dsh-command-goal'
 ```
 
-TUI 应用默认启用完整的持久 goal 栈和此命令。ACP 自动化应用会启用领域与模型工具，但不挂载命令注册表；`goals: false` 会移除该栈。无 UI 的 `agent-spine-demo` 必须显式配置 `goals: {}`，避免无头单次调用方在不知情时从一个物理轮次变为多 round 操作。
+TUI 应用默认启用完整的持久 goal 栈和此命令。ACP（Agent Client Protocol）自动化应用会启用领域与模型工具，但不挂载命令注册表；`goals: false` 会移除该栈。无 UI 的 `agent-spine-demo` 必须显式配置 `goals: {}`，避免无头单次调用方在不知情时从一个物理轮次变为包含多个 Round 的操作。
 
 ## 模型体验
 
@@ -44,15 +44,15 @@ TUI 应用默认启用完整的持久 goal 栈和此命令。ACP 自动化应用
 
 #### Token 影响
 
-读取状态或收到直接命令错误不会增加模型 token。每项已接受变更都会增加 goal 领域保留的完整快照；已启用的同会话驱动器还可能增加后续 goal-round 提示词。
+读取状态或收到直接命令错误不会增加模型 token。每项已接受变更都会增加 goal 领域保留的完整快照；已启用的同会话驱动器还可能增加后续 Goal Round 提示词。
 
 #### KV Cache 影响
 
 命令发现与直接输出不会影响缓存。变更会追加到可复用历史前缀之后；后续压缩可能替换派生历史后缀。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- **仅纯文本交互**：通用命令注册表没有模态编辑表单或替换确认回调；内联 edit 与显式 clear 能在不同适配器中保持确定的破坏性意图。
-- **没有逐命令 round 上限参数**：`defaultMaxGoalRounds` 仍是部署配置；用户直接请求时，可以要求模型通过另行授权的 goal 工具编辑 `max_goal_rounds`。
-- **没有持续状态组件**：裸 `/goal` 是可移植的观察接口；适配器专用徽标和可重新连接的命令输出仍属于未来 UI 工作。
-- **随附应用中只有 TUI 使用此命令**：无头 CLI、ACP 自动化和 JSON-RPC 适配器不消费 `ctx.commands`。如果组合中包含面向模型的 goal 工具，普通提示词仍能授权它们。
+- **仅纯文本交互**：通用命令注册表没有模态编辑表单或替换确认回调；内联 edit 与显式 clear 能在不同适配器中保持明确且一致的破坏性意图。
+- **没有逐命令 Round 上限参数**：`defaultMaxGoalRounds` 仍是部署配置；用户直接请求时，可以要求模型通过另行授权的 goal 工具编辑 `max_goal_rounds`。
+- **没有持续状态组件**：裸 `/goal` 是可移植的观察接口；适配器专用徽标和重连后可恢复的命令输出仍属于未来 UI 工作。
+- **随附应用中只有 TUI 使用此命令**：无头 CLI（命令行界面）、ACP 自动化和 JSON-RPC 适配器不消费 `ctx.commands`。如果组合中包含面向模型的 goal 工具，普通提示词仍能授权它们。
diff --git a/packages/goal/goal-session/README.i18n.yaml b/packages/goal/goal-session/README.i18n.yaml
index e14d52c557..c0c9f24f89 100644
--- a/packages/goal/goal-session/README.i18n.yaml
+++ b/packages/goal/goal-session/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/goal/goal-session/README.md
 README.md: 6a1c3b9455c93762c2458109c753588ce9a08d9a
-README.zh.md: d06d2d96deb845f72c5c88bc47a04a1af621c078
+README.zh.md: 4162411bf2dbeebbec8da6433c71e176057c4277
diff --git a/packages/goal/goal-session/README.zh.md b/packages/goal/goal-session/README.zh.md
index d06d2d96de..4162411bf2 100644
--- a/packages/goal/goal-session/README.zh.md
+++ b/packages/goal/goal-session/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-[`ctx.goals`](../goal/README.md) 的同会话续行驱动器。它通过公开 `Agent` 与会话 seam，把活跃且已激活的目标转换为连续的 [goal round](../../../docs/glossary.md#goal-round)；[同会话驱动器 Agent Note](../../../.agents/notes/implemented/feature/2026-07-19-same-session-goal-round-driver.md)负责竞态和生命周期理由。
+[`ctx.goals`](../goal/README.md) 的同会话续行驱动器。它通过公开 `Agent` 与会话 seam，把 phase 为 active 且已启用续行的目标转换为连续的 [Goal Round](../../../docs/glossary.md#goal-round)；[同会话驱动器 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-same-session-goal-round-driver.md) 记载竞态与生命周期方面的设计理由。
 
 ## 组合
 
@@ -21,9 +21,9 @@
 
 ## Round 契约
 
-当完全相同的活跃 agent 处于 idle 状态，且目标 active、已经激活并有剩余容量时，驱动器先为待处理 goal 变更创建检查点，再预留 `roundsStarted + 1`，对应当前 `{ goalId, revision }`。它会排入一条 `<goal_round>` 提示词，并携带 `GoalMessageSource`。通过 `agent/prompt-submit` 准入时，会在下游提示词 hook 前后同时验证完整的排队记录与当前 goal；只有被接受的 `user/message` 才会增加 `roundsStarted`。因陈旧而被拒绝的预留不会消耗 round 编号。
+当对应的活跃 agent 实例处于 idle 状态，且目标 phase 为 active、已启用续行并有剩余容量时，驱动器先为待处理 goal 变更创建检查点，再预留 `roundsStarted + 1`，对应当前 `{ goalId, revision }`。它会排入一条 `<goal_round>` 提示词，并携带 `GoalMessageSource`。通过 `agent/prompt-submit` 准入时，会在下游提示词钩子前后验证完整的排队记录与当前 goal；只有被接受的 `user/message` 才会增加 `roundsStarted`。因陈旧而被拒绝的预留不会消耗 Round 编号。
 
-一个 goal round 拥有一个普通会话轮次，该轮次可以包含多个模型／工具步骤。驱动器只会把预留与 `message` 轮次配对，且该轮次必须携带完全相同的 `GoalMessageSource`；可通过声明合并扩展的插件轮次触发器不会准入或替换该预留。用户消息仍是普通轮次，不消耗 goal 上限。如果用户工作在预留前进入 inbox，或加入预留的待处理批次，自动工作会让行，直到用户工作结算；混合批次中的待处理自动提示词会被拒绝，只有 agent 再次 idle 后才重新预留。
+一个 Goal Round 对应一个普通会话轮次，该轮次可以包含多个模型／工具步骤。驱动器只会把预留与 `message` 轮次配对，且该轮次必须携带完全相同的 `GoalMessageSource`；可通过声明合并扩展的插件轮次触发器不会准入或替换该预留。用户消息仍是普通轮次，不消耗 goal 上限。如果用户工作在预留前进入 inbox，或加入预留的待处理批次，自动工作会让行，直到用户工作结算；混合批次中的待处理自动提示词会被拒绝，只有 agent 再次 idle 后才重新预留。
 
 保留的提示词会点明经过 JSON 引用的目标与 `round/maxGoalRounds`，将当前工作区、工具结果和持久会话状态视为权威信息，要求在完成前提供证据，并要求在工作仍未完成时保持目标 active。引用可将多行或形似标签的目标文本保留为数据。goal 生命周期变更仍必须通过 `dsh-tool-goal` 的独立权限检查。
 
@@ -31,43 +31,43 @@
 
 | 持久轮次结果 | Goal 操作 | 自动重试 |
 |---|---|---|
-| goal 仍 active 且已激活时的 `completed` | 准入下一 round；达到上限时以代码 `round-limit` 阻塞 | 是 |
-| 已预留／准入 goal round 的取消，或其 `aborted` 结果 | `paused` | 否 |
-| 未尝试 goal round 时取消 | 保留持久 phase；撤销激活 | 否 |
+| goal phase 仍为 active 且已启用续行时的 `completed` | 准入下一 Round；达到上限时以代码 `round-limit` 阻塞 | 是 |
+| 已预留／准入 Goal Round 的取消，或其 `aborted` 结果 | `paused` | 否 |
+| 未尝试 Goal Round 时取消 | 保留持久 phase；撤销激活 | 否 |
 | `error` 且带 `RATE_LIMIT` 或 `QUOTA` | 设为 `blocked`，代码为 `usage-limited` | 否 |
 | 其他 `error`、`max-tokens` 或非陈旧提示词拒绝 | 以诊断代码和消息设为 `blocked` | 否 |
-| 持久性失败、资源释放、中断或未知未来结果 | 撤销激活或阻塞，以便检查 | 否 |
+| 持久性失败、dispose（资源释放）、中断或未知未来结果 | 撤销激活或阻塞，以便检查 | 否 |
 
-某个 goal 在自身 round 中发生的变更，会取代旧 revision 的结算。因此，即使物理轮次随后关闭，完成、暂停、阻塞和编辑仍具有最终决定权。任何异常结果都不会自动重试。
+某个 goal 在自身 Round 中发生的变更，会取代旧 revision 的结算。因此，即使物理轮次随后关闭，完成、暂停、阻塞和编辑仍具有最终决定权。任何异常结果都不会自动重试。
 
 ## 生命周期与持久性
 
-`goal/changed` 会产生持久性义务。排队工作前，驱动器会等待 `ctx.sessions.flush()`，并在等待后重新检查 goal revision 与竞争输入。关闭时的 flush 失败通过 `agent/error` 到达；即使后续一次性注入已经追加另一轮次，驱动器仍会把失败关联到完全相同的已关闭轮次，然后撤销激活，避免另一 round 启动。
+`goal/changed` 会产生持久性义务。排队工作前，驱动器会等待 `ctx.sessions.flush()`，并在等待后重新检查 goal revision 与竞争输入。关闭时的 flush 失败通过 `agent/error` 到达；即使后续一次性注入已经追加另一轮次，驱动器仍会把失败关联到完全相同的已关闭轮次，然后停用续行，避免另一 Round 启动。
 
-此插件加载到现有 agent 上时绝不会继承激活状态。`GoalService.disarm()` 会移除进程本地权限，而不改变持久 phase、revision 或历史；之后由用户明确授权的 resume 会记录重新激活。会话 resume 和 fork 后，goal 领域通过 `agent/session-start` 处理应用相同规则。
+此插件加载到现有 agent 上时绝不会继承续行启用状态。`GoalService.disarm()` 会移除进程本地权限，而不改变持久 phase、revision 或历史；之后由用户明确授权的 resume 会记录重新启用续行。会话 resume 和 fork 后，goal 领域通过 `agent/session-start` 处理应用相同规则。
 
-取消采用先观察、后行动的顺序：具体循环会在清空队列或中止轮次前，发送带类型 cause 的 `agent/cancel-requested`。只有取消操作拥有已预留或已准入的 goal 尝试时，插件才会持久暂停 active goal；取消无关用户工作只会撤销进程本地续行权限。如果 pause 变更失败，驱动器会回退到撤销激活。插件 teardown 会关闭准入，撤销所有活跃 goal 的激活，以 `parent` cause 取消已经准入的 round，并在事件隔离仍安装的情况下等待驱动器和 agent 完全停稳。
+取消采用先观察、后行动的顺序：具体循环会在清空队列或中止轮次前，发送带类型 cause 的 `agent/cancel-requested`。仅当取消操作所针对的是已预留或已准入的 Goal Round 尝试时，插件才会持久暂停 active goal；取消无关用户工作只会撤销进程本地续行权限。如果 pause 变更失败，驱动器会回退到停用续行。插件 teardown 会关闭准入，停用所有活跃 goal 的续行，以 `parent` cause 取消已经准入的 Round，并在事件隔离仍安装的情况下等待驱动器和 agent 完全停稳。
 
 ## 模型体验
 
-### Goal-round 提示词
+### Goal Round 提示词
 
 #### 模型看到的内容
 
-每个已准入 round 都是一段保留的用户角色 `<goal_round>` 块，其中点明完整目标与正 round 编号。更早的用户消息、goal 状态快照、assistant 输出与工具记录仍保留在同一会话历史中。
+每个已准入 Round 都是一段保留的用户角色 `<goal_round>` 块，其中点明完整目标与正数 Round 编号。更早的用户消息、goal 状态快照、assistant 输出与工具记录仍保留在同一会话历史中。
 
 #### Token 影响
 
-每个已准入 round 会增加一个固定指令块和目标。后续请求会重新发送保留的 round，直到压缩将其遮蔽；不会创建新 agent，也不会复制对话前缀。
+每个已准入 Round 会增加一个固定指令块和目标。后续请求会重新发送保留的 Round，直到压缩（compaction）将其遮蔽；不会创建新 agent，也不会复制对话前缀。
 
 #### KV Cache 影响
 
-在一个 epoch 内仅追加：每个已准入 round 都会在可复用前缀后扩展现有对话。压缩可能替换派生历史后缀，并移动可复用边界。
+在一个 epoch 内仅追加：每个已准入 Round 都会在可复用前缀后扩展现有对话。压缩可能替换派生历史后缀，并移动可复用边界。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **没有独立评估器**：面向模型的 goal 策略会判断证据是否足以完成，以及 blocker 在语义上是否未变；评估器支持的认证仍保持暂缓。
-- **只在同一会话执行**：此包有意不 spawn 新 agent、不 fork 会话前缀，也不实现 Ralph 风格的独立尝试；该工作流属于自己的插件层。
-- **已接受队列的卸载竞态**：Cordis 插件卸载是异步的。已经被 agent inbox 接受的 goal 提示词可以在卸载开始前启动并消耗其 round；teardown 随后会取消请求、撤销 goal 激活并等待完全停稳。不会再启动后续 round。
-- **只有 round 上限，不是资源预算**：token、货币、时间与提供方配额策略保持独立；观察到 `RATE_LIMIT` 和 `QUOTA` 时，只会映射为阻塞原因代码 `usage-limited`。
-- **异常情况不自动重试**：短暂的提供方与持久化失败需要之后由用户授权 resume，而不是隐式重试策略。
+- **只在同一会话执行**：此包（package）有意不 spawn 新 agent、不 fork 会话前缀，也不实现 Ralph 风格的独立尝试；该工作流属于单独的插件层。
+- **已接受队列的卸载竞态**：Cordis 插件卸载是异步的。已经被 agent inbox 接受的 goal 提示词可以在卸载开始前启动并消耗其 Round；teardown 随后会取消请求、撤销 goal 激活并等待完全停稳。不会再启动后续 Round。
+- **只有 Round 上限，不是资源预算**：token、货币、时间与提供方配额策略保持独立；观察到 `RATE_LIMIT` 和 `QUOTA` 时，只会映射为阻塞原因代码 `usage-limited`。
+- **异常情况不自动重试**：暂时性的提供方与持久化失败需要之后由用户授权 resume，而不会采用隐式重试策略。
diff --git a/packages/goal/goal/README.i18n.yaml b/packages/goal/goal/README.i18n.yaml
index 88c0530bd2..1c724ef9c0 100644
--- a/packages/goal/goal/README.i18n.yaml
+++ b/packages/goal/goal/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/goal/goal/README.md
 README.md: 45d7ac2e1c8fbd60ab2bc8b917b326d20da02ec1
-README.zh.md: 9936b8af1a5f5b49486eeda6bf137ee534377f06
+README.zh.md: b657d1d52f75657ba44b99306e4621aa82688e96
diff --git a/packages/goal/goal/README.zh.md b/packages/goal/goal/README.zh.md
index 9936b8af1a..b657d1d52f 100644
--- a/packages/goal/goal/README.zh.md
+++ b/packages/goal/goal/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-事件溯源的同会话目标状态。该服务在 agent（智能体）的现有会话中保留一个当前完成目标，同时将继续执行的权限作为进程本地激活状态。[goal 领域 Agent Note](../../../.agents/notes/implemented/feature/2026-07-19-persisted-same-session-goal-domain.md)负责设计理由；[goal 类型目录](../../../docs/core-data-structures/goal.md)记录字面数据形状。
+事件溯源的同会话目标状态。该服务在 agent（智能体）的现有会话中保留一个当前完成目标，同时将继续执行的权限作为进程本地续行启用状态。[goal 领域 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-persisted-same-session-goal-domain.md) 负责设计理由；[goal 类型目录](../../../docs/core-data-structures/goal.md)记录具体的数据形状。
 
 ## 配置
 
@@ -13,25 +13,25 @@
     defaultMaxGoalRounds: 256
 ```
 
-`defaultMaxGoalRounds` 必须是正安全整数。`create()` 会在提交目标前于内部物化这项部署默认值；请求级取值可以覆盖它。
+`defaultMaxGoalRounds` 必须是正的安全整数。`create()` 会在提交目标前于内部物化这项部署默认值；请求级取值可以覆盖它。
 
 ## 服务契约
 
-`ctx.goals` 只接受以对应 id 注册的完全相同的活跃 `Agent` 实例。`get()` 返回分离的 `GoalView`；变更通过 `GoalRef { id, revision }` 比较并设置限制，并拒绝陈旧引用。服务通过生成的[服务目录](../../../docs/cordis-catalog/services.md)公开 create、edit、pause、resume、complete、block 和 clear 动词。创建默认值在内部解析。`disarm()` 是仅供生命周期使用的例外：它移除进程本地续行权限，不写入新 revision，也不发送变更事件。
+`ctx.goals` 只接受以对应 id 注册的完全相同的活跃 `Agent` 实例。`get()` 返回与内部状态脱离的 `GoalView`；变更通过 `GoalRef { id, revision }` 比较并设置屏障，并拒绝陈旧引用。服务通过生成的[服务目录](../../../docs/cordis-catalog/services.md)公开 create、edit、pause、resume、complete、block 和 clear 动词。创建默认值在内部解析。`disarm()` 是仅供生命周期使用的例外：它移除进程本地续行权限，不写入新 revision，也不发出变更。
 
-最多只有一个当前目标。创建操作会生成 revision 为 1 的活跃目标并将其激活。未完成的目标必须编辑、转换或清除；已完成目标可以由拥有全局新 id 的目标替换。编辑会保留 phase、blocker reason 与 activation。暂停、完成、阻塞和清除都会撤销激活。阻塞会记录策略自有的 lower-kebab-case 代码和规范化的自由文本说明；提供方限制、配置预算、执行错误与请求用户输入都使用这一种持久 phase，不会扩增生命周期状态。只有配置的 round 上限仍有剩余容量时，resume 才接受已停止 phase 或撤销激活的 active 目标；它会清除原 blocker reason。活跃且已激活的目标会拒绝冗余操作。
+最多只有一个当前目标。创建操作会生成 revision 为 1、phase 为 active 的目标并启用续行。未完成的目标必须编辑、转换或清除；已完成目标可以由拥有全局未使用过的 id 的目标替换。编辑会保留 phase、blocker reason 与 activation。暂停、完成、阻塞和清除都会停用续行。阻塞会记录策略自有的 lower-kebab-case 代码和规范化的自由文本说明；提供方限制、配置预算、执行错误与请求人工输入都使用这一种持久 phase，不会扩增生命周期状态。只有配置的 Round 上限仍有剩余容量时，resume 才接受已停止 phase 或 phase 为 active 但已停用续行的目标；它会清除原 blocker reason。phase 为 active 且已启用续行的目标会拒绝冗余操作。
 
-每次非 clear 变更都会通过 `agent.inject()` 追加完整的版本化快照；clear 则追加带 revision 的 tombstone。模型可见的 `user/message` 内容与其带类型的 `{ kind: 'goal', change }` 来源必须完全一致。回放会拒绝形状错误、来源／内容漂移、不连续 revision、非法生命周期转换、每目标时间戳非单调，以及不连续的 goal round。墙上时间倒退时，变更时间戳会限制在不早于上一次目标更新的值。
+每次非 clear 变更都会通过 `agent.inject()` 追加完整的版本化快照；clear 则追加带 revision 的 tombstone。模型可见的 `user/message` 内容与其带类型的 `{ kind: 'goal', change }` 来源必须完全一致。回放会拒绝形状错误、来源／内容漂移、不连续 revision、非法生命周期转换、每目标时间戳非单调，以及不连续的 Goal Round。挂钟时间倒退时，变更时间戳会限制在不早于上一次目标更新的值。
 
-注入可以立即追加，也可能在活跃工具批次 FIFO 中等待。服务会在内存中覆盖已经接受但尚待写入的变更，并在每个确切载荷进入日志时执行协调，因此连续的模型工具变更可以看到自身最新 revision，而不会把尚未记录的缓存当作持久状态。可重入追加观察者会且只会看到每项已接受变更一次；增量回放会把游标保留在第一个损坏事件处。追加或入队成功后才触发 `goal/changed`；监听器失败会受到隔离。
+注入可以立即追加，也可能在活跃工具批次 FIFO 中等待。服务会在内存中叠加已接受的待处理变更，并在每个完全一致的载荷进入日志时逐一完成对账，因此连续的模型工具变更可以看到自身最新 revision，而不会把尚未记录的缓存当作持久状态。可重入追加观察者会且只会看到每项已接受变更一次；增量回放会把游标保留在第一个损坏事件处。追加或入队成功后才触发 `goal/changed`；监听器失败会被隔离处理。
 
-激活状态绝不持久化。新缓存与每条 `agent/session-start` 边都会撤销激活，即使回放找到了持久 phase 为 active 的目标。续行驱动器在卸载前或持久性不确定后也会调用 `disarm()`。因此，会话恢复、fork 与驱动器替换会保留目标、phase、revision 和已准入 round 数量，却不会启动工作；之后必须通过显式 resume 变更重新激活续行。
+续行启用状态绝不持久化。新缓存与每次触发 `agent/session-start` 时都会停用续行，即使回放找到了持久 phase 为 active 的目标。续行驱动器在卸载前或持久性不确定后也会调用 `disarm()`。因此，会话恢复、fork 与驱动器替换会保留目标、phase、revision 和已准入 Round 数量，却不会启动工作；之后必须通过显式 resume 变更重新启用续行。
 
-单独发布的 `./invariant` 配套模块会为每个已挂接会话维护独立折叠。它会在候选事件进入持久日志前拒绝错误的 goal 来源变更、模型可见内容漂移、不连续 revision、非法生命周期转换、时间戳回退，以及不连续的已准入 round。
+单独发布的 `./invariant` 配套模块会为每个已挂接会话维护独立折叠。它会在候选事件进入持久日志前拒绝格式错误的 goal 来源变更、模型可见内容漂移、不连续 revision、非法生命周期转换、时间戳回退，以及不连续的已准入 round。
 
 ## 扩展点
 
-策略插件调用服务动词，并响应限定范围的 `goal/changed` 事件。续行消费方将 round 准入为 `user/message` 事件，并携带 `GoalMessageSource`；普通用户轮次绝不会增加 `roundsStarted`。消费方使用 `Agent` 接口和事件，不导入 `dsh-agent-loop`。
+策略插件调用服务动词，并响应限定范围的 `goal/changed` 事件。续行消费方将 Round 准入为 `user/message` 事件，并携带 `GoalMessageSource`；普通的人类轮次绝不会增加 `roundsStarted`。消费方使用 `Agent` 接口和事件，不导入 `dsh-agent-loop`。
 
 ## 模型体验
 
@@ -43,16 +43,16 @@
 
 #### Token 影响
 
-每项保留的变更都会向派生历史增加一份完整快照，直到压缩将其遮蔽。完整快照让每条记录都能独立检查，但会重复目标和生命周期字段。
+每项保留的变更都会向派生历史增加一份完整快照，直到压缩（compaction）将其遮蔽。完整快照让每条记录都能独立检查，但会重复目标和生命周期字段。
 
 #### KV Cache 影响
 
 在一个 epoch 内仅追加：每项变更都位于可复用请求前缀和既有历史之后。压缩可能替换派生历史后缀，并移动可复用边界。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- **只负责状态，不负责任务调度**：此包不决定已激活目标何时继续，不重试异常失败，也不取消活跃轮次；这些策略属于 agent seam 消费方。
-- **只有 round 数量预算**：`maxGoalRounds` 不计量 token、货币、墙上时间或提供方配额。
+- **只负责状态，不负责任务调度**：此包（package）不决定已启用续行的目标何时继续，不重试异常失败，也不取消活跃轮次；这些策略属于 agent seam 消费方。
+- **只有 Round 数量预算**：`maxGoalRounds` 不计量 token、货币、挂钟时间或提供方配额。
 - **没有独立评估器**：记录完成或阻塞的调用方拥有最终决定权；由评估器支持的认证暂缓到独立策略层。
 - **只有一个当前目标**：系统有意不支持并行目标或独立目标数据库；替换或清除后，历史仍可在会话日志中读取。
-- **信任进程内生产方**：能直接访问 `Session` 的插件可以追加伪造的 goal 来源数据。严格回放会检测错误或不一致的记录，并使 goal 访问从该记录起失败，直到日志修复；这是完整性检测，不是插件隔离。
+- **信任进程内生产方**：能直接访问 `Session` 的插件可以追加伪造的 goal 来源数据。严格回放会检测格式错误或不一致的记录，并使 goal 访问从该记录起失败，直到日志修复；这是完整性检测，不是插件隔离。
diff --git a/packages/goal/tool-goal/README.i18n.yaml b/packages/goal/tool-goal/README.i18n.yaml
index a206995b5c..3b9dd9f964 100644
--- a/packages/goal/tool-goal/README.i18n.yaml
+++ b/packages/goal/tool-goal/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/goal/tool-goal/README.md
 README.md: aaed61dd517aeb2f94efa22c34c64d1068155d46
-README.zh.md: ca41900f1b157e51fec6804db6dc79e699bd1db9
+README.zh.md: 5365b64ef65fb3d3f00e19357327479ebd8285a8
diff --git a/packages/goal/tool-goal/README.zh.md b/packages/goal/tool-goal/README.zh.md
index ca41900f1b..5365b64ef6 100644
--- a/packages/goal/tool-goal/README.zh.md
+++ b/packages/goal/tool-goal/README.zh.md
@@ -2,27 +2,27 @@
 
 [English](README.md) | 中文
 
-[`ctx.goals`](../goal/README.md) 的面向模型控制接口：`get_goal`、`create_goal` 和 `update_goal`。[goal 工具 Agent Note](../../../.agents/notes/implemented/feature/2026-07-19-model-facing-goal-tools.md)负责权限拆分与 Codex 风格用户体验。
+[`ctx.goals`](../goal/README.md) 的面向模型控制接口：`get_goal`、`create_goal` 和 `update_goal`。[goal 工具 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-model-facing-goal-tools.md) 负责权限拆分与 Codex 风格用户体验。
 
 ## 工具
 
-- `get_goal()` 返回当前 goal 或 `null`，包括比较并设置 id／revision、持久 phase、已经准入／受限的 goal round、任何 blocker reason，以及当前进程本地激活状态。
-- `create_goal(objective, max_goal_rounds?)` 从顶层用户直接轮次创建一个 goal。模型可以从长期 goal 意图中推断，而无需精确命令短语；非用户轮次和 subagent 会在执行时被拒绝。
+- `get_goal()` 返回当前 goal 或 `null`，包括比较并设置 id／revision、持久 phase、Goal Round 的已准入数／上限、任何 blocker reason，以及当前进程本地续行启用状态。
+- `create_goal(objective, max_goal_rounds?)` 根据顶层人类直接发起的轮次创建一个 goal。模型可以推断长期运行的 goal 意图，而无需精确命令短语；非人类轮次和 subagent 会在执行时被拒绝。
 - `update_goal(goal_id, revision, action, objective?, max_goal_rounds?, blocked_reason?)` 支持 `edit`、`pause`、`resume`、`complete` 和 `blocked`。替换值只属于 `edit`；`blocked_reason` 只有在 action 为 `blocked` 时才必填，并以稳定代码 `model-reported` 持久化。严格 schema 下的空字符串和零填充值视为省略，而有意义的值仍限定到各自 action。
 
 所有调用都互斥，因此模型排序的批次能观察到更早变更及其新 revision。UI 客户端会收到纯通用卡片：`get_goal` 使用 read，变更使用 other。变更卡片选择第一个有意义的 action 值，否则显示 goal id，因此已接受的填充值绝不会产生空输入。
 
 3 个规范值都与已经渲染给 Native 调用方的紧凑 JSON 一致：`{ goal: null }` 或 `{ goal: { id, revision, objective, phase, roundsStarted, maxGoalRounds, blockedReason? }, activation }`。因此，编程消费方无需解析渲染后的 JSON，即可收到相同领域结构。
 
-自主 goal round 成功报告 `complete` 或 `blocked` 时，会用 `concludeTurn()` 标记该次工具执行，使物理轮次在该步骤后停止。用户直接变更绝不会贡献该停止决策：assistant 可以确认变更，并发的用户 steering 仍可进入循环。
+自主 Goal Round 成功报告 `complete` 或 `blocked` 时，会用 `concludeTurn()` 标记该次工具执行，使物理轮次在该步骤后停止。人类直接变更绝不会导致这种停止：assistant 可以确认变更，循环仍可接收并发的人类 steering（中途引导）。
 
 ## 权限
 
 执行要求完全相同的活跃 `exec.agent`、其继承的 `AgentRegistry` initiator、running 状态与开放轮次。create、edit、pause 和 resume 还要求运行时根 agent 的当前轮次中存在已接受的 `{ kind: 'user' }` 消息或 steering 事件。持久 fork 谱系不会降低已恢复根 agent 的等级；活跃 subagent 所有权会降低。
 
-`{ kind: 'user' }` 是宿主证明。`Agent.followup()` 与 `steer()` 会在调用方省略 source 时分配该值，因此插件、调度器与其他非用户生产方必须传入自己的 source，不能继承用户权限。
+`{ kind: 'user' }` 是宿主证明。`Agent.followup()` 与 `steer()` 会在调用方省略 source 时分配该值，因此插件、调度器与其他非人类生产方必须传入自己的 source，不能继承用户权限。
 
-complete 与 blocked 还接受完全相同的当前 goal round：来源为 goal 的 `user/message`，其 id、revision 和 round 与折叠后的当前 goal 相等。在达到 `blockedAfterConsecutiveRounds` 前，goal-round 的 blocked 调用会被机械拒绝；模型判断同一条件是否确实持续，并必须在 `blocked_reason` 中说明。用户直接权限可以立即停止 goal。
+complete 与 blocked 还接受完全一致的当前 Goal Round：来源为 goal 的 `user/message`，其 id、revision 和 Round 编号与折叠后的当前 goal 相等。在达到 `blockedAfterConsecutiveRounds` 前，Goal Round 的 blocked 调用会被机械拒绝；模型判断同一条件是否确实持续，并必须在 `blocked_reason` 中说明。人类直接授权可以立即停止 goal。
 
 ## 配置
 
@@ -33,7 +33,7 @@ complete 与 blocked 还接受完全相同的当前 goal round：来源为 goal
     blockedAfterConsecutiveRounds: 3
 ```
 
-该值必须是正安全整数。它既提供模型自行阻塞的硬下限，也决定模型指引中点名的数量。
+该值必须是正的安全整数。它既提供模型自行报告阻塞的硬下限，也决定模型指引中指明的数值。
 
 ## 模型体验
 
@@ -41,7 +41,7 @@ complete 与 blocked 还接受完全相同的当前 goal round：来源为 goal
 
 #### 模型看到的内容
 
-固定 goal 策略说明何种用户语义意图值得创建 goal，要求更新前先精确读取 ref，解释会话 resume／fork 后如何重新激活，并限制完成／阻塞声明。配置的阈值会插入该指引。
+固定 goal 策略说明何种用户语义意图值得创建 goal，要求更新前先精确读取 ref，解释会话 resume／fork 后如何重新启用续行，并限制完成／阻塞声明。配置的阈值会插入该指引。
 
 ##### Goal 策略
 
@@ -55,26 +55,26 @@ Use goal tools for one long-running completion objective in the current session.
 
 #### KV Cache 影响
 
-插件范围、配置阈值和指引文本不变时，前缀保持稳定。激活、资源释放或配置变更可能使此提示词章节的复用失效。
+插件范围、配置阈值和指引文本不变时，前缀保持稳定。启用、dispose（资源释放）或配置变更可能使此提示词章节的复用失效。
 
 ### 工具 schema 与结果
 
 #### 模型看到的内容
 
-生成的 [`get_goal`、`create_goal` 和 `update_goal` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-goal)。成功结果是紧凑 JSON。变更结果之后是工具批次结束后由 goal 领域产生的原始 `<goal_state>` 快照。结果中的 `activation` 是活跃观察值，绝不会成为回放权限依据。
+生成的 [`get_goal`、`create_goal` 和 `update_goal` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-goal)。成功结果是紧凑 JSON。变更结果之后是工具批次结束后由 goal 领域产生的原始 `<goal_state>` 快照。结果中的 `activation` 是实时观察值，绝不会成为回放权限依据。
 
 #### Token 影响
 
-固定 schema 成本，加上每次调用的一条紧凑结果。变更还会保留领域快照，直到压缩。
+固定 schema 成本，加上每次调用的一条紧凑结果。变更还会保留领域快照，直到压缩（compaction）。
 
 #### KV Cache 影响
 
-Schema 的定义与可见性不变时，前缀保持稳定。调用、结果和生成的 goal 快照会追加到可复用请求前缀之后，不会使更早条目失效。
+schema 的定义与可见性不变时，前缀保持稳定。调用、结果和生成的 goal 快照会追加到可复用请求前缀之后，不会使更早条目失效。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- **语义意图仍由模型判断**：执行只能证明直接用户来源，无法证明请求是否足够重大而值得创建 goal。
-- **阻塞条件是否相同仍由模型判断**：运行时强制执行不同的已准入 round 计数，而不是障碍的语义等价性；独立评估器保持暂缓。
-- **不负责调度或直接用户呈现**：这些工具只变更状态；同会话驱动器与 [`dsh-command-goal`](../command-goal/README.md) 是同一领域的独立消费方。
-- **Goal-round 权限需要驱动器**：除非续行驱动器准入 goal 来源的用户轮次，否则自主 `complete`／`blocked` 路径不会启用；只挂载此工具包不会创建这些轮次。
+- **语义意图仍由模型判断**：执行只能证明人类直接来源，无法证明请求是否足够重大而值得创建 goal。
+- **阻塞条件是否相同仍由模型判断**：运行时强制统计互不重复的已准入 Goal Round，而不判断障碍在语义上是否等价；独立评估器的实现暂缓。
+- **不负责调度或直接面向人类呈现**：这些工具只变更状态；同会话驱动器与 [`dsh-command-goal`](../command-goal/README.md) 是同一领域的独立消费方。
+- **Goal Round 权限需要驱动器**：除非续行驱动器准入 goal 来源的用户轮次，否则自主 `complete`／`blocked` 路径不会启用；只挂载这个包（package）不会创建这些轮次。
 - **提示词注册与过滤相互独立**：某个范围可能隐藏工具，却保留指引，除非部署将两项注册限定在同一范围。
diff --git a/packages/guard/README.i18n.yaml b/packages/guard/README.i18n.yaml
index 30638b2a21..c323a9b295 100644
--- a/packages/guard/README.i18n.yaml
+++ b/packages/guard/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/guard/README.md
 README.md: b7375fd2bb12ae0cec94b13e6a1012c6f143bdad
-README.zh.md: ef218ef62c37ed315ce27541fbc52ffa12ae8e75
+README.zh.md: bba5c144d0663266e4327e388b9915cde176ce08
diff --git a/packages/guard/README.zh.md b/packages/guard/README.zh.md
index ef218ef62c..bba5c144d0 100644
--- a/packages/guard/README.zh.md
+++ b/packages/guard/README.zh.md
@@ -2,10 +2,10 @@
 
 [English](README.md) | 中文
 
-这组行为 guard 插件会监视 agent（智能体）循环中的无效模式，并提醒模型调整方向。这里只有一个**产品** 包，不设接口／实现 seam：guard 是现有核心 seam（`tools/post-execute`、`agent/prompt-submit`、`agent/status`）的自包含消费方，并非可替换能力。
+这组行为 guard 插件会监视 agent loop（智能体循环）中的低效模式，并提醒模型调整方向。这里只有一个**产品**包（package），不设接口／实现 seam：guard 是现有核心 seam（`tools/post-execute`、`agent/prompt-submit`、`agent/status`）的自包含消费方，并非可替换能力。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
-| `repeat-tool-guard/` | 当 agent 对完全相同的工具调用反复循环时给出提示 | （监听 `ctx.tools` 的 waterfall（瀑布式事件）） |
+| `repeat-tool-guard/` | 当 agent 对完全相同的工具调用反复循环时给出提示 | （监听 `ctx.tools` 的 waterfall，即瀑布式事件） |
 
 提示以 `additionalContexts` 形式附在 `tools/post-execute` 决策中传递；agent loop 会在该步骤的工具结果之后，将其追加为有日志记录、来源为插件的 `user/message` 事件（参见[工具包](../core/tools)）。因此，guard 告诉模型的所有内容都能从会话日志中重建。
diff --git a/packages/guard/repeat-tool-guard/README.i18n.yaml b/packages/guard/repeat-tool-guard/README.i18n.yaml
index 14fc1ee3b8..4863dcd147 100644
--- a/packages/guard/repeat-tool-guard/README.i18n.yaml
+++ b/packages/guard/repeat-tool-guard/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/guard/repeat-tool-guard/README.md
 README.md: 226dba10239031e8e79bd5698e77c213688ce579
-README.zh.md: 8e22a5e67d3700024934437c5c9c1a5e969b740d
+README.zh.md: fd6e291d6fe9f286d3dc3f921518f596f3749e3e
diff --git a/packages/guard/repeat-tool-guard/README.zh.md b/packages/guard/repeat-tool-guard/README.zh.md
index 8e22a5e67d..fd6e291d6f 100644
--- a/packages/guard/repeat-tool-guard/README.zh.md
+++ b/packages/guard/repeat-tool-guard/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-这是一个仅提供建议的循环中断器，而非面向模型的工具：它不会出现在工具列表中，不会否决或改写调用，只增加一种行为。它监视每个 agent（智能体）的工具调用流，统计以完全相同的规范化参数连续调用同一工具的次数；达到所配置的连续次数时，它会注入逐级增强的提示，要求模型停止重复、重新阅读上一次结果，并改用其他方案或结束任务。究竟是换一种方式重试、收集更多证据还是完成任务，仍完全由模型决定：合理的重复调用既不会延迟，也不会受阻。决策记录见 [repeat-tool-guard Agent Note](../../../.agents/notes/archived/feature/2026-07-08-repeat-tool-guard.md)。
+这是一个仅提供建议的循环中断器，而非面向模型的工具：它不会出现在工具列表中，不会否决或改写调用，只增加一种行为。它监视每个 agent（智能体）的工具调用流，统计以完全相同的规范化参数连续调用同一工具的次数；达到所配置的连续次数时，它会注入逐级增强的提示，要求模型停止重复、重新阅读上一次结果，并改用其他方案或结束任务。究竟是换一种方式重试、收集更多证据还是完成任务，仍完全由模型决定：合理的重复调用既不会延迟，也不会受阻。决策记录见 [repeat-tool-guard Agent Note（agent 决策记录）](../../../.agents/notes/archived/feature/2026-07-08-repeat-tool-guard.md)。
 
 ## 配置
 
@@ -16,7 +16,7 @@
     argumentsPreviewChars: 500   # default; cap on arguments quoted in the detailed reminder
 ```
 
-插件加载时，`thresholds` 会对错误配置快速失败：空列表、非整数、小于 2 的值或重复值都会抛出错误，绝不静默回退到默认值；`argumentsPreviewChars` 同样只接受大于等于 1 的整数。系统会将列表按升序规范化；第一个阈值只发送简短的通用提醒，后续每个阈值都会发送详细版本，列出工具、连续次数和规范参数。参数内容在 `argumentsPreviewChars` 处从头截断，并附带省略字符数标记，避免循环中的 `write`／`edit` 载荷无限制进入下一次请求（链键始终比较完整的规范字符串；此上限只约束提醒，不影响检测）。
+插件加载时，`thresholds` 会对错误配置快速失败：空列表、非整数、小于 2 的值或重复值都会抛出错误，绝不静默回退到默认值；`argumentsPreviewChars` 同样只接受大于等于 1 的整数。系统会将列表按升序规范化；第一个阈值只发送简短的通用提醒，后续每个阈值都会发送详细版本，列出工具、连续次数和规范参数。参数内容截取前 `argumentsPreviewChars` 个字符，并附带省略字符数标记，避免循环中的 `write`／`edit` 载荷无限制进入下一次请求（链键始终比较完整的规范字符串；此上限只约束提醒，不影响检测）。
 
 `include`／`exclude` 条目支持 `*` 通配符，并针对调用时实际存在的工具执行谓词判断，而不是引用注册表条目。因此，与当前任何已注册工具都不匹配的模式并非错误（未加载 MCP 工具的部署中，`exclude: [mcp_*]` 仍然有效）；这与 `toolOrder` 的引用目标检查不同。
 
@@ -27,8 +27,8 @@
 - **不受跟踪的调用对链透明。** 被 `include`／`exclude` 排除的调用既不递增计数器，也不重置计数器；因此，`grep X → todo_write → grep X` 仍算作连续两次 `grep X`，即使 `todo_write` 已被排除。这正是排除机制的价值：循环中穿插的记录类工具不能掩盖循环。
 - **被拒绝的调用也计数。** 检测位于 `tools/post-execute`；即便调用被 `tools/pre-execute` 监听器拒绝，该事件也会运行。模型反复尝试被拒绝的调用，恰恰是需要打断的循环。
 - **忽略没有 agent 的调用。** 直接调用 `ctx.tools.execute()` 的调用方没有需要提醒的模型，也没有可作为键的活跃 agent 对象。
-- **按 agent 分键。** 工具注册表位于上下文层级，subagent 会交错通过同一个 waterfall，因此每条链使用 `WeakMap<Agent, Chain>`，以活跃 agent 对象为键。一个 agent 的重复调用绝不会触发另一个 agent 的提醒。用户提示词（`agent/prompt-submit`）会重置提交该提示词的 agent 链；对象生命周期会自然限制弱引用条目的寿命，无需资源释放监听器。
-- **仅驻留内存。** 从持久化恢复的会话会从一条全新的链开始：guard 是启发式提醒，并非有日志记录的不变量；这是接受的代价，即后续提醒可能重新开始。
+- **按 agent 分键。** 工具注册表位于上下文层级，subagent 会交错通过同一个 waterfall（瀑布式事件），因此每条链使用 `WeakMap<Agent, Chain>`，以活跃 agent 对象为键。一个 agent 的重复调用绝不会触发另一个 agent 的提醒。用户提示词（`agent/prompt-submit`）会重置提交该提示词的 agent 链；对象生命周期会自然限制弱引用条目的寿命，无需 dispose（资源释放）监听器。
+- **仅驻留内存。** 从持久化恢复的会话会从一条全新的链开始：guard 是启发式提醒，并非有日志记录的不变量；提醒会延后，这是可接受的代价。
 
 ## 提醒传递
 
@@ -36,7 +36,7 @@
 
 ## 测试
 
-单元测试使用 mock 适配器（无网络）驱动真实 agent loop，并对上述链语义实现逐文件 100% 覆盖率。快照层负责 transcript（文本记录）接口：脚本化回放场景会将同一调用重复 5 次，并在 ACP transcript 中固定两个提醒层级，即第 3 次的温和提醒和第 5 次的详细提醒；二者均为注入的 `user/message`。
+单元测试使用 mock 适配器（无网络）驱动真实 agent loop，并对上述链语义实现逐文件 100% 覆盖率。快照层负责 transcript（文本记录）接口：脚本化回放场景会将同一调用重复 5 次，并在 ACP（Agent Client Protocol）的 transcript 中固定两个提醒层级，即第 3 次的温和提醒和第 5 次的详细提醒；二者均为注入的 `user/message`。
 
 ## 模型体验
 
@@ -84,10 +84,10 @@ The repeated calls are not making progress. Do not call this tool with these exa
 
 仅追加；新出现的内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **仅检测精确匹配**：规范化过程会对键进行深度排序，因此近似变体（稍作修改的路径、值内增加的空白）可以绕过链；在没有需求证据前，不采用模糊匹配。
-- **压缩不会重置链**：跨越压缩检查点的链会继续计数。
+- **压缩（compaction）不会重置链**：跨越压缩检查点的链会继续计数。
 - **仅提供建议**：尚未实现达到较高阈值后升级为 `block`，但 `PostToolDecision` 已支持阻止调用。
 - **subagent 之间不共享链**：链始终按 agent 隔离；即使父 agent 与其 subagent 重复相同调用，也不会合并计数。
 - **合理的幂等轮询超过阈值后仍会收到提醒**：可通过 `thresholds`／`exclude` 配置释放压力。
diff --git a/packages/hooks/README.i18n.yaml b/packages/hooks/README.i18n.yaml
index af9d4aa4a8..80f55791c1 100644
--- a/packages/hooks/README.i18n.yaml
+++ b/packages/hooks/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/hooks/README.md
 README.md: 23478fb5e9b813a3370ce465104b1f9db8b0a26a
-README.zh.md: 21c75f0476c76c0be75dc3af25ffb9a2be28dc4e
+README.zh.md: 41024a8bd268550aa07401fd8b21c74db0914796
diff --git a/packages/hooks/README.zh.md b/packages/hooks/README.zh.md
index 21c75f0476..41024a8bd2 100644
--- a/packages/hooks/README.zh.md
+++ b/packages/hooks/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-hooks 子系统让用户可以像使用 Claude Code 和 Codex 一样，在 agent 生命周期节点扩展 agent：把桥接插件指向现有的 `hooks.json`（或 settings），即可忠实运行这些外部 shell hook。规范的扩展表层本身是 harness 的类型化拦截 seam（见[拦截 seam Agent Note](../../.agents/notes/implemented/feature/2026-06-30-interception-seams.md)）；「原生 hook」只是这些 seam 上的普通 cordis 插件。这些包是把外部 shell-hook 协议转换到同一表层的**桥接**，另含它们共同依赖的共享协议格式库。
+hooks 子系统让用户可以像使用 Claude Code 和 Codex 一样，在 agent（智能体）生命周期节点扩展 agent：把桥接插件指向现有的 `hooks.json`（或 settings），即可忠实运行这些外部 shell hook。规范的扩展表层本身是 harness 的类型化拦截 seam（见[拦截 seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-06-30-interception-seams.md)）；「原生 hook」只是这些 seam 上的普通 Cordis 插件。这些包是把外部 shell-hook 协议转换到同一表层的**桥接**，另含它们共同依赖的共享协议格式（wire format）库。
 
 | 包 | 职责 | 形态 |
 |---|---|---|
@@ -10,4 +10,4 @@ hooks 子系统让用户可以像使用 Claude Code 和 Codex 一样，在 agent
 | `hooks-claude/` | Claude Code `hooks.json`／settings 的桥接 | 插件 |
 | `hooks-codex/` | Codex `hooks.json` 的桥接 | 插件 |
 
-Codex 有意重新实现 Claude Code 协议的一个*子集*（`hooks.json` 形状相同、5 个事件而非 CC 的众多事件、仅命令、仅正则表达式 matcher、没有 env／替换），因此 `hook-protocol` 拥有真正相同的原语，每个桥接只拥有不同部分（逐事件 stdin 载荷、env，以及把 hook 的中性结果映射到 harness 类型化 Decision 的方式）。参见 [hook-protocol/README.md](hook-protocol/README.md)。
+Codex 有意重新实现 Claude Code 协议的一个*子集*（`hooks.json` 结构相同、5 个事件而非 CC 的众多事件、仅命令、仅正则表达式 matcher、没有 env／替换），因此 `hook-protocol` 负责真正相同的原语，每个桥接只负责不同部分（逐事件 stdin 载荷、env，以及把 hook 的中性结果映射到 harness 类型化 Decision 的方式）。参见 [hook-protocol/README.md](hook-protocol/README.md)。
diff --git a/packages/hooks/hook-protocol/README.i18n.yaml b/packages/hooks/hook-protocol/README.i18n.yaml
index f74f1bb3a8..65faa23b75 100644
--- a/packages/hooks/hook-protocol/README.i18n.yaml
+++ b/packages/hooks/hook-protocol/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/hooks/hook-protocol/README.md
 README.md: 10cfcdcbf819f318f2ccaf412ae04bba60812397
-README.zh.md: f6fd30c968f68faa46d7ea07188cb22ee5ef3afe
+README.zh.md: 9862d4f332e0e82fb6479fcadf9376403fb9bef8
diff --git a/packages/hooks/hook-protocol/README.zh.md b/packages/hooks/hook-protocol/README.zh.md
index f6fd30c968..9862d4f332 100644
--- a/packages/hooks/hook-protocol/README.zh.md
+++ b/packages/hooks/hook-protocol/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-Claude Code／Codex hook 协议格式的**共享核心**。它不是 cordis 插件：不注册也不注入任何内容。它是一个**库**，提供两个桥接插件（`@deepseek-ai/dsh-hooks-claude`、`@deepseek-ai/dsh-hooks-codex`）导入的方言无关原语，使两者都无需重复实现协议中相同的部分。
+Claude Code／Codex hook 协议格式（wire format）的**共享核心**。它不是 Cordis 插件：不注册也不注入任何内容。它是一个**库**，提供两个桥接插件（`@deepseek-ai/dsh-hooks-claude`、`@deepseek-ai/dsh-hooks-codex`）导入的方言无关原语，使两者都无需重复实现协议中相同的部分。
 
-共享 lib 存在的原因是：Codex 有意重新实现了 Claude Code hook 协议的一个*子集*，包括相同的 `hooks.json` matcher group 形状、相同的退出码／stdout 输出契约以及相同的 command hook 执行模式。真正共享的部分位于此处；每个桥接只拥有不同之处。
+共享 lib 存在的原因是：Codex 有意重新实现了 Claude Code hook 协议的一个*子集*，包括相同的 `hooks.json` matcher group 结构、相同的退出码／stdout 输出契约以及相同的 command hook 执行模式。真正共享的部分位于此处；每个桥接只负责不同的部分。
 
-## 共享内容（此处）与每方言内容（桥接）
+## 共享内容（此处）与各方言内容（桥接）
 
 | 关注点 | 此处（`dsh-hook-protocol`） | 桥接（`dsh-hooks-claude` / `-codex`） |
 |---|---|---|
@@ -15,21 +15,21 @@ Claude Code／Codex hook 协议格式的**共享核心**。它不是 cordis 插
 | 解码输出 | `parseHookOutput(exit, stdout, stderr)` → 中性 `HookOutput` | 将中性 `HookOutput` 映射到 seam 特定的类型化 Decision |
 | 合并 N 个 hook | `mergeHookOutputs(outputs)` → 最严格的 `MergedHookOutcome` | （无） |
 | 持久记录 | `appendHookInvoked` / `appendHookResult`（`hook/*` 会话事件；结果的 `decision`／`stderrSummary` 从此处的 `HookOutput` 派生） | 在每次调用前后调用它们 |
-| 脱离运行完全停稳 | `createDetachedRuns()`：跟踪发射后不再等待的运行链；`drain()` 先 abort，再等待它们 | 将 `signal` 传给每个脱离的 `runHook`，并将 `drain` 注册为 effect disposer |
+| 脱离运行完全停稳 | `createDetachedRuns()`：跟踪触发后不等待的运行链；`drain()` 先 abort，再等待它们 | 将 `signal` 传给每个脱离的 `runHook`，并将 `drain` 注册为 effect disposer |
 
 ## 原语
 
 - **`matchesMatcher(matcher, query, mode)`**：缺失、`''` 或 `'*'` 时匹配全部；`claude` 模式将纯 `[A-Za-z0-9_|]+` pattern 视为字面值（pipe = 精确匹配交替），其他 pattern 视为正则；`codex` 模式始终使用未锚定正则。无效正则不匹配任何内容（绝不抛出异常）。
-- **`runHook(bash, hook, options, now)`**：要求并转发调用方拥有的 `options.signal`，将 `options.payload` 序列化到 hook stdin（当且仅当 `options.trailingNewline` 时添加尾随换行符），在执行器凭证清理后合并 `options.env`（`dsh-bash` 受信任插件表层），遵循 hook 的 `timeoutSec`（否则使用 `options.defaultTimeoutMs`；默认值属于桥接，其配置默认为 lib 的 `DEFAULT_HOOK_TIMEOUT_MS` 10 分钟参考值），再解码结果（将 `options.expectedEventName` 传递给 codec）。因此取消会到达执行器的进程组终止与 join 边界。它绝不抛出异常：执行器拒绝（基础设施故障）会变为 `HookOutput`，其 `exitCode: undefined`（非阻塞错误）。`now` 会被注入，以便测试持续时间。
-- **`parseHookOutput(exitCode, stdout, stderr, expectedEventName?)`** 解码退出状态与结构化 stdout。退出码 2 使用 stderr 阻塞；其他失败不阻塞。匹配的 hook 特定权限决策会覆盖遗留顶层决策；事件判别字段不匹配或缺失只会抑制事件特定字段。顶层字段仍与事件无关，成功但非 JSON 的输出会留给桥接处理。
-- **`mergeHookOutputs(outputs)`**：折叠在一个点上匹配的每个 hook 结果：权限优先级为 **deny > ask > allow**，首个 `continue:false` 使 halt 粘滞，阻塞原因用 `\n\n` 连接，`additionalContext`／`systemMessages` 按顺序累积。
-- **`createDetachedRuns()`**：为脱离运行的 emit 形状点跟踪完全停稳（没有 seam 等待它们）。桥接会跟踪每条运行链，包括 hook 运行及其 continuation，并将 `drain()` 注册为 effect disposer。drain 会触发 tracker 的 abort `signal`（因此仍在运行的 hook 进程会通过 `runHook` 终止，而不是等待到超时），随后在所有已跟踪链结算后 resolve。因此 `fiber.dispose()` resolve 时，没有脱离 hook 工作会留下并触发已 dispose 的上下文（见 [防御模式](../../../docs/defensive-patterns.md)：dispose 必须达到完全停稳）。
+- **`runHook(bash, hook, options, now)`**：要求并转发调用方拥有的 `options.signal`，将 `options.payload` 序列化到 hook stdin（当且仅当 `options.trailingNewline` 时添加尾随换行符），在执行器凭证清理后合并 `options.env`（`dsh-bash` 受信任插件接口），遵循 hook 的 `timeoutSec`（否则使用 `options.defaultTimeoutMs`；默认值属于桥接，其配置默认为 lib 的 `DEFAULT_HOOK_TIMEOUT_MS` 10 分钟参考值），再解码结果（将 `options.expectedEventName` 传递给 codec）。因此取消会到达执行器的进程组终止与 join 边界。它绝不抛出异常：执行器拒绝（基础设施故障）会变为 `HookOutput`，其 `exitCode: undefined`（非阻塞错误）。`now` 会被注入，以便测试持续时间。
+- **`parseHookOutput(exitCode, stdout, stderr, expectedEventName?)`** 解码退出状态与结构化 stdout。退出码为 2 时，会以 stderr 内容阻止执行；其他失败不阻塞。匹配的 hook 特定权限决策会覆盖遗留顶层决策；事件判别字段不匹配或缺失只会抑制事件特定字段。顶层字段仍与事件无关，成功但非 JSON 的输出会留给桥接处理。
+- **`mergeHookOutputs(outputs)`**：折叠在一个点上匹配的每个 hook 结果：权限优先级为 **deny > ask > allow**，从首个 `continue:false` 起，halt 状态保持不变，阻塞原因用 `\n\n` 连接，`additionalContext`／`systemMessages` 按顺序累积。
+- **`createDetachedRuns()`**：跟踪以 emit 形式脱离运行的点是否完全停稳（没有 seam 等待它们）。桥接会跟踪每条运行链，包括 hook 运行及其 continuation，并将 `drain()` 注册为 effect disposer。drain 会触发 tracker 的 abort `signal`（因此仍在运行的 hook 进程会通过 `runHook` 终止，而不是等待到超时），随后在所有已跟踪链结算后 resolve。因此 `fiber.dispose()` resolve 时，不会遗留任何可能作用于已 dispose（资源释放）的上下文的脱离 hook 工作（见 [防御模式](../../../docs/defensive-patterns.md)：dispose 必须达到完全停稳）。
 
 ## `hook/*` 会话事件
 
-通过 declaration merging 合并到 `SessionEventMap`（仅日志，与 `compact/*` 相同；不是 `SurfaceEventType`，没有 `surfaceOp`）：`hook/invoked`（hook 命令已运行）与 `hook/result`（其结果，按 `handlerId` 配对，由 `appendHookResult` 拥有决策规则）。Payload 与每事件 JSDoc 位于生成的 [持久化日志事件目录](../../../docs/persistence-catalog.md)；`stderrSummary` 会截断到记录的 `stderrSummaryMaxChars`（桥接配置，参考默认值 `DEFAULT_STDERR_SUMMARY_MAX_CHARS` = 500；为空时省略）。
+通过 declaration merging 合并到 `SessionEventMap`（仅日志，与 `compact/*` 相同；不是 `SurfaceEventType`，没有 `surfaceOp`）：`hook/invoked`（hook 命令已运行）与 `hook/result`（其结果，按 `handlerId` 配对，决策规则由 `appendHookResult` 负责）。Payload 与每事件 JSDoc 位于生成的 [持久化日志事件目录](../../../docs/persistence-catalog.md)；`stderrSummary` 会截断到记录的 `stderrSummaryMaxChars`（桥接配置，参考默认值 `DEFAULT_STDERR_SUMMARY_MAX_CHARS` = 500；为空时省略）。
 
-Hook 溯源记录必须位于开启轮次内。轮次中点（`PreToolUse`／`PostToolUse`／`Stop`）按构造满足这条由所有方定义的关系。`SessionStart` 与轮次前的 `UserPromptSubmit` 准入 seam 没有 `hook/*` 记录；获准的上下文改由其带来源的 `user/message` 作为证据，详见 hooks Agent Note。
+Hook 溯源记录必须位于一个尚未结束的轮次内。轮次中的点（`PreToolUse`／`PostToolUse`／`Stop`）按构造满足这条由所有者定义的关系。`SessionStart` 与轮次前的 `UserPromptSubmit` 准入 seam 没有 `hook/*` 记录；获准的上下文改由其带来源的 `user/message` 作为证据，详见 hooks Agent Note（agent 决策记录）。
 
 ## 模型体验
 
@@ -37,7 +37,7 @@ Hook 溯源记录必须位于开启轮次内。轮次中点（`PreToolUse`／`Po
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由具名消费方负责。
+不会直接失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
diff --git a/packages/hooks/hooks-claude/README.i18n.yaml b/packages/hooks/hooks-claude/README.i18n.yaml
index c4d7c1bdc9..1eaa331832 100644
--- a/packages/hooks/hooks-claude/README.i18n.yaml
+++ b/packages/hooks/hooks-claude/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/hooks/hooks-claude/README.md
 README.md: 24259c24ea35cd450f8ea27ca2cca423ed4406bd
-README.zh.md: 9f58b782190721de08750e5bd4eac9e5effd5c6a
+README.zh.md: 0a7afc20eba02d59d293124936cb81aeba6d3f0f
diff --git a/packages/hooks/hooks-claude/README.zh.md b/packages/hooks/hooks-claude/README.zh.md
index 9f58b78219..0a7afc20eb 100644
--- a/packages/hooks/hooks-claude/README.zh.md
+++ b/packages/hooks/hooks-claude/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-一个 cordis 插件，在 harness 的规范拦截 seam 上运行 user 现有 **Claude Code** hook 配置（`hooks.json` 或 settings 文件的 `hooks` key）中受支持的 command hook 子集。它是 hooks 子系统的 **CC 方言** 一半，拥有桥接的 CC 形状每事件 stdin payload、CC env + `${CLAUDE_PLUGIN_ROOT}`／`${CLAUDE_PROJECT_DIR}` 替换，以及从 hook 中性结果到 harness 类型化 Decision 的映射。方言无关原语（matcher、退出码／stdout codec、`ctx.bash` 执行、最严格合并、`hook/*` 事件）来自 [`@deepseek-ai/dsh-hook-protocol`](../hook-protocol/README.md)。
+一个 Cordis 插件，在 harness 的规范拦截 seam 上运行用户现有 **Claude Code** hook 配置（`hooks.json` 或 settings 文件的 `hooks` key）中受支持的 command hook 子集。它是 hooks 子系统的 **CC 方言**部分，负责桥接中 CC 格式的逐事件 stdin payload、CC 的 env 和 `${CLAUDE_PLUGIN_ROOT}`／`${CLAUDE_PROJECT_DIR}` 替换，以及将 hook 的中性结果映射为 harness 的类型化 Decision。方言无关原语（matcher、退出码／stdout codec、`ctx.bash` 执行、最严格合并、`hook/*` 事件）来自 [`@deepseek-ai/dsh-hook-protocol`](../hook-protocol/README.md)。
 
-原生 cordis 插件可以完成此桥接的所有工作，功能更强，且具有类型化返回，没有序列化边界。**该桥接只是已映射 CC command hook 子集的兼容路径**；所有定制行为都应当使用相同 seam 上的原生插件（见 [拦截 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-30-interception-seams.md)）。
+原生 Cordis 插件可以完成此桥接的所有工作，功能更强，且具有类型化返回，没有序列化边界。**该桥接只是已映射 CC command hook 子集的兼容路径**；所有定制行为都应当使用相同 seam 上的原生插件（见 [拦截 seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-30-interception-seams.md)）。
 
 ## 配置
 
@@ -28,31 +28,31 @@ const config: Config = {
     projectDir: .
 ```
 
-配置只在加载时解析**一次**。`configPath` 是**进程级** 配置：相对路径在加载时根据进程启动 cwd 解析，因此一份配置应用于整个进程。尚未进行每会话（`session/new.cwd`）配置发现（`TODO(per-session-hook-config)`）。读取／解析失败会被容纳：桥接记录警告且不注册任何内容，而不是使启动崩溃（路径拼写错误不应使 agent 停止）。只运行 shell 形式 `type: 'command'` hook；`http`／`mcp_tool`／`prompt`／`agent` hook 会被解析并跳过，同时记录警告。没有每 hook `timeout` 的 hook 会使用协议参考默认值 `DEFAULT_HOOK_TIMEOUT_MS`（来自 `dsh-hook-protocol`，10 分钟，即 CC 默认值）。
+配置只在加载时解析**一次**。`configPath` 是**进程级**配置：相对路径在加载时根据进程启动 cwd 解析，因此一份配置应用于整个进程。尚未进行每会话（`session/new.cwd`）配置发现（`TODO(per-session-hook-config)`）。读取／解析失败会被隔离处理：桥接记录警告且不注册任何内容，而不是使启动崩溃（路径拼写错误不应使 agent（智能体）停止）。只运行 shell 形式 `type: 'command'` hook；`http`／`mcp_tool`／`prompt`／`agent` hook 会被解析并跳过，同时记录警告。没有每 hook `timeout` 的 hook 会使用协议参考默认值 `DEFAULT_HOOK_TIMEOUT_MS`（来自 `dsh-hook-protocol`，10 分钟，即 CC 默认值）。
 
-hook **本身** 会在 agent 的会话工作区中运行：对 agent scope 点，桥接会将会话 `cwd`（`session/new.cwd`）作为 hook 进程工作目录，因此 hook 的 `pwd`／相对路径／marker 作用于 user 项目树，而非服务器启动目录。
+hook **本身**会在 agent 的会话工作区中运行：对 agent scope 点，桥接会将会话 `cwd`（`session/new.cwd`）作为 hook 进程工作目录，因此 hook 的 `pwd`／相对路径／marker 作用于用户项目树，而非服务器启动目录。
 
 ## Hook 点 → seam Decision
 
 | CC hook | Harness seam | 映射 |
 |---|---|---|
 | `SessionStart` | `agent/session-start`（emit） | additionalContext → `agent.inject()` 到新会话（无法阻塞） |
-| `UserPromptSubmit` | `agent/prompt-submit`（waterfall） | `deny` → `PromptDecision.block`；仅 additionalContext → 通过 `next()` 委托，再将一个单独标记源的上下文前置到下游 `additionalContexts`（后续 listener 仍可阻塞／改写） |
+| `UserPromptSubmit` | `agent/prompt-submit`（waterfall，瀑布式事件） | `deny` → `PromptDecision.block`；仅 additionalContext → 通过 `next()` 委托，再将一个单独标记源的上下文前置到下游 `additionalContexts`（后续 listener 仍可阻塞／改写） |
 | `PreToolUse` | `tools/pre-execute`（waterfall） | `deny` → `PreToolDecision.deny`；`ask` → `PreToolDecision.ask` |
 | `PostToolUse` | `tools/post-execute`（waterfall） | `deny` → 带反馈的 `block`；仅 additionalContext → 通过 `next()` 委托，再将一个单独标记源的上下文前置到下游决策；Code Mode 将子调用上下文延迟到外层 `run_code` 结果 |
 | `Stop` | `agent/turn-stopping`（serial） | 阻塞 Stop hook 通过 `steer()` 送入其原因，强制再执行一步 |
-| `SubagentStart` | `subagent/start`（emit） | additionalContext → `agent.inject()` 到实时同进程 child；远程 child 没有本地注入目标 |
+| `SubagentStart` | `subagent/start`（emit） | additionalContext → `agent.inject()` 到仍在运行的同进程 child；远程 child 没有本地注入目标 |
 | `SubagentStop` | `subagent/end`（emit） | 只观测 |
 
-三个 emit 点都脱离运行：没有 seam 会等待 `SessionStart`／`SubagentStart`／`SubagentStop` hook。每条运行链都会被跟踪；dispose 桥接会中止仍在运行的 hook 进程，再排空 continuation，然后 dispose resolve（`createDetachedRuns`，位于 `dsh-hook-protocol`）。
+三个 emit 点都以分离方式运行：没有 seam 会等待 `SessionStart`／`SubagentStart`／`SubagentStop` hook。每条运行链都会被跟踪；对桥接执行 dispose（资源释放）时，会中止仍在运行的 hook 进程，并在 dispose 完成前排空 continuation（`createDetachedRuns`，位于 `dsh-hook-protocol`）。
 
 matcher subject 是工具名称（`PreToolUse`／`PostToolUse`）、会话源（`SessionStart`），或常量 `agent_type`，其值为 `general-purpose`（`SubagentStart`／`SubagentStop`）。harness subagent seam 不携带每 kind label，因此桥接报告 Claude Code 自身 Task 工具默认值；默认／`*`／空 `agent_type` matcher 会触发，特定 kind matcher 不会触发。`UserPromptSubmit`／`Stop` 忽略 matcher。一个点上文件配置的多个 hook 会**按配置顺序串行运行**，并按最严格方式折叠（`deny > ask > allow`，见 `dsh-hook-protocol`）。串行使每个 hook 的 `hook/invoked`／`hook/result` 对在日志中相邻，决策折叠与顺序无关（见 Agent Note 的「run serially, not concurrently」说明）。
 
-每个 agent scope stdin payload 都携带 `session_id` 与字符串形状的 `transcript_path`。可用时，桥接通过 `ctx.sessionPersistence.locate(session.header)` 解析后者，否则发送 `''`。查找不会创建或 flush 产物，因此第一个轮次结束检查点之前路径可能不存在，也可能省略当前开启轮次。
+每个 agent scope stdin payload 都携带 `session_id` 与字符串形式的 `transcript_path`。可用时，桥接通过 `ctx.sessionPersistence.locate(session.header)` 解析后者，否则发送 `''`。查找不会创建或 flush 产物，因此第一个轮次结束检查点之前路径可能不存在，也可能省略当前开启轮次。
 
 ## 上下文源
 
-注入上下文携带显式 `{ kind: 'plugin', plugin: 'hooks-claude' }` 源。`agent.inject()` 会将缺失源默认为 `{ kind: 'user' }`，这会将插件上下文错误标记为 user 提示词，因此桥接始终标注自身。
+注入上下文携带显式 `{ kind: 'plugin', plugin: 'hooks-claude' }` 源。`agent.inject()` 会将缺失源默认为 `{ kind: 'user' }`，这会将插件上下文错误标记为用户提示词，因此桥接始终标注自身。
 
 ## 模型体验
 
@@ -60,15 +60,15 @@ matcher subject 是工具名称（`PreToolUse`／`PostToolUse`）、会话源（
 
 #### 模型看到的内容
 
-`SessionStart`、已接受提示词、工具后和实时同进程 subagent-start hook 可以添加带源归因的上下文消息；阻塞 `Stop` hook 将原因添加为下一步 steering。远程 child 注入没有本地目标。
+`SessionStart`、已接受提示词、工具后和实时同进程 subagent-start hook 可以添加带源归因的上下文消息；阻塞 `Stop` hook 将原因添加为下一步 steering（中途引导）。远程 child 注入没有本地目标。
 
 #### Token 影响
 
-hook 不返回上下文时没有成本。Hook 文本取决于数据，会被记录，并在后续会话请求中重发，直到压缩。
+hook 不返回上下文时没有成本。Hook 文本取决于数据，会被记录，并在后续会话请求中重发，直到压缩（compaction）。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 已阻塞提示词或工具结果
 
@@ -89,9 +89,9 @@ hook 不返回上下文时没有成本。Hook 文本取决于数据，会被记
 - **不支持的 hook 事件（Claude Code 当前 30 项中的 23 项）：** `Setup`、`InstructionsLoaded`、`UserPromptExpansion`、`MessageDisplay`、`PermissionRequest`、`PostToolUseFailure`、`PostToolBatch`、`PermissionDenied`、`Notification`、`TaskCreated`、`TaskCompleted`、`StopFailure`、`TeammateIdle`、`ConfigChange`、`CwdChanged`、`FileChanged`、`WorktreeCreate`、`WorktreeRemove`、`PreCompact`、`PostCompact`、`SessionEnd`、`Elicitation` 和 `ElicitationResult`。这些事件的配置会被解析，但绝不分派。比较基线是 Claude Code [官方 hook 事件参考](https://code.claude.com/docs/en/hooks#hook-events)。
 - **`SessionStart` 只支持部分功能：** 会消费 JSON `additionalContext`，但不支持纯 stdout 上下文、`initialUserMessage`、`sessionTitle`、`watchPaths`、`reloadSkills` 与 `CLAUDE_ENV_FILE`。hook 脱离运行，因此上下文可能错过第一个请求（`TODO(session-start-gating)`），payload 会省略 `model`、`agent_type` 和 `session_title` 等当前可选字段。
 - **`UserPromptSubmit` 只支持部分功能：** 支持阻塞与 JSON `additionalContext`，但不支持纯 stdout 上下文、`sessionTitle` 和 `suppressOriginalPrompt`。除非被覆盖，否则桥接还会使用自身 600 秒默认值，而非 Claude Code 的事件特定 30 秒 command 超时。
-- **`PreToolUse` 只支持部分功能：** `deny` 与 `ask` 决策可用；`allow` 不会预批准，不支持 `defer`，`additionalContext` 会被忽略，`updatedInput` 会被记录 + 警告但不应用（见 [pre-tool-input-rewrite Agent Note](../../../.agents/notes/proposed/feature/2026-06-30-pre-tool-input-rewrite.md)）。
+- **`PreToolUse` 只支持部分功能：** `deny` 与 `ask` 决策可用；`allow` 不会预审批，不支持 `defer`，`additionalContext` 会被忽略，`updatedInput` 会被记录 + 警告但不应用（见 [pre-tool-input-rewrite Agent Note](../../../.agents/notes/proposed/feature/2026-06-30-pre-tool-input-rewrite.md)）。
 - **`PostToolUse` 只支持部分功能：** 支持阻塞反馈与 JSON `additionalContext`，但不支持 `updatedToolOutput` 和 `updatedMCPToolOutput`，`tool_response` 会展平为文本。
-- **`SubagentStart` 与 `SubagentStop` 只支持部分功能：** 两者均报告常量 `agent_type`，其值为 `general-purpose`，并在 Claude Code 报告父会话的位置使用 child 会话 id。Start 上下文是尽力而为，且只能到达实时同进程 child；stop 只观测，无法阻塞 subagent 或向其提供上下文。Start 省略 `transcript_path`；stop 还省略 `agent_transcript_path`、`last_assistant_message`、`background_tasks` 和 `session_crons`，并始终报告 `stop_hook_active: false`。
+- **`SubagentStart` 与 `SubagentStop` 只支持部分功能：** 两者均报告常量 `agent_type`，其值为 `general-purpose`，并在 Claude Code 报告父会话的位置使用 child 会话 id。Start 上下文是尽力而为，且只能到达仍在运行的同进程 child；stop 只观测，无法阻塞 subagent 或向其提供上下文。Start 省略 `transcript_path`；stop 还省略 `agent_transcript_path`、`last_assistant_message`、`background_tasks` 和 `session_crons`，并始终报告 `stop_hook_active: false`。
 - **`Stop` 只支持部分功能：** 阻塞会强制另一个模型轮次，但 `stop_hook_active` 始终为 `false`，会省略 `last_assistant_message`、`background_tasks` 和 `session_crons`，且未实现连续阻塞上限（`TODO(stop-loop-guard)`）。因此，无条件阻塞 hook 会在每个步骤中强制 continuation，除非它自我限制。
 - **通用 payload 与输出字段只支持部分功能：** 已映射事件会省略 Claude Code 原本会提供的 `prompt_id`、`transcript_path`、`permission_mode` 和 `effort`。`systemMessage` 会被记录 + 警告但不呈现；`{"continue": false}` 会被记录但不会停止运行；不会应用 `suppressOutput`、`stopReason` 和 `terminalSequence`（`TODO(hook-continue-false)`）。
-- **Handler 与配置只支持部分功能：** 只运行 shell 形式 command handler。会跳过 `http`、`mcp_tool`、`prompt` 和 `agent` handler；不遵循 `args`、`async`、`asyncRewake`、`shell`、`if`、`once` 和 `statusMessage` 等 command handler 选项。匹配 handler 串行运行且不去重，而 Claude Code 会并行运行并对相同 handler 去重。一个进程级 `configPath` 会在加载时解析一次；尚未实现 Claude Code 的分层项目、user、plugin 与 policy 发现和实时重新加载（`TODO(per-session-hook-config)`）。
+- **Handler 与配置只支持部分功能：** 只运行 shell 形式 command handler。会跳过 `http`、`mcp_tool`、`prompt` 和 `agent` handler；不遵循 `args`、`async`、`asyncRewake`、`shell`、`if`、`once` 和 `statusMessage` 等 command handler 选项。匹配 handler 串行运行且不去重，而 Claude Code 会并行运行并对相同 handler 去重。一个进程级 `configPath` 会在加载时解析一次；尚未实现 Claude Code 的分层项目、用户、插件与策略发现和实时重新加载（`TODO(per-session-hook-config)`）。
diff --git a/packages/hooks/hooks-codex/README.i18n.yaml b/packages/hooks/hooks-codex/README.i18n.yaml
index bb102c814b..74155768be 100644
--- a/packages/hooks/hooks-codex/README.i18n.yaml
+++ b/packages/hooks/hooks-codex/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/hooks/hooks-codex/README.md
 README.md: fd57762c6fb91e0ea47ec57c30bf9850bc488a33
-README.zh.md: 367d6acd0fec486cb0f9fb50023ad2ed4cca7217
+README.zh.md: 58b387e22f0e56770e4ea779f184c6c82a38ff98
diff --git a/packages/hooks/hooks-codex/README.zh.md b/packages/hooks/hooks-codex/README.zh.md
index 367d6acd0f..58b387e22f 100644
--- a/packages/hooks/hooks-codex/README.zh.md
+++ b/packages/hooks/hooks-codex/README.zh.md
@@ -2,17 +2,17 @@
 
 [English](README.md) | 中文
 
-一个 cordis 插件，在 harness 的规范拦截 seam 上运行 user 现有 **Codex** hook 配置的受支持子集。它是 hooks 子系统的 **Codex 方言** 一半。方言无关原语来自 [`@deepseek-ai/dsh-hook-protocol`](../hook-protocol/README.md)；该桥接拥有 Codex 形状 payload、matcher 模式和决策映射。
+一个 Cordis 插件，在 harness 的规范拦截 seam 上运行用户现有 **Codex** hook 配置的受支持子集。它是 hooks 子系统中采用 **Codex 方言** 的一侧。方言无关原语来自 [`@deepseek-ai/dsh-hook-protocol`](../hook-protocol/README.md)；该桥接负责处理 Codex 形状的 payload、matcher 模式和决策映射。
 
 该桥接实现 Codex 当前 hook 协议的一个明确子集：
 
 - **10 个 hook 点中的 5 个：** `PreToolUse`、`PostToolUse`、`SessionStart`、`UserPromptSubmit` 和 `Stop`。
 - **只使用正则 matcher**（没有字面快速路径；matcher 始终是未锚定正则）。
-- **snake_case stdin payload**，携带 `turn_id`／`model` 额外字段，写入时**不带** 尾随换行符。
+- **snake_case stdin payload**，携带 `turn_id`／`model` 额外字段，写入时**不带**尾随换行符。
 - **没有 Codex 插件 env 注入，也没有配置时 placeholder 替换**（命令仍会接收执行器环境，并通过其 shell 运行）。
-- **没有工具前批准或改写路径**：hook 可以阻塞，但桥接不会预批准或替换工具输入。
+- **没有工具前审批或改写路径**：hook 可以阻塞，但桥接不会预审批或替换工具输入。
 
-原生 cordis 插件可以完成此桥接的所有工作，并且功能更强；该桥接只是已映射 Codex 子集的兼容路径（见 [拦截 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-30-interception-seams.md)）。
+原生 Cordis 插件可以完成此桥接的所有工作，并且功能更强；该桥接只是已映射 Codex 子集的兼容路径（见 [拦截 seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-30-interception-seams.md)）。
 
 ## 配置
 
@@ -34,25 +34,25 @@ const config: Config = {
     model: deepseek-v4
 ```
 
-配置只在加载时解析**一次**。`configPath` 是**进程级** 配置：相对路径在加载时根据进程启动 cwd 解析，而非每会话解析（`TODO(per-session-hook-config)`）。读取／解析失败会被容纳（记录 + 不注册任何内容）。只运行同步 `type: 'command'` hook；非 command 或 `async: true` hook 会被解析并跳过，同时记录警告。hook 接受 `timeout` 或 `timeoutSec` alias；两者都未设置时，使用协议参考默认值 `DEFAULT_HOOK_TIMEOUT_MS`（来自 `dsh-hook-protocol`，10 分钟）。五个桥接支持点之外的事件会在解析时丢弃。
+配置只在加载时解析**一次**。`configPath` 是**进程级**配置：相对路径在加载时根据进程启动 cwd 解析，而非每会话解析（`TODO(per-session-hook-config)`）。读取／解析失败会被隔离处理（记录 + 不注册任何内容）。只运行同步 `type: 'command'` hook；非 command 或 `async: true` hook 会被解析并跳过，同时记录警告。hook 接受 `timeout` 或 `timeoutSec` alias；两者都未设置时，使用协议参考默认值 `DEFAULT_HOOK_TIMEOUT_MS`（来自 `dsh-hook-protocol`，10 分钟）。五个桥接支持点之外的事件会在解析时丢弃。
 
-hook 本身会在 agent 的会话工作区中运行：对 agent scope 点，桥接会将会话 `cwd` 作为 hook 进程工作目录，因此 hook 作用于 user 项目树，而非服务器启动目录。
+hook 本身会在 agent（智能体）的会话工作区中运行：对 agent scope 点，桥接会将会话 `cwd` 作为 hook 进程工作目录，因此 hook 作用于用户项目树，而非服务器启动目录。
 
 ## Hook 点 → seam Decision
 
 | Codex hook | Harness seam | 映射 |
 |---|---|---|
 | `SessionStart` | `agent/session-start`（emit） | 纯 stdout hook 的输出 → additionalContext → `agent.inject()` |
-| `UserPromptSubmit` | `agent/prompt-submit`（waterfall） | `block`（退出码 2）→ `PromptDecision.block`；仅 additionalContext → 通过 `next()` 委托，再将一个单独标记源的上下文前置到下游 `additionalContexts` |
+| `UserPromptSubmit` | `agent/prompt-submit`（waterfall，瀑布式事件） | `block`（退出码 2）→ `PromptDecision.block`；仅 additionalContext → 通过 `next()` 委托，再将一个单独标记源的上下文前置到下游 `additionalContexts` |
 | `PreToolUse` | `tools/pre-execute`（waterfall） | `block` → `PreToolDecision.deny`（没有 `allow`／`ask`） |
 | `PostToolUse` | `tools/post-execute`（waterfall） | `block` → 带反馈的 `block`；仅 additionalContext → 通过 `next()` 委托，再将一个单独标记源的上下文前置到下游决策；Code Mode 将子调用上下文延迟到外层 `run_code` 结果 |
 | `Stop` | `agent/turn-stopping`（serial） | 阻塞 Stop hook 通过 `steer()` 送入其原因，强制再执行一步 |
 
 工具调用的 payload 携带真实 `tool_name`（matcher 测试的相同值）与 Codex `tool_input: { command }` 形状（存在 `command` arg 时使用该值，否则使用 `''`）。matcher subject 是工具名称（`PreToolUse`／`PostToolUse`）或会话源（`SessionStart`）；`UserPromptSubmit`／`Stop` 忽略 matcher。
 
-每个 agent scope stdin payload 都携带 `session_id` 和 `transcript_path`。可用时，桥接通过 `ctx.sessionPersistence.locate(session.header)` 解析后者，否则发送 `null`，保留 Codex `string | null` 形状。查找不会创建或 flush 产物，因此第一个轮次结束检查点之前路径可能不存在，也可能省略当前开启轮次。
+每个 agent scope stdin payload 都携带 `session_id` 和 `transcript_path`。可用时，桥接通过 `ctx.sessionPersistence.locate(session.header)` 解析后者，否则发送 `null`，保留 Codex `string | null` 形状。查找不会创建或 flush 产物，因此在第一个轮次结束检查点之前，路径可能尚不存在，或其指向的 transcript 可能尚未包含当前未结束的轮次。
 
-`SessionStart` 是唯一的 emit 点，它会脱离运行。每条运行链都会被跟踪；dispose 桥接会中止仍在运行的 hook 进程，再排空 continuation，然后 dispose resolve（`createDetachedRuns`，位于 `dsh-hook-protocol`）。
+`SessionStart` 是唯一的 emit 点，它会脱离运行。每条运行链都会被跟踪；对桥接执行 dispose（资源释放）会中止仍在运行的 hook 进程，再排空 continuation，之后 dispose 才会完成（`createDetachedRuns`，位于 `dsh-hook-protocol`）。
 
 ## 上下文源
 
@@ -64,15 +64,15 @@ hook 本身会在 agent 的会话工作区中运行：对 agent scope 点，桥
 
 #### 模型看到的内容
 
-`SessionStart`、已接受提示词和工具后 hook 可以添加带源归因的上下文消息；阻塞 `Stop` hook 将其原因添加为下一步 steering。
+`SessionStart`、已接受提示词和工具后 hook 可以添加带源归因的上下文消息；阻塞 `Stop` hook 将其原因添加为下一步 steering（中途引导）。
 
 #### Token 影响
 
-hook 不返回上下文时没有成本。Hook 文本取决于数据，会被记录，并重发直到压缩。
+hook 不返回上下文时没有成本。Hook 文本取决于数据，会被记录，并重发直到压缩（compaction）。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 已阻塞提示词或工具结果
 
@@ -96,5 +96,5 @@ hook 不返回上下文时没有成本。Hook 文本取决于数据，会被记
 - **`PreToolUse` 只支持部分功能：** 支持阻塞，但会忽略 `additionalContext`、`permissionDecision: "allow"` 和 `updatedInput`。每个工具都表示为 `tool_input: { command }`，因此非 shell 工具参数不会如实公开给 hook。
 - **`PostToolUse` 只支持部分功能：** 支持阻塞反馈与 JSON `additionalContext`，但不会强制执行 `{"continue": false}`，非 shell 工具参数会缩减为 `{ command }`，结构化工具输出会在 `tool_response` 中展平为文本。
 - **`Stop` 只支持部分功能：** 阻塞会强制另一个模型轮次，但 `stop_hook_active` 始终为 `false`，`last_assistant_message` 始终为 `null`，且不会强制执行 `{"continue": false}`。因此，无条件阻塞 hook 会在每个步骤中强制 continuation，除非它自我限制（`TODO(stop-loop-guard)`）。
-- **通用 payload 与输出字段只支持部分功能：** 每个已映射事件都报告 `transcript_path: null`、静态配置的 `model` 与 `permission_mode: "default"`，而非当前 Codex 运行时值。`systemMessage` 会被记录 + 警告但不呈现，`{"continue": false}` 会被记录但不会应用 Codex 事件特定停止行为（`TODO(hook-continue-false)`）。
-- **配置加载与执行只支持部分功能：** 一个进程级 `configPath` 会在加载时解析；尚未实现 Codex 的活动 user、project、session、system/managed 和 plugin 分层、信任控制与内联 `config.toml` hook 形式（`TODO(per-session-hook-config)`）。只运行同步 `command` handler，忽略 `statusMessage` 与 `commandWindows` 等当前元数据，匹配 handler 串行运行，而非使用 Codex 的并发启动语义。
+- **通用 payload 与输出字段只支持部分功能：** 每个已映射事件都报告 `transcript_path: null`、静态配置的 `model` 与 `permission_mode: "default"`，而非当前 Codex 运行时值。`systemMessage` 会被记录并触发警告，但不呈现，`{"continue": false}` 会被记录但不会应用 Codex 事件特定停止行为（`TODO(hook-continue-false)`）。
+- **配置加载与执行只支持部分功能：** 一个进程级 `configPath` 会在加载时解析；尚未实现 Codex 的活动用户层、项目层、会话层、系统／托管层和插件层、信任控制与内联 `config.toml` hook 形式（`TODO(per-session-hook-config)`）。只运行同步 `command` handler，忽略 `statusMessage` 与 `commandWindows` 等当前元数据，匹配 handler 串行运行，而非使用 Codex 的并发启动语义。
diff --git a/packages/host/apiproxy/README.i18n.yaml b/packages/host/apiproxy/README.i18n.yaml
index 73b0845370..6f22ce2466 100644
--- a/packages/host/apiproxy/README.i18n.yaml
+++ b/packages/host/apiproxy/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/host/apiproxy/README.md
-README.md: ca4471454f5be5d3fcba38ce665d4fb3fbd85e74
-README.zh.md: 953539e1198a52b2bf7cdd9ca1b0d263cc2ae6f9
+README.md: b5f80dcb3a077a411db3b721737a9c16b56fcecf
+README.zh.md: 1c3c7486f7d500f8c2d36028d47f29b112d9b5ae
diff --git a/packages/host/apiproxy/README.md b/packages/host/apiproxy/README.md
index ca4471454f..b5f80dcb3a 100644
--- a/packages/host/apiproxy/README.md
+++ b/packages/host/apiproxy/README.md
@@ -12,7 +12,7 @@ The layering/protocol decisions are recorded in the [GUI layering and RPC protoc
 
 `session.history`'s tail page (`beforeSeq` absent) additionally carries an optional `projections` block — the watermark snapshot of every unit registered on `ctx.sessionProjections` (`@deepseek-ai/dsh-session-projection`), with `asOfSeq` = the last event seq the values reflect (`-1` on an empty log). The gateway also subscribes to the registry's change feed and mints a `session/projection` mux frame per changed unit (`{sessionId, key, value, seq}` — live push state, never logged; clients hold one generic per-session value store under higher-seq-wins). The carrier holds zero domain knowledge (each value passed its unit's own schema inside the registry; the wire schemas keep `values`/`value` wide); loadOlder pages never carry the block, and a composition without the registry serves histories without either surface.
 
-Session titles ride the generic projection pair like every other domain — the history-tail `projections` block plus `session/projection` frames under the `title` key (the bespoke `session/title` frame is retired). Titles do not join `session.list`; cold sessions remain metadata-only there until opening or resuming attaches their logs.
+Session titles ride the generic projection pair like every other domain — the history-tail `projections` block plus `session/projection` frames under the `title` key (the bespoke `session/title` frame is retired). Titles do not join `session.list`; cold sessions remain metadata-only there until opening or resuming attaches their logs. `session.rename` accepts an explicit user title (resuming a cold session first), delegating to `ctx.sessionTitle.rename` — the accepted `session/title` event pins the title against automatic regeneration — and returns the normalized title plus its event seq so a client settles its `title` projection cell ahead of the push frame; a title that normalizes to empty returns `title-invalid`.
 
 Session model routing is a session-domain contract. `session.models` returns the selected provider/model/reasoning target with provider-grouped advisory models, exact-route reasoning metadata, and provider-local lookup failures. `session.selectModel` validates the optional adapter-owned reasoning effort and replaces the complete target selected for the next prompt-assembly boundary. Catalog membership is not validation: an adapter may resolve an unlisted model, while an unavailable route or unsupported effort returns `model-unavailable`.
 
diff --git a/packages/host/apiproxy/README.zh.md b/packages/host/apiproxy/README.zh.md
index 953539e119..1c3c7486f7 100644
--- a/packages/host/apiproxy/README.zh.md
+++ b/packages/host/apiproxy/README.zh.md
@@ -12,7 +12,7 @@
 
 `session.history` 的尾页（不带 `beforeSeq`）额外携带一个可选的 `projections` 块——`ctx.sessionProjections`（`@deepseek-ai/dsh-session-projection`）上每个已注册单元的水位线快照，`asOfSeq` = 这些值共同反映到的最后一个事件 seq（空日志为 `-1`）。网关还订阅注册表的变更流，为每个状态发生变化的单元铸造一个 `session/projection` mux 帧（`{sessionId, key, value, seq}`——实时推送状态，绝不入日志；客户端按 seq 高者胜维护一个按会话的通用值仓）。载体不持有任何领域知识（每个值在注册表内部已过其单元自己的 schema；协议 schema 对 `values`/`value` 保持宽松）；loadOlder 页永不携带该块，未装注册表的组合则两个面都不提供。
 
-会话标题与其他所有领域一样搭乘这对通用投影机制——历史尾页的 `projections` 块外加 `title` 键下的 `session/projection` 帧（专设的 `session/title` 帧已下线）。标题不会加入 `session.list`；冷会话在其中仍只有元数据，直到打开或恢复操作附加其日志。
+会话标题与其他所有领域一样搭乘这对通用投影机制——历史尾页的 `projections` 块外加 `title` 键下的 `session/projection` 帧（专设的 `session/title` 帧已下线）。标题不会加入 `session.list`；冷会话在其中仍只有元数据，直到打开或恢复操作附加其日志。`session.rename` 接受用户显式标题（冷会话先恢复），委托给 `ctx.sessionTitle.rename`——被接受的 `session/title` 事件将标题钉住、不再被自动生成覆盖——并返回规范化后的标题及其事件 seq，让 client 在推送帧到达前就结算自己的 `title` 投影格；规范化后为空的标题返回 `title-invalid`。
 
 会话模型路由属于会话领域契约。`session.models` 返回选中的提供方／模型／推理（reasoning）目标，以及按提供方分组的建议性模型、精确路由推理元数据和逐提供方查询失败记录。`session.selectModel` 校验由适配器持有的可选推理强度，并替换将在下一提示词组装边界使用的完整目标。目录成员关系不构成校验：适配器可以解析未列出的模型，而不可用路由或不受支持的推理强度会返回 `model-unavailable`。
 
diff --git a/packages/host/apiproxy/package.json b/packages/host/apiproxy/package.json
index 8cbdb91cf4..8b4a895774 100644
--- a/packages/host/apiproxy/package.json
+++ b/packages/host/apiproxy/package.json
@@ -51,6 +51,7 @@
     "@deepseek-ai/dsh-session-persistence": "workspace:^",
     "@deepseek-ai/dsh-session-projection": "workspace:^",
     "@deepseek-ai/dsh-session-projection-cache": "workspace:^",
+    "@deepseek-ai/dsh-session-title": "workspace:^",
     "@deepseek-ai/dsh-skill": "workspace:^",
     "@deepseek-ai/dsh-tools": "workspace:^",
     "@deepseek-ai/dsh-user-approval": "workspace:^",
diff --git a/packages/host/apiproxy/src/api-proxy.ts b/packages/host/apiproxy/src/api-proxy.ts
index f178bfefd0..4f5336323c 100644
--- a/packages/host/apiproxy/src/api-proxy.ts
+++ b/packages/host/apiproxy/src/api-proxy.ts
@@ -38,6 +38,8 @@ import type { GoalRef as CoreGoalRef } from '@deepseek-ai/dsh-goal'
 // Type-only edges: resolve `ctx.get('commands')`, the `commands/change` event, and `ctx.get('skills')`.
 import type {} from '@deepseek-ai/dsh-commands'
 import type {} from '@deepseek-ai/dsh-skill'
+// Value edge: the rename impl narrows the title service's validation failure; the import also resolves `ctx.get('sessionTitle')`.
+import { SessionTitleInvalidError } from '@deepseek-ai/dsh-session-title'
 import type { CallId } from '@deepseek-ai/dsh-llm/brand'
 import type { ApprovalOutcome, ApprovalRequestId } from '@deepseek-ai/dsh-user-approval'
 // Side-effect type import: resolves the `approval/request` waterfall and
@@ -1049,6 +1051,36 @@ export function createApiProxy(ctx: Context, defaults: ApiProxyDefaults): ApiPro
         }
       },
 
+      async rename(request) {
+        const { sessionId, title } = request.payload
+        const found = await agentFor(sessionId)
+        if ('error' in found) return err(request, found.error)
+        const titles = ctx.get('sessionTitle')
+        if (titles === undefined) {
+          return err(request, { code: 'internal', message: 'renaming is unavailable: this deployment mounts no session-title service', details: {} })
+        }
+        try {
+          const accepted = titles.rename(found.agent.session, title)
+          return ok(request, { title: accepted.title, seq: accepted.eventSeq })
+        } catch (error: unknown) {
+          // Only the input's fault maps to title-invalid (the message is
+          // product-user-visible in the rename dialog); liveness and disposal
+          // races are deployment trouble, not a bad title.
+          if (error instanceof SessionTitleInvalidError) {
+            return err(request, {
+              code: 'title-invalid',
+              message: error.message,
+              details: { sessionId },
+            })
+          }
+          return err(request, {
+            code: 'internal',
+            message: `failed to rename session "${sessionId}": ${String(error)}`,
+            details: {},
+          })
+        }
+      },
+
       async prompt(request) {
         const { sessionId, mode, content } = request.payload
         const found = await agentFor(sessionId)
diff --git a/packages/host/apiproxy/src/api/rpc-map.ts b/packages/host/apiproxy/src/api/rpc-map.ts
index 5394691248..8b26b1fc2a 100644
--- a/packages/host/apiproxy/src/api/rpc-map.ts
+++ b/packages/host/apiproxy/src/api/rpc-map.ts
@@ -23,6 +23,7 @@ export interface RpcMethodMap {
   'session.history': SessionsApi['history']
   'session.models': SessionsApi['models']
   'session.selectModel': SessionsApi['selectModel']
+  'session.rename': SessionsApi['rename']
   'session.prompt': SessionsApi['prompt']
   'session.cancel': SessionsApi['cancel']
   'host.describe': HostApi['describe']
diff --git a/packages/host/apiproxy/src/api/rpc.schema.ts b/packages/host/apiproxy/src/api/rpc.schema.ts
index 58d3a17126..db5435a757 100644
--- a/packages/host/apiproxy/src/api/rpc.schema.ts
+++ b/packages/host/apiproxy/src/api/rpc.schema.ts
@@ -49,6 +49,7 @@ export const rpcErrorSchema: z.ZodType<RpcError> = z.discriminatedUnion('code',
   z.object({ code: z.literal('agent-busy'), message: z.string(), details: z.object({ reason: z.string() }) }),
   z.object({ code: z.literal('command-error'), message: z.string(), details: z.object({}) }),
   z.object({ code: z.literal('unknown-command'), message: z.string(), details: z.object({}) }),
+  z.object({ code: z.literal('title-invalid'), message: z.string(), details: z.object({ sessionId: z.string() }) }),
   z.object({ code: z.literal('internal'), message: z.string(), details: z.object({}) }),
 ]) as unknown as z.ZodType<RpcError>
 
diff --git a/packages/host/apiproxy/src/api/rpc.ts b/packages/host/apiproxy/src/api/rpc.ts
index fd99d015ee..8daf898693 100644
--- a/packages/host/apiproxy/src/api/rpc.ts
+++ b/packages/host/apiproxy/src/api/rpc.ts
@@ -48,6 +48,7 @@ export interface RpcErrorDetailsMap {
   'command-error': {}
   /** A leading-/ prompt named no registered command; the message names the token. */
   'unknown-command': {}
+  'title-invalid': { sessionId: SessionId }
   'internal': {}
 }
 
diff --git a/packages/host/apiproxy/src/api/sessions.schema.ts b/packages/host/apiproxy/src/api/sessions.schema.ts
index d1bd4b9a2d..29dd88c89b 100644
--- a/packages/host/apiproxy/src/api/sessions.schema.ts
+++ b/packages/host/apiproxy/src/api/sessions.schema.ts
@@ -73,6 +73,18 @@ export const sessionCreateValueSchema = z.object({
   sessionId: sessionIdSchema,
 }) satisfies z.ZodType<Wire<ResponseValue<'session.create'>>>
 
+/** session.rename request payload (raw title; host-side normalization decides acceptance). */
+export const sessionRenameRequestSchema = z.object({
+  sessionId: sessionIdSchema,
+  title: z.string(),
+}) satisfies z.ZodType<Wire<RequestPayload<'session.rename'>>>
+
+/** session.rename response value (the normalized accepted title and its event seq). */
+export const sessionRenameValueSchema = z.object({
+  title: z.string().min(1),
+  seq: z.number().int().nonnegative(),
+}) satisfies z.ZodType<Wire<ResponseValue<'session.rename'>>>
+
 /** session.history request payload (beforeSeq/maxMessages page backwards from the window tail). */
 export const sessionHistoryRequestSchema = z.object({
   sessionId: sessionIdSchema,
@@ -211,4 +223,3 @@ export const sessionCancelRequestSchema = z.object({
 export const sessionCancelValueSchema = z.object({
   accepted: z.literal(true),
 }) satisfies z.ZodType<Wire<ResponseValue<'session.cancel'>>>
-
diff --git a/packages/host/apiproxy/src/api/sessions.ts b/packages/host/apiproxy/src/api/sessions.ts
index f952c7c542..3a824c8261 100644
--- a/packages/host/apiproxy/src/api/sessions.ts
+++ b/packages/host/apiproxy/src/api/sessions.ts
@@ -208,6 +208,16 @@ export interface SessionsApi {
   }>):
   Promise<RpcResponse<{ selected: ModelTarget }>>
 
+  /**
+   * Renames a session: appends a `session/title` event with the `user`
+   * source, which pins the title against automatic regeneration. The
+   * normalized accepted title and the title event's seq return so the caller
+   * can settle its projection cell without waiting for the push frame. A
+   * title that normalizes to empty fails with `title-invalid`.
+   */
+  rename(request: RpcRequest<{ sessionId: SessionId; title: string }>):
+  Promise<RpcResponse<{ title: string; seq: number }>>
+
   /**
    * Sends a message. content is core's ContentBlock[] verbatim; mode maps 1:1 — queue→send, steer→steer.
    * A prompt whose content is exactly one text block starting with '/' is a slash command: the host
diff --git a/packages/host/apiproxy/src/fetch/client.ts b/packages/host/apiproxy/src/fetch/client.ts
index 8bcc6f7c3e..5b81d1c4ea 100644
--- a/packages/host/apiproxy/src/fetch/client.ts
+++ b/packages/host/apiproxy/src/fetch/client.ts
@@ -24,6 +24,7 @@ import {
   sessionListValueSchema,
   sessionModelsValueSchema,
   sessionPromptValueSchema,
+  sessionRenameValueSchema,
   sessionSelectModelValueSchema,
 } from '../api/sessions.schema.ts'
 import {
@@ -66,6 +67,7 @@ export interface IApiClient {
     history(payload: RequestPayload<'session.history'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'session.history'>>>
     models(payload: RequestPayload<'session.models'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'session.models'>>>
     selectModel(payload: RequestPayload<'session.selectModel'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'session.selectModel'>>>
+    rename(payload: RequestPayload<'session.rename'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'session.rename'>>>
     prompt(payload: RequestPayload<'session.prompt'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'session.prompt'>>>
     cancel(payload: RequestPayload<'session.cancel'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'session.cancel'>>>
   }
@@ -116,6 +118,7 @@ const UNARY_VALUE_SCHEMAS: { [K in keyof RpcMethodMap]: z.ZodType<Wire<ResponseV
   'session.history': sessionHistoryValueSchema,
   'session.models': sessionModelsValueSchema,
   'session.selectModel': sessionSelectModelValueSchema,
+  'session.rename': sessionRenameValueSchema,
   'session.prompt': sessionPromptValueSchema,
   'session.cancel': sessionCancelValueSchema,
   'host.describe': hostDescribeValueSchema,
@@ -327,6 +330,7 @@ export abstract class AbstractApiClient implements IApiClient {
     history: (payload, signal) => this.callUnary('session.history', payload, signal),
     models: (payload, signal) => this.callUnary('session.models', payload, signal),
     selectModel: (payload, signal) => this.callUnary('session.selectModel', payload, signal),
+    rename: (payload, signal) => this.callUnary('session.rename', payload, signal),
     prompt: (payload, signal) => this.callUnary('session.prompt', payload, signal),
     cancel: (payload, signal) => this.callUnary('session.cancel', payload, signal),
   }
diff --git a/packages/host/apiproxy/src/fetch/handler.ts b/packages/host/apiproxy/src/fetch/handler.ts
index e1340aad5b..46e3703f3a 100644
--- a/packages/host/apiproxy/src/fetch/handler.ts
+++ b/packages/host/apiproxy/src/fetch/handler.ts
@@ -21,6 +21,7 @@ import {
   sessionListRequestSchema,
   sessionModelsRequestSchema,
   sessionPromptRequestSchema,
+  sessionRenameRequestSchema,
   sessionSelectModelRequestSchema,
 } from '../api/sessions.schema.ts'
 import {
@@ -68,6 +69,7 @@ const UNARY_ROUTES: UnaryRoutes = {
   'session.history': { schema: sessionHistoryRequestSchema, invoke: (api, r) => api.sessions.history(r) },
   'session.models': { schema: sessionModelsRequestSchema, invoke: (api, r) => api.sessions.models(r) },
   'session.selectModel': { schema: sessionSelectModelRequestSchema, invoke: (api, r) => api.sessions.selectModel(r) },
+  'session.rename': { schema: sessionRenameRequestSchema, invoke: (api, r) => api.sessions.rename(r) },
   'session.prompt': { schema: sessionPromptRequestSchema, invoke: (api, r) => api.sessions.prompt(r) },
   'session.cancel': { schema: sessionCancelRequestSchema, invoke: (api, r) => api.sessions.cancel(r) },
   'host.describe': { schema: hostDescribeRequestSchema, invoke: (api, r) => api.host.describe(r) },
diff --git a/packages/host/apiproxy/tests/api-proxy-rename.spec.ts b/packages/host/apiproxy/tests/api-proxy-rename.spec.ts
new file mode 100644
index 0000000000..0e3f0ffe18
--- /dev/null
+++ b/packages/host/apiproxy/tests/api-proxy-rename.spec.ts
@@ -0,0 +1,129 @@
+/**
+ * sessions.rename delegation through the composed SessionTitleService. The
+ * agent factory is a structural stub whose createAgent forwards seed/meta into
+ * the real SessionStore, and whose resume never runs (every source here is
+ * already attached). Cold-session resolution is the shared `agentFor` path —
+ * api-proxy-cold.spec.ts owns the resume evidence for every unary that rides
+ * it, rename included.
+ */
+
+import { describe, expect, it } from 'vitest'
+import { Context } from 'cordis'
+import SessionStore from '@deepseek-ai/dsh-session'
+import AgentRegistry from '@deepseek-ai/dsh-agent'
+import type { Agent, AgentHandle, CreateAgentOptions } from '@deepseek-ai/dsh-agent'
+import { createUserMessage } from '@deepseek-ai/dsh-llm'
+import SessionTitleService from '@deepseek-ai/dsh-session-title'
+import UserInteractionService from '@deepseek-ai/dsh-user-interaction'
+import type { Session, SessionId } from '@deepseek-ai/dsh-session'
+import type { RpcRequest } from '@deepseek-ai/dsh-host-apiproxy/api/rpc'
+import { RpcId } from '@deepseek-ai/dsh-host-apiproxy/api/rpc'
+import { createApiProxy } from '@deepseek-ai/dsh-host-apiproxy'
+
+const sid = (id: string): SessionId => id as SessionId
+
+let nextRpc = 1
+function request<P>(payload: P): RpcRequest<P> {
+  return { rpcId: RpcId(`fr-${String(nextRpc++)}`), payload }
+}
+
+async function composed(withTitles = true): Promise<Context> {
+  const ctx = new Context()
+  await ctx.plugin(SessionStore)
+  await ctx.plugin(AgentRegistry)
+  await ctx.plugin(UserInteractionService)
+  if (withTitles) {
+    await ctx.plugin(SessionTitleService, { fallbackMaxWords: 5, fallbackMaxBytes: 40, maxTitleBytes: 40 })
+  }
+  // Store-backed structural factory: create builds the session with the
+  // forwarded seed/meta (the store validates the balanced prefix) and
+  // registers an idle agent stub over it.
+  ctx.agents.setFactory({
+    createAgent: (ownerCtx: Context, options: CreateAgentOptions): Promise<AgentHandle> => {
+      const session = ctx.sessions.create(options.sessionId, {
+        ...options.seed === undefined ? {} : { seed: [...options.seed] },
+        ...options.meta === undefined ? {} : { meta: options.meta },
+      })
+      const agent = { id: session.id, session, status: 'idle', ctx: ownerCtx } as Agent
+      ctx.agents.register(agent)
+      return Promise.resolve({ agent, dispose: () => Promise.resolve() })
+    },
+    resume: () => Promise.reject(new Error('resume must not run: every source is attached')),
+  })
+  return ctx
+}
+
+/** Register one live agent whose log holds `turns` completed turns. */
+function liveAgent(ctx: Context, id: string, turns: number): Session {
+  const session = ctx.sessions.create(sid(id), { meta: { cwd: '/proj' } })
+  for (let turn = 1; turn <= turns; turn++) {
+    session.append('turn/start', { turn, trigger: { kind: 'message', source: { kind: 'user' } } })
+    session.append('user/message', createUserMessage({
+      content: [{ type: 'text', text: `prompt ${String(turn)}` }],
+      source: { kind: 'user' },
+    }), { surfaceOp: 'append' })
+    session.append('turn/end', { turn, reason: { kind: 'completed' } })
+  }
+  ctx.agents.register({ id: session.id, session, status: 'idle', ctx } as Agent)
+  return session
+}
+
+const api = (ctx: Context) => createApiProxy(ctx, { provider: 'p', model: 'm', cwd: '/tmp', workspaceRoot: '/tmp' })
+
+describe('sessions.rename', () => {
+  it('accepts through the composed title service: normalized user-source event, echoed seq', async () => {
+    const ctx = await composed()
+    const source = liveAgent(ctx, 'session-rename', 1)
+
+    const renamed = await api(ctx).sessions.rename(request({ sessionId: source.id, title: '  new   name  ' }))
+    expect(renamed.result.ok).toBe(true)
+    if (!renamed.result.ok) return
+    expect(renamed.result.value.title).toBe('new name')
+    const event = source.events.findLast(item => item.type === 'session/title')
+    expect(event?.seq).toBe(renamed.result.value.seq)
+    expect(event?.data).toMatchObject({ title: 'new name', source: { kind: 'user' } })
+  })
+
+  it('maps only an empty-normalizing title to title-invalid, with a presentable message', async () => {
+    const ctx = await composed()
+    const source = liveAgent(ctx, 'session-rename-bad', 1)
+
+    // U+200B passes a client-side trim gate but normalizes to empty host-side.
+    const response = await api(ctx).sessions.rename(request({ sessionId: source.id, title: ' ​ ' }))
+    expect(response.result.ok).toBe(false)
+    if (!response.result.ok) {
+      expect(response.result.error).toMatchObject({
+        code: 'title-invalid',
+        details: { sessionId: source.id },
+      })
+      // The message renders verbatim in the rename dialog's alert.
+      expect(response.result.error.message).toBe('session title must contain visible characters')
+    }
+  })
+
+  it('maps a non-validation rename failure (stale session object) to internal, not title-invalid', async () => {
+    const ctx = await composed()
+    // The registered agent holds a session object from another store: the
+    // title service's liveness check throws a plain Error, which must not
+    // read as the user's fault.
+    const foreign = await composed(false)
+    const stale = liveAgent(foreign, 'session-rename-stale', 1)
+    ctx.agents.register({ id: stale.id, session: stale, status: 'idle', ctx } as Agent)
+
+    const response = await api(ctx).sessions.rename(request({ sessionId: stale.id, title: 'name' }))
+    expect(response.result.ok).toBe(false)
+    if (!response.result.ok) expect(response.result.error.code).toBe('internal')
+  })
+
+  it('answers internal when the composition mounts no session-title service', async () => {
+    const ctx = await composed(false)
+    const source = liveAgent(ctx, 'session-no-titles', 1)
+
+    const response = await api(ctx).sessions.rename(request({ sessionId: source.id, title: 'name' }))
+    expect(response.result.ok).toBe(false)
+    if (!response.result.ok) {
+      expect(response.result.error.code).toBe('internal')
+      expect(response.result.error.message).toMatch(/mounts no session-title service/)
+    }
+  })
+})
diff --git a/packages/host/apiproxy/tests/client-handler.spec.ts b/packages/host/apiproxy/tests/client-handler.spec.ts
index d807557d00..4f56f17dfd 100644
--- a/packages/host/apiproxy/tests/client-handler.spec.ts
+++ b/packages/host/apiproxy/tests/client-handler.spec.ts
@@ -46,6 +46,7 @@ function scriptedApi(overrides: {
       selectModel: r => ok(r, {
         selected: { provider: r.payload.provider, model: r.payload.model },
       }),
+      rename: r => ok(r, { title: 'renamed', seq: 0 }),
       prompt: r => ok(r, { accepted: true as const }),
       cancel: r => ok(r, { accepted: true as const }),
       ...overrides.sessions,
diff --git a/packages/host/apiproxy/tests/fetch-carrier.spec.ts b/packages/host/apiproxy/tests/fetch-carrier.spec.ts
index d7528072f7..1e975d6008 100644
--- a/packages/host/apiproxy/tests/fetch-carrier.spec.ts
+++ b/packages/host/apiproxy/tests/fetch-carrier.spec.ts
@@ -67,6 +67,9 @@ function fakeApi(overrides: Partial<{ muxFrames: MuxFrame[]; hostFrames: HostFra
           },
         }
       },
+      async rename(request) {
+        return { rpcId: request.rpcId, result: { ok: true, value: { title: request.payload.title, seq: 0 } } }
+      },
       async prompt(request) {
         return { rpcId: request.rpcId, result: { ok: true, value: { accepted: true as const } } }
       },
@@ -224,6 +227,8 @@ describe('unary round trip (handler ⇄ client, no network)', () => {
         },
       },
     })
+    const renamed = await c.sessions.rename({ sessionId: 's' as never, title: 'named' })
+    expect(renamed.result).toMatchObject({ ok: true, value: { title: 'named', seq: 0 } })
     expect((await c.sessions.prompt({ sessionId: 's' as never, mode: 'queue', content: [{ type: 'text', text: 'x' }] })).result.ok).toBe(true)
     expect((await c.sessions.cancel({ sessionId: 's' as never })).result.ok).toBe(true)
     expect((await c.host.describe({})).result.ok).toBe(true)
diff --git a/packages/host/apiproxy/tests/rpc-schemas.spec.ts b/packages/host/apiproxy/tests/rpc-schemas.spec.ts
index fd1f107a80..fe2780eadc 100644
--- a/packages/host/apiproxy/tests/rpc-schemas.spec.ts
+++ b/packages/host/apiproxy/tests/rpc-schemas.spec.ts
@@ -70,11 +70,13 @@ describe('rpcErrorSchema', () => {
     expect(rpcErrorSchema.parse({ code: 'agent-busy', message: 'm', details: { reason: 'r' } }).code).toBe('agent-busy')
     expect(rpcErrorSchema.parse({ code: 'command-error', message: 'm', details: {} }).code).toBe('command-error')
     expect(rpcErrorSchema.parse({ code: 'unknown-command', message: 'm', details: {} }).code).toBe('unknown-command')
+    expect(rpcErrorSchema.parse({ code: 'title-invalid', message: 'm', details: { sessionId: 's' } }).code).toBe('title-invalid')
     expect(rpcErrorSchema.parse({ code: 'internal', message: 'm', details: {} }).code).toBe('internal')
   })
 
   it('rejects a known code with missing details', () => {
     expect(() => rpcErrorSchema.parse({ code: 'agent-busy', message: 'm', details: {} })).toThrow()
+    expect(() => rpcErrorSchema.parse({ code: 'title-invalid', message: 'm', details: {} })).toThrow()
     expect(() => rpcErrorSchema.parse({ code: 'command-error', message: 'm' })).toThrow()
     expect(() => rpcErrorSchema.parse({ code: 'nope', message: 'm', details: {} })).toThrow()
   })
diff --git a/packages/host/apiproxy/tsconfig.json b/packages/host/apiproxy/tsconfig.json
index 7e1e83e39b..5185081283 100644
--- a/packages/host/apiproxy/tsconfig.json
+++ b/packages/host/apiproxy/tsconfig.json
@@ -41,6 +41,9 @@
     {
       "path": "../../session-projection/session-projection-cache"
     },
+    {
+      "path": "../../session-title/session-title"
+    },
     {
       "path": "../../skill/skill"
     },
diff --git a/packages/llm/README.i18n.yaml b/packages/llm/README.i18n.yaml
index c5f5de580a..0dfa7fc0e6 100644
--- a/packages/llm/README.i18n.yaml
+++ b/packages/llm/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/llm/README.md
 README.md: 66b7beabd73cc3fec7230f209a9da0da48a37c95
-README.zh.md: 48c54358ce3e8e21e33a6ef5b75a7e095b6581d5
+README.zh.md: 3f38c94fb4a43bed007d06b8c8558b5bd330a41c
diff --git a/packages/llm/README.zh.md b/packages/llm/README.zh.md
index 48c54358ce..3f38c94fb4 100644
--- a/packages/llm/README.zh.md
+++ b/packages/llm/README.zh.md
@@ -1,15 +1,15 @@
-# llm/：LLM 能力家族
+# llm/：LLM（大语言模型）能力家族
 
 [English](README.md) | 中文
 
-LLM seam 及其提供方适配器。接口包（`llm`）拥有抽象服务、内容块词汇和流分片组装器；适配器是在 `ctx.llm` 上注册的具体实现。这些全是**产品** 包。
+LLM seam 及其提供方适配器。接口包（`llm`）拥有抽象服务、内容块词汇和流分片组装器；适配器是在 `ctx.llm` 上注册的具体实现。这些全是**产品**包（package）。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
 | `llm/` | 抽象 LLM 服务 + 内容块词汇 + 分片组装器 | `ctx.llm` |
-| `token-meter/` | 感知回放的请求与表层 token 测量 | `ctx.tokenMeter` |
-| `llm-retry/` | 确切提供方的 normal 或无界请求重试策略 | （监听 `agent/request-error`） |
-| `llm-deepseek/` | DeepSeek API 适配器（直接 fetch + eventsource-parser SSE） | （注册到 `ctx.llm`） |
+| `token-meter/` | 感知回放的请求 token 与表层 token 测量 | `ctx.tokenMeter` |
+| `llm-retry/` | 确切提供方的常规或无界请求重试策略 | （监听 `agent/request-error`） |
+| `llm-deepseek/` | DeepSeek API 适配器，直接使用 fetch + eventsource-parser 和 SSE（Server-Sent Events） | （注册到 `ctx.llm`） |
 | `llm-pi-ai/` | 通过 `@earendil-works/pi-ai` 实现的多提供方适配器 | （注册到 `ctx.llm`） |
 
-接口位于 `llm/llm/`；适配器、重试策略和可复用的 token 计量器都是该分组下的扁平兄弟包。请求按 `provider` 路由，而 `model` 会原样传给选中的适配器。拥有路由的适配器提供重试策略，并解析可用的确切模型身份、上下文容量和推理元数据；重试执行器与 token 计量器仍与提供方无关。新的提供方适配器只需在 `ctx.llm` 上注册一个或多个提供方路由，无需改动消费方。两个已交付实现见[双生 LLM 适配器](../../.agents/notes/implemented/architecture/2026-06-13-twin-llm-adapters.md)，测量归属见[回放 token 计量器 Agent Note](../../.agents/notes/implemented/architecture/2026-07-15-replay-token-meter-service.md)，容量与压缩策略归属见[路由模型上下文 Agent Note](../../.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.md)。
+接口位于 `llm/llm/`；适配器、重试策略和可复用的 token 计量器以扁平结构并列在该分组下。请求按 `provider` 路由，而 `model` 会原样传给选中的适配器。负责该路由的适配器提供重试策略，并解析可用的确切模型身份、上下文容量和推理元数据；重试执行器与 token 计量器仍与提供方无关。新的提供方适配器只需在 `ctx.llm` 上注册一个或多个提供方路由，无需改动消费方。两个已交付实现见[双生 LLM 适配器](../../.agents/notes/implemented/architecture/2026-06-13-twin-llm-adapters.md)，测量归属见[回放 token 计量器 Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-15-replay-token-meter-service.md)，容量与压缩（compaction）策略归属见[路由模型上下文 Agent Note](../../.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.md)。
diff --git a/packages/llm/llm-deepseek/README.i18n.yaml b/packages/llm/llm-deepseek/README.i18n.yaml
index 7f7619b003..594600e8c3 100644
--- a/packages/llm/llm-deepseek/README.i18n.yaml
+++ b/packages/llm/llm-deepseek/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/llm/llm-deepseek/README.md
 README.md: 7a2314ea2ca0fb6606310a4961fcbc658240a7b8
-README.zh.md: 523bbbfd29b4598c024a1fff4a7121a7cb88bf41
+README.zh.md: d73145f8b8a32d8a515b6b4c0e916f0a11cd4771
diff --git a/packages/llm/llm-deepseek/README.zh.md b/packages/llm/llm-deepseek/README.zh.md
index 523bbbfd29..d73145f8b8 100644
--- a/packages/llm/llm-deepseek/README.zh.md
+++ b/packages/llm/llm-deepseek/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-harness LLM seam 的 DeepSeek chat-completions 适配器：直接 `fetch` + SSE（由 `eventsource-parser` 分帧），将官方协议格式（真源：API 文档 guides/thinking_mode、guides/tool_calls、api/create-chat-completion）转换为 `StreamChunk` 协议。
+harness LLM（大语言模型）seam 的 DeepSeek chat-completions 适配器：直接 `fetch` + SSE（Server-Sent Events，由 `eventsource-parser` 分帧），将官方协议格式（wire format；真源：API 文档 guides/thinking_mode、guides/tool_calls、api/create-chat-completion）转换为 `StreamChunk` 协议。
 
-同一 seam 的第二个库支持实现位于 `@deepseek-ai/dsh-llm-pi-ai`。本包始终拥有 `deepseek` 提供方路由；在同一上下文中装载 `provider: deepseek` 的 pi-ai profile 会按设计抛出 `LlmError('DUPLICATE_ADAPTER')`。
+同一 seam 的第二个基于库的实现位于 `@deepseek-ai/dsh-llm-pi-ai`。本包（package）始终负责 `deepseek` 提供方路由；在同一上下文中装载 `provider: deepseek` 的 pi-ai profile 会按设计抛出 `LlmError('DUPLICATE_ADAPTER')`。
 
-包根目录公开 Cordis 插件契约与 `DeepSeekAdapter`；协议序列化、SSE 解析与 chunk 转换 helper 不属于该根契约。
+包根入口导出 Cordis 插件契约与 `DeepSeekAdapter`；协议序列化、SSE 解析与分片转换 helper 不属于该根契约。
 
 ## 配置
 
@@ -36,33 +36,33 @@ harness LLM seam 的 DeepSeek chat-completions 适配器：直接 `fetch` + SSE
 
 该插件注册唯一提供方路由 `deepseek`，同时注册解析后的 `retryPolicy`。请求使用 `provider: deepseek` 选择该路由；其 `model` 会作为协议 `model` 字符串原样传递，因此更改 DeepSeek 模型不需要生命周期时注册。省略 `models` 会公布 `deepseek-v4-flash`（名称为 `DeepSeek-V4-Flash`）和 `deepseek-v4-pro`（名称为 `DeepSeek-V4-Pro`），两者的上下文窗口均为 256,000 token；显式列表会替换这些默认值，`models: []` 则不公布任何模型。Catalog 配置项通过 `ctx.llm.listModels('deepseek')` 公开给 ACP（Agent Client Protocol）编辑器和 Web 选择器等客户端，但仍只提供建议：未列出模型 id 仍原样传递。省略配置项 name 默认为其 id。
 
-`contextWindow` 对每个已配置模型都可选，不会通过建议 catalog 公开。`ctx.llm.resolveModelInfo('deepseek', model).context` 先返回精确模型值，再对不含容量的配置项或未列出原样传递 id 返回 `defaultContextWindow`。两者都不存在时，`context` 字段缺失但不会使路由失效。因此，压力敏感插件可以获得部署拥有的容量，不会将模型 selector 视为权威。为 `deepseek` 注册另一个适配器会抛出 `LlmError('DUPLICATE_ADAPTER')`。
+`contextWindow` 对每个已配置模型都可选，不会通过建议 catalog 公开。`ctx.llm.resolveModelInfo('deepseek', model).context` 先返回精确模型值，再对不含容量的配置项或未列出原样传递 id 返回 `defaultContextWindow`。两者都不存在时，`context` 字段缺失但不会使路由失效。因此，压力敏感插件可以获得由部署决定的容量，不会将模型 selector 视为权威。为 `deepseek` 注册另一个适配器会抛出 `LlmError('DUPLICATE_ADAPTER')`。
 
-同一确切模型结果会在部署策略允许思考时，为每个原样传递模型在 `reasoning` 下公开有序的 `off`、`high` 和 `max` 推理强度。`reasoningEffort` 选择部署默认值，省略时回退为 `high`。`agent/request` 可以在每个会话步骤替换它；解析后的值会记录在 `request/header`。`high` 和 `max` 会启用思考，并序列化为官方顶层 `reasoning_effort`；适配器持有的 `off` 则序列化为 `thinking.type: disabled`，且省略 `reasoning_effort`。不支持的值会在网络 I/O 前以 `UNSUPPORTED_REASONING_EFFORT` 失败。
+同一确切模型结果会在部署策略允许思考时，为每个原样传递模型在 `reasoning` 下公开有序的 `off`、`high` 和 `max` 推理（reasoning）强度。`reasoningEffort` 选择部署默认值，省略时回退为 `high`。`agent/request` 可以在每个会话步骤替换它；解析后的值会记录在 `request/header`。`high` 和 `max` 会启用思考，并序列化为官方顶层 `reasoning_effort`；适配器持有的 `off` 则序列化为 `thinking.type: disabled`，且省略 `reasoning_effort`。不支持的值会在网络 I/O 前以 `UNSUPPORTED_REASONING_EFFORT` 失败。
 
-`thinking: disabled` 是部署锁定：它只公布 `off`，并以 `off` 为默认值。省略 `reasoningEffort` 或将其配置为 `off` 均有效；配置 `high` 或 `max` 会使插件加载失败，直接按请求启用思考也会在网络 I/O 前失败。携带 `GenerateOptions.purpose: 'session-title'` 的请求也会强制禁用思考并省略已解析的推理强度，将有界输出保留给可见标题文本，不改变会话或压缩默认值。
+`thinking: disabled` 是部署锁定：它只公布 `off`，并以 `off` 为默认值。省略 `reasoningEffort` 或将其配置为 `off` 均有效；配置 `high` 或 `max` 会使插件加载失败，直接按请求启用思考也会在网络 I/O 前失败。携带 `GenerateOptions.purpose: 'session-title'` 的请求也会强制禁用思考并省略已解析的推理强度，将有界输出保留给可见标题文本，不改变会话或压缩（compaction）默认值。
 
-`streamIdleTimeoutMs` 会限制每次未完成提供方读取，包括初始 `fetch`，但不计入消费方在 chunk 间花费的时间。一个稳定 abort 信号会在整个调用中达到请求与 body reader；过期会停止传输并抛出 `LlmError('TIMEOUT')`，较早的调用方 abort 则抛出 `LlmError('ABORTED')`。适配器每次 `stream()` 调用精确发起一次提供方请求；它把已配置策略注册为提供方元数据，再由 `dsh-llm-retry` 在持久 agent 步骤边界单独执行该策略。
+`streamIdleTimeoutMs` 会限制每次未完成提供方读取，包括初始 `fetch`，但不计入消费方在分片间花费的时间。同一个稳定的 abort 信号会在整个调用期间传递给请求与 body reader；过期会停止传输并抛出 `LlmError('TIMEOUT')`，较早的调用方 abort 则抛出 `LlmError('ABORTED')`。适配器每次 `stream()` 调用恰好发起一次提供方请求；它把已配置策略注册为提供方元数据，再由 `dsh-llm-retry` 在持久化的 agent（智能体）步骤边界单独执行该策略。
 
 ## 应用归因
 
-每个请求都携带 dsh-llm `attributionHeaders()` 的共享归因标头，即用于识别 harness 的必需 `User-Agent` 基线（见 [dsh-llm § 应用归因](../llm/README.md#app-attribution-attributionts)）。在该适配器契约下，直接 DeepSeek 请求与 OpenAI 兼容 gateway 请求都不会获得提供方特定应用归因标头；OpenRouter 应用归因暂缓到未来的显式 OpenRouter 适配器或模式。`GenerateOptions.purpose` 为 `compaction` 的请求（dsh-compact-basic 的辅助摘要调用）还会携带 `x-deepseek-harness-compact: 1`，让宿主可以将压缩流量与会话请求分开。
+每个请求都携带 dsh-llm `attributionHeaders()` 的共享归因标头，即用于识别 harness 的必需 `User-Agent` 基线（见 [dsh-llm § 应用归因](../llm/README.md#app-attribution-attributionts)）。在该适配器契约（adapter contract）下，直接 DeepSeek 请求与 OpenAI 兼容 gateway 请求都不会获得提供方特定应用归因标头；OpenRouter 应用归因暂缓到未来的显式 OpenRouter 适配器或模式。`GenerateOptions.purpose` 为 `compaction` 的请求（dsh-compact-basic 的辅助摘要调用）还会携带 `x-deepseek-harness-compact: 1`，让宿主可以将压缩流量与会话请求分开。
 
 ## 协议格式说明（已通过实时请求与官方文档验证）
 
-- 只支持流式输出（`stream_options.include_usage` 始终开启）。`usage` 可能附着在 finish chunk 上，也可能作为尾随仅 usage chunk 到达；转换器会将两者都延迟到 `[DONE]`，因此 `usage` 始终位于 `finish` 之前，`finish` 之后不会出现任何内容。
-- 适配器持有的 `off` 推理强度映射为 `thinking: {type: 'disabled'}`，绝不会以 `reasoning_effort: 'off'` 跨越协议。
-- 第一个 thinking 模式 chunk 携带 `reasoning_content: ""`，系统会处理它（不会产生多余 reasoning 块）。
-- **Reasoning 回传规则**：对携带工具调用的 assistant 轮次，会将 `reasoning_content` 序列化回历史（thinking 模式 API 必需）；对不含工具调用的轮次，它会被丢弃（不会使用，可节省 token）。
+- 只支持流式输出（`stream_options.include_usage` 始终开启）。`usage` 可能附着在 finish 分片上，也可能作为尾随的纯 usage 分片到达；转换器会将两者都延迟到 `[DONE]`，因此 `usage` 始终位于 `finish` 之前，`finish` 之后不会出现任何内容。
+- 适配器持有的 `off` 推理强度映射为 `thinking: {type: 'disabled'}`，绝不会以 `reasoning_effort: 'off'` 通过协议发送。
+- 第一个思考模式分片携带 `reasoning_content: ""`，系统会处理它（不会产生多余 reasoning 块）。
+- **推理回传规则**：对携带工具调用的 assistant 轮次，会将 `reasoning_content` 序列化回历史（思考模式 API 必需）；对不含工具调用的轮次，它会被丢弃（不会使用，可节省 token）。
 - Cache 计量：`cacheReadTokens` ← `prompt_cache_hit_tokens` / `prompt_tokens_details.cached_tokens`；DeepSeek 不报告 cache-write 指标。
 
 ## 错误
 
-非 2xx 响应会抛出稳定 code 的 `LlmError`：`AUTH`（401/403）、`QUOTA`（提供方详细信息标识配额、余额或点数耗尽的响应）、`RATE_LIMIT`（其他 429）、`CONTEXT_WINDOW_EXCEEDED`（提供方 code、type 或 message 标识上下文溢出的 400）、`INVALID_REQUEST`（其他 400）、`SERVER`（5xx），其他情况为 `HTTP_<status>`。其可序列化 `failure` 保留 HTTP 状态，以及有效的正 `Retry-After` 秒数／日期延迟和存在时的 `x-request-id` / `x-deepseek-request-id`。响应前传输失败（DNS、连接被拒绝、TLS、proxy）会抛出命名已配置端点的 `TRANSPORT`，并将原始拒绝链接为 `cause`；调用方 abort 抛出 `ABORTED`，loop 的取消信号仍最具权威。协议违例抛出 `STREAM_CLOSED`（没有 `[DONE]`）或 `MALFORMED_RESPONSE`（JSON payload 错误）。未知协议 `finish_reason`（例如 `content_filter`、`insufficient_system_resource`）会变为 `finish {kind: 'error', failure}` chunk；已完成流如果使用 `stop`（或缺失）finish 但没有开启内容块，就会变为 `finish {kind: 'error'}`，code 为 `EMPTY_RESPONSE`（默认策略会重试）。
+非 2xx 响应会抛出稳定 code 的 `LlmError`：`AUTH`（401/403）、`QUOTA`（提供方详细信息标识配额、余额或点数耗尽的响应）、`RATE_LIMIT`（其他 429）、`CONTEXT_WINDOW_EXCEEDED`（提供方 code、type 或 message 标识上下文溢出的 400）、`INVALID_REQUEST`（其他 400）、`SERVER`（5xx），其他情况为 `HTTP_<status>`。其可序列化 `failure` 保留 HTTP 状态，以及有效的正 `Retry-After` 秒数／日期延迟和存在时的 `x-request-id` / `x-deepseek-request-id`。响应前传输失败（DNS、连接被拒绝、TLS、proxy）会抛出命名已配置端点的 `TRANSPORT`，并将原始拒绝作为 `cause`；调用方 abort 抛出 `ABORTED`，仍以 loop 的取消信号为准。协议违例抛出 `STREAM_CLOSED`（没有 `[DONE]`）或 `MALFORMED_RESPONSE`（JSON payload 格式错误）。未知协议 `finish_reason`（例如 `content_filter`、`insufficient_system_resource`）会变为 `finish {kind: 'error', failure}` 分片；已完成流如果使用 `stop`（或缺失）finish 但没有开启内容块，就会变为 `finish {kind: 'error'}`，code 为 `EMPTY_RESPONSE`（默认策略会重试）。
 
 ## 测试
 
-单元套件使用本地 `node:http` mock SSE 服务器（无网络），覆盖动态 `high`／`off`／`max` 选择、结构化 HTTP 事实、格式错误／截断流、调用方 abort、连接失败，以及 idle 超时确实会 abort 实际 body 的证明。真实 API 覆盖位于 `tests/adapter.e2e.ts`（`pnpm run test:e2e`，由 key 调节）：V4 Flash + V4 Pro，覆盖 thinking 启用／禁用与两种官方 effort 级别，包括 thinking + 工具往返与 reasoning 回传。
+单元套件使用本地 `node:http` mock SSE 服务器（无网络），覆盖动态 `high`／`off`／`max` 选择、结构化 HTTP 事实、格式错误／截断流、调用方 abort、连接失败，以及 idle 超时确实会 abort 实际 body 的证明。真实 API 覆盖位于 `tests/adapter.e2e.ts`（`pnpm run test:e2e`，需有 key 才会运行）：V4 Flash + V4 Pro，覆盖思考启用／禁用与两种官方 effort 级别，包括思考 + 工具往返与推理回传。
 
 ## 模型体验
 
@@ -70,21 +70,21 @@ harness LLM seam 的 DeepSeek chat-completions 适配器：直接 `fetch` + SSE
 
 #### 模型看到的内容
 
-所选 DeepSeek 模型会收到 harness 系统提示词、消息历史、工具 schema、stop sequence 和调用配置，不含适配器撰写的提示词文本。当之前的 assistant 轮次包含工具调用时，会按要求回传其 reasoning 内容；不含工具调用的轮次会省略 reasoning。
+所选 DeepSeek 模型会收到 harness 系统提示词、消息历史、工具 schema、stop sequence 和调用配置，不含适配器撰写的提示词文本。当之前的 assistant 轮次包含工具调用时，会按要求回传其推理内容；不含工具调用的轮次会省略推理。
 
 #### Token 影响
 
-精确输入取决于提供方 tokenization。有条件 reasoning 回传会增加工具往返上下文，丢弃其他 reasoning 则避免再次支付这些 token；可用时会报告 cache-read 用量。
+精确输入取决于提供方 tokenization。有条件推理回传会增加工具往返上下文，丢弃其他推理则避免再次支付这些 token；可用时会报告 cache-read 用量。
 
 #### KV Cache 影响
 
-未更改的已组装前缀可使用 DeepSeek cache 复用，适配器会在 usage 中报告它。模型路由变更，或任何上游提示词、schema、前缀或历史变更，都可能使从第一个改变 token 起的复用失效；reasoning 回传会在工具往返期间追加。
+未更改的已组装前缀可使用 DeepSeek cache 复用，适配器会在 usage 中报告它。模型路由变更，或任何上游提示词、schema、前缀或历史变更，都可能使从首个发生变化的 token 起的复用失效；推理回传会在工具往返期间追加。
 
 ### DeepSeek 响应
 
 #### 模型看到的内容
 
-Reasoning、文本与原始字符串工具参数会转换为 harness chunk，供 loop 记录和组装。
+推理、文本与原始字符串工具参数会转换为 harness 分片，供 loop 记录和组装。
 
 #### Token 影响
 
@@ -98,4 +98,4 @@ loop 保留的响应块会追加到下一个请求，并保留其较早可复用
 
 - **未映射 `tool_choice`**：它不属于核心词汇（MVP 取舍，与 pi-ai twin 共享）。
 - **请求使用原始 `fetch`，而非 `@cordisjs/plugin-http`**：没有共享 proxy／拦截配置；采用暂缓到第二个适配器需要该功能时（`TODO(http)`）。
-- **序列化会将 user 与工具结果内容展平为文本块**：会跳过插件添加的块类型，空工具输出会以字面 `(no output)` 跨越协议。
+- **序列化会将 user 与工具结果内容展平为文本块**：会跳过插件添加的块类型，空工具输出会以字面 `(no output)` 通过协议发送。
diff --git a/packages/llm/llm-pi-ai/README.i18n.yaml b/packages/llm/llm-pi-ai/README.i18n.yaml
index 38b9c093f6..fb47822ca5 100644
--- a/packages/llm/llm-pi-ai/README.i18n.yaml
+++ b/packages/llm/llm-pi-ai/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/llm/llm-pi-ai/README.md
 README.md: ac47cf6a21285fc887948a5a7798a9f1cb9157b0
-README.zh.md: a3d864ed9068d8bdaff4c5b73a4b7b339802ae05
+README.zh.md: 650ec7a578549cec6bd001e15ff5be4dea86e942
diff --git a/packages/llm/llm-pi-ai/README.zh.md b/packages/llm/llm-pi-ai/README.zh.md
index a3d864ed90..650ec7a578 100644
--- a/packages/llm/llm-pi-ai/README.zh.md
+++ b/packages/llm/llm-pi-ai/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-基于 [`@earendil-works/pi-ai`](https://www.npmjs.com/package/@earendil-works/pi-ai) 的 harness LLM seam 通用多提供方适配器。一个插件实例拥有显式提供方 profile 列表；每个请求使用 `GenerateOptions.provider` 选择 profile，并从 pi-ai 已安装 catalog 中动态解析 `GenerateOptions.model`。
+基于 [`@earendil-works/pi-ai`](https://www.npmjs.com/package/@earendil-works/pi-ai) 的 harness LLM（大语言模型）seam 通用多提供方适配器。一个插件实例拥有显式提供方 profile 列表；每个请求使用 `GenerateOptions.provider` 选择 profile，并从 pi-ai 已安装 catalog 中动态解析 `GenerateOptions.model`。
 
-包根目录公开 Cordis 插件契约与 `PiAiAdapter`；profile 解析、模型构造、回放转换和流转换保留在包内部。
+包（package）根入口导出 Cordis 插件契约与 `PiAiAdapter`；profile 解析、模型构造、回放转换和流转换保留在包内部。
 
 ## 配置
 
@@ -39,39 +39,39 @@
 
 适配器通过 `ctx.llm.listModels(provider)` 公开每个已配置提供方已安装的 pi-ai 模型。这是从 `getModels(provider)` 派生的提供方无关 selector 元数据；请求时解析仍会执行权威 catalog 查找，因此发现不会创建第二个模型注册表。`ctx.llm.resolveModelInfo(provider, model)` 会执行一次精确 descriptor 查找，并返回其身份、上下文窗口和可选思考级别，让权威元数据保留在拥有路由的适配器上，而非消费方。
 
-`reasoning.efforts` 列表是 pi-ai 有序的 `getSupportedThinkingLevels(model)` 结果，不经筛选或规范化，其中包括 `off`，以及模型对 `xhigh` 或 `max` 的特定支持。Harness 将每个规范 pi-ai 级别公开为不透明 ID；提供方／模型协议拼写仍保留在 pi-ai 的 `thinkingLevelMap` 中。因此，不具备推理（reasoning）能力的模型也会公开 pi-ai 的 `off` 选项。配置 profile 的 `reasoning` 值（包括 `off`）在存在时是部署默认值；省略它会保留提供方默认值。每次请求的 `GenerateOptions.reasoningEffort` 优先；任何未出现在确切模型能力中的显式值都会在网络 I/O 前以 `UNSUPPORTED_REASONING_EFFORT` 失败，而不会被自动调整。pi-ai 的通用流选项通过省略 `reasoning` 表示 `off`。
+`reasoning.efforts` 列表是 pi-ai 有序的 `getSupportedThinkingLevels(model)` 结果，不经筛选或规范化，其中包括 `off`，以及模型对 `xhigh` 或 `max` 的特定支持。Harness 将每个规范 pi-ai 级别公开为不透明 ID；提供方／模型在协议格式中的表示仍保留在 pi-ai 的 `thinkingLevelMap` 中。因此，不具备推理（reasoning）能力的模型也会公开 pi-ai 的 `off` 选项。配置 profile 的 `reasoning` 值（包括 `off`）在存在时是部署默认值；省略它会保留提供方默认值。每次请求的 `GenerateOptions.reasoningEffort` 优先；任何未出现在确切模型能力中的显式值都会在网络 I/O 前以 `UNSUPPORTED_REASONING_EFFORT` 失败，而不会被自动调整。pi-ai 的通用流选项通过省略 `reasoning` 表示 `off`。
 
-受支持的 profile 字段是 `provider`、`apiKey`、`baseURL`、`headers`、`reasoning`、`thinkingBudgets`、`cacheRetention`、`transport`、`timeoutMs`、`websocketConnectTimeoutMs`、`streamIdleTimeoutMs` 和 `retryPolicy`。每个 profile 的可选重试策略都会与该提供方路由一同捕获；省略时使用有界的 normal 默认值。流 idle 间隔必须是正的有限 Node 定时器延迟，默认为五分钟，且只覆盖未完成提供方读取，不包括消费方思考时间。Harness 应用归因会胜过名称冲突的已配置标头。
+受支持的 profile 字段是 `provider`、`apiKey`、`baseURL`、`headers`、`reasoning`、`thinkingBudgets`、`cacheRetention`、`transport`、`timeoutMs`、`websocketConnectTimeoutMs`、`streamIdleTimeoutMs` 和 `retryPolicy`。每个 profile 的可选重试策略都会与该提供方路由一同捕获；省略时使用有界的常规默认值。流空闲间隔必须是正的有限 Node 定时器延迟，默认为五分钟，且只覆盖未完成提供方读取，不包括消费方思考时间。若已配置标头中有同名项，则以 Harness 应用归因为准。
 
-适配器强制 pi-ai SDK `maxRetries` 为零，因此一次 `stream()` 调用只会发起一次提供方请求。已移除 profile 字段 `maxRetries` 和 `maxRetryDelayMs` 会使加载失败，而不是静默倍增或隐藏单独组合的 agent 级重试预算。Idle 过期会 abort SDK 的稳定请求信号，并以 `TIMEOUT` 呈现；较早的调用方 abort 仍为 `ABORTED`。
+适配器强制 pi-ai SDK `maxRetries` 为零，因此一次 `stream()` 调用只会发起一次提供方请求。已移除 profile 字段 `maxRetries` 和 `maxRetryDelayMs` 会使加载失败，而不是静默倍增或隐藏单独组合的 agent（智能体）级重试预算。空闲超时会 abort SDK 的稳定请求信号，并以 `TIMEOUT` 呈现；较早的调用方 abort 仍为 `ABORTED`。
 
 ## 提供方／模型路由与回放
 
 所选 pi-ai catalog descriptor 提供协议实现。这包括原生 API 差异，例如 descriptor 使用 Responses API 而非 Chat Completions 的 OpenAI 模型；harness 适配器不会按模型名称硬编码端点选择。
 
-成功的 assistant 响应会在自身持久提供方／模型溯源旁存储经版本化的无损 JSON 回放状态。请求时，`LlmService` 只有在历史提供方路由与目标提供方路由当前由同一个 `PiAiAdapter` 实例拥有时，才会传递回放状态。即使目标提供方或模型改变，适配器也会验证状态并恢复 pi-ai 响应 id 与提供方 signature；随后由 pi-ai 判定目标 API 可以复用哪些元数据。没有回放状态的历史会被转换为外来提供方无关内容，绝不伪装为原生 pi-ai 响应。
+成功的 assistant 响应会在自身持久提供方／模型溯源旁存储经版本化的无损 JSON 回放状态。请求时，`LlmService` 只有在历史提供方路由与目标提供方路由当前由同一个 `PiAiAdapter` 实例拥有时，才会传递回放状态。即使目标提供方或模型改变，适配器也会验证状态并恢复 pi-ai 响应 id 与提供方 signature；随后由 pi-ai 判定目标 API 可以复用哪些元数据。没有回放状态的历史会被转换为外来的、与提供方无关的内容，绝不伪装为原生 pi-ai 响应。
 
 如果 listener 改写已组装 assistant 内容，loop 会在记录消息前丢弃回放状态，因为其提供方元数据不再描述该内容。无效版本、格式错误元数据、溯源提供方／模型不匹配，以及内容／块不匹配都会显式以 `LlmError('INVALID_REPLAY_STATE')` 失败。
 
 ## 词汇差异
 
 - pi-ai 工具调用参数是已解析对象；harness 存储原始 JSON 字符串。适配器会解析输入，并将输出重新字符串化。
-- pi-ai 将失败报告为流内错误事件；它们会映射到 `finish {kind:'error'|'aborted', failure}` chunk。提供方特定错误文本会区分终端 `QUOTA` 与短暂 `RATE_LIMIT`，针对已解析模型上下文窗口评估的文本与 usage 信号则将溢出规范化为 `CONTEXT_WINDOW_EXCEEDED`。携带零个内容块消息的终止 `stop` 会映射为 `finish {kind:'error'}`，code 为 `EMPTY_RESPONSE`（默认策略会重试），而非成功空消息。
-- pi-ai 将 reasoning token 折叠到输出 usage 中；没有可映射的独立 reasoning 计数。
-- pi-ai 的 `off` thinking 级别会原样穿过 Harness 能力 seam，并在分派时变为被省略的 pi-ai 通用 `reasoning` 选项。
-- `GenerateOptions.stop` 会以 `UNSUPPORTED_OPTION` 被拒绝，因为 pi-ai 的通用流式输出表层无法保证所有提供方都支持它。
+- pi-ai 将失败报告为流内错误事件；它们会映射到 `finish {kind:'error'|'aborted', failure}` 分片。提供方特定错误文本会区分终止型 `QUOTA` 与暂时型 `RATE_LIMIT`，针对已解析模型上下文窗口评估的文本与 usage 信号则将溢出规范化为 `CONTEXT_WINDOW_EXCEEDED`。终止时的 `stop` 若消息不含内容块，则会映射为 `finish {kind:'error'}`，code 为 `EMPTY_RESPONSE`（默认策略会重试），而非成功空消息。
+- pi-ai 将推理 token 折叠到输出 usage 中；没有可映射的独立推理计数。
+- pi-ai 的 `off` 思考级别会原样穿过 Harness 能力 seam，并在分派时变为被省略的 pi-ai 通用 `reasoning` 选项。
+- `GenerateOptions.stop` 会以 `UNSUPPORTED_OPTION` 被拒绝，因为 pi-ai 的通用流式输出接口无法保证所有提供方都支持它。
 
 ## 应用归因
 
 每个请求都携带 dsh-llm `attributionHeaders()` 的共享归因标头，并通过 pi-ai `headers` 流选项合并。不会合成提供方特定应用归因标头。详见 [dsh-llm § 应用归因](../llm/README.md#app-attribution-attributionts)。
 
-## 依赖重量
+## 依赖体量
 
-pi-ai 会安装多个提供方 SDK，并延迟加载 catalog 模型所选的 SDK。依赖重量隔离在该可选适配器包中。
+pi-ai 会安装多个提供方 SDK，并延迟加载 catalog 模型所选的 SDK。该可选适配器包将依赖体量隔离在自身范围内。
 
 ## 测试
 
-单元测试使用重定向到本地 mock 服务器的 pi-ai catalog 模型，覆盖提供方／profile 路由、每次适配器调用一次协议请求、idle-timeout 响应终止、调用方 abort、原生 API 选择、端点覆盖、归因、转换、回放状态验证，以及一个适配器实例内的跨提供方／模型回放。真实 API 覆盖仍位于由 key 调节的 `pnpm run test:e2e` 下。
+单元测试使用重定向到本地 mock 服务器的 pi-ai catalog 模型，覆盖提供方／profile 路由、每次适配器调用只发起一个协议请求、idle-timeout 响应终止、调用方 abort、原生 API 选择、端点覆盖、归因、转换、回放状态验证，以及一个适配器实例内的跨提供方／模型回放。真实 API 覆盖仍需 key 才会启用，并通过 `pnpm run test:e2e` 运行。
 
 ## 模型体验
 
@@ -87,17 +87,17 @@ pi-ai 会安装多个提供方 SDK，并延迟加载 catalog 模型所选的 SDK
 
 #### KV Cache 影响
 
-转换保留逻辑请求顺序，不添加文本；复用取决于所选提供方的序列化与回放状态。更改适配器实例、提供方、模型或任何上游请求 token，都可能使从第一个差异起的复用失效。
+转换保留逻辑请求顺序，不添加文本；复用取决于所选提供方的序列化与回放状态。更改适配器实例、提供方、模型或任何上游请求 token，都可能使复用从首个出现差异的 token 起失效。
 
 ### 提供方响应
 
 #### 模型看到的内容
 
-pi-ai 事件会变为 harness reasoning、文本、工具调用、usage 与 finish chunk。已解析工具参数以原始 JSON 字符串形式跨越 harness 边界。
+pi-ai 事件会变为 harness 推理、文本、工具调用、usage 与 finish 分片。已解析工具参数以原始 JSON 字符串形式通过 harness 边界传递。
 
 #### Token 影响
 
-只有在 loop 记录生成内容后，它才会影响后续输入。提供方不单独报告 reasoning token 时，pi-ai 会将其折叠到输出 usage 中。
+只有在 loop 记录生成内容后，它才会影响后续输入。提供方不单独报告推理 token 时，pi-ai 会将其折叠到输出 usage 中。
 
 #### KV Cache 影响
 
@@ -109,4 +109,4 @@ pi-ai 事件会变为 harness reasoning、文本、工具调用、usage 与 fini
 - **不支持 `GenerateOptions.stop`**：pi-ai 的通用流选项无法保证所有提供方都支持 stop sequence，因此适配器会拒绝该字段。
 - **历史中的 `system` 消息使用 pi-ai 通用上下文转换**：提供方特定位置由 pi-ai 决定，而非由 harness 拥有的协议覆盖决定。
 - **无法获取提供方 HTTP 状态**：pi-ai 错误事件不会在所有提供方上公开稳定 HTTP 状态；失败只公开稳定 harness 错误 code。
-- **重试策略由提供方持有，而不是 SDK 重试**：每个提供方 profile 都可以配置嵌套的 `retryPolicy`，由 `dsh-llm-retry` 在 agent 的失败步骤 seam 上执行；pi-ai SDK 重试仍保持禁用，因此持久 agent 步骤与 `llm/retry` 事件拥有每次可见尝试，直接 `ctx.llm.stream()` 调用仍只尝试一次。
+- **重试策略由提供方持有，而不是 SDK 重试**：每个提供方 profile 都可以配置嵌套的 `retryPolicy`，由 `dsh-llm-retry` 在 agent 的失败步骤 seam 上执行；pi-ai SDK 重试仍保持禁用，因此持久化的 agent 步骤与 `llm/retry` 事件记录每次可见尝试，直接 `ctx.llm.stream()` 调用仍只尝试一次。
diff --git a/packages/llm/llm-retry/README.i18n.yaml b/packages/llm/llm-retry/README.i18n.yaml
index 50ae377010..8a0ecf3434 100644
--- a/packages/llm/llm-retry/README.i18n.yaml
+++ b/packages/llm/llm-retry/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/llm/llm-retry/README.md
 README.md: 7a86652a794e70c4dfd00ab7427730387e3ec949
-README.zh.md: 6255cca8c3b669ddec401496acb1e79ad54a8b3e
+README.zh.md: c66c04806597c11b5c64dcb01443bb84f489e0f5
diff --git a/packages/llm/llm-retry/README.zh.md b/packages/llm/llm-retry/README.zh.md
index 6255cca8c3..c66c048065 100644
--- a/packages/llm/llm-retry/README.zh.md
+++ b/packages/llm/llm-retry/README.zh.md
@@ -4,13 +4,13 @@
 
 一个函数插件，通过 agent loop（智能体循环）在已关闭步骤上触发的 `agent/request-error` waterfall（瀑布式事件）应用确切提供方重试策略。它不包装 `ctx.llm.stream()`：每次适配器调用仍是一次提供方尝试，每次重试都会开启新的编号轮次。
 
-每个提供方适配器都拥有可选的嵌套 `retryPolicy`；路由在 `ctx.llm` 上注册时会捕获该策略，任何到达该注册最终适配器边界的调用都会携带它。如果之后释放或替换路由，进行中的失败仍会保留为其提供服务的策略；在选中任何最终适配器前发生的失败没有提供方策略，会继续委托。省略策略时使用 normal mode：为 `EMPTY_RESPONSE`、`RATE_LIMIT`、`SERVER`、`TIMEOUT` 和 `TRANSPORT` 重试两次，并采用从 500 ms 到 10 秒的有界指数退避与 10% jitter。`EMPTY_RESPONSE` 是适配器对未产生任何持久内容的退化提供方完成所作的分类，因此可安全重复。normal 策略可以更改其有限预算、合格 code 和退避配置。always mode 会先请求下游恢复，再无次数上限地重试每个模型请求失败；成功、取消或插件 dispose（资源释放）会在活跃的委托恢复完全停稳后终止它。
+每个提供方适配器都拥有可选的嵌套 `retryPolicy`；路由在 `ctx.llm` 上注册时会捕获该策略，任何到达该注册最终适配器边界的调用都会携带它。如果之后释放或替换路由，进行中的失败仍会保留当时为其提供服务的策略；在选中任何最终适配器前发生的失败没有提供方策略，会继续委托。省略策略时使用 normal mode：为 `EMPTY_RESPONSE`、`RATE_LIMIT`、`SERVER`、`TIMEOUT` 和 `TRANSPORT` 重试两次，并采用从 500 ms 到 10 秒的有界指数退避与 10% jitter。`EMPTY_RESPONSE` 是适配器对未产生任何持久内容的退化提供方完成所作的分类，因此可安全重复。normal 策略可以更改其有限预算、符合条件的 code 和退避配置。always mode 会先请求下游恢复，再无次数上限地重试每个模型请求失败；成功、取消或插件 dispose（资源释放）会在活跃的委托恢复完全停稳后终止它。
 
 两种 mode 都使用带对称 jitter 的有界指数退避。有效 `providerRetryAfterMs` 不超过 `maxDelayMs` 时会替换本地退避，并且不加 jitter。超出上限的提供方延迟会使 normal mode 继续委托；always mode 则改用已配置的本地退避，避免该指令终止重试。
 
-等待前，插件会追加一条不进入表层的 `llm/retry` 事件，其中包含提供方、mode、规范的解析策略 key、失败和计划延迟。该 key 包含所有影响行为的字段，并对 normal mode 的 code 排序，因为合格性采用集合成员关系判断。只有提供方与完整策略 key 都相同的事件才会延续重试编号；因此，用限制、code 成员关系或退避不同的路由替换后，会开始自己的历史。normal 事件包含有限上限；always 事件省略该上限，UI 会渲染 `∞`。等待结束后，监听器返回 `{ kind: 'retry' }`，循环关闭失败轮次，并在同一持久历史上开启重试轮次。取消与插件 dispose 会中止活跃退避，在应用中止前排空活跃的委托恢复，并使 dispose 前捕获的 callback 只能以失败结束。
+等待前，插件会追加一条不进入表层的 `llm/retry` 事件，其中包含提供方、mode、已解析策略的规范 key、失败和计划延迟。该 key 包含所有影响行为的字段，并对 normal mode 的 code 排序，因为合格性采用集合成员关系判断。只有提供方与完整策略 key 都相同的事件才会延续重试编号；因此，用限制、code 成员关系或退避不同的路由替换后，会开始自己的历史。normal 事件包含有限上限；always 事件省略该上限，UI 会渲染 `∞`。等待结束后，监听器返回 `{ kind: 'retry' }`，循环关闭失败轮次，并在同一持久历史上开启重试轮次。取消与插件 dispose 会中止活跃退避，在应用中止前排空活跃的委托恢复，并使 dispose 前捕获的 callback 只能以失败结束。
 
-单独发布的 `./invariant` 配套模块会检查每个重试记录是否指向当前开启轮次及其最新已关闭步骤，是否与失败请求的持久提供方匹配，是否携带非空的提供方与策略身份，是否满足 mode 特定边界，是否拥有唯一步骤记录和正确的提供方策略重试编号，以及是否携带有界定时器延迟。完整 jitter 可以在下界调度为零毫秒。
+单独发布的 `./invariant` 配套模块会检查每个重试记录是否指向当前开启轮次及其最新已关闭步骤，是否与失败请求的持久提供方匹配，是否携带非空的提供方与策略标识，是否满足 mode 特定边界，是否拥有唯一步骤记录和正确的提供方策略重试编号，以及是否携带有界定时器延迟。完整 jitter 可以在下界调度为零毫秒。
 
 ```yaml
 - name: '@deepseek-ai/dsh-llm-deepseek'
@@ -46,8 +46,8 @@
 
 ## 已知限制与暂缓事项
 
-- **Agent 轮次是唯一重试边界**：直接 `ctx.llm.stream()` 消费方仍只尝试一次，因为原始流无法将已发出 chunk 持久分隔为不同尝试。
+- **agent 轮次是唯一重试边界**：直接 `ctx.llm.stream()` 消费方仍只尝试一次，因为原始流无法持久地区分各次尝试已经发出的分片。
 - **always mode 会重试永久性失败**：身份验证、配额、无效请求、协议和无法恢复的上下文错误都会继续重试，直至成功、取消或 dispose；部署负责提供方特定的成本与延迟控制。
-- **有限插件预算可叠加**：normal mode 只统计已配置 code 和确切提供方策略，上下文溢出压缩则拥有独立预算。未来如有重叠策略，必须记录并测试注册顺序行为。
+- **有限插件预算可叠加**：normal mode 只统计已配置 code 和确切提供方策略，上下文溢出压缩（compaction）则拥有独立预算。未来如有重叠策略，必须记录并测试注册顺序行为。
 - **恢复策略按 waterfall 顺序组合**：always mode 会先接受下游重试，再应用自己的回退。后续策略如果忽略取消且永不结算，也会阻止回退、轮次完全停稳和插件 dispose 完成。
 - **`llm/retry` 记录调度，不是完成**：后续步骤与轮次事件用于确立成功、耗尽或取消。
diff --git a/packages/llm/llm/README.i18n.yaml b/packages/llm/llm/README.i18n.yaml
index a4716026cd..49ff6d7c48 100644
--- a/packages/llm/llm/README.i18n.yaml
+++ b/packages/llm/llm/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/llm/llm/README.md
 README.md: d343449d1530bf70a3a8c57f883894e29c42d18f
-README.zh.md: 6ac57b1e6010b58c45b516f13ec6361d47ca8d12
+README.zh.md: 4dc4a0ca06378116d05fdb4b9b048738930511fd
diff --git a/packages/llm/llm/README.zh.md b/packages/llm/llm/README.zh.md
index 6ac57b1e60..4dc4a0ca06 100644
--- a/packages/llm/llm/README.zh.md
+++ b/packages/llm/llm/README.zh.md
@@ -2,30 +2,30 @@
 
 [English](README.md) | 中文
 
-提供方无关的 LLM 词汇与抽象服务。本包定义 agent loop、会话日志和每个插件使用的规范语言。
+提供方无关的 LLM（大语言模型）词汇与抽象服务。本包（package）定义 agent loop（智能体循环）、会话日志和每个插件使用的规范语言。
 
 ## 服务：`LlmService`（ctx key：`llm`）
 
-一个适配器注册表加单一流式调用表层，可通过 waterfall 事件拦截。
+一个适配器注册表加单一流式调用接口，可通过 waterfall（瀑布式事件）拦截。
 
 ### 公开 API
 
-- `ctx.llm.registerAdapter(providers: string[], adapter: LlmAdapter): () => void` 为给定提供方路由注册一个适配器实例。注册要么全部成功，要么全部不生效，并且会随调用 fiber dispose。
+- `ctx.llm.registerAdapter(providers: string[], adapter: LlmAdapter): () => void` 为给定提供方路由注册一个适配器实例。注册要么全部成功，要么全部不生效，并且会随调用 fiber 一起 dispose（资源释放）。
 - `ctx.llm.listProviders(): LlmProviderInfo[]` 按注册顺序描述已注册提供方路由。
 - `ctx.llm.providerRetryPolicy(provider: string): ResolvedRetryPolicy` 返回注册时捕获的提供方重试策略，并解析 normal 默认值。
 - `ctx.llm.listModels(provider: string): Promise<LlmModelInfo[]>` 发现某个已注册提供方当前公布的模型。
 - `ctx.llm.resolveModelInfo(provider: string, model: string, signal?: AbortSignal): Promise<LlmResolvedModelInfo>` 从拥有精确路由的适配器解析经校验的确切模型身份、可用上下文和推理（reasoning）元数据；异步适配器可选地支持取消。
 - `ctx.llm.resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig>` 校验显式推理强度，并填入适配器配置的默认值，但不自动调整。
 - `ctx.llm.prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<PreparedLlmCall>` 解析配置并将其当前适配器注册捕获为一次可取消、一次性调用。
-- `ctx.llm.stream(options: GenerateOptions): AsyncIterable<StreamChunk>` 将一次模型调用流式输出为原始 chunk（token 级 delta）。消费方使用 `BlockAssembler` 将 chunk 组装为块／消息。
+- `ctx.llm.stream(options: GenerateOptions): AsyncIterable<StreamChunk>` 将一次模型调用流式输出为原始分片（token 级增量）。消费方使用 `BlockAssembler` 将分片组装为块／消息。
 
-`LlmService` 保留来自最终适配器选择、同步 dispatch、iterator 构造与迭代的错误，并将其溯源绑定到该次模型调用返回的精确流句柄。`isLlmAdapterFailure(stream, value)` 只报告该调用最终适配器边界的错误；`llmFailureOf(stream, value)` 返回相邻的不可变 `LlmFailure`；`llmRetryPolicyOf(stream)` 返回在该边界选中的确切注册所对应的不可变策略，即使之后释放或替换路由也不变。未到达最终适配器的调用没有服务策略。嵌套模型调用、`llm/stream` middleware 和下游消费方失败对外层调用仍未分类。分类绝不替换或更改适配器的原始编码 `Error`。
+`LlmService` 保留来自最终适配器选择、同步 dispatch、iterator 构造与迭代的错误，并将其溯源绑定到该次模型调用返回的精确流句柄。`isLlmAdapterFailure(stream, value)` 只报告该调用最终适配器边界的错误；`llmFailureOf(stream, value)` 返回关联的不可变 `LlmFailure`；`llmRetryPolicyOf(stream)` 返回在该边界选中的确切注册所对应的不可变策略，即使之后释放或替换路由也不变。未到达最终适配器的调用没有服务策略。嵌套模型调用、`llm/stream` middleware 和下游消费方失败对外层调用仍未分类。分类绝不替换或更改适配器原有的带代码 `Error`。
 
-提供方与模型元数据是发现表层，不是路由白名单。`registerAdapter()` 仍拥有提供方排他性，并为每条路由捕获适配器的重试策略；适配器则可以接受 `listModels()` 中不存在的模型 id，消费方禁止因模型未列出而拒绝请求。返回的 selector 元数据与输入脱离，无效或重复适配器配置项会以 `INVALID_ADAPTER` 或 `INVALID_CATALOG` 失败。
+提供方与模型元数据是发现接口，不是路由白名单。`registerAdapter()` 仍拥有提供方排他性，并为每条路由捕获适配器的重试策略；适配器则可以接受 `listModels()` 中不存在的模型 id，消费方禁止因模型未列出而拒绝请求。返回的 selector 元数据与输入脱离，无效或重复适配器配置项会以 `INVALID_ADAPTER` 或 `INVALID_CATALOG` 失败。
 
 确切模型元数据是独立的正确性查询，不是 catalog 装饰或全局 LLM 设置。`resolveModelInfo()` 会向拥有精确提供方／模型路由的适配器查询一次；适配器可以描述未列出的动态模型，缺少 `context` 或 `reasoning` 字段只表示相应能力不可用。无效的身份、上下文或推理元数据会以 `INVALID_MODEL_INFO`、`INVALID_MODEL_CONTEXT` 或 `INVALID_MODEL_REASONING` 失败。
 
-推理标识符是由适配器持有的不透明字符串，而非核心枚举。适配器会公布有序可选列表；模型能力 API 提供 `off` id 时，列表也会包含它。`resolveCallConfig()` 只接受与已公布标识符完全一致的值，在存在 `defaultEffort` 时填入它，否则保留提供方默认值。异步模型解析器会接收调用方的 signal，并且必须在取消后迅速完成结算。`prepareCall()` 还会让精确适配器注册跨越请求头记录和最终分派，因此 HMR（热模块替换）不会将一个适配器的能力结果与另一个适配器的请求混用；复用其一次性句柄或更改调用配置字段会以 `INVALID_PREPARED_CALL` 失败。不支持的显式或配置推理强度会在提供方 I/O 前以 `UNSUPPORTED_REASONING_EFFORT` 失败。
+推理标识符是由适配器持有的不透明字符串，而非核心枚举。适配器会公布有序可选列表；模型能力 API 提供 `off` id 时，列表也会包含它。`resolveCallConfig()` 只接受与已公布标识符完全一致的值，在存在 `defaultEffort` 时填入它，否则保留提供方默认值。异步模型解析器会接收调用方的 signal，并且必须在取消后迅速结束。`prepareCall()` 还会让精确适配器注册跨越请求头记录和最终分派，因此 HMR（热模块替换）不会将一个适配器的能力结果与另一个适配器的请求混用；复用其一次性句柄或更改调用配置字段会以 `INVALID_PREPARED_CALL` 失败。不支持的显式或配置推理强度会在提供方 I/O 前以 `UNSUPPORTED_REASONING_EFFORT` 失败。
 
 ### 事件
 
@@ -36,42 +36,42 @@
 ### 扩展点
 
 - 继承 `LlmAdapter` 并调用 `ctx.llm.registerAdapter(providers, adapter)`，添加一条或多条提供方路由。`GenerateOptions.provider` 选择适配器；`GenerateOptions.model` 属于适配器，可以动态解析。覆盖 `providerRetryPolicy()` 以提供由提供方持有的恢复配置，覆盖 `providerInfo()` 和异步 `listModels()` 以公开 selector 元数据；精确身份、容量或可选推理强度可用时，实现 `resolveModel()`；异步解析器必须响应其可选的取消 signal。默认实现使用有界的 normal 重试策略，将路由和模型 id 用作名称，不公布模型，也不返回容量或推理元数据。
-- 包装 `llm/stream` 时，通过 `ctx.on()` waterfall listener 实现缓存、日志或路由。发出 chunk 后重试的包装层没有持久尝试边界；因此已发布 agent 重试策略改用 `agent/request-error`。
+- 包装 `llm/stream` 时，通过 `ctx.on()` waterfall listener 实现缓存、日志或路由。发出分片后重试的包装层没有持久尝试边界；因此已发布 agent 重试策略改用 `agent/request-error`。
 
 ### 消息（`message.ts`）与内容块（`types.ts`）
 
 `Message` 是投递、持久历史和模型请求共享的不可变值。每条消息从创建起都必须具有 `MessageId`、角色、内容和带类型的来源。`createMessage(input)` 生成标识，并返回与输入分离且深度冻结的值；`createUserMessage({ content, source })` 固定 user 角色；`createAssistantMessage({ content, source })` 固定 assistant 角色与模型来源类别；`createToolResultMessage({ callId, content, isError })` 固定 user 角色，并将工具来源与其结果块耦合；`freezeMessage(message)` 导入已有标识，绝不将其替换。改写消息时会保留标识，并产生另一个冻结值。浏览器端代码会从依赖最少的 `@deepseek-ai/dsh-llm/message` 入口导入这些值构造函数，而不是从包含服务的包根入口导入。
 
-消息内容是类型化内容块数组：`text`、`reasoning`、`tool-call`、`tool-result`。联合从可合并扩展的 `ContentBlockMap` 派生，因此插件可以通过 declaration merging 添加块类型。assistant 消息使用模型来源，其中携带提供方／模型溯源与可选适配器私有回放状态。dispatch 前，`LlmService` 只在历史提供方路由与目标提供方路由当前由完全相同的适配器实例拥有时才保留该状态；随后由适配器判定能否在模型／提供方间恢复或转换该状态。核心块集只包含每条已发布路径都支持的块。多模态内容（图像、音频等）没有核心块类型；需要它的功能会通过 map 添加，并一并添加支持它的适配器／UI／压缩实现。
+消息内容是类型化内容块数组：`text`、`reasoning`、`tool-call`、`tool-result`。联合从可合并扩展的 `ContentBlockMap` 派生，因此插件可以通过 declaration merging 添加块类型。assistant 消息使用模型来源，其中携带提供方／模型溯源与可选适配器私有回放状态。dispatch 前，`LlmService` 只在历史提供方路由与目标提供方路由当前由完全相同的适配器实例拥有时才保留该状态；随后由适配器判定能否在模型／提供方间恢复或转换该状态。核心块集只包含每条已发布路径都支持的块。多模态内容（图像、音频等）没有核心块类型；需要它的功能会通过 map 添加，并一并添加相应的适配器／UI／压缩（compaction）支持。
 
-流式输出是原始 chunk 协议（`block-start`、`text-delta`、`reasoning-delta`、`tool-call-delta`、`block-end`、`usage`、`finish`）。`BlockAssembler` 是将 chunk 组装为块／消息的唯一共享实现。
+流式输出是原始分片协议（`block-start`、`text-delta`、`reasoning-delta`、`tool-call-delta`、`block-end`、`usage`、`finish`）。`BlockAssembler` 是将分片组装为块／消息的唯一共享实现。
 
 ### 调用配置（`call-config.ts`）
 
-`LlmCallConfig` 是一个会话请求的提供方、模型、可选的适配器持有推理强度和采样标量（`provider`、`model`、`reasoningEffort`、`temperature`、`maxTokens`、`stop`，每个都与同名 `GenerateOptions` 字段 1:1 映射）。它是作为请求标头一部分记录在会话日志中的每会话状态（见 dsh-session `request/header` 事件），绝不是可静默调整的每次调用旋钮：`agent/request` waterfall 会提议替换，`prepareCall()` 在轮次 signal 控制下校验并填入默认值，loop 随后记录生效值，再使用准备完成调用的注册绑定流。`callConfigEquals(a, b)` 是逐字段真实变更检测器；`deepFreeze(value)` 是 loop 在 dispatch 前对每个已构建请求应用的所有权 helper（`llm/stream` listener 与适配器只读，绝不改写）。`markAgentLoopRequest()` 为该精确对象添加进程本地 loop 溯源，`isAgentLoopRequest()` 让观测方可以将其与同样可能冻结并关联会话、但独立记录的辅助调用区分。`GenerateOptions.purpose` 对已记录辅助压缩与会话标题调用分类，让适配器可以应用目的特定传输策略，而不改变普通会话请求。
+`LlmCallConfig` 是一个会话中各次请求的提供方、模型、可选的适配器持有推理强度和采样标量（`provider`、`model`、`reasoningEffort`、`temperature`、`maxTokens`、`stop`，每个都与同名 `GenerateOptions` 字段 1:1 映射）。它是作为请求标头一部分记录在会话日志中的每会话状态（见 dsh-session `request/header` 事件），绝不是可静默调整的每次调用旋钮：`agent/request` waterfall 会提议替换，`prepareCall()` 在轮次 signal 控制下校验并填入默认值，loop 随后记录生效值，再使用已准备调用中与注册绑定的流。`callConfigEquals(a, b)` 是逐字段真实变更检测器；`deepFreeze(value)` 是 loop 在 dispatch 前对每个已构建请求应用的所有权 helper（`llm/stream` listener 与适配器只读，绝不改写）。`markAgentLoopRequest()` 为该精确对象添加进程本地 loop 溯源，`isAgentLoopRequest()` 让观测方可以将其与同样可能冻结并关联会话、但独立记录的辅助调用区分。`GenerateOptions.purpose` 对已记录辅助压缩与会话标题调用分类，让适配器可以应用目的特定传输策略，而不改变普通会话请求。
 
 ### 应用归因（`attribution.ts`）
 
-每个产品适配器都会在提供方 HTTP 请求上发送应用身份。`attributionHeaders(identity?)` 构建标准 `User-Agent`，默认为公开 `APP_IDENTITY`；白标部署可以替换它，但不能抑制它。适配器会直接验证 wire 标头，或通过自身库 hook 验证。详见 [归因 Agent Note](../../../.agents/notes/implemented/architecture/2026-06-21-mandatory-app-attribution-headers.md)。
+每个产品适配器都会在提供方 HTTP 请求上发送应用身份。`attributionHeaders(identity?)` 构建标准 `User-Agent`，默认为公开 `APP_IDENTITY`；白标部署可以替换它，但不能抑制它。适配器会直接验证 wire 标头，或通过自身库 hook 验证。详见 [归因 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-21-mandatory-app-attribution-headers.md)。
 
 ### 类
 
 - `LlmAdapter`：提供方适配器的抽象基类。唯一必需方法是 `stream()`。
-- `BlockAssembler`：将原始 chunk 逐步组装为完整内容块，并能据此创建带标识且冻结的 assistant 消息。agent loop 向它提供原始 chunk（同时记录以供回放），并读取已组装块以构建历史。
-- `HarnessError`：harness 错误分类体系的基类，包含稳定 `code` 字符串（与面向人的 `message` 不同）加 `cause` 链接。它位于所有其他包都导入的叶子包中，因此可以共享单一基类，无需新的依赖边。每包错误（`LlmError`、`ToolArgsError`、`InvariantError` 等）都会扩展它。`isHarnessError(value)` 在 seam 处收窄类型。
-- `LlmError`：扩展 `HarnessError`；其稳定 `code` 字符串（`NO_ADAPTER`、`DUPLICATE_ADAPTER` 与 `AUTH`／`RATE_LIMIT` 等适配器 code）与冻结可序列化 `failure.code` 匹配。Payload 还可以保留已验证状态、`Retry-After` 和品牌化提供方请求 id 事实；策略位于错误之外。
+- `BlockAssembler`：将原始分片逐步组装为完整内容块，并能据此创建带标识且冻结的 assistant 消息。agent loop 向它提供原始分片（同时记录以供回放），并读取已组装块以构建历史。
+- `HarnessError`：harness 错误分类体系的基类，包含稳定 `code` 字符串（与面向人的 `message` 不同）加 `cause` 链接。它位于所有其他包都从中导入的叶子包中，因此可以共享单一基类，无需新的依赖边。各包的错误（`LlmError`、`ToolArgsError`、`InvariantError` 等）都继承自它。`isHarnessError(value)` 在 seam 处收窄类型。
+- `LlmError`：继承自 `HarnessError`；其稳定 `code` 字符串（`NO_ADAPTER`、`DUPLICATE_ADAPTER` 与 `AUTH`／`RATE_LIMIT` 等适配器 code）与冻结可序列化 `failure.code` 匹配。Payload 还可以保留已验证状态、`Retry-After` 和品牌化提供方请求 id 事实；策略位于错误之外。
 - `errorChain(value)`：渲染抛出值的完整 `cause` 链与 AggregateError 成员，供诊断表层使用，包括 UI 通知、logger 行和持久 `turn/end` 消息。因此 undici 的 `TypeError: fetch failed` 等传输包装层会显示底层 `ECONNREFUSED`／DNS／TLS 详细信息，而不是将其遮蔽。该函数只负责渲染：请按 `code` 路由，绝不解析结果。
-- `CONTEXT_WINDOW_EXCEEDED_CODE`：当请求超过模型上下文窗口时，无论通过抛出 HTTP 还是带内 finish 交付，两个 DeepSeek 适配器都使用的提供方无关 code。`isContextWindowExceededError(detail)` 是它们针对 OpenAI 兼容提供方详细信息的共享保守分类器。
+- `CONTEXT_WINDOW_EXCEEDED_CODE`：当请求超过模型上下文窗口时，无论通过 HTTP 异常抛出还是带内 finish 交付，两个 DeepSeek 适配器都使用的提供方无关 code。`isContextWindowExceededError(detail)` 是它们针对 OpenAI 兼容提供方详细信息的共享保守分类器。
 - `QUOTA_EXCEEDED_CODE`：帐户配额、余额、点数、预算或用量限制耗尽时使用的非短暂提供方无关 code。`isQuotaExceededError(detail)` 使这些失败与请求速率限制保持区分。
-- `EMPTY_RESPONSE_CODE`：对退化提供方完成使用的提供方无关 code，两个适配器均使用：一个不携带任何内容块的终止 `stop`。它会被分类为错误 finish（而非成功空消息），因为尝试未产生持久内容；`dsh-llm-retry` 默认重试它。
+- `EMPTY_RESPONSE_CODE`：两个适配器都使用的提供方无关 code，用于表示退化的提供方生成结果：一个未携带任何内容块的终止 `stop`。它会被分类为错误 finish（而非成功空消息），因为尝试未产生持久内容；`dsh-llm-retry` 默认重试它。
 
 ### 真实适配器
 
-两个适配器使用不同内部机制实现 `LlmAdapter`：[`@deepseek-ai/dsh-llm-deepseek`](../llm-deepseek) 针对 `deepseek` 路由使用直接 fetch 加 `eventsource-parser` SSE 分帧，[`@deepseek-ai/dsh-llm-pi-ai`](../llm-pi-ai) 则通过 `@earendil-works/pi-ai` 动态解析已配置提供方／模型对。两者都遵循 `StreamChunk` 约定，定义见 `types.ts`：usage 先于 finish，工具参数保持原始字符串，错误使用两种已批准路径之一。设计理由见 [双 LLM 适配器](../../../.agents/notes/implemented/architecture/2026-06-13-twin-llm-adapters.md)。
+两个适配器使用不同内部机制实现 `LlmAdapter`：[`@deepseek-ai/dsh-llm-deepseek`](../llm-deepseek) 针对 `deepseek` 路由使用直接 fetch 加 `eventsource-parser` SSE（Server-Sent Events）分帧，[`@deepseek-ai/dsh-llm-pi-ai`](../llm-pi-ai) 则通过 `@earendil-works/pi-ai` 动态解析已配置提供方／模型对。两者都遵循 `StreamChunk` 约定，定义见 `types.ts`：usage 先于 finish，工具参数保持原始字符串，错误使用两种已批准路径之一。设计理由见 [双 LLM 适配器](../../../.agents/notes/implemented/architecture/2026-06-13-twin-llm-adapters.md)。
 
 ## 模型体验
 
-无。服务不添加或更改任何模型边界文本、schema 或消息；它只会填入并记录适配器配置的推理强度。
+无。服务不添加任何与模型绑定的文本、schema 或消息；它只会填入并记录适配器配置的推理强度。
 
 #### KV Cache 影响
 
@@ -79,9 +79,9 @@
 
 ## 已知限制与暂缓事项
 
-- **本服务不执行重试、缓存或速率限制**：提供方注册会存储重试策略，但 `llm/stream` 仍是单次尝试调用包装 seam。agent loop 会将已验证模型请求失败单独提供给 `agent/request-error`，其默认行为是保留原始失败；`@deepseek-ai/dsh-llm-retry` 是共享示例 spine 加载的可选执行器。
+- **本服务不执行重试、缓存或速率限制**：提供方注册会存储重试策略，但 `llm/stream` 仍是单次尝试调用包装 seam。agent loop 会将已验证模型请求失败单独提供给 `agent/request-error`，其默认行为是保留原始失败；`@deepseek-ai/dsh-llm-retry` 是共享示例主干加载的可选执行器。
 - **`GenerateOptions` 采样只包含 `temperature`／`maxTokens`／`stop`**：没有 `tool_choice`、`top_p` 或 penalty 字段；有产生方落地时词汇才会增长（见 [已删除惰性旋钮](../../../.agents/notes/archived/simplification/2026-07-04-drop-inert-request-knobs.md)）。
-- **由产生方调节的变体在实际产生前保持在外**：`prefill`、每工具 `strict`、块 `cache` 提示与 `agent` 消息源变体因没有产生方而被剪除（见 [Agent Note](../../../.agents/notes/archived/simplification/2026-07-04-prune-producerless-vocabulary-variants.md)）。
-- **`BlockAssembler` 只处理核心块 kind**：如果插件添加块类型的流从未由 `block-end` 关闭，`blocks()` 会抛出异常。
+- **受产生方约束的变体在实际产生前不会加入**：`prefill`、每工具 `strict`、块 `cache` 提示与 `agent` 消息源变体因没有产生方而被剪除（见 [Agent Note](../../../.agents/notes/archived/simplification/2026-07-04-prune-producerless-vocabulary-variants.md)）。
+- **`BlockAssembler` 只处理核心块类型**：如果插件添加块类型的流从未由 `block-end` 关闭，`blocks()` 会抛出异常。
 - **`APP_IDENTITY.url` 指向一个尚不存在的仓库**：`FIXME`：创建公开 `deepseek-ai/deepseek-harness-sdk` 仓库是首次发布的前置条件。
 - **`GenerateOptions.sessionId` 是本地声明的品牌类型**：导入 dsh-session 的 `SessionId` 会产生循环；未来拥有 id 的包可以消除该权宜之计。
diff --git a/packages/llm/token-meter/README.i18n.yaml b/packages/llm/token-meter/README.i18n.yaml
index 702406747b..a4ad4aef87 100644
--- a/packages/llm/token-meter/README.i18n.yaml
+++ b/packages/llm/token-meter/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/llm/token-meter/README.md
 README.md: 578728ded9cf51a12abcd70d541404e995028f26
-README.zh.md: 51518e98c43e740822970c1a39154f550ea962dc
+README.zh.md: 9d54ddb4792e6c897af7d57a6f8ae98204caa11d
diff --git a/packages/llm/token-meter/README.zh.md b/packages/llm/token-meter/README.zh.md
index 51518e98c4..9d54ddb479 100644
--- a/packages/llm/token-meter/README.zh.md
+++ b/packages/llm/token-meter/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-通过单例 `ctx.tokenMeter` 服务进行感知回放的 token 测量。它从持久日志为每个会话推进一个隔离 fold，因此压缩与其他压力敏感插件可以共享计量，无需依赖 `CompactService`。
+通过单例 `ctx.tokenMeter` 服务进行具备回放感知能力的 token 测量。它从持久日志为每个会话推进一个隔离 fold，因此压缩（compaction）与其他压力敏感插件可以共享计量，无需依赖 `CompactService`。
 
 ## 配置
 
-估算器没有设置。它有意使用一项固定启发式规则：每个 token 按四个字符估算，再加上角色、块与请求 envelope 字段的结构开销。任何 key 都会被拒绝，包括已废弃的全局 `contextWindow`；模型容量属于拥有精确提供方／模型路由的适配器，可通过 `ctx.llm.resolveModelInfo().context` 获取。
+估算器没有配置项。它有意使用一项固定启发式规则：每个 token 按四个字符估算，再加上角色、块与请求 envelope 字段的结构开销。任何配置键都会被拒绝，包括已废弃的全局 `contextWindow`；模型容量属于拥有精确提供方／模型路由的适配器，可通过 `ctx.llm.resolveModelInfo().context` 获取。
 
 ## 测量契约
 
@@ -15,11 +15,11 @@
 - `measure(session, requestHeader?)` 在同一个已消费日志 revision 上返回请求压力与当前已计价表层。
 - `estimateMessage(message)` 使用固定启发式规则为一条消息计价。
 
-`measure()` 会同步一次，返回一个与输入脱离、深度不可变的快照。`totalTokens` 是请求与响应压力，`surfaceTokens` 是仅表层启发式总量，等于 `nodes[].tokens` 之和。`requestHeader` 覆盖只影响压力字段；表层字段仍描述当前会话。每次调用都会克隆带位置的节点，因此测量是 O(surface)。
+`measure()` 会同步一次，并返回一个独立且深度不可变的快照。`totalTokens` 是请求与响应压力，`surfaceTokens` 是仅表层启发式总量，等于 `nodes[].tokens` 之和。`requestHeader` 覆盖只影响压力字段；表层字段仍描述当前会话。每次调用都会克隆带位置的节点，因此测量是 O(surface)。
 
-fold 跟踪完整请求标头快照、步骤边界、表层追加与替换、成功 assistant 消息、提供方用量和 assistant chunk 溯源。只有当最新成功调用的规范请求 envelope 与已测量 envelope 匹配，且其总量不低于该调用的完整启发式锚点时，才会复用提供方用量；后续成功会替换较早锚点。否则估算完整当前 envelope 与表层。表层变更保持相对于匹配锚点的带符号值，包括缩减替换后的负 delta。
+fold 跟踪完整请求标头快照、步骤边界、表层追加与替换、成功 assistant 消息、提供方用量和 assistant 分片溯源。只有当最新成功调用的规范请求 envelope 与已测量 envelope 匹配，且其总量不低于该调用的完整启发式锚点时，才会复用提供方用量；后续成功会替换较早锚点。否则会对当前 envelope 与表层进行完整估算。表层变更保持相对于匹配锚点的带符号值，包括缩减替换后的负 delta。
 
-用量计量会求和不重叠的输入、cache-read、cache-write 与输出 bucket；不会再次添加 reasoning。每次成功调用都会记录一个 assistant 锚点，包括无内容调用。显式空溯源列表表示已知空提供方流，而缺失的遗留溯源会保守地将持久 assistant 输出视为提供方输出。
+用量计量会求和不重叠的输入、cache-read、cache-write 与输出 bucket；不会再次添加推理（reasoning）。每次成功调用都会记录一个 assistant 锚点，包括无内容调用。显式空溯源列表表示已知空提供方流，而遗留溯源缺失时，fold 会保守地将持久 assistant 输出视为提供方输出。
 
 ## 组合
 
@@ -28,7 +28,7 @@ fold 跟踪完整请求标头快照、步骤边界、表层追加与替换、成
 - name: '@deepseek-ai/dsh-compact-basic'
 ```
 
-两个插件都有可用默认值。meter 保持与模型路由和可选压缩无关。部署会在 LLM 适配器上配置容量，并在 `dsh-compact-basic` 上配置压缩策略。
+两个插件都有可用默认值。meter 保持与模型路由和可选压缩无关。部署会在 LLM（大语言模型）适配器上配置容量，并在 `dsh-compact-basic` 上配置压缩策略。
 
 ## 模型体验
 
@@ -36,11 +36,11 @@ fold 跟踪完整请求标头快照、步骤边界、表层追加与替换、成
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由具名消费方负责。
+不会直接失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
 - **固定启发式规则是近似值**：没有可复用提供方用量的内容按字符数加结构开销计价，而不是使用精确提供方 tokenizer 或请求 serializer。
-- **每次测量都会克隆当前表层**：连贯不可变快照使读取成为 O(surface)，包括低于阈值的压力检查。
+- **每次测量都会克隆当前表层**：一致且不可变的快照使读取成为 O(surface)，包括低于阈值的压力检查。
 - **提供方用量只能为完全相同的规范 envelope 复用**：提示词、前缀、工具、提供方、模型或调用配置变更都会有意回退到完整启发式估算。
-- **遗留溯源采取保守策略**：没有 `sourceEventSeqs` 的 assistant 消息无法区分提供方输出与 listener 改写，因此 fold 不会声称已知空流或精确 chunk 流。
+- **遗留溯源采取保守策略**：没有 `sourceEventSeqs` 的 assistant 消息无法区分提供方输出与 listener 改写，因此 fold 不会声称已知空流或精确分片流。
diff --git a/packages/lsp/README.i18n.yaml b/packages/lsp/README.i18n.yaml
index 96991875e8..d27247d06d 100644
--- a/packages/lsp/README.i18n.yaml
+++ b/packages/lsp/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/lsp/README.md
 README.md: 7b5d9e0f50e733113539cf1ea1ed72e8651ad94c
-README.zh.md: 9b4dbb9f3ab40cf631a40d0c6d5c654689cf7061
+README.zh.md: 0d068b2a78a11faa2d138d9c46d0b6bd705f8ea4
diff --git a/packages/lsp/README.zh.md b/packages/lsp/README.zh.md
index 9b4dbb9f3a..0d068b2a78 100644
--- a/packages/lsp/README.zh.md
+++ b/packages/lsp/README.zh.md
@@ -2,14 +2,14 @@
 
 [English](README.md) | 中文
 
-语言服务器能力 seam：抽象 LSP 接口、通用 stdio 提供方，以及面向模型的 `lsp` 工具。这些全是**产品** 包。
+语言服务器能力 seam：抽象 LSP 接口、通用 stdio 提供方，以及面向模型的 `lsp` 工具。这些全是**产品**包。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
 | `lsp/` | 抽象 LSP seam（按品牌化 id + 扩展名映射组织的提供方注册表、逐查询选择、词汇、`LspError`） | `ctx.lsp` |
-| `lsp-local/` | 通用多服务器本地后端（spawn、JSON-RPC、临时打开查询） | （在 `ctx.lsp` 上注册提供方） |
+| `lsp-local/` | 通用多服务器本地后端（spawn、JSON-RPC、查询时临时打开文档） | （在 `ctx.lsp` 上注册提供方） |
 | `tool-lsp/` | 面向模型的 `lsp` 工具（四种操作、从 1 开始的 UTF-16 光标坐标） | （注册到 `ctx.tools`） |
 
-接口位于 `lsp/lsp/`。该 seam 恰好公开四种语义操作：`goToDefinition`、`findReferences`、`goToImplementation`、`hover`，且不提供通用 JSON-RPC 逃生口；因此，替换提供方不会改变模型请求导航的方式，也不会让协议载荷或未经评审的修改进入模型契约。提供方注册的是**能力** 而非工具；`tool-lsp` 是面向模型名称、schema、提示词指引和呈现的唯一 owner。
+接口位于 `lsp/lsp/`。该 seam 恰好公开四种语义操作：`goToDefinition`、`findReferences`、`goToImplementation`、`hover`，且不提供通用 JSON-RPC 逃生口；因此，替换提供方不会改变模型请求导航的方式，也不会让协议载荷或未经评审的修改进入模型契约。提供方注册的是**能力**而非工具；`tool-lsp` 是面向模型名称、schema、提示词指引和呈现的唯一 owner。
 
-设计原理见 [LSP 能力 seam Agent Note](../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)，其中也解释了文档为何在每次查询时临时打开、本地主机为何通过 Node API 而非 `ctx.fs` 读取，以及扩展名归属为何在同一运行时内互斥。
+设计原理见 [LSP 能力 seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)，其中也解释了文档为何在每次查询时临时打开、本地主机为何通过 Node API 而非 `ctx.fs` 读取，以及扩展名归属为何在同一运行时内互斥。
diff --git a/packages/lsp/lsp-local/README.i18n.yaml b/packages/lsp/lsp-local/README.i18n.yaml
index a459427a78..ffc3bf1d27 100644
--- a/packages/lsp/lsp-local/README.i18n.yaml
+++ b/packages/lsp/lsp-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 85254eea2bb74df277be6fd5de1529b5da4ea178
-README.zh.md: 290601d455ba2012d0c1f5c97503ae87304881cb
+#   pnpm run verify-translation-pairing --write packages/lsp/lsp-local/README.md
+README.md: 40515ad173dec8cfe6a0937525612b7f18d22fb8
+README.zh.md: f8b17957dbd0d239aa2c0b0aeadf6136427788aa
diff --git a/packages/lsp/lsp-local/README.md b/packages/lsp/lsp-local/README.md
index 85254eea2b..40515ad173 100644
--- a/packages/lsp/lsp-local/README.md
+++ b/packages/lsp/lsp-local/README.md
@@ -23,7 +23,7 @@ The `servers` record key is the stable provider id reserved on `ctx.lsp`; each v
 |---|---|---|
 | `command` | (required) | Executable to spawn — absolute, or resolved on the child PATH at load. Launch uses no shell. |
 | `args` | `[]` | Arguments passed to the executable. |
-| `env` | `{}` | Extra env merged on top of the credential-scrubbed ambient env (vars matching `KEY`/`SECRET`/`TOKEN` are not forwarded); an explicit `DSH_*` entry merges after the seam's scrub of ambient ones. |
+| `env` | `{}` | Extra env merged on top of the credential-scrubbed ambient env (vars matching `KEY`/`PASSWORD`/`SECRET`/`TOKEN` are not forwarded); an explicit `DSH_*` entry merges after the seam's scrub of ambient ones. |
 | `extensionToLanguage` | (required) | Lowercase leading-dot extension → LSP language id (e.g. `{ '.ts': 'typescript' }`). |
 | `initializationOptions` | `null` | Static `initialize` options forwarded to the server. |
 | `configuration` | `null` | Static answer to every `workspace/configuration` item. |
diff --git a/packages/lsp/lsp-local/README.zh.md b/packages/lsp/lsp-local/README.zh.md
index 290601d455..f8b17957db 100644
--- a/packages/lsp/lsp-local/README.zh.md
+++ b/packages/lsp/lsp-local/README.zh.md
@@ -2,17 +2,17 @@
 
 [English](README.md) | 中文
 
-`ctx.lsp` 的**通用本地 stdio 语言服务器后端**。一个插件实例接受一张命名服务器表，并逐配置项注册一个隔离的提供方。这是通用主机，而不是语言服务器目录或安装器：部署需要显式配置命令与映射，preset 应放在 `cordis.yml` overlay 中。
+`ctx.lsp` 的**通用本地 stdio 语言服务器后端**。一个插件实例接受一张命名服务器表，并逐配置项注册一个隔离的提供方。这是通用主机，而不是语言服务器目录或安装器：部署需要显式配置命令与映射，预设应放在 `cordis.yml` overlay 中。
 
 Namespace 插件（`name`／`inject`／`Config`／`apply`，无默认导出）。
 
 ## 功能
 
 - 在注册前解析每项服务器局部设置；无效映射或注册冲突会回滚较早配置项，因此加载失败不会留下提供方路由。
-- 每个 `(server id, canonical workspace realpath)` 惰性 single-flight 一个服务器进程。存活服务器错误不会回放；如果选中的池化传输在只读查询之前或期间失败，提供方会等待其释放，并在新进程上重试该查询一次。
-- 每次查询都使用兼容性优先的**临时打开** 序列：通过 Node API 规范化并读取源文件、`textDocument/didOpen`（版本 1、完整文本）、所请求操作，然后执行 `textDocument/didClose`，该操作位于 `finally` 中。写入 `didOpen` 失败或取消时，会先终止实例再允许池复用。文档在每次调用后关闭，因此第一版不需要 `didChange`、内容 cache 或文档 LRU。
+- 每个 `(server id, canonical workspace realpath)` 惰性 single-flight 一个服务器进程。存活服务器错误不会回放；如果选中的池化传输在只读查询之前或期间失败，提供方会等待其 dispose（资源释放）完成，并在新进程上重试该查询一次。
+- 每次查询都使用兼容性优先的**临时打开**序列：通过 Node API 规范化并读取源文件、`textDocument/didOpen`（版本 1、完整文本）、所请求操作，然后执行 `textDocument/didClose`，该操作位于 `finally` 中。写入 `didOpen` 失败或取消时，会在池复用该实例前将其终止。文档在每次调用后关闭，因此第一版不需要 `didChange`、内容 cache 或文档 LRU。
 - 通过一条逐 Workspace、可中止的队列，串行执行每个源读取／打开／查询／关闭生命周期，因此排队调用只会在轮到自身时读取当前源；不同 Workspace 并行运行。
-- 协议 shutdown 失败后，经由进程管理器 seam 终止服务器后代树（POSIX 进程组信号；Windows `taskkill /T /F`）。树终止的投递结果与所有进程组信号一样被就地吸收，不向外抛出（投递与服务器退出存在竞态）；服务器是否完全停稳，由句柄的进程树存活等待确认，而非由这次终止自身的结果确认。
+- 协议 shutdown 失败后，经由子进程 seam 终止服务器后代树（POSIX 进程组信号；Windows `taskkill /T /F`）。树终止的投递结果与所有进程组信号一样被就地吸收，不向外抛出（投递与服务器退出存在竞态）；服务器是否完全停稳，由句柄的进程树存活等待确认，而非由这次终止自身的结果确认。
 - 通过子进程 host namespace 中的 Node 文件系统 API 读取源文件，绝不使用 `ctx.fs`，也不发出 `fs/observed`：只有 LSP 结果对模型可见，因此查询不满足先读后写策略。
 
 ## 配置
@@ -23,13 +23,13 @@ Namespace 插件（`name`／`inject`／`Config`／`apply`，无默认导出）
 |---|---|---|
 | `command` | （必填） | 要 spawn 的可执行文件：绝对路径，或在加载时从子进程 PATH 解析。不使用 shell 启动。 |
 | `args` | `[]` | 传给可执行文件的参数。 |
-| `env` | `{}` | 合并到已清理 credential 的环境之上的额外 env（匹配 `KEY`／`SECRET`／`TOKEN` 的变量不会转发）；显式 `DSH_*` 条目在 seam 清除环境中同名值之后合并。 |
+| `env` | `{}` | 合并到已清理 credential 的环境之上的额外 env（匹配 `KEY`／`PASSWORD`／`SECRET`／`TOKEN` 的变量不会转发）；显式 `DSH_*` 条目在 seam 清除环境中同名值之后合并。 |
 | `extensionToLanguage` | （必填） | 小写、以点开头的扩展名 → LSP language id（例如 `{ '.ts': 'typescript' }`）。 |
 | `initializationOptions` | `null` | 转发给服务器的静态 `initialize` 选项。 |
 | `configuration` | `null` | 每个 `workspace/configuration` 配置项的静态答案。 |
 | `maxMessageBytes` | `16000000` | 从服务器接受的单条 framed 消息最大大小。 |
 | `maxStderrBytes` | `1000000` | 为诊断保留的 stderr 尾部最大大小。 |
-| `maxDocumentBytes` | `4000000` | 该主机可打开的最大源文件。 |
+| `maxDocumentBytes` | `4000000` | 该主机可打开的源文件大小上限。 |
 | `shutdownTimeoutMs` | `5000` | 升级前用于优雅 `shutdown`／`exit` 的预算。 |
 | `killGraceMs` | `2000` | 请求取消及 SIGTERM→SIGKILL 升级的宽限期。 |
 
@@ -37,11 +37,11 @@ Namespace 插件（`name`／`inject`／`Config`／`apply`，无默认导出）
 
 ## 协议行为
 
-初始化会声明 `general.positionEncodings: ['utf-16']`、`workspace: { workspaceFolders: true, configuration: true }`、`textDocument.hover.contentFormat: ['markdown', 'plaintext']`，以及定义与实现使用的 `linkSupport: true`，且不进行动态注册。服务器返回的能力具有最终决定权：不受支持的操作，或缺少临时打开／关闭的同步方式，会使查询失败。服务器省略 `positionEncoding` 时默认为 `utf-16`；其他值都属于协议错误。客户端通过静态配置回答 `workspace/configuration`，接受生命周期记账请求，并拒绝 `workspace/applyEdit`：它绝不应用编辑或运行命令。导航直接映射 `Location`，并从 `LocationLink` 的 `targetUri` + `targetSelectionRange` 映射；hover 规范化会取得有效的 `MarkupContent.value`，保留 string `MarkedString`，把带 language tag 的值渲染为围栏代码，并用一个空行连接数组。缺失结果、格式错误的范围或位置，以及格式错误的 hover 编码，都会作为结构化 `LSP_MALFORMED_RESPONSE` 错误失败。
+初始化会声明 `general.positionEncodings: ['utf-16']`、`workspace: { workspaceFolders: true, configuration: true }`、`textDocument.hover.contentFormat: ['markdown', 'plaintext']`，以及定义与实现使用的 `linkSupport: true`，且不进行动态注册。服务器返回的能力具有最终决定权：不受支持的操作，或缺少临时打开／关闭的同步方式，会使查询失败。服务器省略 `positionEncoding` 时默认为 `utf-16`；其他值都属于协议错误。客户端通过静态配置回答 `workspace/configuration`，接受生命周期记账请求，并拒绝 `workspace/applyEdit`：它绝不应用编辑或运行命令。导航直接映射 `Location`，并从 `LocationLink` 的 `targetUri` + `targetSelectionRange` 映射；hover 规范化会取得有效的 `MarkupContent.value`，保留 string `MarkedString`，把带 language tag 的值渲染为围栏代码，并用一个空行连接数组。缺失结果、格式错误的范围或位置，以及格式错误的 hover 编码，都会以结构化 `LSP_MALFORMED_RESPONSE` 错误的形式失败。
 
 ## 安全边界
 
-提供方信任其配置的服务器，不声明任何沙箱限制。它通过 Node API 规范化并读取源文件，拒绝缺失、非普通文件、非 UTF-8、过大，或规范路径位于规范 Workspace 外部的源文件（符号链接别名共享一个实例）。结果位置可以在外部，但外部路径不能成为查询源。因此，第一版要求可信的主机本地部署；受限、远程或虚拟 Workspace 需要另一个提供方。
+提供方信任其配置的服务器，不提供任何沙箱隔离。它通过 Node API 规范化并读取源文件，拒绝缺失、非普通文件、非 UTF-8、过大，或规范路径位于规范 Workspace 外部的源文件（符号链接别名共享一个实例）。结果位置可以在外部，但外部路径不能成为查询源。因此，第一版要求可信的主机本地部署；受限、远程或虚拟 Workspace 需要另一个提供方。
 
 ## 模型体验
 
@@ -53,6 +53,6 @@ Namespace 插件（`name`／`inject`／`Config`／`apply`，无默认导出）
 
 ## 已知限制与暂缓事项
 
-- **仅限可信主机本地环境**：没有沙箱限制，也没有私有 cache／temp 写入契约；支持不受信任 binary 或受限／远程／虚拟 Workspace，需要后续的进程／文件系统契约及不同提供方（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)）。限制逻辑先解析 `realpath`，再通过一个带 `O_NOFOLLOW | O_NONBLOCK` 的 handle 打开源文件（最终组件符号链接防护，并以非阻塞方式拒绝 FIFO），同时进行有界读取；并发修改方如果在解析与打开之间把*祖先*目录替换为符号链接，会造成残余 TOCTOU。在该可信部署模型下接受此风险，不使用不可移植的 `openat` 逐 segment 遍历来封闭。
+- **仅限可信主机本地环境**：没有沙箱隔离，也没有私有 cache／temp 写入契约；支持不受信任 binary 或受限／远程／虚拟 Workspace，需要后续的进程／文件系统契约及不同提供方（见 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)）。限制逻辑先解析 `realpath`，再通过一个带 `O_NOFOLLOW | O_NONBLOCK` 的 handle 打开源文件（最终组件符号链接防护，并以非阻塞方式拒绝 FIFO），同时进行有界读取；并发修改方如果在解析与打开之间把*祖先*目录替换为符号链接，会造成残余 TOCTOU。在该可信部署模型下接受此风险，不使用不可移植的 `openat` 逐 segment 遍历来封闭。
 - **临时打开兼容性下限**：同步能力省略打开／关闭（或声明 `None`）的服务器不受支持，即使关闭文档查询能够工作；固定的 TypeScript e2e 只建立一项兼容性下限，不代表跨语言承诺。
-- **逐服务器／Workspace 串行化延迟**：共享同一个服务器与 Workspace 的并行 agent 会在一个进程后排队；长生命周期 Workspace 进程会占用内存直到释放。
+- **逐服务器／Workspace 串行化延迟**：共享同一个服务器与 Workspace 的并行 agent（智能体）会在一个进程后排队；长生命周期 Workspace 进程会占用内存直到 dispose。
diff --git a/packages/lsp/lsp/README.i18n.yaml b/packages/lsp/lsp/README.i18n.yaml
index e63252c7f8..0f4dc26e63 100644
--- a/packages/lsp/lsp/README.i18n.yaml
+++ b/packages/lsp/lsp/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/lsp/lsp/README.md
 README.md: f96fc67ec8cb95f423eff9b312b7b591ec9d3008
-README.zh.md: 13ae9700e284ff238147538a571622066efc5747
+README.zh.md: d9404f09396ed5fe7d8cccb13bc634c04785e64b
diff --git a/packages/lsp/lsp/README.zh.md b/packages/lsp/lsp/README.zh.md
index 13ae9700e2..d9404f0939 100644
--- a/packages/lsp/lsp/README.zh.md
+++ b/packages/lsp/lsp/README.zh.md
@@ -4,11 +4,11 @@
 
 **LSP 能力 seam**：抽象 `LspService`（`ctx.lsp`）定义 harness 具备哪些语义代码导航能力（转到定义、查找引用、查找实现、悬停），并通过语言服务器提供方实现，不把模型契约绑定到本地子进程。
 
-该包是 LSP 能力中负责接口的三分之一：
+该包（package）是 LSP 能力中负责接口的三分之一：
 
 | 包 | 职责 |
 |---|---|
-| `@deepseek-ai/dsh-lsp`（本包） | 接口：服务、以品牌化 id + 扩展名映射为 key 的提供方注册表、逐查询选择、请求／结果词汇、`LspError` 分类体系 |
+| `@deepseek-ai/dsh-lsp`（本包） | 接口：服务、以带品牌类型的 id 和扩展名映射为键的提供方注册表、逐查询选择、请求／结果词汇、`LspError` 分类体系 |
 | `@deepseek-ai/dsh-lsp-local` | 通用本地后端，注册已配置的 stdio 语言服务器提供方 |
 | `@deepseek-ai/dsh-tool-lsp` | 面向模型的 `lsp` 工具，基于 `ctx.lsp` |
 
@@ -18,16 +18,16 @@
 
 | 成员 | 语义 |
 |---|---|
-| `registerProvider(provider)` | 注册后端，以原子方式保留其品牌化 `id` 与每个规范化文件扩展名。任何无效输入或冲突都不会发布内容，并抛出 `LspError`（`LSP_INVALID_PROVIDER`／`LSP_CONFLICT`）。返回释放所有保留项的 disposer。随调用 fiber 释放。 |
+| `registerProvider(provider)` | 注册后端，以原子方式保留其带品牌类型的 `id` 与每个规范化文件扩展名。任何无效输入或冲突都不会发布内容，并抛出 `LspError`（`LSP_INVALID_PROVIDER`／`LSP_CONFLICT`）。返回释放所有保留项的 disposer。随调用 fiber 一同 dispose（资源释放）。 |
 | `query(request, signal?)` | 按文件最终扩展名选择提供方，从该提供方的映射派生 `languageId`，并运行一次查询。没有匹配项时抛出 `LspError` `LSP_UNAVAILABLE`。 |
 
-选择逐查询进行且与顺序无关：一个提供方独占一组扩展名，因此注册和 HMR 顺序绝不会改变路由。扩展名 key 规范化为小写且以点开头；`languageId` 只用于同步临时文档，绝不参与选择。第一版没有 glob、language-id 或显式路由 selector。
+选择逐查询进行且与顺序无关：一个提供方独占一组扩展名，因此注册和 HMR（热模块替换）顺序绝不会改变路由。扩展名 key 规范化为小写且以点开头；`languageId` 只用于同步临时文档，绝不参与选择。第一版没有 glob、language-id 或显式路由 selector。
 
-提供方注册的是**能力** 而非工具。`dsh-tool-lsp` 是面向模型名称、描述、提示词指引、schema 和呈现的唯一 owner。
+提供方注册的是**能力**，而非工具。`dsh-tool-lsp` 是面向模型名称、描述、提示词指引、schema 和呈现的唯一 owner。
 
 ## 词汇
 
-`LspQueryRequest`（`operation`、`filePath`、`position`、`workspaceRoot`）：每个字段都必填，因此没有字段需要实现默认值，也不存在 `resolve()` 步骤。位置与范围使用从零开始的 UTF-16，与协议一致；工具拥有从 1 开始的光标约定。`findReferences` 始终包含声明，提供方在内部强制执行，因此调用方没有 flag。`LspQueryResult` 是封闭的判别联合：导航使用 `{ kind: 'locations'; locations; resolvedWorkspaceRoot }`，悬停使用 `{ kind: 'hover'; hover }`（内容或 `null`）；消费方通过 `switch` 实现穷尽检查，因此新增分支会使编译失败，直到完成处理。`resolvedWorkspaceRoot` 是提供方对请求 `workspaceRoot` 的规范形式，也是其 `file:` URI 所相对的根；调用方把显示路径相对化时使用该值，而非可能含符号链接的请求根。完整契约见 `src/types.ts`；`src/index.ts` 给出 `LspError` code，包括 `LSP_DISPOSED` 和 `LSP_MALFORMED_RESPONSE`。
+`LspQueryRequest`（`operation`、`filePath`、`position`、`workspaceRoot`）：每个字段都必填，因此没有字段需要实现默认值，也不存在 `resolve()` 步骤。位置与范围使用从零开始的 UTF-16，与协议一致；工具负责从 1 开始的光标约定。`findReferences` 始终包含声明，提供方在内部强制执行，因此调用方没有 flag。`LspQueryResult` 是封闭的判别联合：导航使用 `{ kind: 'locations'; locations; resolvedWorkspaceRoot }`，悬停使用 `{ kind: 'hover'; hover }`（内容或 `null`）；消费方通过 `switch` 实现穷尽检查，因此新增分支会使编译失败，直到完成处理。`resolvedWorkspaceRoot` 是提供方对请求 `workspaceRoot` 的规范形式，也是其 `file:` URI 所相对的根；调用方把显示路径相对化时使用该值，而非可能含符号链接的请求根。完整契约见 `src/types.ts`；`src/index.ts` 给出 `LspError` 代码，包括 `LSP_DISPOSED` 和 `LSP_MALFORMED_RESPONSE`。
 
 ## 模型体验
 
@@ -39,6 +39,6 @@
 
 ## 已知限制与暂缓事项
 
-- **同一运行时内扩展名归属互斥**：两个提供方不能同时声明 `.ts`，即使 language id 不同；重叠会使注册失败。预期扩展是在注册之上增加部署配置的 selector；它可以放宽互斥保留，而无需把提供方选择加入模型输入（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)）。
+- **同一运行时内扩展名归属互斥**：两个提供方不能同时声明 `.ts`，即使语言 ID 不同；重叠会使注册失败。预期扩展是在注册之上增加部署配置的 selector；它可以放宽互斥保留，而无需把提供方选择加入模型输入（见 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)）。
 - **仅四种操作**：symbol 与 call hierarchy 暂缓（它们需要不同 schema）；diagnostics 需要独立的新鲜度／累积规则；修改操作（rename、code action、formatting）需要独立工具，并集成预览、权限和写入策略。
-- **没有观测表层**：可用性只能通过运行 `query()` 并按抛出的 `LspError` code 路由来观测；没有提供方变更事件或能力状态查询。
+- **没有观测接口**：可用性只能通过运行 `query()` 并按抛出的 `LspError` 代码进行路由来观测；没有提供方变更事件或能力状态查询。
diff --git a/packages/lsp/tool-lsp/README.i18n.yaml b/packages/lsp/tool-lsp/README.i18n.yaml
index e27d2f9344..0254413924 100644
--- a/packages/lsp/tool-lsp/README.i18n.yaml
+++ b/packages/lsp/tool-lsp/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/lsp/tool-lsp/README.md
 README.md: 9b4130015ddf7e1cad6fa9a0e131be86f3bd4bcc
-README.zh.md: e08ffcef0ff272a46d58ea17c032bb46153b0d96
+README.zh.md: 570fd353a3db4265eae9447448a7716cde2efe9a
diff --git a/packages/lsp/tool-lsp/README.zh.md b/packages/lsp/tool-lsp/README.zh.md
index e08ffcef0f..570fd353a3 100644
--- a/packages/lsp/tool-lsp/README.zh.md
+++ b/packages/lsp/tool-lsp/README.zh.md
@@ -8,13 +8,13 @@ Namespace 插件（`name`／`inject`／`Config`／`apply`，无默认导出）
 
 ## 工具
 
-`lsp` 接受 `operation`（`goToDefinition` | `findReferences` | `goToImplementation` | `hover`）、`file_path`、`line` 和 `character`。`line` 与 `character` 是正的、从 1 开始的 UTF-16 光标坐标；工具将其转换为 seam 从零开始的位置，并把渲染位置转换回来。`findReferences` 包含声明，因此影响分析不会遗漏定义位置。提供方、language id、Workspace 根、限制、超时、初始化和可执行文件均不进入模型输入。
+`lsp` 接受 `operation`（`goToDefinition` | `findReferences` | `goToImplementation` | `hover`）、`file_path`、`line` 和 `character`。`line` 与 `character` 是正的、从 1 开始的 UTF-16 光标坐标；工具将其转换为 seam 从零开始的位置，并把渲染位置转换回来。`findReferences` 包含声明，因此影响分析不会遗漏定义位置。提供方、语言 ID、工作区根目录、限制、超时、初始化和可执行文件均不进入模型输入。
 
-该工具要求从会话 `header.cwd` 取得 Workspace 根，没有回退值：缺失时会在查询前以 `LSP_WORKSPACE_REQUIRED` 失败。其规范结果是完整的已规范化 seam 联合：`{ kind: "locations", locations, resolvedWorkspaceRoot }` 或 `{ kind: "hover", hover }`；Code Mode 可以直接检查每个已取得的位置和从零开始的范围。原生渲染随后投影按文件稳定分组的 `path:line:character` 配置项，并相对于结果的 `resolvedWorkspaceRoot`（提供方的规范根）而非会话 cwd；因此，即使 cwd 包含符号链接，Workspace 内结果仍渲染为相对路径。`file:` URI 在内部时成为 Workspace 相对路径，在外部时成为绝对路径，其他 URI 保持原样。空位置和 `null` hover 都是成功的无结果响应；格式错误的提供方载荷仍是结构化错误。
+该工具要求从会话 `header.cwd` 取得工作区根目录，没有回退值：缺失时会在查询前以 `LSP_WORKSPACE_REQUIRED` 失败。其规范结果是完整的已规范化 seam 联合：`{ kind: "locations", locations, resolvedWorkspaceRoot }` 或 `{ kind: "hover", hover }`；Code Mode 可以直接检查每个已取得的位置和从零开始的范围。Native 渲染随后投影出稳定的、按文件分组的 `path:line:character` 条目，并相对于结果的 `resolvedWorkspaceRoot`（提供方的规范根目录）而非会话 cwd；因此，即使 cwd 包含符号链接，工作区内的结果仍渲染为相对路径。`file:` URI 位于工作区内时成为工作区相对路径，位于工作区外时成为绝对路径，其他 URI 保持原样。空位置和 `null` hover 都是成功的无结果响应；格式错误的提供方载荷仍是结构化错误。
 
 ## 配置
 
-| Key | 默认值 | 含义 |
+| 配置键 | 默认值 | 含义 |
 |---|---|---|
 | `maxLocations` | `100` | 出现省略标记前可渲染位置的最大数量。 |
 | `maxResultChars` | `16000` | 完整渲染结果的最大长度，包括截断元数据。 |
@@ -40,7 +40,7 @@ Use search/read for ordinary navigation. Use lsp when textual matches are ambigu
 
 #### KV Cache 影响
 
-只要插件 scope 与指引文本不变，前缀就保持稳定；激活或释放可能使从该区段起的复用失效。
+只要插件作用域与指引文本不变，前缀就保持稳定；激活或 dispose（资源释放）可能使从该区段起的复用失效。
 
 ### 工具 schema
 
@@ -54,13 +54,13 @@ Use search/read for ordinary navigation. Use lsp when textual matches are ambigu
 
 #### KV Cache 影响
 
-只要可见工具定义与顺序不变，前缀就保持稳定；注册生命周期或 scope 限制可能使从第一个变化的 schema token 起的复用失效。
+只要可见工具定义与顺序不变，前缀就保持稳定；注册生命周期或作用域限制可能使从第一个变化的 schema token 起的复用失效。
 
 ### 结果
 
 #### 模型看到的内容
 
-按文件分组的 `path:line:character` 位置行或规范化 hover 文本，先由 `maxLocations` 限制，再由 `maxResultChars` 限制；省略与截断标记计入完整字符上限。这些上限只影响原生／模型呈现，不影响规范值。空结果使用不同的 `No results.`／`No hover information.` 行。
+按文件分组的 `path:line:character` 位置行或规范化 hover 文本，先由 `maxLocations` 限制，再由 `maxResultChars` 限制；省略与截断标记计入完整字符上限。这些上限只影响 Native／模型呈现，不影响规范值。空结果使用不同的 `No results.`／`No hover information.` 行。
 
 #### Token 影响
 
@@ -74,7 +74,7 @@ Use search/read for ordinary navigation. Use lsp when textual matches are ambigu
 
 #### 模型看到的内容
 
-无。客户端渲染通用搜索卡片：`{ card: 'generic', kind: 'search', title, locations: [{ path, line }] }`；从 args 派生的标题携带操作与从 1 开始的光标，跟随焦点对准查询行，标题则保留列号。
+无。客户端渲染通用搜索卡片：`{ card: 'generic', kind: 'search', title, locations: [{ path, line }] }`；从 args 派生的标题携带操作与从 1 开始的光标，编辑器跟随定位会聚焦所查询的行，标题则保留列号。
 
 #### Token 影响
 
@@ -86,5 +86,5 @@ Use search/read for ordinary navigation. Use lsp when textual matches are ambigu
 
 ## 已知限制与暂缓事项
 
-- **UTF-16 光标坐标**：列坐标与协议精确一致，但模型难以在非 BMP 字符周围计数；非 symbol 位置可能返回空结果，因此提示词解释了该约定，但不会鼓励宽泛使用 LSP（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)）。
+- **UTF-16 光标坐标**：列坐标与协议精确一致，但模型难以在非 BMP 字符周围计数；不在符号上的位置可能返回空结果，因此提示词解释了该约定，但不会鼓励宽泛使用 LSP（见 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-15-lsp-capability-seam.md)）。
 - **不承诺跨服务器完整性**：受支持的服务器仍可能根据索引就绪情况返回空或部分结果；该工具不承诺跨语言或服务器的完整性。
diff --git a/packages/mcp/mcp-client/README.i18n.yaml b/packages/mcp/mcp-client/README.i18n.yaml
index 5e1728972d..487742d907 100644
--- a/packages/mcp/mcp-client/README.i18n.yaml
+++ b/packages/mcp/mcp-client/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/mcp/mcp-client/README.md
 README.md: 82d974635cb35878d6f0365b1aa7a9745436240e
-README.zh.md: 006e662ad69c61a011d92998f0daeb8a1b55a1c9
+README.zh.md: e687c620158955f954dcfebf682503225ee710de
diff --git a/packages/mcp/mcp-client/README.zh.md b/packages/mcp/mcp-client/README.zh.md
index 006e662ad6..e687c62015 100644
--- a/packages/mcp/mcp-client/README.zh.md
+++ b/packages/mcp/mcp-client/README.zh.md
@@ -29,7 +29,7 @@ MCP 客户端桥接插件：连接外部 [Model Context Protocol](https://modelc
       Authorization: !!js '`Bearer ${process.env.MCP_TOKEN}`'
 ```
 
-模型会看到 `mcp__github__create_issue`、`mcp__web__search` 等工具，这与 Claude Code 和 Codex 使用的服务器限定形状相同。HMR 会热替换：编辑配置项会触发断开 + 重新连接，无需重启进程；`serverName` 不变时会生成完全相同的工具名称。
+模型会看到 `mcp__github__create_issue`、`mcp__web__search` 等工具，这与 Claude Code 和 Codex 使用的服务器限定形状相同。HMR（热模块替换）支持热替换：编辑配置项会触发断开 + 重新连接，无需重启进程；`serverName` 不变时会生成完全相同的工具名称。
 
 ## 配置
 
@@ -39,7 +39,7 @@ MCP 客户端桥接插件：连接外部 [Model Context Protocol](https://modelc
 | `serverName` | 两者 | 是 | 该服务器面向模型工具名称的 namespace；`[A-Za-z0-9_-]{1,32}`，在存活实例中唯一 |
 | `command` | stdio | 是 | 要 spawn 的可执行文件 |
 | `args` | stdio | 否 | 传给命令的参数 |
-| `env` | stdio | 否 | 合并到已清理环境之上的额外环境变量 |
+| `env` | stdio | 否 | 合并到已清理环境中的额外环境变量 |
 | `cwd` | stdio | 否 | 子进程工作目录 |
 | `url` | http | 是 | MCP 服务器 URL |
 | `headers` | http | 否 | 额外标头（例如认证 token） |
@@ -52,7 +52,7 @@ MCP 客户端桥接插件：连接外部 [Model Context Protocol](https://modelc
 - 发布相同原始名称（例如 `search`）的两个服务器会在各自 namespace 下共存。
 - 存活实例中的重复 `serverName` 会使后加载的插件实例失败。
 - 服务器在工具列表中两次列出同一工具名称时，该列表会作为无效工具列表被拒绝。
-- 外部注册抢占该服务器 namespace 时，会回滚整个世代（绝不保留部分集合），并高声报错。
+- 外部注册抢占该服务器 namespace 时，会回滚整个世代（绝不保留部分集合），并明确报错。
 
 ## 行为
 
@@ -60,7 +60,7 @@ MCP 客户端桥接插件：连接外部 [Model Context Protocol](https://modelc
 - 监听 `notifications/tools/list_changed` → 重新同步；同步失败时保留上一世代的注册。
 - 工具执行：`client.callTool({ name: rawName, arguments }, { signal })`，支持超时 + 中止；公开名称绝不会发给服务器。
 - 规范成功值是 `{ content: JsonValue[], structuredContent? }`；完整的 JSON MCP 块会保留给编程调用方。受支持且已声明的 `outputSchema` 会验证 `structuredContent`；不受支持的 schema 词汇会回退为不受约束的 `JsonValue`。
-- 原生／模型渲染保留现有文本投影：文本块以换行连接，图片、音频、资源和不受支持的块会变成占位符。
+- Native／模型渲染保留现有文本投影：文本块以换行连接，图片、音频、资源和不受支持的块会变成占位符。
 - 断开／崩溃时：注销所有工具；不自动重新连接。
 
 ## 消费的服务
@@ -75,7 +75,7 @@ MCP 客户端桥接插件：连接外部 [Model Context Protocol](https://modelc
 
 #### 模型看到的内容
 
-初始发现成功后，每个已声明的 MCP 工具都会显示为名为 `mcp__<serverName>__<rawName>`（或其确定性规范化形式）的原生工具，并携带服务器提供的描述和输入 schema。成功的重新同步会替换整个世代；插件释放会移除它。
+初始发现成功后，每个已声明的 MCP 工具都会显示为名为 `mcp__<serverName>__<rawName>`（或其确定性规范化形式）的原生工具，并携带服务器提供的描述和输入 schema。成功的重新同步会替换整个世代；对插件执行 dispose（资源释放）会移除该世代。
 
 #### Token 影响
 
@@ -89,20 +89,20 @@ MCP 客户端桥接插件：连接外部 [Model Context Protocol](https://modelc
 
 #### 模型看到的内容
 
-公开工具名称和 JSON 参数会保留在 assistant 历史中。文本结果块会以换行连接为一个保留的原生文本结果；图片、音频、资源和不受支持的块在其中变为简短占位符。它们的完整 JSON 块及可选结构化内容保留在执行局部的规范值中；MCP `isError` 会通过注册表的错误路径拒绝调用。
+公开工具名称和 JSON 参数会保留在 assistant 历史中。文本结果块会以换行连接为一个保留的 Native 文本结果；图片、音频、资源和不受支持的块在其中变为简短占位符。它们的完整 JSON 块及可选结构化内容保留在执行局部的规范值中；MCP `isError` 会通过注册表的错误路径拒绝调用。
 
 #### Token 影响
 
-参数和映射后的文本会保留到压缩发生时。二进制与资源载荷会被丢弃，而不会加入上下文。
+参数和映射后的文本会保留到压缩（compaction）发生时。二进制与资源载荷会被丢弃，而不会加入上下文。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
 - **初始发现是异步的**：插件加载不会等待连接和 `listTools()`，因此在启动或 HMR 后立即开始的轮次可能在 MCP 工具注册前完成组装。
-- **只桥接 MCP 的工具能力**：资源和提示词没有 harness 消费表层，暂缓实现。
+- **只桥接 MCP 的工具能力**：资源和提示词没有 harness 消费接口，暂缓实现。
 - **崩溃恢复需要手动触发**：传输关闭会注销服务器工具，但重新连接需要 HMR 重载或重启 harness。
-- **原生非文本渲染有损**：图片、音频与资源载荷在模型上下文中会变成占位符，即使执行局部的规范值保留了其 JSON 块。更丰富的原生多媒体投影暂缓实现。
+- **Native 非文本渲染有损**：图片、音频与资源载荷在模型上下文中会变成占位符，即使执行局部的规范值保留了其 JSON 块。更丰富的 Native 多媒体投影暂缓实现。
 - **不强制执行不受支持的 MCP 输出 schema**：已声明 schema 使用 harness 子集之外的词汇时，`structuredContent` 会回退到 `JsonValue`。
diff --git a/packages/plan/README.i18n.yaml b/packages/plan/README.i18n.yaml
index f38be94b01..39eca60b79 100644
--- a/packages/plan/README.i18n.yaml
+++ b/packages/plan/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/plan/README.md
 README.md: eeb58703c34acae0eb2146b87b01a56815e1362a
-README.zh.md: cedb8dac138fff15595b4b935000334f8a36c262
+README.zh.md: 4afe39e0eb638257c5907ddce1532abbaf720a9b
diff --git a/packages/plan/README.zh.md b/packages/plan/README.zh.md
index cedb8dac13..4afe39e0eb 100644
--- a/packages/plan/README.zh.md
+++ b/packages/plan/README.zh.md
@@ -2,10 +2,10 @@
 
 [English](README.md) | 中文
 
-Plan mode 是一种按 agent 分开记录到日志的协作状态。它是单一 **产品** 包（package），而非通用 mode 注册表或能力 seam 三包组合。
+Plan mode 是一种按 agent（智能体）分开记录到日志的协作状态。它是单一**产品**包（package），而非通用模式注册表或能力 seam 三包组合。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `plan-mode/` | `plan/mode` 词汇与折叠、在边界生效的状态、`plan:policy` 引导段、`/plan [message]` 进入命令与 `/plan off` 退出命令，以及面向模型的 `exit_plan_mode` 评审工具 | `ctx.planMode` |
 
-活跃状态是会话日志的纯函数，因此恢复和 fork 无需额外机制即可还原该状态。部署通过 Cordis 配置提供 plan 指令，而 `exit_plan_mode` 在 plan mode 未激活时仍保持注册，以稳定请求工具目录。交互式适配器使用插件拥有的 `/plan` 命令；沙箱模式和批准策略仍是独立的强制执行设置。设计详见 [plan 专用协作状态](../../.agents/notes/implemented/simplification/2026-07-22-plan-specific-collaboration-state.md)。
+活跃状态是会话日志的纯函数，因此恢复和 fork 无需额外机制即可还原该状态。部署通过 Cordis 配置提供 plan 引导内容，而 `exit_plan_mode` 在 Plan mode 未激活时仍保持注册，以稳定请求工具目录。交互式适配器使用插件拥有的 `/plan` 命令；沙箱模式和审批策略仍是独立的强制执行设置。设计详见 [plan 专用协作状态](../../.agents/notes/implemented/simplification/2026-07-22-plan-specific-collaboration-state.md)。
diff --git a/packages/plan/plan-mode/package.json b/packages/plan/plan-mode/package.json
index 70ed87889d..6507ea4a0c 100644
--- a/packages/plan/plan-mode/package.json
+++ b/packages/plan/plan-mode/package.json
@@ -39,6 +39,7 @@
     "@deepseek-ai/dsh-agent": "^0.0.1",
     "@deepseek-ai/dsh-commands": "^0.0.1",
     "@deepseek-ai/dsh-invariants": "^0.0.1",
+    "@deepseek-ai/dsh-llm": "^0.0.1",
     "@deepseek-ai/dsh-session": "^0.0.1",
     "@deepseek-ai/dsh-session-projection": "^0.0.1",
     "@deepseek-ai/dsh-system-prompt": "^0.0.1",
diff --git a/packages/pty/README.i18n.yaml b/packages/pty/README.i18n.yaml
index ef4c7b5c13..c773791907 100644
--- a/packages/pty/README.i18n.yaml
+++ b/packages/pty/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/pty/README.md
 README.md: a9121455519a5f83a63a005cb857fec0f0e06b92
-README.zh.md: 9fc262787b960d5bf03a59cd01bf36bd5c76614b
+README.zh.md: 4e36ef934d31861d72748b9b9585695f86749171
diff --git a/packages/pty/README.zh.md b/packages/pty/README.zh.md
index 9fc262787b..4e36ef934d 100644
--- a/packages/pty/README.zh.md
+++ b/packages/pty/README.zh.md
@@ -2,12 +2,12 @@
 
 [English](README.md) | 中文
 
-`PTY` 的全称是 **Pseudo-Terminal（伪终端）**。这项能力提供持久且限定所有者范围的终端会话，适用于需要跨工具调用保留状态或使用交互式 stdin 的工作流。PTY 是单次 bash 与文件系统工具的补充，不会取代后两者更严格的逐操作契约。
+`PTY` 的全称是 **Pseudo-Terminal**（伪终端）。这项能力提供持久且限定所有者范围的终端会话，适用于需要跨工具调用保留状态或使用交互式 stdin 的工作流。PTY 是单次 bash 与文件系统工具的补充，不会取代后两者更严格的逐操作契约。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
-| [`pty`](pty/README.md)（`@deepseek-ai/dsh-pty`） | 后端注册表、品牌化 id、精确的 Agent 所有权、会话操作与等待完成的清理 | `ctx.pty` |
+| [`pty`](pty/README.md)（`@deepseek-ai/dsh-pty`） | 后端注册表、品牌化 id、精确到 agent（智能体）的所有权、会话操作与等待清理完成的机制 | `ctx.pty` |
 | `pty-local`（`@deepseek-ai/dsh-pty-local`） | 本地 `node-pty` 后端、就绪检测、有界终端状态、沙箱与进程会话监管 | 注册到 `ctx.pty` |
-| `tool-pty`（`@deepseek-ai/dsh-tool-pty`） | 6 个面向模型的工具，并为后台发送集成通用任务 | 注册到 `ctx.tools` |
+| `tool-pty`（`@deepseek-ai/dsh-tool-pty`） | 6 个面向模型的工具，以及用于后台发送的通用任务集成 | 注册到 `ctx.tools` |
 
-设计与暂缓边界记录在[持久 PTY Agent Note](../../.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md) 中。
+设计与暂缓边界记录在[持久 PTY Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-16-persistent-pty-sessions.md) 中。
diff --git a/packages/pty/pty-local/README.i18n.yaml b/packages/pty/pty-local/README.i18n.yaml
index 04e34c3f3c..97c0b693e3 100644
--- a/packages/pty/pty-local/README.i18n.yaml
+++ b/packages/pty/pty-local/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/pty/pty-local/README.md
 README.md: 6f243a6edf3ab8bc228cfda3f6b3b774dabadcbb
-README.zh.md: c417828a443d227b5bfcdc1c788cf4ab6751b2ee
+README.zh.md: 0cf2981332fa41dab22a496de5ae302066fbdba8
diff --git a/packages/pty/pty-local/README.zh.md b/packages/pty/pty-local/README.zh.md
index c417828a44..0cf2981332 100644
--- a/packages/pty/pty-local/README.zh.md
+++ b/packages/pty/pty-local/README.zh.md
@@ -6,11 +6,11 @@
 
 ## 插件（`pty-local`）
 
-该插件注入 `pty`、`sandbox` 和 `sandboxPolicy`，然后注册所配置的后端类型（`shell`）。`danger-full-access` 会直接启动 shell；受限模式则通过 `ctx.sandbox` 包装确切的 shell argv。系统在 spawn 时解析会话的实际模式。当某个所有者存在开放的 PTY 或正在进行 spawn 时，如果配置变更会得到不同的实际模式，系统会在对应 `sandbox/mode` 事件提交前拒绝该变更。该限制绑定到确切所有者，因此即使本地提供方重新加载并保留现有会话，它仍然有效。更改模式前，请等待创建结算并关闭会话，避免以更宽权限打开的终端在权限降级后继续存在。
+该插件注入 `pty`、`sandbox` 和 `sandboxPolicy`，然后注册所配置的后端类型（`shell`）。`danger-full-access` 会直接启动 shell；受限模式则通过 `ctx.sandbox` 包装确切的 shell argv。系统在 spawn 时解析会话的实际模式。当某个所有者存在开放的 PTY 或正在进行 spawn 时，如果配置变更会得到不同的实际模式，系统会在对应 `sandbox/mode` 事件提交前拒绝该变更。该限制绑定到确切所有者，因此即使本地提供方重新加载并保留现有会话，它仍然有效。更改模式前，请等待创建完成并关闭会话，避免以更宽权限打开的终端在权限降级后继续存在。
 
-Linux 的就绪检测结合以下机制：由前台状态验证的私有 bash 提示符标记、前台进程组 syscall 检查、静默回退和绝对超时。macOS 没有 `/proc` syscall 接口，因此使用经过验证的提示符标记以及静默／超时。当可打印的提示符文本尚未到达时，即使 OSC 标记和 `PS1` 被拆到多个数据回调中，系统也不会把标记视为就绪。如果 bash 在内核发布其重新取得前台进程组的状态前打印标记，轮询会在普通静默上限之后再保留该候选状态 `handoffGraceMs`，使恰好同时发生的前台交接有机会胜出；该宽限至少要覆盖一个 `pollIntervalMs`，否则加载时即被拒绝。因此，继承 `PROMPT_COMMAND` 的交互式子进程无法持续压制推断空闲就绪，最多只能延续到绝对超时。无法识别或读取的进程状态绝不会作为精确空闲的正向信号。同样，一次 send 之前就已存在的前台进程组 stdin 等待并不代表写入后就绪：必须先观察到同一进程组脱离该等待，之后再次进入等待才能使该次 send 完成；前台进程组发生变化则构成新的证据。尚未发布的启动过程中，回退路径要求已经观察到输出；零输出静默不能发布空会话，超时则拒绝 spawn。取消操作会关闭尚未发布的 shell，并以调用方提供的确切中止原因拒绝，即使当时还无法观察其前台进程组。如果关闭失败，`PtyBackendCleanupError` 会单独保留清理失败，供注册表释放资源时处理。未完成的终端控制序列受 `maxReadBytes` 限制；超过上限后，系统会丢弃内容直到其终止符。末尾的回车会跨回调保留，使拆分的 CRLF 合并为一个换行。
+Linux 的就绪检测结合以下机制：由前台状态验证的私有 bash 提示符标记、前台进程组 syscall 检查、静默回退和绝对超时。macOS 没有 `/proc` syscall 接口，因此使用经过验证的提示符标记以及静默／超时。当可打印的提示符文本尚未到达时，即使 OSC 标记和 `PS1` 被拆到多个数据回调中，系统也不会把标记视为就绪。如果 bash 在内核发布其重新取得前台进程组的状态前打印标记，轮询会在普通静默上限之后再保留该候选状态 `handoffGraceMs`，使恰好同时发生的前台交接有机会胜出；该宽限至少要覆盖一个 `pollIntervalMs`，否则加载时即被拒绝。因此，继承 `PROMPT_COMMAND` 的交互式子进程无法一直抑制推断空闲就绪直至绝对超时。无法识别或读取的进程状态绝不会作为精确空闲的正向信号。同样，一次 send 之前就已存在的前台进程组 stdin 等待并不代表写入后就绪：必须先观察到同一进程组脱离该等待，之后再次进入等待才能使该次 send 完成；前台进程组发生变化则构成新的证据。尚未发布的启动过程中，回退路径要求已经观察到输出；零输出静默不能发布空会话，超时则拒绝 spawn。取消操作会关闭尚未发布的 shell，并以调用方提供的确切中止原因拒绝，即使当时还无法观察其前台进程组。如果关闭失败，`PtyBackendCleanupError` 会单独保留清理失败，供注册表 dispose（资源释放）时处理。未完成的终端控制序列受 `maxReadBytes` 限制；超过上限后，系统会丢弃内容直到其终止符。末尾的回车会跨回调保留，使拆分的 CRLF 合并为一个换行。
 
-取消发送时，系统会解析当前前台进程组并发送真正的 `SIGINT`；它绝不会通过写入 `\x03` 模拟中断，因此原始模式程序仍可取消。关闭操作先向后代发送 `SIGTERM` 并等待，再向已捕获的存活进程与新扫描到的后代之并集发送 `SIGKILL`，防止进程通过重新设定父进程而逃避清理。系统确认每个保留的进程身份都已消失；在 Linux 上，非执行中的僵尸进程也视为完全停稳，并会随 shell 退出而回收。如果仍有进程存活，失败结果不会缓存成永久拒绝的关闭操作；后续关闭仍会重试清理。
+取消发送时，系统会解析当前前台进程组并发送真正的 `SIGINT`；它绝不会通过写入 `\x03` 模拟中断，因此，即使程序运行在 raw 模式下，也仍可取消。关闭操作先向后代发送 `SIGTERM` 并等待，再向已捕获的存活进程与新扫描到的后代之并集发送 `SIGKILL`，防止进程通过重新设定父进程而逃避清理。在停止 shell 前，系统会确认每个保留的进程身份都已消失，或者在 Linux 上已成为不再执行的僵尸进程；僵尸进程条目视为完全停稳，并会随 shell 退出而回收。如果仍有进程存活，失败结果不会缓存成永久拒绝的关闭操作；后续关闭仍会重试清理。
 
 ## 模型体验
 
@@ -22,15 +22,15 @@ Linux 的就绪检测结合以下机制：由前台状态验证的私有 bash 
 
 #### Token 影响
 
-消费方返回有界的后端输出前没有影响。此包不会把保留的 PTY scrollback 放入模型历史。
+消费方返回有界的后端输出前没有影响。此包（package）不会把保留的 PTY scrollback 放入模型历史。
 
 #### KV Cache 影响
 
-不会直接失效；提示词、schema 与追加结果由消费方负责。
+不会直接使 KV Cache 失效；提示词、schema 与追加结果由消费方负责。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - 输出按行规范化；不支持全屏备用缓冲区交互。
 - Linux 精确探针支持 x64 与 arm64 UAPI 表；其他架构使用提示符标记和静默／超时就绪机制。
 - 如果后代进程在清理前守护化并重新设定父进程，它会脱离已捕获的进程树；清理绝不会扩大到启动器 PID 所属的整个 POSIX 会话，因为其中可能包含无关进程。
-- 会话无法跨 harness 进程退出保留。
+- harness 进程退出后，会话无法继续存在。
diff --git a/packages/pty/pty/README.i18n.yaml b/packages/pty/pty/README.i18n.yaml
index 063950ec0c..97b38bacd6 100644
--- a/packages/pty/pty/README.i18n.yaml
+++ b/packages/pty/pty/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/pty/pty/README.md
 README.md: 0f8b8e499dc81ce91e249f44bb38c8cc1af89d3f
-README.zh.md: 9158afd6a7c82db27820d94cdbebd4eaa537daab
+README.zh.md: 95c6345a8077d072cd22e421d3cfbbc9bf2e89b6
diff --git a/packages/pty/pty/README.zh.md b/packages/pty/pty/README.zh.md
index 9158afd6a7..95c6345a80 100644
--- a/packages/pty/pty/README.zh.md
+++ b/packages/pty/pty/README.zh.md
@@ -2,20 +2,20 @@
 
 [English](README.md) | 中文
 
-限定所有者范围的持久 PTY seam。`PtyService` 注册为 `ctx.pty`，生成不透明的会话 id，通过具名后端路由创建操作，将每个操作限制在完全相同的活跃 `Agent` 内，并在该 agent 或服务释放资源时等待后端完全停稳。
+限定所有者范围的持久 PTY seam。`PtyService` 注册为 `ctx.pty`，生成不透明的会话 id，通过具名后端路由创建操作，将每个操作限制在完全相同的活跃 `Agent` 内，并在该 agent（智能体）或服务 dispose（资源释放）时等待后端完全停稳。
 
 ## 契约
 
-- 后端注册一个稳定的 `type`，并返回尚未发布的 `PtyBackendSession`；失败或取消的设置过程必须清理部分资源。若清理失败，则以 `PtyBackendCleanupError` 拒绝，使注册表能在取消之后继续保留该资源。
-- spawn 取消会保留调用方提供的确切中止原因。后端设置完成后，服务资源释放与所有者消失仍分别对应可供机器路由的不同失败。
-- 所有者与服务的资源释放会通过服务所有的信号中止尚未发布的设置，并等待后端结算和回滚后才返回。
-- 如果回滚关闭失败，或后端报告启动清理失败，资源释放生命周期会以拒绝结束，不会声称已经完全停稳。调用方触发的取消仍收到其确切原因；生命周期触发的回滚失败也会拒绝待完成的 spawn。
-- 调用方取消后发生的后端清理失败仍算作所有者活动，直到所有者或服务释放资源并消费、报告该失败，避免生命周期策略把失败的清理误判为完全停稳。
+- 后端注册一个稳定的 `type`，并返回尚未发布的 `PtyBackendSession`；失败或取消的设置过程必须清理部分资源。若清理失败，则以 `PtyBackendCleanupError` 拒绝，使注册表能在取消后继续保留该清理失败。
+- spawn 取消会保留调用方提供的确切中止原因。后端设置完成后，服务 dispose 与所有者消失仍分别对应可供机器路由的不同失败。
+- 所有者与服务的 dispose 会通过服务持有的信号中止尚未发布的设置，并等待后端结算和回滚后才返回。
+- 如果回滚关闭失败，或后端报告启动清理失败，dispose 生命周期会以拒绝结束，不会声称已经完全停稳。调用方触发的取消仍收到其确切原因；生命周期触发的回滚失败也会拒绝待完成的 spawn。
+- 调用方取消后发生的后端清理失败仍算作所有者活动，直到所有者或服务 dispose 并消费、报告该失败，避免生命周期策略把失败的清理误判为完全停稳。
 - `hasOwnerActivity(owner)` 覆盖从尚未发布的设置到最终关闭的全过程，使生命周期策略能精确限制对应所有者，不受发布竞态影响。
 - 成功的 spawn 会发布一个 `PtySessionId`。可选的 `name` 只是所有者本地的显示元数据，绝不代表权限。
 - 一个会话最多接受一个活跃的发送操作。读取和信号操作可以观察该发送；在当前操作结算前，另一项发送会失败。
 - `PtySendResult.waitReason` 与 `sessionStatus` 相互独立。`session_exit` 描述顶层 PTY 进程，而不是任意前台命令。
-- `kill()` 与资源释放只会在后端捕获的进程树完全停稳后完成。清理失败会以拒绝结束，而非声称成功；同时它会清除匹配的后端和注册表限制，使后续关闭能够重试，且不会干扰较新的尝试。
+- `kill()` 与 dispose 只会在后端捕获的进程树完全停稳后完成。清理失败会以拒绝结束，而非声称成功；同时它会清除匹配的后端和注册表限制，使后续关闭能够重试，且不会干扰较新的尝试。
 
 该 seam 不包含 `node-pty`、沙箱、工具 schema、提示词、任务或终端渲染策略。实现负责终端机制；消费方负责模型呈现和可选的后台任务注册。
 
@@ -25,7 +25,7 @@
 
 #### 模型看到的内容
 
-没有直接可见内容。此包不注册提示词或工具；可见 schema 和结果文本由 `@deepseek-ai/dsh-tool-pty` 负责。
+没有直接可见内容。此包（package）不注册提示词或工具；可见 schema 和结果文本由 `@deepseek-ai/dsh-tool-pty` 负责。
 
 #### Token 影响
 
@@ -33,9 +33,9 @@
 
 #### KV Cache 影响
 
-不会直接失效；由具名消费方负责请求前缀变更。
+不会直接失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - 会话只存在于进程本地，harness 重启后不会恢复。
 - 系统有意不支持跨 agent 共享；未来的共享会话设计需要独立的权限契约。
diff --git a/packages/pty/tool-pty/README.i18n.yaml b/packages/pty/tool-pty/README.i18n.yaml
index ea1d0cc895..e5b19d9432 100644
--- a/packages/pty/tool-pty/README.i18n.yaml
+++ b/packages/pty/tool-pty/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/pty/tool-pty/README.md
 README.md: 417cf7bd0e7881f6ffefc47f5200164464b516f6
-README.zh.md: cdce3c25a6364676a6cd69c30dd877cbc8091778
+README.zh.md: 72eb13957c62cbb08dcecce5971bb064d5085470
diff --git a/packages/pty/tool-pty/README.zh.md b/packages/pty/tool-pty/README.zh.md
index cdce3c25a6..72eb13957c 100644
--- a/packages/pty/tool-pty/README.zh.md
+++ b/packages/pty/tool-pty/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-基于 `ctx.pty` 提供 6 个面向模型的工具：`terminal_open`、`terminal_send`、`terminal_read`、`terminal_signal`、`terminal_close` 和 `terminal_list`。每项操作都要求提供完全相同的发起 `Agent`，因此即使模型获知另一个 agent 的 id，也无法操作其终端。
+基于 `ctx.pty` 提供 6 个面向模型的工具：`terminal_open`、`terminal_send`、`terminal_read`、`terminal_signal`、`terminal_close` 和 `terminal_list`。每项操作都要求提供完全相同的发起 `Agent`，因此即使模型获知另一个 agent（智能体）的 id，也无法操作其终端。
 
-`terminal_send(run_in_background: true)` 会复用 `ctx.tasks`；任务预检和 PTY 服务对每会话发送的独占预留都发生在返回 task id 之前。系统通过 `task_output` 收集完成结果，`task_kill` 则向前台进程组发送 `SIGINT`。前台发送使用终端调用／结果卡片。后台发送使用通用执行卡片；打开、读取、发送信号、关闭和列出操作则分别使用通用 `execute`、`read`、`execute`、`delete` 和 `read` 卡片。所有操作都不声明源位置。
+`terminal_send(run_in_background: true)` 会复用 `ctx.tasks`；任务预检和 PTY 服务对每个会话的独占发送预留都发生在返回 task id 之前。系统通过 `task_output` 收集完成结果，`task_kill` 则向前台进程组发送 `SIGINT`。前台发送使用终端调用／结果卡片。后台发送使用通用执行卡片；打开、读取、发送信号、关闭和列出操作则分别使用通用 `execute`、`read`、`execute`、`delete` 和 `read` 卡片。所有操作都不声明源位置。
 
 ## 配置
 
@@ -13,7 +13,7 @@
 | `enableRunInBackground` | `true` | 公开并接受 `run_in_background`；设为 false 时，schema 会省略该字段，并拒绝强行传入未声明的参数 |
 | `maxResultBytes` | `262144` | 每个完整终端结果或 PTY 任务输出的 UTF-8 上限（最小值 `64`）；在等待、会话、分页、截断和任务状态元数据全部加入后计算 |
 
-两个值都会在加载时验证。最小结果上限可保证注册表签发的每个会话或 task id 都能出现在创建确认中。结果超过 `maxResultBytes` 时，只要空间允许，渲染会为控制元数据和截断标记预留空间；截断会保留 UTF-8 边界。每个终端定义的最终内容回调都会应用同一个上限，涵盖经过规范化的 pre-execute、around-execute 与 post-execute 策略失败、拒绝、短路、替换或阻止；结构化的多块策略结果保留其形状。
+两个值都会在加载时验证。最小结果上限可保证注册表签发的每个会话或 task id 都能出现在创建确认中。结果超过 `maxResultBytes` 时，只要空间允许，渲染会为控制元数据和截断标记预留空间；截断会保留 UTF-8 边界。每个终端定义的最终内容回调都会应用同一个上限，涵盖经过规范化的 pre-execute、around-execute 与 post-execute 策略失败、拒绝、短路、替换或阻止；结构化的多块策略结果保留其结构。
 
 ## 模型体验
 
@@ -55,7 +55,7 @@ Use a terminal session only when work needs persistent terminal state or interac
 
 #### 模型看到的内容
 
-spawn 会返回 id 和有界 MOTD。发送／读取会返回有界终端文本以及就绪／历史标记。后台模式返回通用 task id。所有终端自身或策略产生的单文本结果，在经过规范化的工具或流水线错误、拒绝、短路、替换、阻止与通用任务状态文本之后，都受 `maxResultBytes` 限制。结构化的多块策略结果保留其形状。结果会保留在会话历史中直到压缩；增量任务读取不会重复已经消费的输出。编程调用方会收到带类型的会话快照、有界的提供方读取／发送 DTO、信号与关闭结果，或 `{ kind: "background", taskId }`；Native 渲染会应用上述呈现上限。
+spawn 会返回 id 和有界 MOTD。发送／读取会返回有界终端文本以及就绪／历史标记。后台模式返回通用 task id。所有终端自身或策略产生的单文本结果，在经过规范化的工具或流水线错误、拒绝、短路、替换、阻止与通用任务状态文本之后，都受 `maxResultBytes` 限制。结构化的多块策略结果保留其结构。结果会保留在会话历史中直到压缩（compaction）；增量任务读取不会重复已经消费的输出。编程调用方会收到带类型的会话快照、有界的提供方读取／发送 DTO、信号与关闭结果，或 `{ kind: "background", taskId }`；Native 渲染会应用上述呈现上限。
 
 #### Token 影响
 
@@ -65,7 +65,7 @@ spawn 会返回 id 和有界 MOTD。发送／读取会返回有界终端文本
 
 仅追加；新结果位于可复用请求前缀之后。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - 不公开具名按键序列、TUI、BEL、调整大小、自动启动或跨 agent 共享 schema。
 - 后台模式同时依赖 `@deepseek-ai/dsh-tasks` 及其面向模型的控制接口。
diff --git a/packages/sandbox/README.i18n.yaml b/packages/sandbox/README.i18n.yaml
index 4e60781469..29c5f4e7a6 100644
--- a/packages/sandbox/README.i18n.yaml
+++ b/packages/sandbox/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sandbox/README.md
 README.md: 4a651f262ee817edf5e9620d1cbf911b2d51b90b
-README.zh.md: 63eaef2488a0bd478ec1f7f34cd96085faf63e45
+README.zh.md: d782d1b6391d3057efbe4ee742d86b751ebd5967
diff --git a/packages/sandbox/README.zh.md b/packages/sandbox/README.zh.md
index 63eaef2488..d782d1b639 100644
--- a/packages/sandbox/README.zh.md
+++ b/packages/sandbox/README.zh.md
@@ -2,14 +2,14 @@
 
 [English](README.md) | 中文
 
-[能力 seam 拆分](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)中负责限制的一半：抽象提供方接口、平台后端和共享策略归属位置。消费方把即将 spawn 的精确 argv 交给 `ctx.sandbox`，改为 spawn 返回的已包装 argv；完整的 `SandboxExecutionPolicy`（模式 + Workspace 根）随每次能力调用传递，其中受限制的子集成为提供方的 `SandboxPolicy`。因此，不同会话与消费方可以同时按不同策略施加限制。这些全是**产品** 包。
+[能力 seam 拆分](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)中负责限制的一半：抽象提供方接口、平台后端和共享策略归属位置。消费方把即将 spawn 的精确 argv 交给 `ctx.sandbox`，改为 spawn 返回的已包装 argv；完整的 `SandboxExecutionPolicy`（模式 + 工作区根目录）随每次能力调用传递，其中受限制的子集成为提供方的 `SandboxPolicy`。因此，不同会话与消费方可以同时按不同策略施加限制。这些均为**产品**包（package）。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
-| `sandbox/` | 抽象进程沙箱 seam（`SandboxProvider` 契约 + 模式／强制执行／策略词汇），加共享 ESCALATION 工具包（`approveEscalation`、严格变宽的阶梯、拒绝／提示标记），以及所有强制执行方言共享的 `writableRoots` 派生 | `ctx.sandbox` |
-| `sandbox-local/` | 按平台链选择的本地后端：Linux 使用 `bwrap`，否则使用 `landlock-run` launcher（通过 npm 分发的 [`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run) 家族，在自身仓库构建发布）；darwin 使用 `sandbox-exec`／Seatbelt。多候选链会执行功能探测，唯一候选项直接选择，结论缓存，快速失败 | （注册 `ctx.sandbox`） |
-| `sandbox-policy/` | 策略解析器：部署回退值，加每个会话的持久模式与不可变 cwd 根。两个强制执行家族都消费完整的逐调用结果，因此 bash 与 fs 不会限制到不同根目录 | `ctx.sandboxPolicy` |
+| `sandbox/` | 抽象进程沙箱 seam（`SandboxProvider` 契约 + 模式／强制执行／策略词汇），加共享 ESCALATION 工具包（`approveEscalation`、权限逐级严格扩大的阶梯、拒绝／提示标记），以及所有强制执行方言共享的 `writableRoots` 派生 | `ctx.sandbox` |
+| `sandbox-local/` | 按平台链选择的本地后端：Linux 使用 `bwrap`，否则使用 `landlock-run` launcher（通过 npm 分发的 [`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run) 家族，在自身仓库构建发布）；darwin 使用 `sandbox-exec`／Seatbelt。多候选链会执行功能探测，唯一候选项直接选择，结论缓存，失败时默认拒绝 | （注册 `ctx.sandbox`） |
+| `sandbox-policy/` | 策略解析器：部署回退值，加每个会话的持久模式与不可变 cwd 根目录。两个强制执行家族都消费完整的逐调用结果，因此 bash 与 fs 不会限制到不同根目录 | `ctx.sandboxPolicy` |
 
-该 seam 只限制与宿主共享文件系统和内核的子进程。容器、microVM 和远程执行器都不是这里的后端：它们会以环境一致的分组替换整个能力实现（`ctx.bash`、`ctx.fs`）；边界记录在[沙箱 Agent Note](../../.agents/notes/implemented/feature/2026-07-06-sandbox.md) 中。
+该 seam 只限制与宿主共享文件系统和内核的子进程。容器、microVM 和远程执行器都不是这里的后端：它们会以环境一致的分组替换整个能力实现（`ctx.bash`、`ctx.fs`）；边界记录在[沙箱 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)中。
 
-当前消费方：[`bash/bash-sandbox`](../bash/bash-sandbox/)（包装 `['bash', '-c', command]` 并通过 `ctx.sandbox` 执行）和 [`fs/fs-sandbox`](../fs/fs-sandbox/)（进程内路径隔离，而非 argv 包装层；读取 `ctx.sandboxPolicy`，对写入／编辑强制执行共享模式）。跨家族边界是沙箱 Agent Note 的[跨家族 fs 沙箱](../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)阶段；共享词汇使两个家族可以向模型传授同一种拒绝标记与升权流程。
+当前消费方：[`bash/bash-sandbox`](../bash/bash-sandbox/)（包装 `['bash', '-c', command]` 时会调用 `ctx.sandbox`）和 [`fs/fs-sandbox`](../fs/fs-sandbox/)（进程内路径隔离，而非 argv 包装层；读取 `ctx.sandboxPolicy`，对写入／编辑强制执行共享模式）。跨家族边界是沙箱 Agent Note 的[跨家族 fs 沙箱](../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)阶段；共享词汇使两个家族可以向模型传授同一种拒绝标记与升权流程。
diff --git a/packages/sandbox/sandbox-local/README.i18n.yaml b/packages/sandbox/sandbox-local/README.i18n.yaml
index e07ea66030..1bcf308a49 100644
--- a/packages/sandbox/sandbox-local/README.i18n.yaml
+++ b/packages/sandbox/sandbox-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sandbox/sandbox-local/README.md
 README.md: 923d983be8c2ccd60ed6eabcf9212dd89ef9bce3
-README.zh.md: 9c5a79ed8df36281932959af7b0ad0b801dbaa55
+README.zh.md: 5c02b0ce3fac1e39a16f8fc32517856996c30e23
diff --git a/packages/sandbox/sandbox-local/README.zh.md b/packages/sandbox/sandbox-local/README.zh.md
index 9c5a79ed8d..5c02b0ce3f 100644
--- a/packages/sandbox/sandbox-local/README.zh.md
+++ b/packages/sandbox/sandbox-local/README.zh.md
@@ -4,17 +4,17 @@
 
 [`dsh-sandbox`](../sandbox/) seam 的本地实现。它选择并缓存一个平台 runner：Linux 优先选择可工作的 `bwrap`，否则选择 Landlock；macOS 使用 Seatbelt。多个候选项会按顺序探测，只有一个候选项时则直接选择。
 
-包根导出默认及命名的 `LocalSandboxProvider` 插件、`Config` 和公共测试注入 seam；平台 profile builder 保持内部状态。
+包（package）根目录导出默认及命名的 `LocalSandboxProvider` 插件、`Config` 和公共测试注入 seam；平台 profile builder 仍为内部实现。
 
-不受支持的平台和不可用 runner 会以 `SANDBOX_UNAVAILABLE` 快速失败；执行绝不会静默回退为不受限制。每次包装都携带 runner 失败签名，使消费方能够区分损坏的沙箱与命令失败。[沙箱 Agent Note](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)拥有选择原理与 profile 差异。
+不受支持的平台和不可用 runner 会以 `SANDBOX_UNAVAILABLE` 拒绝执行；执行绝不会静默回退为不受限制。每次包装都携带 runner 失败签名，使消费方能够区分损坏的沙箱与命令失败。[沙箱 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)负责说明选择依据与 profile 差异。
 
-策略逐调用传入；提供方只存储机制与缓存的 runner 结论。每次包装都会报告强制执行完整度，以及后端专用的拒绝和 runner 失败签名。`runnerCommand` 是操作方对 bwrap 形状 runner 的断言，会跳过探测；但命令缺失或不可执行时，执行仍会快速失败。由于其机制未知，它会同时携带两种 Linux 拒绝方言。`probeTimeoutMs` 限制功能探测。[沙箱 Agent Note](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)拥有选择与失败语义。
+策略逐调用传入；提供方只存储机制与缓存的 runner 结论。每次包装都会报告强制执行完整度，以及后端专用的拒绝和 runner 失败签名。`runnerCommand` 是操作方对 bwrap 形式 runner 的断言，会跳过探测；但命令缺失或不可执行时，执行仍会被拒绝。由于其机制未知，它会同时携带两种 Linux 拒绝方言。`probeTimeoutMs` 限制功能探测。[沙箱 Agent Note](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)负责说明选择与失败语义。
 
-Seatbelt profile 默认允许，但带 `(deny file-write*)` 和写入 allow-list，因此恰好治理对应模式承诺的文件 effect：`read-only` 只授予 `/dev/null` 字面路径；`workspace-write` 另加 Workspace 根、`/tmp` 和逐用户 darwin 临时目录（`os.tmpdir()`，即平台供 mkstemp 家族工具使用的真实临时区域）。每个根都经过规范化，因为 Seatbelt 匹配解析后的路径（`/tmp` 就是 `/private/tmp`）。Apple 将 `sandbox-exec` CLI 标为 deprecated，但每个 macOS 仍会提供它；若情况发生变化，功能探测会快速失败。
+Seatbelt profile 默认允许，但带 `(deny file-write*)` 和写入 allow-list，因此恰好约束相应模式承诺的文件操作：`read-only` 只授予 `/dev/null` 字面路径；`workspace-write` 另加工作区根目录、`/tmp` 和逐用户 darwin 临时目录（`os.tmpdir()`，即平台供 mkstemp 家族工具使用的真实临时区域）。每个根目录都经过规范化，因为 Seatbelt 匹配解析后的路径（`/tmp` 就是 `/private/tmp`）。Apple 将 `sandbox-exec` CLI（命令行界面）标为 deprecated，但所有 macOS 系统仍会提供它；若情况发生变化，功能探测会使执行被拒绝。
 
-[`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run)提供平台 launcher、功能探测和 CLI 参数词汇。该提供方只拥有模式到授权的映射与 runner 选择。把路径解析和探测解析保留在带版本的 binary 中，可防止契约漂移。
+[`node-addon-landlock-run`](https://www.npmjs.com/package/node-addon-landlock-run)提供平台 launcher、功能探测和 CLI 参数词汇。该提供方只负责模式到授权的映射与 runner 选择。把路径解析和探测解析保留在带版本的 binary 中，可防止契约漂移。
 
-每个阶梯都有会自行跳过的无密钥 world-effect 测试；CI 在真实内核上运行平台 job，并拒绝所有测试静默跳过。打包安装测试通过纯 Node 消费方运行 registry launcher 与可执行模式。
+每个权限层级都有会自行跳过的无密钥 world-effect 测试；CI 在真实内核上运行平台 job，并拒绝所有测试静默跳过。打包安装测试通过纯 Node 消费方运行注册表 launcher 与可执行模式。
 
 ```yaml
 - id: sandbox
@@ -25,16 +25,16 @@ Seatbelt profile 默认允许，但带 `(deny file-write*)` 和写入 allow-list
 
 ## 模型体验
 
-通过 [`dsh-bash-sandbox`](../../bash/bash-sandbox/README.md) 和 [`dsh-tool-bash`](../../bash/tool-bash/README.md) 间接影响；它们渲染该提供方的强制执行与拒绝事实，而 [`dsh-sandbox`](../sandbox/README.md) seam 拥有 `SANDBOX_UNAVAILABLE` 文本，runner 选择与 profile 则不进入上下文。
+通过 [`dsh-bash-sandbox`](../../bash/bash-sandbox/README.md) 和 [`dsh-tool-bash`](../../bash/tool-bash/README.md) 间接影响；它们渲染该提供方的强制执行与拒绝事实，而 [`dsh-sandbox`](../sandbox/README.md) seam 负责定义 `SANDBOX_UNAVAILABLE` 文本，runner 选择与 profile 则不进入上下文。
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责。
+不会直接使 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **Windows 没有 runner**：`win32` 以 `SANDBOX_UNAVAILABLE` 快速失败；AppContainer 家族后端暂缓实现。
+- **Windows 没有 runner**：`win32` 以 `SANDBOX_UNAVAILABLE` 拒绝执行；AppContainer 家族后端暂缓实现。
 - **Landlock 可能只实现部分强制执行**：较旧且受支持的内核 ABI 只能限制自身公开的访问类别，因此报告 `enforcement: 'partial'`，不会夸大为完整强制执行。
-- **Seatbelt 依赖 deprecated 的 `sandbox-exec`**：macOS 仍会提供它，但若 Apple 移除该私有策略引擎，该提供方无法替换或探测。
+- **Seatbelt 依赖已弃用的 `sandbox-exec`**：macOS 仍会提供它，但若 Apple 移除该私有策略引擎，该提供方无法替换或探测。
 - **runner 选择在提供方生命周期内缓存**：安装、移除或修复 runner 后，必须重载插件才能改变选择。
-- **`runnerCommand` 是操作方断言**：配置的自定义 runner 会跳过功能探测，并假定它诚实实现 bwrap 形状 profile。
+- **`runnerCommand` 是操作方断言**：配置的自定义 runner 会跳过功能探测，并假定它诚实实现 bwrap 形式的 profile。
diff --git a/packages/sandbox/sandbox-policy/README.i18n.yaml b/packages/sandbox/sandbox-policy/README.i18n.yaml
index c324986a78..b21c8d885a 100644
--- a/packages/sandbox/sandbox-policy/README.i18n.yaml
+++ b/packages/sandbox/sandbox-policy/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/sandbox/sandbox-policy/README.md
 README.md: dca54330bc888af9ecac21aa92019d8a2b0140bd
-README.zh.md: abf2d9fb8830fdcaf7f1357b393b434de4a5ad8d
+README.zh.md: a201d48c81f563fc3d85495e964bb67432517a3c
diff --git a/packages/sandbox/sandbox-policy/README.zh.md b/packages/sandbox/sandbox-policy/README.zh.md
index abf2d9fb88..a201d48c81 100644
--- a/packages/sandbox/sandbox-policy/README.zh.md
+++ b/packages/sandbox/sandbox-policy/README.zh.md
@@ -2,30 +2,30 @@
 
 [English](README.md) | 中文
 
-沙箱策略解析的唯一 owner：部署默认 [`SandboxMode`](../sandbox/README.md) 与回退根目录，加上每个会话的持久模式覆盖和不可变 Workspace 根。每个执行强制限制的能力家族在每次调用时收到一项解析完成的模式与根策略。
+沙箱策略解析的唯一归属位置：部署默认 [`SandboxMode`](../sandbox/README.md) 与回退根目录，加上每个会话的持久模式覆盖和不可变工作区根目录。每个强制执行策略的能力家族在每次调用时都会收到一项解析完成的模式与根目录策略。
 
 ## 为何需要共享归属位置
 
-两个家族强制执行同一套模式词汇：沙箱化 bash 执行器（`@deepseek-ai/dsh-bash-sandbox`）与沙箱化文件系统提供方（`@deepseek-ai/dsh-fs-sandbox`）。如果两者各自解析 `mode` + `workspaceRoot`，就可能漂移成分裂世界：bash 限制在一个根目录，fs 却隔离另一个根目录，正是[沙箱 RFC](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)所警告的情况。两个工具层都通过 `ctx.sandboxPolicy` 解析策略，两个执行后端也都消费完整的逐调用结果。[跨家族 fs 沙箱 RFC](../../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)记录了共享策略决策。
+两个家族强制执行同一套模式词汇：沙箱化 bash 执行器（`@deepseek-ai/dsh-bash-sandbox`）与沙箱化文件系统提供方（`@deepseek-ai/dsh-fs-sandbox`）。如果两者各自解析 `mode` + `workspaceRoot`，就可能漂移成分裂世界：bash 限制在一个根目录，fs 却隔离另一个根目录，正是[沙箱 RFC](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)所警告的情况。两个工具层都通过 `ctx.sandboxPolicy` 解析策略，两个强制执行后端也都消费完整的逐调用结果。[跨家族 fs 沙箱 RFC](../../../.agents/notes/implemented/feature/2026-07-14-cross-family-fs-sandbox.md)记录了共享策略决策。
 
 ## 配置
 
 - `mode`：部署默认 `SandboxMode`（`read-only`／`workspace-write`／`danger-full-access`），加载时验证。默认为 `read-only`（故障安全）。
-- `workspaceRoot`：agentless 调用或没有 cwd 的会话在 `workspace-write` 下可写入的回退目录。默认为 `process.cwd()`，两种情况下都会解析为其绝对文件系统标识。普通 agent 调用改用其会话头中不可变的 `cwd`。
+- `workspaceRoot`：无 agent（智能体）的调用或没有 cwd 的会话在 `workspace-write` 下可写入的回退目录。默认为 `process.cwd()`；无论显式配置还是采用默认值，都会解析为其绝对文件系统标识。普通 agent 调用改用其会话头中不可变的 `cwd`。
 
-## 表层
+## 接口
 
 - `ctx.sandboxPolicy.resolve({ session?, mode? })`：解析一项完整的逐调用策略。显式批准的模式优先于会话最后一条 `sandbox/mode` 事件，后者又优先于 `defaultMode`；会话不可变的 `cwd` 会先按文件系统语义规范化，再成为 `workspaceRoot`，否则使用配置的回退值。规范化先于词法归一化，因此 `symlink/..` 与进程工作目录解析保持一致。
-- `ctx.sandboxPolicy.defaultMode`／`ctx.sandboxPolicy.workspaceRoot`：`resolve()` 使用的部署默认值与回退根。
+- `ctx.sandboxPolicy.defaultMode`／`ctx.sandboxPolicy.workspaceRoot`：`resolve()` 使用的部署默认值与回退根目录。
 - `effectiveSandboxMode(events)`：会话 `sandbox/mode` 事件的纯 fold（最后一次切换胜出，没有则为 `undefined`），在 `resolve()` 内使用。
 - `setSandboxMode(session, mode)`：逐会话覆盖的唯一写入路径：恰好追加一条 `sandbox/mode` 事件。切换本身就是事件；不会在带外修改模式。
 - `SANDBOX_MODES`：所有模式，用于选项展示与运行时验证。
 
-可选的 `./invariant` 配套组件会拒绝伪造的持久 `sandbox/mode` 事件，只要其值不在该封闭词汇中；Session 与其配套组件拥有周围的存储与核心执行封闭规则。
+可选的 `./invariant` 配套组件会拒绝伪造的持久 `sandbox/mode` 事件，只要其值不在该封闭词汇中；Session 与其配套组件负责相关存储与核心执行封闭规则。
 
-## 逐会话 store
+## 逐会话存储
 
-运行时切换是在对应会话日志中追加的一条 `sandbox/mode` 事件。`effective = explicit grant ?? fold(events) ?? deployment default`，因此覆盖会通过回放跨重启保留，两个会话也绝不会看到彼此状态。Workspace 标识无需另一条事件：创建时记录的不可变 `SessionHeader.cwd` 是该会话每次调用使用的根。该事件只进入日志（沿用 `approval/*` 先例）：模型通过强制执行工具的拒绝标记获知模式，绝不会从事件获知。
+运行时切换是在对应会话日志中追加的一条 `sandbox/mode` 事件。`effective = explicit grant ?? fold(events) ?? deployment default`，因此覆盖会通过回放跨重启保留，两个会话也绝不会看到彼此状态。工作区标识无需另一条事件：创建时记录的不可变 `SessionHeader.cwd` 是该会话每次调用使用的根。该事件只进入日志（沿用 `approval/*` 先例）：模型通过强制执行工具的拒绝标记获知模式，绝不会从事件获知。
 
 ## 模型体验
 
@@ -33,9 +33,9 @@
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责，且提示词有意不包含模式。
+不会直接使 KV Cache 失效；请求前缀变更由上述消费方负责，且提示词有意不包含模式。
 
 ## 已知限制与暂缓事项
 
-- **每个会话只有一个主要 Workspace 根**：策略解析 `SessionHeader.cwd`；额外可写根不属于 `SandboxExecutionPolicy`。
-- **只有文件 effect 模式**：`SandboxMode` 治理文件 effect；网络和进程策略不在其词汇中，因此这里没有限制它们的旋钮。
+- **每个会话只有一个主要工作区根目录**：策略解析 `SessionHeader.cwd`；额外可写根目录不属于 `SandboxExecutionPolicy`。
+- **仅限文件操作模式**：`SandboxMode` 管控文件操作；网络和进程策略不在其词汇中，因此这里没有限制它们的旋钮。
diff --git a/packages/sandbox/sandbox/README.i18n.yaml b/packages/sandbox/sandbox/README.i18n.yaml
index 6619765ed7..83b28c41b8 100644
--- a/packages/sandbox/sandbox/README.i18n.yaml
+++ b/packages/sandbox/sandbox/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sandbox/sandbox/README.md
 README.md: 99f0641560937f66df6db76ae55c90595329792f
-README.zh.md: 402c24135359137b5acde10e11aa676c2460584a
+README.zh.md: 5d7dfb303ea2405e725d162698c3ab3d39485636
diff --git a/packages/sandbox/sandbox/README.zh.md b/packages/sandbox/sandbox/README.zh.md
index 402c241353..5d7dfb303e 100644
--- a/packages/sandbox/sandbox/README.zh.md
+++ b/packages/sandbox/sandbox/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-抽象进程沙箱 seam。拥有 `ctx.sandbox` 服务契约（[`SandboxProvider`](src/index.ts)）与 harness 共享的限制词汇：`SandboxMode`（`read-only`／`workspace-write`／`danger-full-access`，仅限文件 effect）、`SandboxEnforcement`（`full`／`partial`，逐内核 ABI）、`SandboxExecutionPolicy`（完整的逐调用模式 + Workspace 根）、`SandboxPolicy`（其中受限制的子集），以及快速失败的 `SANDBOX_UNAVAILABLE` 错误。它是[能力 seam 拆分](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)的接口包：只依赖 cordis（及 harness 错误基类），绝不依赖后端。
+抽象进程沙箱 seam。负责定义 `ctx.sandbox` 服务契约（[`SandboxProvider`](src/index.ts)）与 harness 共享的限制词汇：`SandboxMode`（`read-only`／`workspace-write`／`danger-full-access`，仅限文件操作）、`SandboxEnforcement`（`full`／`partial`，针对每种内核 ABI）、`SandboxExecutionPolicy`（每次调用的完整模式及工作区根目录）、`SandboxPolicy`（其中受限制的子集），以及故障时拒绝放行的 `SANDBOX_UNAVAILABLE` 错误。它是[能力 seam 拆分](../../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)的接口包（package）：只依赖 cordis（及 harness 错误基类），绝不依赖后端。
 
-用一句话概括契约：`ctx.sandbox.confine(argv, policy)` 返回应当取代自有 argv 用于 spawn 的 argv。返回值经过包装，使进程及其 spawn 的一切都在限制下运行；另附所选后端的两个事实：它达到的强制执行完整度，以及拒绝方言（`denialSignatures`，即内核在文件 effect 被拒绝时打印到 stderr 的子字符串；通过 stderr 推断的消费方会匹配这些字符串，而不是跨后端联合）。没有可用后端时，它会抛出异常，绝不会原样传递 argv 使其不受限制地运行。
+用一句话概括契约：`ctx.sandbox.confine(argv, policy)` 返回用于 spawn、应当取代调用方原始 argv 的 argv。返回值经过包装，使进程及其派生的所有进程都在限制下运行；另附所选后端的两个事实：它达到的强制执行完整度，以及拒绝方言（`denialSignatures`，即内核在文件操作被拒绝时打印到 stderr 的子字符串；通过 stderr 推断的消费方会匹配这些字符串，而不是统一的跨后端签名集合）。没有可用后端时，它会抛出异常，绝不会原样传递 argv 使其不受限制地运行。
 
-策略随调用传递，而不属于提供方：两个消费方可以同时按不同策略施加限制（bash 使用 `read-only`，而受限制子 agent 保持其状态目录可写）；获批的升权重试只是使用更宽策略发起的新调用。
+策略随调用传递，而不属于提供方：两个消费方可以同时按不同策略施加限制（bash 使用 `read-only`，而受限制的子 agent（智能体）保持其状态目录可写）；获批的升权重试只是使用更宽策略发起的新调用。
 
-**只支持与宿主共享文件系统和内核的限制。** 后端与宿主共享文件系统和内核（`bwrap`、Landlock、Seatbelt）；`workspaceRoot` 指向文件系统规范化后的真实主机目录。系统先解析 Workspace 标识，再做词法规范化，因此包含 `symlink/..` 的有效 cwd 会授权 `chdir` 实际到达的目录，而非无关的词法父目录。容器、microVM 与远程执行器都不是该 seam 的后端：它们会以环境一致的分组替换整个能力实现（`ctx.bash`、`ctx.fs`）。边界及其原理见[沙箱 Agent Note](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)。
+**只支持与宿主共享文件系统和内核的限制。** 后端与宿主共享文件系统和内核（`bwrap`、Landlock、Seatbelt）；`workspaceRoot` 指向文件系统规范化后的真实主机目录。系统先解析工作区所指的目录，再做词法规范化，因此包含 `symlink/..` 的有效 cwd 会授权 `chdir` 实际到达的目录，而非无关的词法父目录。容器、microVM 与远程执行器都不是该 seam 的后端：它们会以环境一致的分组替换整个能力实现（`ctx.bash`、`ctx.fs`）。边界及其设计理由见[沙箱 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-06-sandbox.md)。
 
-实现：[`@deepseek-ai/dsh-sandbox-local`](../sandbox-local/)（Linux：`bwrap`，否则使用逐平台 Landlock launcher；macOS：`sandbox-exec`／Seatbelt）。消费方：[`@deepseek-ai/dsh-bash-sandbox`](../../bash/bash-sandbox/)（包装 `['bash', '-c', command]`）。
+实现：[`@deepseek-ai/dsh-sandbox-local`](../sandbox-local/)（Linux：`bwrap`，否则使用相应平台的 Landlock launcher；macOS：`sandbox-exec`／Seatbelt）。消费方：[`@deepseek-ai/dsh-bash-sandbox`](../../bash/bash-sandbox/)（包装 `['bash', '-c', command]`）。
 
 ## 模型体验
 
@@ -18,7 +18,7 @@
 
 #### 模型看到的内容
 
-通过 [`dsh-bash-sandbox`](../../bash/bash-sandbox/README.md) 和 [`dsh-tool-bash`](../../bash/tool-bash/README.md)，无法强制执行所请求模式时会生成 code `SANDBOX_UNAVAILABLE` 及以下精确错误。执行期 runner 失败会追加 ` Runner failure: <detail>`。
+通过 [`dsh-bash-sandbox`](../../bash/bash-sandbox/README.md) 和 [`dsh-tool-bash`](../../bash/tool-bash/README.md)，无法强制执行所请求模式时会产生错误码 `SANDBOX_UNAVAILABLE` 及以下精确错误。执行期 runner 失败会追加 ` Runner failure: <detail>`。
 
 ##### 精确错误
 
@@ -28,15 +28,15 @@ sandbox mode "<mode>" is requested but no sandbox backend is usable on this host
 
 #### Token 影响
 
-条件性错误文本对该次调用可见，并保留在历史中直到压缩。
+条件性错误文本对该次调用可见，并保留在历史中直到压缩（compaction）。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
-- **文件 effect 是完整的策略词汇**：该 seam 不表达网络、进程、syscall、设备或 credential 限制。
+- **文件操作是完整的策略词汇**：该 seam 不表达网络、进程、系统调用、设备或凭据限制。
 - **只支持与宿主共享文件系统和内核的限制**：容器、microVM 与远程执行需要替换能力实现，而不是在此处增加提供方。
 - **拒绝报告是一种 stderr 方言**：该 seam 返回后端签名，而非类型化运行时拒绝通道，因此需要分类的消费方必须从子进程输出推断。
 - **每个上下文只有一个提供方**：同时组合不同沙箱机制需要提供方级阶梯或独立 Cordis 上下文；调用方逐调用选择策略，而非后端标识。
diff --git a/packages/sdk/README.i18n.yaml b/packages/sdk/README.i18n.yaml
index e433146312..0602769f1c 100644
--- a/packages/sdk/README.i18n.yaml
+++ b/packages/sdk/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sdk/README.md
 README.md: 0dcf3655fc4d20452981f83000f9b9b586ede17b
-README.zh.md: 1a692de2f7dc56cc3dbc7e40b8e2369d0bf2e8c5
+README.zh.md: 215e2ddc47151d8828cbfaee7eae48d517bdad47
diff --git a/packages/sdk/README.zh.md b/packages/sdk/README.zh.md
index 1a692de2f7..215e2ddc47 100644
--- a/packages/sdk/README.zh.md
+++ b/packages/sdk/README.zh.md
@@ -1,19 +1,19 @@
-# SDK 包
+# SDK 包（package）
 
 [English](README.md) | 中文
 
 用于创建、编辑、构建和运行 DeepSeek Harness 项目的开发者工具，外加从另一进程驱动 harness 运行时的客户端 SDK 栈。
 
-[功能 Agent Note](../../.agents/notes/proposed/feature/2026-07-14-sdk-developer-projects.md)负责开发者工作流；[架构 Agent Note](../../.agents/notes/proposed/architecture/2026-07-15-sdk-project-editing-architecture.md)负责包与项目编辑边界；[TypeScript SDK Agent Note](../../.agents/notes/implemented/feature/2026-07-27-typescript-sdk-and-sdk-subagent-backend.md)负责客户端 SDK 栈。
+[功能 Agent Note（agent 决策记录）](../../.agents/notes/proposed/feature/2026-07-14-sdk-developer-projects.md)负责开发者工作流；[架构 Agent Note](../../.agents/notes/proposed/architecture/2026-07-15-sdk-project-editing-architecture.md)负责包与项目编辑边界；[TypeScript SDK Agent Note](../../.agents/notes/implemented/feature/2026-07-27-typescript-sdk-and-sdk-subagent-backend.md)负责客户端 SDK 栈。
 
 | 包 | 职责 |
 |---|---|
 | [`helper`](helper/README.md) | 项目聚合、编辑会话、内置功能、项目文档、模板、包管理器与提示词抽象 |
 | [`scripts`](scripts/README.md) | `dsh-sdk` 启动器：`start`、`dev`、`build` 和交互式 `config` |
 | [`create-sdk`](create-sdk/README.md) | `npm create @deepseek-ai/sdk` 初始化器 |
-| [`sdk-protocol`](sdk-protocol/README.md) | 共享的 SDK 运行时线协议：按换行分帧的 JSON-RPC 传输 + 具名请求/通知类型 |
-| [`sdk-client`](sdk-client/README.md) | TypeScript 客户端 SDK：走 stdio JSON-RPC 驱动 harness 运行时子进程（Python SDK 的设计孪生） |
+| [`sdk-protocol`](sdk-protocol/README.md) | 共享的 SDK 运行时通信协议：以换行符分隔的 JSON-RPC 传输 + 具名请求/通知类型 |
+| [`sdk-client`](sdk-client/README.md) | TypeScript 客户端 SDK：通过 stdio JSON-RPC 驱动 harness 运行时子进程（Python SDK 的设计孪生） |
 
 `@deepseek-ai/create-sdk` 是仓库 `@deepseek-ai/dsh-*` 命名规则的唯一例外：npm 的 scoped initializer 约定要求使用该名称，才能支持 `npm create @deepseek-ai/sdk`。
 
-生成的项目始终以 `cordis.yml` 作为唯一运行时插件树。`dsh-sdk dev` 只是在同一文件周围增加 TypeScript 与本地工作区解析，不会创建仅供开发环境使用的配置。
+生成的项目始终以 `cordis.yml` 作为唯一运行时插件树。`dsh-sdk dev` 只是围绕同一文件增加 TypeScript 与本地工作区解析，不会创建仅供开发环境使用的配置。
diff --git a/packages/sdk/create-sdk/README.i18n.yaml b/packages/sdk/create-sdk/README.i18n.yaml
index 9fe9c61cb9..27609d9a11 100644
--- a/packages/sdk/create-sdk/README.i18n.yaml
+++ b/packages/sdk/create-sdk/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sdk/create-sdk/README.md
 README.md: aa09236832a50abdcd2b158e0561db26bce19cf6
-README.zh.md: 7900f243f944b61cdf114bf63ec7fd5c177806d7
+README.zh.md: a4685e66bc3eb7924dd247a7b6b420f6bcc70990
diff --git a/packages/sdk/create-sdk/README.zh.md b/packages/sdk/create-sdk/README.zh.md
index 7900f243f9..a4685e66bc 100644
--- a/packages/sdk/create-sdk/README.zh.md
+++ b/packages/sdk/create-sdk/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-用于 `npm create @deepseek-ai/sdk [directory]` 的交互式初始化器。目录／名称／描述都提供可见且可编辑的默认值。树形选择器用于选择功能；可选项通过 Right／Left 导航配置，只有选中相应选项后才会询问密钥文本。本地插件创建提供 none／plugin／tool 三选一。
+用于 `npm create @deepseek-ai/sdk [directory]` 的交互式初始化器。目录／名称／描述都提供可见且可编辑的默认值。树形选择器用于选择功能，并通过 Right／Left 导航配置取值有限的选项；只有选中相应选项后才会询问密钥文本。本地插件创建提供 none／plugin／tool 三选一。
 
 受支持的包接口是 `create-sdk` bin。包根不导出任何符号，也不导出 workflow、bin、source 或 package-manifest 子路径。
 
@@ -10,7 +10,7 @@
 
 公开标志包括 `[directory]`、`--description`、`--provider`、`--base-url`、`--api-key`、`--model`、`--interface`、`--pm`、`--install`／`--no-install`，以及无头模式标志 `--config <path>`／`--config-json <json>` 和 `--json`。交互式标志会预填对应问题；无头 spec（`--config`／`--config-json`）会预先提供所有答案和功能方案，因此创建过程无需 TTY，并通过 `HeadlessPromptPort` 驱动；若缺少任何必填答案，该端口会明确失败。`--json` 会发送 NDJSON 生命周期事件（`done`／`action-required`／`error`），使 agent（智能体）能够补充其中点名的缺失输入并重新运行。
 
-提供方可以选择 DeepSeek，也可以选择由 `llm-pi-ai` 支持的自定义端点。选择 DeepSeek 时只询问 API key，并使用公共端点与 `deepseek-v4-flash`；自定义端点还会询问 base URL。密钥为空时必须确认；系统会创建包含注释和空 `.env` 变量的文件，使提供方在填写变量前启动时明确失败。现有插件的默认值会被省略；必填 SDK 预设仍按所属包的 Config 保持类型约束。
+提供方可以选择 DeepSeek，也可以选择由 `llm-pi-ai` 支持的自定义端点。选择 DeepSeek 时只询问 API key，并使用公共端点与 `deepseek-v4-flash`；自定义端点还会询问 base URL。密钥为空时必须确认；系统会在 `.env` 中创建一个被注释掉的空变量，从而使提供方在为该变量填入值之前启动时明确失败。现有插件的默认值会被省略；必填 SDK 预设仍按所属包的 Config 保持类型约束。
 
 ## 模型体验
 
@@ -18,8 +18,8 @@
 
 #### KV Cache 影响
 
-不会直接失效；由具名消费方负责请求前缀变更。
+不会直接导致 KV Cache 失效；由具名消费方负责请求前缀变更。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **无头本地插件**：无头 spec 会提供项目答案和功能方案；目前还不能在 spec 中表达本地插件脚手架（交互式 none／plugin／tool 选择），默认使用 none。
diff --git a/packages/sdk/helper/README.i18n.yaml b/packages/sdk/helper/README.i18n.yaml
index 4e25e5c082..8842e34d26 100644
--- a/packages/sdk/helper/README.i18n.yaml
+++ b/packages/sdk/helper/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sdk/helper/README.md
 README.md: 8c6ed9e87be0a97af67849793edb7fa30ffb33ab
-README.zh.md: 6798962acd4f282068394cae9bc40aace2288879
+README.zh.md: e8844aab14dd0494144251de5e0a9f5cfcaf37be
diff --git a/packages/sdk/helper/README.zh.md b/packages/sdk/helper/README.zh.md
index 6798962acd..e8844aab14 100644
--- a/packages/sdk/helper/README.zh.md
+++ b/packages/sdk/helper/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-供 `create-sdk` 与 `dsh-sdk config` 共用的项目领域和基础设施。`SdkProject` 是只读快照；`ProjectEditSession` 是唯一的变更与提交边界。设计理由由 [SDK 架构 Agent Note](../../../.agents/notes/proposed/architecture/2026-07-15-sdk-project-editing-architecture.md)负责。
+供 `create-sdk` 与 `dsh-sdk config` 共用的项目领域和基础设施。`SdkProject` 是只读快照；`ProjectEditSession` 是唯一的变更与提交边界。设计理由由 [SDK 架构 Agent Note（agent 决策记录）](../../../.agents/notes/proposed/architecture/2026-07-15-sdk-project-editing-architecture.md) 负责。
 
-该包负责内置的类型化 spec 目录、提供方／应用行为实体、结构化项目文件对象、helper 自有项目模板、共享的类型化 `TextTemplate` 渲染器、包管理器策略、本地插件蓝图、类型化问题，以及 clack 提示词适配器。它绝不会启动 Cordis 应用。
+该包（package）负责内置的类型化 spec 目录、提供方／应用行为实体、结构化项目文件对象、helper 自有项目模板、共享的类型化 `TextTemplate` 渲染器、包管理器策略、本地插件蓝图、类型化问题，以及 clack 交互提示适配器。它绝不会启动 Cordis 应用。
 
 所有业务验证与文档验证都会在提交写入任何受影响文件前完成。提交会检测编辑会话打开后发生的外部修改，但在开始写入后，有意不提供跨文件回滚。
 
-内置功能包括 provider、bash、app、persistence、HMR、filesystem、todo、skill、web、subagent、workflow、compaction、hooks、repeat-tool guard、timeout policy 和 ask-user。目录负责功能选项、必填和非默认 Cordis 插件配置、功能依赖、资源贡献与往返标记；create 与 config 使用同一注册表和配置器。ACP 应用选项只贡献自动化桥；交互式服务属于 TUI 或 Web 组合。
+内置功能包括提供方、bash、app、持久化、HMR（热模块替换）、filesystem、todo、skill（技能）、web、subagent、工作流、压缩（compaction）、钩子、repeat-tool guard、timeout policy 和 ask-user。目录负责功能选项、必需和非默认 Cordis 插件配置、功能依赖、资源贡献与往返标记；create 与 config 使用同一注册表和配置器。ACP（Agent Client Protocol）应用选项只贡献自动化桥；交互式服务属于 TUI 或 Web 组合。
 
-`SdkProject.open()` 只要求根目录下的 `package.json` 和 `cordis.yml` 可读。Cordis 配置项用于锚定功能安装；如果某个包只存在于链接的 NPM 依赖闭包中，则该功能仍视为不存在。一旦所属的 Cordis 配置项存在，资源形状不完整就是 `inconsistent`，无法自动修改。
+`SdkProject.open()` 只要求根目录下的 `package.json` 和 `cordis.yml` 可读。Cordis 配置项用于锚定功能安装；如果某个包只存在于链接的 NPM 依赖闭包中，则该功能仍视为不存在。一旦所属的 Cordis 配置项存在，资源结构不完整就是 `inconsistent`，无法自动修改。
 
 `.env.example` 跟随当前所选功能。`.env` 仅追加：helper 可以补充缺失且名称不同的变量，但绝不会更新或删除现有内容。
 
@@ -18,12 +18,12 @@
 
 ## 模型体验
 
-无。项目领域只编辑文件，绝不会挂载活跃 agent 或模型请求。
+无。项目领域只编辑文件，绝不会挂载运行中的 agent（智能体），也不会发起模型请求。
 
 #### KV Cache 影响
 
 无；此包既不组装也不发送提供方请求。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **提交不具备跨文件事务性**：每次写入前都会检测外部修改，但后续失败不会回滚已经写入的文件。
diff --git a/packages/sdk/helper/src/package-managers/package-manager.ts b/packages/sdk/helper/src/package-managers/package-manager.ts
index 3d67d194d3..5c070afce7 100644
--- a/packages/sdk/helper/src/package-managers/package-manager.ts
+++ b/packages/sdk/helper/src/package-managers/package-manager.ts
@@ -5,7 +5,7 @@
  */
 
 import { execFile, spawn } from 'node:child_process'
-import { scrubbedParentEnv } from '@deepseek-ai/dsh-subprocess'
+import { scrubbedParentEnv, SENSITIVE_ENV_PATTERN } from '@deepseek-ai/dsh-subprocess'
 import { promisify } from 'node:util'
 import type { PackageJsonFile } from '../documents/package-json-file.ts'
 import { PnpmWorkspaceFile } from '../documents/pnpm-workspace-file.ts'
@@ -60,7 +60,7 @@ export async function probePackageManagerVersion(name: PackageManagerName, cwd:
  */
 export function scrubEnvironment(environment?: NodeJS.ProcessEnv): NodeJS.ProcessEnv {
   if (environment === undefined) return scrubbedParentEnv()
-  return Object.fromEntries(Object.entries(environment).filter(([name]) => !/(?:KEY|SECRET|TOKEN)/i.test(name)))
+  return Object.fromEntries(Object.entries(environment).filter(([name]) => !SENSITIVE_ENV_PATTERN.test(name)))
 }
 
 /** Node child-process command runner with inherited stdio and quiescent completion. */
diff --git a/packages/sdk/helper/tests/documents.spec.ts b/packages/sdk/helper/tests/documents.spec.ts
index 1e4b944f00..6ef18b12a2 100644
--- a/packages/sdk/helper/tests/documents.spec.ts
+++ b/packages/sdk/helper/tests/documents.spec.ts
@@ -309,7 +309,12 @@ describe('package manager strategies', () => {
     await expect(npm.install('/tmp', failed)).rejects.toThrow('exited with code 2')
     const killed: CommandRunner = { run: async () => ({ exitCode: null, signal: 'SIGTERM' }) }
     await expect(npm.build('/tmp', killed)).rejects.toThrow('killed by SIGTERM')
-    expect(scrubEnvironment({ PATH: '/bin', API_KEY: 'secret', TOKEN_VALUE: 'secret' })).toEqual({ PATH: '/bin' })
+    expect(scrubEnvironment({
+      PATH: '/bin',
+      API_KEY: 'secret',
+      DB_PASSWORD: 'secret',
+      TOKEN_VALUE: 'secret',
+    })).toEqual({ PATH: '/bin' })
   })
 
   it('probes versions and runs real child-process boundaries', async () => {
diff --git a/packages/sdk/scripts/README.i18n.yaml b/packages/sdk/scripts/README.i18n.yaml
index 651cd50e81..b148eb76ab 100644
--- a/packages/sdk/scripts/README.i18n.yaml
+++ b/packages/sdk/scripts/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sdk/scripts/README.md
 README.md: 9a696bf5a4de9a80f0741f07a7e753733bc2f998
-README.zh.md: 85c2e52c7bca0a9b7c40ebd6ede7c40f113168de
+README.zh.md: cf9fc2cbbe467eb86e89d9eb77e9bc34860b5a3a
diff --git a/packages/sdk/scripts/README.zh.md b/packages/sdk/scripts/README.zh.md
index 85c2e52c7b..cf9fc2cbbe 100644
--- a/packages/sdk/scripts/README.zh.md
+++ b/packages/sdk/scripts/README.zh.md
@@ -10,19 +10,19 @@
 | `dsh-sdk dev [target] [-- args…]` | 注册 TypeScript 与本地工作区源代码解析，然后进入 start 路径 |
 | `dsh-sdk build [args…]` | 使用项目参数调用项目已安装的 tsdown |
 | `dsh-sdk config` | 打开一个交互式编辑会话，审阅累计变更，统一提交一次；NPM 依赖变化时只安装一次 |
-| `dsh-sdk create <source>` | 从原生包管理器来源（`pkg@version` 或 `github:owner/repo#ref`）添加外部 Cordis 插件：确认后执行 `<pm> add <source>`，再将解析出的依赖挂载到 `cordis.yml`。不使用 giget／pacote；由包管理器解析并固定来源（GitHub 依赖会在管理器策略下通过自身 `prepare` 构建） |
+| `dsh-sdk create <source>` | 从包管理器原生支持的来源（`pkg@version` 或 `github:owner/repo#ref`）添加外部 Cordis 插件：确认后执行 `<pm> add <source>`，再将解析出的依赖挂载到 `cordis.yml`。不使用 giget／pacote；由包管理器解析并固定来源（GitHub 依赖会在管理器策略下通过自身 `prepare` 构建） |
 
 `ProjectBuild(tsdownConfig)` 与 `PluginBuild(tsdownConfig)` 只从 `@deepseek-ai/dsh-scripts/dev/tsdown-config` 导出。开发环境与生产环境读取同一个 `cordis.yml`。
 
-生成项目的脚本通过 `dsh-sdk` 执行 dev、build、start 和 config；类型检查直接运行 `tsc -b`。HMR 始终是显式的 `cordis.yml` 功能，并由 dev 与 start 同时加载。
+生成项目的脚本通过 `dsh-sdk` 执行 dev、build、start 和 config；类型检查直接运行 `tsc -b`。HMR（热模块替换）始终是显式的 `cordis.yml` 功能，并由 dev 与 start 同时加载。
 
 运行时库导出 `startSDK(source)`，用于加载 `.env` 和 `cordis.yml` 并返回活跃上下文；还导出 `runSDK(target)`，用于导入项目模块并调用其 `main(bootContext)`（不带目标的 `runSDK()` 会委派给 `startSDK('./cordis.yml')`）。`SdkBootContext` 携带原样转发的 `argv`、通用 `args`、启动器的绝对 `cwd`，以及 `start`／`dev` 模式。启动器不声明项目选项：Node `parseArgs()` 使用空 schema 运行，因此带值的标志写作 `--key=value`，裸标志变为布尔值，`--no-cache` 变为 `args.cache = false`，选项名称保留 Node 的拼写（`--max-depth=3` → `args['max-depth']`）。
 
 `start` 绝不构建。`dev` 注册项目已安装的 tsx 转换，并建立从 `plugins/*/package.json` 中的精确包名到各自 `src/index.ts` 的映射，然后沿用相同的 start 路径。`build` 调用项目已安装的 tsdown 并转发其参数；缺少 tsdown 配置时视为成功且不执行操作。
 
-`config` 要求 TTY。一个功能树用于选择期望的启用集合；变更行会高亮，Right 用于修改有限功能选项，必填行无法取消选择，不一致行会显示诊断，自定义／手动 Cordis 配置项支持启用／禁用。工作流会将该目标协调到一个编辑会话中。Review & Apply 只提交一次；之后，如果 NPM 依赖有变更，则触发一次包管理器安装。安装失败不会撤销已提交文件。
+`config` 要求 TTY。一个功能树用于选择期望的启用集合；变更行会高亮，Right 用于修改取值有限的功能选项，必需行无法取消选择，不一致行会显示诊断，自定义／手动 Cordis 配置项支持启用／禁用。工作流会在一个编辑会话中将配置协调至该目标状态。Review & Apply 只提交一次；之后，如果 NPM 依赖有变更，则触发一次包管理器安装。安装失败不会撤销已提交文件。
 
-根库导出 `startSDK`、`runSDK` 以及 `SdkBootArgs`／`SdkBootContext` 类型；命令组合仍由 bin 私有持有。不导出 `src/*`、bin 或 package-manifest 子路径。
+根库导出 `startSDK`、`runSDK` 以及 `SdkBootArgs`／`SdkBootContext` 类型；命令组合仍是 bin 的私有实现。不导出 `src/*`、bin 或 package-manifest 子路径。
 
 ## 模型体验
 
@@ -30,8 +30,8 @@
 
 #### KV Cache 影响
 
-不会直接失效；由具名消费方负责请求前缀变更。
+不会直接导致 KV Cache 失效；由具名消费方负责请求前缀变更。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - **启动器参数没有 schema**：`start` 和 `dev` 会保留 Node `parseArgs()` 输出，而不会验证项目专用标志。
diff --git a/packages/sdk/sdk-client/README.i18n.yaml b/packages/sdk/sdk-client/README.i18n.yaml
index 30fa3a1a99..c0bf2706ff 100644
--- a/packages/sdk/sdk-client/README.i18n.yaml
+++ b/packages/sdk/sdk-client/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/sdk/sdk-client/README.md
 README.md: 3ac4de540401f6f40dab3e84f7005f91d024aee8
-README.zh.md: 1d9f8fbded8b477d519a5244735179fba581cdd0
+README.zh.md: 95fb6b2887a7bc748a610f7fedae4be1aa2af623
diff --git a/packages/sdk/sdk-client/README.zh.md b/packages/sdk/sdk-client/README.zh.md
index 1d9f8fbded..95fb6b2887 100644
--- a/packages/sdk/sdk-client/README.zh.md
+++ b/packages/sdk/sdk-client/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-以子进程方式驱动 DeepSeek Harness 运行时、走 stdio JSON-RPC 的 TypeScript 客户端 SDK——[Python SDK](../../../python/README.md)（`deepseek-harness`）的设计孪生，共享同一个运行时对端、协议与分层：`DeepSeekHarness` 是高层回合 API，`HarnessClient` 是低层协议客户端。包（package）根枚举消费方接口：两层客户端、面向调用方的类型和 `JsonRpcResponseError`；源模块、规范化辅助函数与订阅投递机制不供消费方导入。纯库：不在任何 Cordis 上下文注册；它所生成的运行时进程是一个完整 harness，其组成由自己的 `cordis.yml` 决定。
+以子进程方式驱动 DeepSeek Harness 运行时、走 stdio JSON-RPC 的 TypeScript 客户端 SDK——[Python SDK](../../../python/README.md)（`deepseek-harness`）的设计孪生，共享同一个运行时对端、协议与分层：`DeepSeekHarness` 是高层轮次 API，`HarnessClient` 是低层协议客户端。包（package）根枚举消费方接口：两层客户端、面向调用方的类型和 `JsonRpcResponseError`；源模块、规范化辅助函数与订阅投递机制不供消费方导入。纯库：不在任何 Cordis 上下文注册；它所 spawn 的运行时进程是一个完整 harness，其组成由自己的 `cordis.yml` 决定。
 
-与 Python SDK 不同，启动规格完全显式（`command`/`args`）：本包面向仓库近旁的 TypeScript 消费者——[`dsh-subagent-dsh-sdk`](../../subagent/subagent-dsh-sdk/README.md) 后端、测试、自动化——它们知道自己要启动哪个运行时。捆绑运行时解析（寻找打包可执行文件）仍归 Python 发行版负责。
+与 Python SDK 不同，启动规格完全显式（`command`/`args`）：本包面向仓库近旁的 TypeScript 消费方——[`dsh-subagent-dsh-sdk`](../../subagent/subagent-dsh-sdk/README.md) 后端、测试、自动化——它们知道自己要启动哪个运行时。捆绑运行时解析（寻找打包可执行文件）仍归 Python 发行版负责。
 
 ## DeepSeekHarness
 
@@ -21,31 +21,31 @@ const result = await harness.run('say hi')
 console.log(result.status, result.finalResponse)
 ```
 
-子进程在首次使用时惰性启动，并在多次 `run()` 之间持续归实例所有；必须 `close()`（或 `await using`），子进程才总能被收割。`start()` 记忆化 `initialize` 握手（工作区 cwd——在跨越线之前解析为绝对路径——加 provider/model 路由和可选的正整数 `maxTokens` 输出上限）；握手失败会收割运行时并换入全新客户端，后续调用用新子进程重试（直到终结性的 `close()`）。该上限作用于根 agent 的每次请求，并由进程内后代继承；压缩插件单独持有摘要上限。`session(id?)` 打开具名或全新的会话句柄；`run(input, { sessionId?, onNotification? })` 发送一个 prompt 回合，在配对的 `session.finished` 到达时尘埃落定，返回 `TurnResult`：`status`（按部署映射的 `ok`/`error`）、结构化 `reason`（`TurnEndReason`）、`finalResponse`（最后一条助手消息文本）、根会话的 `events`，以及该会话和通过 `subagent.started` 发现的后代的原始 `notifications`，均按线序排列。模型层失败是 `status: 'error'` 的结果，绝不是拒绝；拒绝意味着传输丢失、超时或协议违例。
+子进程在首次使用时惰性启动，并在多次 `run()` 之间持续归实例所有；必须 `close()`（或 `await using`），子进程才总能被回收。`start()` 记忆化 `initialize` 握手（工作区 cwd——在通过协议传输之前解析为绝对路径——加 provider/model 路由和可选的正整数 `maxTokens` 输出上限）；握手失败会回收运行时并换入全新客户端，后续调用用新子进程重试（直到终结性的 `close()`）。该上限作用于根 agent（智能体）的每次请求，并由进程内后代继承；压缩（compaction）插件单独持有摘要上限。`session(id?)` 打开具名或全新的会话句柄；`run(input, { sessionId?, onNotification? })` 发送一个提示词轮次，在配对的 `session.finished` 到达时完成，并返回 `TurnResult`：`status`（按部署映射的 `ok`/`error`）、结构化 `reason`（`TurnEndReason`）、`finalResponse`（最后一条助手消息文本）、根会话的 `events`，以及该会话和通过 `subagent.started` 发现的后代的原始 `notifications`，均按协议传输顺序排列。模型层失败会返回 `status: 'error'` 的结果，绝不会导致 Promise 被拒绝；Promise 被拒绝意味着传输丢失、超时或协议违例。
 
 ## HarnessClient
 
-回合 API 之下的协议客户端：显式 `start()`/`initialize()`/`prompt()`/`request()`/`close()`，外加通知订阅。`subscribe(filter?)` 返回 `NotificationSubscription`（可等待的 `next()`、非阻塞 `tryNext()`、异步迭代）；`subscribeSessionTree(id)` 把范围限定到一个会话及从 `subagent.started` 血缘边发现的后代——运行时对上下文内每个会话都发通知，范围限定在客户端完成，与 Python SDK 完全一致。错误表面有类型且由本包导出：`JsonRpcResponseError`（线上错误响应，保留 code/data）、`RequestTimeoutError`（配置的时限已到；线上没有取消方法，请求在服务端继续运行直到 close）、`SdkProtocolError`（响应超出文档化协议）、`TransportClosedError`（运行时已消失——消息携带退出码与有界 stderr 尾部）。
+轮次 API 之下的协议客户端：显式 `start()`/`initialize()`/`prompt()`/`request()`/`close()`，外加通知订阅。`subscribe(filter?)` 返回 `NotificationSubscription`（可等待的 `next()`、非阻塞 `tryNext()`、异步迭代）；`subscribeSessionTree(id)` 把范围限定到一个会话及从 `subagent.started` 血缘边发现的后代——运行时对上下文内每个会话都发通知，范围限定在客户端完成，与 Python SDK 完全一致。本包导出有明确类型的错误：`JsonRpcResponseError`（协议错误响应，保留 code/data）、`RequestTimeoutError`（配置的时限已到；协议层没有取消机制，请求在服务端继续运行直到 close）、`SdkProtocolError`（响应超出文档化协议）、`TransportClosedError`（运行时已消失——消息携带退出码与有界 stderr 尾部）。
 
-`close()` 先请求协议 `shutdown`（受 `shutdownTimeoutMs` 约束，默认 1000 毫秒），然后走 stdin-EOF → SIGTERM → SIGKILL 阶梯（`disposeEofGraceMs` 默认 6000，`disposeGraceMs` 默认 3000）直到进程真正退出。该阶梯为本客户端私有：它运行在任何 harness 上下文之外，无法搭乘 [`dsh-subprocess`](../../subprocess/README.md) 服务——即该接缝记载的 SDK 托管传输例外。幂等，已关闭的客户端拒绝复用。
+`close()` 先请求协议 `shutdown`（受 `shutdownTimeoutMs` 约束，默认 1000 毫秒），然后走 stdin-EOF → SIGTERM → SIGKILL 阶梯（`disposeEofGraceMs` 默认 6000，`disposeGraceMs` 默认 3000）直到进程真正退出。该阶梯为本客户端私有：它运行在任何 harness 上下文之外，无法搭乘 [`dsh-subprocess`](../../subprocess/README.md) 服务——即该 seam 所记录的 SDK 托管传输例外。幂等，已关闭的客户端拒绝复用。
 
-`HarnessClientOptions.env` 给定时整体替换子环境（`undefined` 原样继承父环境）；凭据策略归调用方——`dsh-subprocess` 的 `scrubbedParentEnv` 是面向隔离启动的共享擦除基底。
+`HarnessClientOptions.env` 给定时整体替换子进程环境（`undefined` 原样继承父进程环境）；凭据策略归调用方——`dsh-subprocess` 的 `scrubbedParentEnv` 是面向隔离启动的共享擦除基底。
 
 ## 测试
 
-免密钥单元测试通过真实 stdio 驱动一个脚本化伪运行时子进程（`tests/fake-runtime.ts`，纯协议、环境变量脚本化）：回合循环、会话树范围限定、超时/死亡/畸形响应表面、处置阶梯。[SDK 快照套件](../../../examples/jsonrpc-agent/tests/sdk.snapshot.ts)经由 `llm-replay` 免密钥地通过本客户端驱动真实 `dsh-jsonrpc-agent` 运行时，钉住通知流、回合结果与持久化日志；`DSH_SNAPSHOT=record` 对真实 API 重录。
+免密钥单元测试通过真实 stdio 驱动一个脚本化伪运行时子进程（`tests/fake-runtime.ts`，纯协议、环境变量脚本化）：轮次循环、会话树范围限定、超时、进程死亡和响应畸形场景，以及 dispose（资源释放）阶梯。[SDK 快照套件](../../../examples/jsonrpc-agent/tests/sdk.snapshot.ts) 经由 `llm-replay` 免密钥地通过本客户端驱动真实 `dsh-jsonrpc-agent` 运行时，固定通知流、轮次结果与持久化日志；`DSH_SNAPSHOT=record` 对真实 API 重录。
 
-## Model Experience
+## 模型体验
 
-None, as this is a client-process library; the model runs in the spawned runtime, whose experience is owned by the plugins its `cordis.yml` composes.
+无，因为这是一个客户端进程库；模型运行在 spawn 出的运行时中，其体验由该运行时的 `cordis.yml` 所组合的插件决定。
 
-#### KV Cache effect
+#### KV Cache 影响
 
-None; this package neither assembles nor sends a provider request.
+无；本包既不组装也不发送提供方请求。
 
-## Known Limitations and Deferred Work
+## 已知限制与暂缓事项
 
-- **无捆绑运行时解析** —— 调用方显式指定运行时可执行文件；打包可执行文件的发现留在 Python 侧，直到出现 TypeScript 发行版消费者。
-- **无回合中取消** —— 线上没有 prompt 取消方法；放弃回合意味着关闭运行时（见协议的 [Known Limitations](../sdk-protocol/README.md)）。
-- **每会话同时只有一个在途 prompt** —— 服务端规则，本客户端将其呈现为 `JsonRpcResponseError`；相互独立的会话可在同一运行时上并发。
-- **client→server 通知与 server→client 请求**在线两端都未实现；传输层为未来审批流保留了承载能力。
+- **无捆绑运行时解析**——调用方显式指定运行时可执行文件；打包可执行文件的发现留在 Python 侧，直到出现 TypeScript 发行版消费方。
+- **无轮次中取消**——协议层没有提示词取消方法；放弃轮次意味着关闭运行时（见协议的 [已知限制](../sdk-protocol/README.md)）。
+- **每会话同时只有一个在途提示词**——服务端规则，本客户端将其呈现为 `JsonRpcResponseError`；相互独立的会话可在同一运行时上并发。
+- **客户端→服务端通知与服务端→客户端请求**在协议两端都未实现；传输层为未来审批流保留了承载能力。
diff --git a/packages/sdk/sdk-protocol/README.i18n.yaml b/packages/sdk/sdk-protocol/README.i18n.yaml
index 072917a67f..7eec4f64dd 100644
--- a/packages/sdk/sdk-protocol/README.i18n.yaml
+++ b/packages/sdk/sdk-protocol/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/sdk/sdk-protocol/README.md
 README.md: 62b26d4a82d358fa4efcb7ab84036e5f4848057f
-README.zh.md: c08c32b2fbf0b63b28826bae8fd6c2a73bfe09d0
+README.zh.md: 11677c6119c7da407d95ee38ad9f8f7a552c15de
diff --git a/packages/sdk/sdk-protocol/README.zh.md b/packages/sdk/sdk-protocol/README.zh.md
index c08c32b2fb..11677c6119 100644
--- a/packages/sdk/sdk-protocol/README.zh.md
+++ b/packages/sdk/sdk-protocol/README.zh.md
@@ -2,38 +2,38 @@
 
 [English](README.md) | 中文
 
-DeepSeek Harness SDK 运行时的共享线协议：一个按换行分帧的 JSON-RPC 2.0 传输类，加上线两端共同使用的具名请求、结果与通知类型。包（package）根枚举协议消费方接口；源模块不以深层导入形式导出。服务端是 [`dsh-jsonrpc`](../../ui/jsonrpc/README.md) 插件；客户端是 [`dsh-sdk-client`](../sdk-client/README.md)（TypeScript）与 [Python SDK](../../../python/README.md)（后者镜像这些形状但不导入它们）。纯库——无插件、无 Config、无注册。
+DeepSeek Harness SDK 运行时的共享协议格式（wire format）：一个按换行分帧的 JSON-RPC 2.0 传输类，加上协议两端共同使用的具名请求、结果与通知类型。包（package）根枚举协议消费方接口；源模块不支持深层导入。服务端是 [`dsh-jsonrpc`](../../ui/jsonrpc/README.md) 插件；客户端是 [`dsh-sdk-client`](../sdk-client/README.md)（TypeScript）与 [Python SDK](../../../python/README.md)（后者复现这些结构但不导入它们）。纯库——无插件、无 Config、无注册。
 
 ## 传输
 
-`JsonRpcLineTransport` 在调用方持有的字节流上为 JSON-RPC 2.0 分帧，每行一个紧凑 JSON 帧、以 `\n` 结尾。带 `id` 与 `method` 的帧是请求，仅 `id` 是响应，仅 `method` 是通知；非法 JSON 行被忽略。`start()` 挂接流监听器，`close()` 摘除监听器并拒绝挂起请求、但不销毁流。缺失请求处理器时应答 `-32601`；处理器拒绝则应答携带错误消息的 `-32603`。错误响应会以 `JsonRpcResponseError` 拒绝挂起的 `request()`，保留线上的 `code` 与可选 `data`。`JsonRpcTransportPeer` 是服务器类所依赖的出站表面（request/notify）。
+`JsonRpcLineTransport` 在调用方持有的字节流上为 JSON-RPC 2.0 分帧，每行一个紧凑 JSON 帧、以 `\n` 结尾。带 `id` 与 `method` 的帧是请求，仅 `id` 是响应，仅 `method` 是通知；非法 JSON 行被忽略。`start()` 挂接流监听器，`close()` 移除监听器并拒绝挂起请求，但不销毁流。缺失请求处理器时应答 `-32601`；处理器返回的 Promise 被拒绝时，则应答携带错误消息的 `-32603`。错误响应会以 `JsonRpcResponseError` 拒绝挂起的 `request()` Promise，并保留协议格式中的 `code` 与可选 `data`。`JsonRpcTransportPeer` 是服务器类据以进行类型声明的出站接口（request/notify）。
 
-## 线类型
+## 协议类型
 
 `types.ts` 为 `HarnessSdkServer` 所服务协议的每个载荷命名：
 
 | 方向 | 方法 | 类型 |
 |---|---|---|
 | client→server | `initialize` | `InitializeParams` → `InitializeResult` |
-| client→server | `session/prompt` | `SessionPromptParams` → `SessionPromptResult`（仅在回合尘埃落定后应答） |
+| client→server | `session/prompt` | `SessionPromptParams` → `SessionPromptResult`（仅在轮次结算完成后应答） |
 | client→server | `shutdown` | 无参数 → `{}` |
 | server→client | `session.event` | `SessionEventNotification`（运行时内每个会话，不过滤） |
-| server→client | `session.finished` | `SessionFinishedNotification`（每个被接受的 prompt 一条） |
+| server→client | `session.finished` | `SessionFinishedNotification`（每个获准的提示词请求一条） |
 | server→client | `subagent.started` | `SubagentStartedNotification` |
-| server→client | `subagent.finished` | `SubagentFinishedNotification`（仅进程内 run） |
+| server→client | `subagent.finished` | `SubagentFinishedNotification`（仅进程内运行） |
 
-`HarnessSdkRequestMap` 与 `HarnessSdkNotificationMap` 按方法名索引这些类型。`InitializeParams.maxTokens` 是可选的正安全整数，用于限制 SDK 创建的 agent 及其进程内后代每次对话模型输出；省略时由提供方默认值控制。通知载荷类型依赖 `SessionEvent`（`dsh-session`）、`ContentBlock`（`dsh-llm`）与 `SubagentStopReason`（`dsh-subagent`）——协议以完整会话日志封套进行流式传输，因此会话词汇表是线契约的一部分。`serverInfo.name` 保持线上稳定值 `deepseek-harness-sdk-runtime`。
+`HarnessSdkRequestMap` 与 `HarnessSdkNotificationMap` 按方法名索引这些类型。`InitializeParams.maxTokens` 是可选的正的安全整数，用于限制 SDK 创建的 agent（智能体）及其进程内后代的每次对话模型输出；省略时由提供方默认值控制。通知载荷类型依赖 `SessionEvent`（`dsh-session`）、`ContentBlock`（`dsh-llm`）与 `SubagentStopReason`（`dsh-subagent`）——协议以完整会话日志封套进行流式传输，因此会话词汇是协议格式契约的一部分。`serverInfo.name` 的协议值固定为 `deepseek-harness-sdk-runtime`。
 
-## Model Experience
+## 模型体验
 
-None, as this package defines the client-facing wire protocol; the model-visible surfaces belong to the runtime plugins composed behind the serving [`dsh-jsonrpc`](../../ui/jsonrpc/README.md) entry.
+无，因为此包定义面向客户端的协议格式；模型可见接口属于组合在对外服务入口 [`dsh-jsonrpc`](../../ui/jsonrpc/README.md) 后方的运行时插件。
 
-#### KV Cache effect
+#### KV Cache 影响
 
-None; this package neither assembles nor sends a provider request.
+无；此包既不组装也不发送提供方请求。
 
-## Known Limitations and Deferred Work
+## 已知限制与暂缓事项
 
-- **无协议版本协商** —— 握手只携带 `serverInfo.version`（`0.0.1`，客户端不校验）；预发布立场，无兼容承诺。
-- **无取消与会话关闭方法** —— 客户端放弃回合的方式是关闭运行时进程；见 [`dsh-jsonrpc` README](../../ui/jsonrpc/README.md)。
-- **server→client 请求是死能力** —— 传输层支持，但服务器从不发送；Python SDK 的应答表面为未来审批流预留。
+- **无协议版本协商**——握手只携带 `serverInfo.version`（`0.0.1`，客户端不校验）；处于预发布阶段，无兼容承诺。
+- **无取消与会话关闭方法**——客户端放弃轮次的方式是关闭运行时进程；见 [`dsh-jsonrpc` README](../../ui/jsonrpc/README.md)。
+- **server→client 请求是未使用的功能**——传输层支持，但服务器从不发送；Python SDK 的应答接口为未来审批流程预留。
diff --git a/packages/sdk/telemetry/README.i18n.yaml b/packages/sdk/telemetry/README.i18n.yaml
index 039d5a94cb..b7bd3df0d5 100644
--- a/packages/sdk/telemetry/README.i18n.yaml
+++ b/packages/sdk/telemetry/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/sdk/telemetry/README.md
 README.md: 1d33915f36e0af10eedac5f9ab34f2534268a327
-README.zh.md: 3b60040f02ceaf17292bcbd1c046d82bdf86dafe
+README.zh.md: d71aa36be66a250056dd20a1ca9a88836b28c184
diff --git a/packages/sdk/telemetry/README.zh.md b/packages/sdk/telemetry/README.zh.md
index 3b60040f02..d71aa36be6 100644
--- a/packages/sdk/telemetry/README.zh.md
+++ b/packages/sdk/telemetry/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-用于 dsh-sdk 工具链的启动器侧 telemetry 原语。这是启动器在每个命令周围导入的普通库；它**不是** Cordis 插件，因为 `build` 与首次初始化的 `create` 从不启动 Cordis。将 reporter 接入启动器命令分发，并把 telemetry consent 功能加入 `dsh-helper` 目录，属于各自所属包的职责，而不是此包的职责。
+用于 dsh-sdk 工具链的启动器侧 telemetry 原语。这是启动器在执行每个命令时导入的普通库；它**不是** Cordis 插件，因为 `build` 与首次初始化的 `create` 从不启动 Cordis。将 reporter 接入启动器命令分发，并把 telemetry consent 功能加入 `dsh-helper` 目录，属于各自所属包（package）的职责，而不是此包的职责。
 
 | 导出 | 职责 |
 |---|---|
-| `SecretRedactor` | 保守的安全后备：在已解析值（`redactValue`）与原始文本（`redactText`）中，将形似密钥的值（密钥式键名、已知 token 形状、PEM 块、URL 凭据、高熵不透明 token）替换为占位符。绝不删除字段或行。 |
-| `ConsentResolver` | 解析项目 `cordis.yml`（绝不启动），读取 telemetry 配置项的启用／禁用状态作为 consent；`DO_NOT_TRACK`／CI 环境会强制彻底退出。 |
-| `buildTelemetryPayload` | 组装 `{command, durationMs, success, cordisYmlContent, packageJsonContent}`，对完整的 `cordis.yml` 与 `package.json` 文本运行 redactor。绝不读取 `.env`；发送 `package.json` 的前提是同时存在 `cordis.yml`，因此在非 SDK 目录运行的命令不会上传该目录中无关的 manifest。 |
+| `SecretRedactor` | 保守的安全后备：在已解析值（`redactValue`）与原始文本（`redactText`）中，将形似密钥的值（疑似密钥的键名、已知 token 格式、PEM 块、URL 凭据、高熵不透明 token）替换为占位符。绝不删除字段或行。 |
+| `ConsentResolver` | 解析项目 `cordis.yml`（绝不启动），读取 telemetry 配置项的启用／禁用状态作为 consent；`DO_NOT_TRACK`／CI 环境会强制完全停止上报。 |
+| `buildTelemetryPayload` | 组装 `{command, durationMs, success, cordisYmlContent, packageJsonContent}`，对完整的 `cordis.yml` 与 `package.json` 文本运行 redactor。绝不读取 `.env`；发送 `package.json` 的前提是同时存在 `cordis.yml`，因此在非 SDK 目录运行的命令不会上传该目录中无关的 manifest（元数据清单）。 |
 | `getOrCreateAnonymousId` | 将随机 UUID 持久化到 [`@deepseek-ai/dsh-paths`](../../util/paths/README.md) 解析出的 harness home（`$DSH_HOME` > `~/.dsh`）；其范围限定为该 home，而不是整台机器，且绝不从 git 派生。 |
-| `TelemetryReporter` | 即发即弃发送：`report()` 绝不阻塞或抛出；所有路径都会结算发送；`flush()` 可以在上限内排空进行中的发送。 |
+| `TelemetryReporter` | 即发即弃发送：`report()` 绝不阻塞或抛出；无论经过哪条路径，发送操作最终都会结束；`flush()` 可以在上限内排空进行中的发送。 |
 
 Consent 由 `cordis.yml` 中的 telemetry 配置项承载，因此禁用 telemetry 就是禁用该配置项。telemetry 默认上报，只有已经存在的 telemetry 配置项被显式设为 `disabled` 时才关闭：缺少 `cordis.yml`（首次 `create`）、配置项已启用，或 `cordis.yml` 中没有 telemetry 配置项时都会上报。`DO_NOT_TRACK`／CI 始终拒绝。无配置与缺少配置项的默认值可以通过 `ConsentResolver` 配置。
 
@@ -24,7 +24,7 @@ Consent 由 `cordis.yml` 中的 telemetry 配置项承载，因此禁用 telemet
 
 无；此包既不组装也不发送提供方请求。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- **占位端点**：`DSH_TELEMETRY_ENDPOINT` 指向 `.invalid`，直到配置真实端点。
+- **占位端点**：`DSH_TELEMETRY_ENDPOINT` 指向 `.invalid`，直到设置真实端点。
 - **脱敏依赖启发式规则**：这只是保守后备，不是保证；密钥应存放于 `.env`，而该文件绝不会被读取或上报。
diff --git a/packages/session-persistence/README.i18n.yaml b/packages/session-persistence/README.i18n.yaml
index 842251bd2d..43b0fcfdb6 100644
--- a/packages/session-persistence/README.i18n.yaml
+++ b/packages/session-persistence/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-persistence/README.md
 README.md: ac4e0a8310152b9d2ba5daae61fbbf1eb0ed54ec
-README.zh.md: cae5cbd83bac5ceed59067217635b98b194949ce
+README.zh.md: fe311556b3cdc518db10e513baa81d6a0002165d
diff --git a/packages/session-persistence/README.zh.md b/packages/session-persistence/README.zh.md
index cae5cbd83b..fe311556b3 100644
--- a/packages/session-persistence/README.zh.md
+++ b/packages/session-persistence/README.zh.md
@@ -1,14 +1,14 @@
-# session-persistence/：持久化功能家族
+# session-persistence/：持久化能力家族
 
 [English](README.md) | 中文
 
-持久会话持久化 seam 及其存储后端。接口包负责抽象 `SessionPersistence` 服务和共享写入协调器；后端是注册到 `ctx.sessionPersistence` 的具体实现。全部都是**产品** 包。
+持久会话的持久化 seam 及其存储后端。接口包（package）负责抽象 `SessionPersistence` 服务和共享写入协调器；后端是注册到 `ctx.sessionPersistence` 的具体实现。全部都是**产品**包。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `session-persistence/` | 持久化 seam + 共享写入协调器 | `ctx.sessionPersistence` |
-| `session-checkpoint-policy/` | agent 请求和工具执行的语义持久性屏障 | （包装 `ctx.llm` / `ctx.tools`，监听 agent 事件） |
-| `session-persistence-jsonl/` | JSONL sidecar 持久化后端 | （注册 `ctx.sessionPersistence`） |
-| `session-persistence-sqlite/` | SQLite 持久化后端 | （注册 `ctx.sessionPersistence`） |
+| `session-checkpoint-policy/` | agent（智能体）请求和工具执行的语义持久性屏障 | （包装 `ctx.llm` / `ctx.tools`，监听 agent 事件） |
+| `session-persistence-jsonl/` | JSONL 伴随文件持久化后端 | （注册到 `ctx.sessionPersistence`） |
+| `session-persistence-sqlite/` | SQLite 持久化后端 | （注册到 `ctx.sessionPersistence`） |
 
-接口位于 `session-persistence/session-persistence/`；后端是平级同级包。新存储后端在此加入，并注册到 `ctx.sessionPersistence`。详见[会话持久化](../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md)。
+接口位于 `session-persistence/session-persistence/`；后端是同级包。新存储后端归入此处，并注册到 `ctx.sessionPersistence`。详见[会话持久化](../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md)。
diff --git a/packages/session-persistence/session-checkpoint-policy/README.i18n.yaml b/packages/session-persistence/session-checkpoint-policy/README.i18n.yaml
index ad8dc1eed6..2ec207d786 100644
--- a/packages/session-persistence/session-checkpoint-policy/README.i18n.yaml
+++ b/packages/session-persistence/session-checkpoint-policy/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-persistence/session-checkpoint-policy/README.md
 README.md: aba0f00a2960eca3db06dda267cc954f602c1b09
-README.zh.md: ea17fb00ec22cb3a94ae22dc4a37aa7e2f73a225
+README.zh.md: ae403302f86b9fadc964c7963d3a454dd14180d3
diff --git a/packages/session-persistence/session-checkpoint-policy/README.zh.md b/packages/session-persistence/session-checkpoint-policy/README.zh.md
index ea17fb00ec..ae403302f8 100644
--- a/packages/session-persistence/session-checkpoint-policy/README.zh.md
+++ b/packages/session-persistence/session-checkpoint-policy/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-持久化 agent 的语义持久性策略。它会在模型适配器收到请求前、顶层工具正文可产生外部副作用前，以及每个 `agent/step` 边界为事件溯源会话创建检查点，使前一响应与有序工具结果在下一个请求前持久。
+已持久化的 agent（智能体）的语义持久性策略。它会在模型适配器收到请求前、顶层工具正文可产生外部副作用前，以及每个 `agent/step` 边界为事件溯源会话创建检查点，使前一响应与有序工具结果在下一个请求前已持久化。
 
 ## 插件（命名空间：`session-checkpoint-policy`）
 
@@ -16,30 +16,30 @@
   name: '@deepseek-ai/dsh-session-checkpoint-policy'
 ```
 
-持久化与检查点调度刻意拆分为独立 Cordis 插件。持久化后端会立即写入 `session/event` 追加，并把每个已请求 `session/flush` 变成观测屏障；该策略选择请求、工具分派和下一步骤屏障。不带此策略加载后端是有效的，但崩溃可能丢失最新的立即缓冲事件。第一方持久化应用和运行时显式挂载两个插件；专用部署可以刻意省略或替换策略。
+持久化与检查点调度刻意拆分为独立 Cordis 插件。持久化后端会主动写入追加的 `session/event`，并把每个已请求 `session/flush` 变成观测屏障；该策略选择请求、工具分派和下一步骤屏障。不带此策略加载后端是有效的，但崩溃可能丢失最新的已缓冲事件。第一方持久化应用和运行时显式挂载两个插件；专用部署可以刻意省略或替换策略。
 
-策略延迟包装 `llm/stream`，因此下游流只会在实时会话缓冲请求事件持久后构造。它在预执行策略和保护后包装 `tools/execute`；只有在已记录调用持久后，顶层工具正文才会运行。如果取消在 flush 等待期间到达，包装层会返回规范 `ABORTED_BEFORE_DISPATCH` 结果，不进入工具正文。嵌套工具分派重用外层模型可见调用的检查点。`agent/step` 在派生请求前持久前一响应/结果批次。
+策略延迟包装 `llm/stream`，因此下游流只会在活动会话中缓冲的请求事件已持久化后构造。它在预执行策略和防护机制之后包装 `tools/execute`；只有在已记录调用已持久化后，顶层工具正文才会运行。如果取消在 flush 等待期间到达，包装层会返回规范的 `ABORTED_BEFORE_DISPATCH` 结果，不进入工具正文。嵌套工具分派重用外层模型可见调用的检查点。`agent/step` 在派生请求前持久化前一响应/结果批次。
 
-在模型和工具边界，检查点拒绝会快速失败：适配器和顶层工具正文都不运行。步骤边界拒绝会在另一个请求开始前使轮次失败。并发工具检查点共享会话存储的串行持久化 drain，无法复制序列号。
+在模型和工具边界，检查点被拒绝时会按失败即阻止原则处理：适配器和顶层工具正文都不运行。步骤边界处的检查点被拒绝会在另一个请求开始前使轮次失败。并发工具检查点共享会话存储的串行持久化排空流程，不会产生重复的序列号。
 
 ## 模型体验
 
 ### 中断调用
 
-#### 模型所见
+#### 模型看到的内容
 
 插件不添加提示词或工具 schema。工具检查点后、结果前的硬崩溃会留下持久的未匹配调用；会话恢复会提供模型可见的 `TOOL_OUTCOME_UNKNOWN` 结果，该结果由 `dsh-session` 负责。该消息允许重试只读或幂等工作，并要求对可能有副作用的调用验证状态或请求用户确认。
 
 #### Token 影响
 
-成功检查点不添加 token，也不改变请求。恢复会添加一条短工具结果消息，以平衡中断 transcript。
+成功检查点不添加 token，也不改变请求。恢复会添加一条短工具结果消息，以平衡中断的 transcript（文本记录）。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 修复结果追加在可重用前缀之后，因此不会使较早的缓存条目失效。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- 该策略持久记录执行意图，而非通用的精确一次副作用。当提供方支持时，有副作用的工具应将 `exec.callId` 作为幂等键转发。
+- 该策略以持久方式记录执行意图，而非为通用副作用提供恰好一次保证。当提供方支持时，有副作用的工具应将 `exec.callId` 作为幂等键转发。
 - 流式 `assistant/chunk` 事件没有每分片检查点。它们在下一个语义检查点到达存储，因此硬崩溃可能丢失当前部分响应。
-- 持久调用没有结果时，无法证明其外部副作用是否完成。因此，恢复会记录未知结果，而不是自动重试。
+- 已持久化的调用没有结果时，无法证明其外部副作用是否完成。因此，恢复会记录未知结果，而不是自动重试。
diff --git a/packages/session-persistence/session-persistence-jsonl/README.i18n.yaml b/packages/session-persistence/session-persistence-jsonl/README.i18n.yaml
index f817c87919..5368aabeb3 100644
--- a/packages/session-persistence/session-persistence-jsonl/README.i18n.yaml
+++ b/packages/session-persistence/session-persistence-jsonl/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-persistence/session-persistence-jsonl/README.md
 README.md: ab6ecd28f12bd167aeac789d1565705e167d60f4
-README.zh.md: 97d387a04fa4c658217e28619410a49b7e6d4ec0
+README.zh.md: 16e0afa98c08710d298aa1487ddd8e51e1dc55c9
diff --git a/packages/session-persistence/session-persistence-jsonl/README.zh.md b/packages/session-persistence/session-persistence-jsonl/README.zh.md
index 97d387a04f..16e0afa98c 100644
--- a/packages/session-persistence/session-persistence-jsonl/README.zh.md
+++ b/packages/session-persistence/session-persistence-jsonl/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-JSONL 持久会话持久化后端：一个具体 `SessionPersistence`（`dsh-session-persistence` seam）。每个会话有一个仅追加逻辑 JSONL 日志，默认存储为 `.jsonl.zstd`；禁用压缩时使用原始 `.jsonl`。
+JSONL 持久会话存储后端：`SessionPersistence` 的一个具体实现（`dsh-session-persistence` seam）。每个会话有一个仅追加的逻辑 JSONL 日志，默认存储为 `.jsonl.zstd`；禁用压缩时使用原始 `.jsonl`。
 
 ## 磁盘布局
 
@@ -16,22 +16,22 @@ JSONL 持久会话持久化后端：一个具体 `SessionPersistence`（`dsh-ses
 
 - 第一个逻辑行是不可变的 `SessionHeader`，标记为 `{ type: 'session', version, id, cwd?, createdAt, parentSession?, seedLength?, delegationDepth }`。`delegationDepth` 在磁盘上必需，顶层会话为 `0`；缺失或无效值会拒绝日志。后续每个逻辑行是一条存储记录；`assistant/chunk` 事件绝不丢弃，且 `seq` 在解码日志中保持连续（`events[i].seq === i`）。
 - 存储记录是原样 `SessionEvent` JSON，或在 `packChunks` 已启用且连续段符合条件时写入的**打包分片行**（`text-chunks` / `reasoning-chunks` / `tool-call-chunks`；像 header 的 `session` 一样不带斜杠，因此行 tag 不会与事件类型混淆）：一行保存至少 3 个连续同 block `assistant/chunk` delta 事件，`seq0`/`time0` 和每成员 `dt` 间隔精确重建每个成员的 `seq`/`time`。无损 codec 位于 `@deepseek-ai/dsh-session`（`packChunkRuns`/`decodeStorageRecord`），并使用精确形态 allowlist：任何未识别内容原样存储。读取与布局无关：`load` 始终解码行，因此打包、非打包和混合文件加载结果一致。
-- 项目目录保留规范化 cwd 可读，并限制在文件系统组件上限内。分隔符替换和截断刻意有损，因此规范化相同的 cwd 字符串共享项目目录；会话 id 仍选择不同会话目录。在不区分大小写的文件系统上，只有文件系统规范化将两种写法解析到同一 transcript 时，身份验证才接受备选路径写法。配置根仍由部署控制：可以是项目本地、共享、临时或集中式。[项目会话目录决策](../../../.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md) 记录这项取舍。
-- 会话 id 是未验证的品牌化字符串，因此在使用前单射转义为一个安全路径段（无遍历、无冲突）。结果目录保留给其他会话自有产物；发现只读取固定 transcript 文件名。
+- 项目目录保留规范化 cwd 可读，并限制在文件系统组件上限内。分隔符替换和截断刻意有损，因此规范化相同的 cwd 字符串共享项目目录；会话 id 仍选择不同会话目录。在不区分大小写的文件系统上，只有文件系统规范化将两种写法解析到同一 transcript（文本记录）时，身份验证才接受备选路径写法。配置根仍由部署控制：可以是项目本地、共享、临时或集中式。[项目会话目录决策](../../../.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md) 记录这项取舍。
+- 会话 id 是未验证的带品牌类型的字符串，因此在使用前单射转义为一个安全路径段（无遍历、无冲突）。结果目录保留给其他会话自有产物；发现只读取固定 transcript 文件名。
 
 ## 配置
 
 | 键 | 类型 | 说明 |
 |---|---|---|
-| `root` | `string` (required) | 所有会话文件的根目录。**无默认值**：`process.cwd()` 默认值会随进程 cwd 变更（bash 调用、子进程）而分散文件。现有根必须是可读目录；缺失根在第一次实体化时创建。 |
-| `packChunks` | `boolean` (default `true`) | 将符合条件的 delta 分片连续段写为打包行（在真实编码会话上测得逻辑日志约小 60%）。设为 `false` 可用于每事件一行诊断；无论该写入侧开关如何，都能读取打包行。 |
+| `root` | `string`（必需） | 所有会话文件的根目录。**无默认值**：`process.cwd()` 默认值会随进程 cwd 变更（bash 调用、子进程）而分散文件。现有根必须是可读目录；缺失根在第一次实体化时创建。 |
+| `packChunks` | `boolean`（默认 `true`） | 将符合条件的 delta 分片连续段写为打包行（在真实编码会话上测得逻辑日志约小 60%）。设为 `false` 可用于每事件一行诊断；无论该写入侧开关如何，都能读取打包行。 |
 | `compression` | `'zstd' \| 'none'` | 默认 `'zstd'`；`'none'` 保留换行分隔 UTF-8 文本。 |
 
-`locate(meta)` 返回已解析项目/会话目录内固定 transcript 的 `{ kind: 'jsonl', path }`。它不执行文件系统 I/O：可以在目录或文件存在前返回目标，现有文件也只包含最后 flush 前缀。
+`locate(meta)` 返回已解析项目/会话目录内固定 transcript 的 `{ kind: 'jsonl', path }`。它不执行文件系统 I/O：可以在目录或文件存在前返回目标，现有文件也只包含最近一次 flush 完成的前缀。
 
 ## 物理编码
 
-默认产物是独立 [Zstandard frame](../../../.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md) 的标准连接：一个仅包含 header 行的带 checksum frame，后跟每个持久 append 批次一个带 checksum frame。后端使用 Node 内置 Zstandard API 和默认压缩级别，不提供级别开关。列表只读取并验证 header frame。`compression: 'none'` 在原始表示中保留相同逻辑行。
+默认产物是独立 [Zstandard frame](../../../.agents/notes/implemented/architecture/2026-07-19-zstandard-jsonl-session-logs.md) 的标准拼接：一个仅包含 header 行的带 checksum frame，后跟每个持久 append 批次一个带 checksum frame。后端使用 Node 内置 Zstandard API 和默认压缩级别，不提供级别开关。列表只读取并验证 header frame。`compression: 'none'` 在原始表示中保留相同逻辑行。
 
 一个根只属于一种编码。启动发现和定向查找会拒绝相反 suffix，错误会命名不兼容产物，并指示调用方选择匹配 mode 或独立根。平铺 `<project>/<id>.jsonl*` 产物也会被拒绝，而不是忽略。不提供迁移、混合根回退或双写。
 
@@ -39,37 +39,37 @@ JSONL 持久会话持久化后端：一个具体 `SessionPersistence`（`dsh-ses
 
 - **绑定存储身份。** 查找要求可读项目目录中只有一个匹配会话目录，然后验证 header id 等于请求 id，且 header id/cwd 派生所选 transcript 路径。列表应用同一路径检查，并拒绝重复 id。身份失败发生在修复或 append 前。
 - **延迟实体化。**`create(meta)` 不写入；第一次 `append` 将编码 header 和第一批写入临时文件并执行 `fsync`。POSIX 通过硬链接无覆盖发布，并对父目录 `fsync`。Windows 通过 `MoveFileExW(..., MOVEFILE_WRITE_THROUGH)` 无覆盖发布，并通过同一 write-through pattern 创建缺失目录。已创建但从未 append 的会话不留下磁盘内容，不在 `list` 中。
-- **仅追加。** 已 flush 事件绝不重写。后续原始批次 append 行；压缩批次 append 一个 frame。两条路径都执行 `fsync`，并在捕获写入或同步失败时回滚到之前字节长度。
-- **崩溃恢复：保留有效尾部工作。**`load` 验证每个完整压缩 frame，并扫描解压 JSONL。最后 frame 结构不完整时，读取器保留其完整解码记录，从 frame 开头截断，并使用共享[持久化契约](../../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md) 需要的合成工具、步骤和轮次 closer 重新编码这些记录。原始 mode 从第一个不完整行截断。完整 frame 中的 checksum/解压失败，或最后已提交 `turn/end` 之前或当时的缺陷属于损坏，会被拒绝。
+- **仅追加。** 已 flush 事件绝不重写。后续原始批次 append 行；压缩批次 append 一个 frame。两条路径都执行 `fsync`，并在捕获到写入或同步失败时回滚到之前字节长度。
+- **崩溃恢复：保留有效尾部工作。**`load` 验证每个完整压缩 frame，并扫描解压 JSONL。最后 frame 结构不完整时，读取器保留其完整解码记录，从 frame 开头截断，并使用共享[持久化契约](../../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md) 需要的合成工具、步骤和轮次 closer 重新编码这些记录。原始 mode 从第一个不完整行截断。完整 frame 中的 checksum/解压失败，或位于最后已提交的 `turn/end` 处或之前的缺陷属于损坏，会被拒绝。
 - **非变更检查。**`inspect()` 返回脱离的有效前缀，不截断不完整尾部或关闭中断轮次，并保持轻量修订不变。
 - **连续 seq。**`append` 拒绝第一个 `seq` 不继续已存储日志的批次，并拒绝非 JSON 可序列化 `event.data`，同时命名违规事件类型。
-- **轻量修订。**`listSnapshots(signal?)` 使用 device、inode、size 和纳秒时间戳标识日志，避免解析完整日志，同时在 append、修复、替换或存储变更后改变。它通过产物发现转发精确信号，并在每个 `stat` 前后检查取消；由于文件系统 `stat` 不可中断，取消会等待活动调用结算，然后在不启动另一次调用的情况下拒绝。
+- **轻量修订。**`listSnapshots(signal?)` 使用 device、inode、size 和纳秒时间戳标识日志，避免解析完整日志；该标识会在 append、修复、替换或存储变更后改变。它通过产物发现转发精确信号，并在每个 `stat` 前后检查取消；由于文件系统 `stat` 不可中断，取消会等待活动调用完成，然后在不启动另一次调用的情况下拒绝。
 
 ## 写入路径
 
-插件将冻结会话事件复制到每个实时会话的一个 controller，并启动急切 drain。并发事件共享当前写入；期间接纳的事件形成后续批次，`session/flush` 则等待当前和 pending 批次持久。每会话游标防止恢复会话重新 append 已存储事件，插件加载时会为实时会话播种。所属后端实例串行化单会话操作；dispose 在拆卸前 drain 每个保留 controller。
+插件将冻结的会话事件复制到每个活动会话各自的 controller，并启动主动排空流程。并发事件共享当前写入；期间接纳的事件形成后续批次，`session/flush` 则等待当前和待处理批次完成持久化。每会话游标防止恢复后的会话重新 append 已存储事件，插件加载时会为活动会话设置初始状态。所属后端实例串行化单会话操作；dispose（资源释放）会在拆卸前排空每个保留的 controller。
 
 ## 模型体验
 
 ### 恢复的对话历史
 
-#### 模型所见
+#### 模型看到的内容
 
-JSONL 存储不贡献实时提示词或 schema。加载恢复已存储接口历史，并保留之前的请求 header 用于重建；新 loop 组合当前 envelope。恢复将无持久调用的 assistant 请求平衡为 `TOOL_NOT_STARTED`；有持久调用但无结果时变为 `TOOL_OUTCOME_UNKNOWN`，它要求模型只重试只读或幂等工作，并验证可能副作用或请求用户。原始 `assistant/chunk` 记录不重复消息。
+JSONL 存储不影响当前提示词或 schema。加载会恢复已存储的呈现历史，并保留之前的请求 header 用于重建；新 loop 组合当前 envelope。恢复会用 `TOOL_NOT_STARTED` 平衡没有已持久化调用的 assistant 请求；已有已持久化调用但无结果时则变为 `TOOL_OUTCOME_UNKNOWN`，它要求模型只重试只读或幂等工作，并验证可能的副作用或询问用户。原始 `assistant/chunk` 记录不会重复生成消息。
 
 #### Token 影响
 
-实时请求为零 token。恢复 agent 支付已保留历史和当前 envelope，以及每个中断调用的引用修复结果。
+当前请求不会新增 token。恢复后的 agent（智能体）会因保留的历史、当前 envelope，以及每个中断调用的前述修复结果文本而消耗 token。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 JSONL 存储不修改实时请求前缀。只有重建历史、当前 envelope 和模型路由匹配时，恢复 loop 才能重用提供方缓存；崩溃修复结果仅追加。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **只加载已配置编码和当前 `SESSION_FORMAT_VERSION` (v0)**：更改压缩需要独立/全新根，或选择遗留原始 mode；预发布格式没有迁移。
 - **平铺文件存储布局不加载**：加载前使用独立根，或将预发布产物移入项目/会话目录布局。
-- **压缩文件不能直接按行读取**：使用后端加载；或在写入新根前选择 `compression: 'none'`，以便文本 fixture 或外部行 reader 使用。
+- **压缩文件不能直接按行读取**：使用后端加载；或在写入新根前选择 `compression: 'none'`，以便文本 fixture（测试前置数据）或外部行 reader 使用。
 - **不删除会话文件**：日志在 `root` 下累积，直到外部移除（seam 无删除接口）。
 - **每会话一个实时 writer**：append 和修复只在所属后端实例内协调。在 owner 完全停稳 dispose 前，其他后端实例或进程不得写入同一会话；初始同 id 发布仍通过 POSIX 无覆盖硬链接或 Windows 无替换 write-through rename 保持冲突安全。
 - **POSIX 实体化需要硬链接支持**：第一次 append 使用 `link()`，使同 id 竞态失败，而不覆盖已提交日志；Windows 使用无替换 write-through rename。
diff --git a/packages/session-persistence/session-persistence-sqlite/README.i18n.yaml b/packages/session-persistence/session-persistence-sqlite/README.i18n.yaml
index b232652911..8dea06ca1f 100644
--- a/packages/session-persistence/session-persistence-sqlite/README.i18n.yaml
+++ b/packages/session-persistence/session-persistence-sqlite/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-persistence/session-persistence-sqlite/README.md
 README.md: 394b10a70fc757d75f19178050c0d63699a59e54
-README.zh.md: f186d71912e61c5eb664973195ec1d05070a3cef
+README.zh.md: 6d2a47aa64cee59c35c558e1923a112b0d72f58b
diff --git a/packages/session-persistence/session-persistence-sqlite/README.zh.md b/packages/session-persistence/session-persistence-sqlite/README.zh.md
index f186d71912..6d2a47aa64 100644
--- a/packages/session-persistence/session-persistence-sqlite/README.zh.md
+++ b/packages/session-persistence/session-persistence-sqlite/README.zh.md
@@ -2,17 +2,17 @@
 
 [English](README.md) | 中文
 
-SQLite 持久会话持久化后端：第二个 `SessionPersistence` 实现（见[会话持久化](../../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md)），用于验证抽象 seam 和共享 `runPersistenceContract` 套件真正与后端无关。它满足与 `dsh-session-persistence-jsonl` 相同的契约（仅追加、连续 seq、延迟实体化、在 load 时关闭中断轮次），但用 `node:sqlite` 行而非文件字节表达。
+SQLite 持久会话存储后端：第二个 `SessionPersistence` 实现（见[会话持久化](../../../.agents/notes/implemented/architecture/2026-06-14-session-persistence.md)），用于验证抽象 seam 和共享 `runPersistenceContract` 套件真正与后端无关。它满足与 `dsh-session-persistence-jsonl` 相同的契约（仅追加、连续 seq、延迟实体化、在 load 时关闭中断轮次），但用 `node:sqlite` 行而非文件字节表达。
 
-`locate(meta)` 返回 `undefined`：所有会话共享一个数据库，因此不存在真实的独立每会话 transcript 路径。
+`locate(meta)` 返回 `undefined`：所有会话共享一个数据库，因此不存在真实、独立的逐会话 transcript（文本记录）路径。
 
-> **TODO：** 该后端直接调用 `node:sqlite`。如果采用 Cordis 数据库服务（`cordis/db` / `@cordisjs` SQL driver 插件），应改为通过该服务路由，而不在此保持原始 `DatabaseSync`；契约接口（`SessionPersistence`）不会变，只更换存储 driver。
+> **TODO：** 该后端直接调用 `node:sqlite`。如果采用 Cordis 数据库服务（`cordis/db` / `@cordisjs` SQL driver 插件），应改为通过该服务路由，而不在此直接持有 `DatabaseSync`；契约接口（`SessionPersistence`）不会变，只更换存储驱动。
 
 ## 存储模型
 
-每个 `SessionEvent` 1:1 映射到 `events` 表中的一行 `(session_id, seq, type, time, data, source_event_seqs, surface_op)`；`data` 是作为 JSON 文本的事件 payload，因此行形态就是原样事件（包括 `assistant/chunk`，保持 `seq` 连续）。两个 `TEXT` 列 `source_event_seqs` 和 `surface_op` 可为空，存储事件可选接口元数据字段（见[会话接口](../../../.agents/notes/implemented/architecture/2026-06-18-session-surface.md)）。日志外元数据（`SessionHeader`）、每实体化 incarnation id 和每日志单调修订位于 `sessions` 行；`createdAt` 是存储在 strict `INTEGER` 列中的非负安全整数。单例状态行携带不可变存储 id。`sessions` 行只由第一次 `append` 写入，其存在性是延迟实体化信号（`list` 精确报告有行的会话）。
+每个 `SessionEvent` 1:1 映射到 `events` 表中的一行 `(session_id, seq, type, time, data, source_event_seqs, surface_op)`；`data` 是作为 JSON 文本的事件 payload，因此行结构就是原始事件本身（包括 `assistant/chunk`，保持 `seq` 连续）。两个 `TEXT` 列 `source_event_seqs` 和 `surface_op` 可为空，存储事件可选接口元数据字段（见[会话接口](../../../.agents/notes/implemented/architecture/2026-06-18-session-surface.md)）。日志外元数据（`SessionHeader`）、每实体化 incarnation id 和每日志单调修订位于 `sessions` 行；`createdAt` 是存储在 strict `INTEGER` 列中的非负安全整数。单例状态行携带不可变存储 id。`sessions` 行只由第一次 `append` 写入，其存在性是延迟实体化信号（`list` 精确报告有行的会话）。
 
-仓库支持的 Node 范围可不加 flag 使用 `node:sqlite`。数据库启用外键，并使用已配置 journal mode（默认 `wal`；WAL 共享内存文件不适用时使用 rollback mode）。`PRAGMA application_id` 标识规范持久化数据库，`PRAGMA user_version` 存储布局版本。新数据库必须没有 application identity 或用户定义 schema 对象；初始化在一个事务中创建全部表并盖上两个 pragma。非 pristine 无版本数据库、外部 application identity 和所有非当前版本在 journal-mode 变更前拒绝，因为该未发布格式无迁移。
+仓库支持的 Node 范围可不加 flag 使用 `node:sqlite`。数据库启用外键，并使用已配置 journal mode（默认 `wal`；WAL 共享内存文件不适用时使用 rollback mode）。`PRAGMA application_id` 标识规范持久化数据库，`PRAGMA user_version` 存储布局版本。新数据库必须没有 application identity 或用户定义 schema 对象；初始化在一个事务中创建全部表并盖上两个 pragma。非 pristine 无版本数据库、外部 application identity 和所有非当前版本在 journal-mode 变更前均会被拒绝，因为该未发布格式无迁移。
 
 在具有 POSIX mode 的文件系统上，后端为缺失目录请求 mode `0700`，并在 SQLite 打开前以 mode `0600` 排他创建缺失数据库；进程 umask 可进一步限制两者。新 WAL、共享内存和持久 rollback-journal sidecar 获得数据库最终的仅所有者 mode。现有目录、数据库文件和 sidecar 保留原 mode；除已存在数据库外的文件系统设置错误会使初始化失败。这些默认值防止宽松进程 umask 造成的意外暴露，但当其他 principal 能替换父目录中的数据库条目时，不保护数据库机密性或完整性。
 
@@ -35,27 +35,27 @@ interface Config {
 
 ## 写入路径
 
-与 JSONL 后端一样，插件将每个冻结 `session/event` 复制到每个实时会话的一个 controller，并启动急切 drain。并发事件共享当前事务；期间接纳的事件形成后续批次，`session/flush` 则等待当前和 pending 批次持久。Controller 对 fork 种子持久一次，保留写入游标，使 resume 绝不重新 append 已存储事件，并在 apply 时为实时会话播种，因为 HMR 不回放 `session/created`。Dispose 在关闭数据库前 drain 每个保留 controller。
+与 JSONL 后端一样，插件将每个冻结的 `session/event` 复制到每个活动会话各自的 controller，并启动主动排空流程。并发事件共享当前事务；期间接纳的事件形成后续批次，`session/flush` 则等待当前和待处理批次完成持久化。Controller 会持久化一次 fork 种子，并保留写入游标，使恢复操作绝不重新 append 已存储事件；它还会在 apply 时为活动会话设置初始状态，因为 HMR（热模块替换）不回放 `session/created`。dispose（资源释放）会在关闭数据库前排空每个保留的 controller。
 
 ## 模型体验
 
 ### 恢复的对话历史
 
-#### 模型所见
+#### 模型看到的内容
 
-SQLite 存储不贡献实时提示词或 schema。加载恢复与 JSONL 相同的接口历史，并保留之前的 header 用于重建；新 loop 组合当前 envelope。恢复将无持久调用的 assistant 请求平衡为 `TOOL_NOT_STARTED`；有持久调用但无结果时变为 `TOOL_OUTCOME_UNKNOWN`，它要求模型只重试只读或幂等工作，并验证可能副作用或请求用户。行元数据和原始分片不是消息。
+SQLite 存储不影响当前提示词或 schema。加载会恢复与 JSONL 相同的呈现历史，并保留之前的 header 用于重建；新 loop 组合当前 envelope。恢复会用 `TOOL_NOT_STARTED` 平衡没有已持久化调用的 assistant 请求；已有已持久化调用但无结果时则变为 `TOOL_OUTCOME_UNKNOWN`，它要求模型只重试只读或幂等工作，并验证可能的副作用或询问用户。行元数据和原始分片不会成为消息。
 
 #### Token 影响
 
-实时请求为零 token。Resume 恢复已保留历史并支付当前 envelope，以及每个中断调用的引用修复结果。
+当前请求不会新增 token。恢复会还原已保留的历史，并产生当前 envelope 以及每个中断调用所附修复结果文本的 token 开销。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-SQLite 存储不修改实时请求前缀。只有重建历史、当前 envelope 和模型路由匹配时，恢复 loop 才能重用提供方缓存；崩溃修复结果仅追加。
+SQLite 存储不修改当前请求前缀。只有重建历史、当前 envelope 和模型路由匹配时，恢复 loop 才能重用提供方缓存；崩溃修复结果会追加到末尾。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **`DatabaseSync` 是同步的**：每个 append 事务在整个期间阻塞事件 loop；对本地存储可接受，对繁忙多会话服务器是吞吐上限。
+- **`DatabaseSync` 是同步的**：每个 append 事务在整个期间阻塞事件循环；对本地存储可接受，对繁忙多会话服务器是吞吐上限。
 - **写入争用无等待或重试策略**：后端不设置 busy timeout，也不重试 locked-database 错误，因此其他连接持有写事务时操作立即拒绝。
-- **只打开 pristine 新数据库或当前自有 `SCHEMA_VERSION`**：无版本 schema 对象、外部 application identity 和所有其他 schema 版本被拒绝，而不是迁移（未发布软件，无持久用户数据需要保留）。
-- **不删除已存储会话**：行会累积，直到外部移除（seam 无删除接口；`ON DELETE CASCADE` 已为这种带外清理接线）。
+- **只有 pristine 新数据库或当前自有 `SCHEMA_VERSION` 才能打开**：无版本 schema 对象、外部 application identity 和所有其他 schema 版本被拒绝，而不是迁移（未发布软件，无持久用户数据需要保留）。
+- **不删除已存储会话**：行会累积，直到外部移除（seam 无删除接口；`ON DELETE CASCADE` 已为这种带外清理配置）。
diff --git a/packages/session-query/README.i18n.yaml b/packages/session-query/README.i18n.yaml
index aadf733553..f1f511b9c8 100644
--- a/packages/session-query/README.i18n.yaml
+++ b/packages/session-query/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-query/README.md
 README.md: 38edbf6b0303e3d5a7bd0dc1d180cd127c60f9dc
-README.zh.md: 448fae8470e11ebd5a3f7e9d30e1eb82ab995e7b
+README.zh.md: f50b93076a55e7530420400089d48672d3ea6b40
diff --git a/packages/session-query/README.zh.md b/packages/session-query/README.zh.md
index 448fae8470..f50b93076a 100644
--- a/packages/session-query/README.zh.md
+++ b/packages/session-query/README.zh.md
@@ -1,13 +1,13 @@
-# session-query/：会话取回功能家族
+# session-query/：会话检索能力家族
 
 [English](README.md) | 中文
 
 针对实时和持久会话日志提供可信的精确读取、关系跟踪、与提供方无关的语义过滤和 SQLite 全文搜索。
 
-| 包 | 职责 | ctx 键 |
+| 包（package） | 职责 | ctx 键 |
 |---|---|---|
 | [`session-query/`](session-query/README.md) | 组合式服务契约：提供具体的逻辑语料库读取、跟踪和语义过滤，以及抽象全文方法 | `ctx.sessionQuery` |
 | [`session-query-sqlite/`](session-query-sqlite/README.md) | 具体服务后端：使用 SQLite FTS5 持久基库和实时覆盖层 | `ctx.sessionQuery` |
 | [`tool-session-query/`](tool-session-query/README.md) | 工作区授权的面向模型搜索、血缘、关系和精确事件工具 | 无 |
 
-查询服务与压缩无关：它读取规范血缘、接口操作、已记录来源信息和语义事件文本，但不参与压缩策略或执行。一个抽象服务组合全部查询操作；一个具体后端负责全文生命周期，无需提供方注册表或协调器；消费方将过大的纯文本结果交给通用执行后 spill 策略。
+查询服务与压缩（compaction）无关：它读取规范血缘、接口操作、已记录来源信息和语义事件文本，但不参与压缩策略或执行。一个抽象服务组合全部查询操作；一个具体后端负责全文生命周期，无需提供方注册表或协调器；消费方将过大的纯文本结果交给通用执行后 spill 策略。
diff --git a/packages/session-query/session-query-sqlite/README.i18n.yaml b/packages/session-query/session-query-sqlite/README.i18n.yaml
index 9c5f95f8ce..ce6efd6cfc 100644
--- a/packages/session-query/session-query-sqlite/README.i18n.yaml
+++ b/packages/session-query/session-query-sqlite/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-query/session-query-sqlite/README.md
 README.md: ceffb3ac25bc8b5252d6cc40cd6389839dfce1e2
-README.zh.md: 4e11ae9c9b8012045a7f3bab5d5c45724e553303
+README.zh.md: 33afe99c3e065e231692e162549aef8e28e542e3
diff --git a/packages/session-query/session-query-sqlite/README.zh.md b/packages/session-query/session-query-sqlite/README.zh.md
index 4e11ae9c9b..33afe99c3e 100644
--- a/packages/session-query/session-query-sqlite/README.zh.md
+++ b/packages/session-query/session-query-sqlite/README.zh.md
@@ -2,29 +2,29 @@
 
 [English](README.md) | 中文
 
-具体 `ctx.sessionQuery` 后端。`SessionQuerySqlite` 从接口包继承精确读取、跟踪和与提供方无关的过滤，并使用 SQLite FTS5 实现其两个全文方法。搜索使用实时优先的逻辑会话语料库，并按其匹配最强的事件对跨会话结果分组。
+具体 `ctx.sessionQuery` 后端。`SessionQuerySqlite` 从接口包（package）继承精确读取、跟踪和提供方无关的过滤，并使用 SQLite FTS5 实现其两个全文方法。搜索使用实时优先的逻辑会话语料库，并按每个会话中匹配度最高的事件对跨会话结果分组。
 
 ## 搜索契约
 
-`searchSessions(request, exec?)` 返回跨语料库的 `SessionSearchHit` 分页结果；`searchEvents(request, exec?)` 返回单个会话内的 `SessionEventSearchHit` 分页结果。查询不得省略，会被修剪并将空白规范化为字面短语。引号、`OR`、`NEAR` 和 `*` 等 FTS5 语法被视为数据，而非可执行 MATCH 语法。元数据过滤器是在排名前应用的参数化 SQL 谓词。为使 SQLite FTS5 MATCH 保持在受支持的外层谓词上下文中，跨会话请求最多可编译 14 个组合会话与事件过滤谓词；会话内请求最多可编译 13 个过滤谓词，因为固定目标会话谓词占用一个槽位。每个范围端点编译为一个谓词。请求超过任一谓词预算，或超过 SQLite 可移植的 32,766 总绑定上限（包括固定查询和分页值）时，会在准备语句前以 `SESSION_QUERY_INVALID_FILTER` 失败。
+`searchSessions(request, exec?)` 返回跨语料库的 `SessionSearchHit` 分页结果；`searchEvents(request, exec?)` 返回单个会话内的 `SessionEventSearchHit` 分页结果。查询不得省略，首尾空白会被移除，内部空白会被规范化，并按字面短语处理。引号、`OR`、`NEAR` 和 `*` 等 FTS5 语法被视为数据，而非可执行 MATCH 语法。元数据过滤器是在排名前应用的参数化 SQL 谓词。为使 SQLite FTS5 MATCH 保持在受支持的外层谓词上下文中，跨会话请求最多可编译 14 个组合会话与事件过滤谓词；会话内请求最多可编译 13 个过滤谓词，因为固定目标会话谓词占用一个槽位。每个范围端点编译为一个谓词。请求超过任一谓词预算，或超过 SQLite 可移植的 32,766 总绑定上限（包括固定查询和分页值）时，会在准备语句前以 `SESSION_QUERY_INVALID_FILTER` 失败。
 
-持久表和 TEMP 表之间的相关性可比：先按实际 FTS5 高亮匹配 span 数降序，再按已存储文档码点长度升序。事件时间、适用时的会话 id 和 seq 打破其余平局。跨会话结果将所选事件公开为 `bestMatch`；两种范围都从 FTS5 高亮位置派生空白规范化的纯文本，并按 Unicode 码点限制长度。游标是不透明的品牌化值，绑定到规范化请求和服务实例，并在相关世代变更时失败。会话内游标可在不相关会话变更后延续使用；跨会话游标则不能。
+持久表和 TEMP 表之间的相关性排名可直接比较：先按实际 FTS5 高亮匹配 span 数降序，再按已存储文档码点长度升序。事件时间、适用时的会话 id 和 seq 打破其余平局。跨会话结果将所选事件公开为 `bestMatch`；两种范围都从 FTS5 高亮位置派生空白规范化的纯文本，并按 Unicode 码点限制长度。游标是带品牌类型的不透明值，绑定到规范化请求和服务实例，并在相关世代变更时失败。会话内游标可在不相关会话变更后延续使用；跨会话游标则不能。
 
 默认可搜索全部三种接口（`current`、`shadowed` 和 `log-only`）。传入接口过滤器可缩小范围。
 
 ## 来源与索引生命周期
 
-该服务需要 `ctx.sessions`，并动态观察可选的 `ctx.sessionPersistence`。一个串行化状态机比较来源限定的轻量持久化快照修订，以非变更方式只检查新日志或已更改日志，提取共享语义文档，以事务方式对账变更，然后运行查询。会话查询绝不会调用持久化后端会修复崩溃的 `load()`；检查期间附加的 owner 无法修改其日志，稳定观察重试使结果优先使用实时来源。TEMP 实时行仍会记录持久化可用性，而持久基库会在该实时 owner 脱离后刷新。重复查询和未变的同存储重新打开不会执行完整持久化日志检查；切换存储，或观察到新增、已更改、已删除或经外部 load 修复的来源时，会在下次稳定观察时对账。来源或事务失败不会提交任何内容，下一次搜索会重试。
+该服务需要 `ctx.sessions`，并动态观察可选的 `ctx.sessionPersistence`。一个串行化状态机比较来源限定的轻量持久化快照修订，仅以不修改日志的方式检查新日志或已更改日志，提取共享语义文档，以事务方式对账变更，然后运行查询。会话查询绝不会调用持久化后端会修复崩溃的 `load()`；检查期间接入的活动所有者无法修改其日志，稳定观察重试使结果优先使用实时来源。TEMP 实时行仍会记录持久化可用性，而持久基库会在该活动所有者脱离后刷新。重复查询和未变的同存储重新打开不会执行完整持久化日志检查；切换存储，或观察到新增、已更改、已删除或经外部 load 修复的来源时，会在下次稳定观察时对账。来源或事务失败不会提交任何内容，下一次搜索会重试。
 
-持久化 FTS 行位于专用派生数据库中。连接本地 TEMP 表保存实时行，这些行会遮蔽同一会话的持久化基库，并在实时 owner 消失后使其重新可见。卸载持久化会隐藏持久行，但不会丢弃缓存；重新挂载会对账缓存。关闭或重新打开数据库会删除全部实时覆盖层，但保留持久行。
+持久化 FTS 行位于专用派生数据库中。连接本地 TEMP 表保存实时行，这些行会遮蔽同一会话的持久化基库，并在实时所有者消失后使其重新可见。卸载持久化会隐藏持久行，但不会丢弃缓存；重新挂载会对账缓存。关闭或重新打开数据库会删除全部实时覆盖层，但保留持久行。
 
-该数据库可丢弃，但 reset 受到保护：每个已识别 schema 版本都会在修改 journal mode 前拒绝未知用户表；只有包含派生表的已识别不兼容 schema 才会原地重建。不相关数据库或规范数据库将被拒绝。绝不能将 `path` 指向 session-persistence 数据库。在具有 POSIX mode 的文件系统上，缺失的目录和数据库会以仅所有者可访问的方式创建（进程 umask 前为 `0700` 和 `0600`），SQLite sidecar 继承数据库 mode；现有 mode 保持不变。每个派生索引路径在一个进程中只能由一个服务拥有；不支持外部写入者或第二个进程，因为世代和 TEMP 遮蔽状态归连接所有。
+该数据库虽可丢弃重建，但 reset 操作受到保护：每个已识别 schema 版本都会在修改 journal mode 前拒绝未知用户表；只有包含派生表的已识别不兼容 schema 才会原地重建。不相关数据库或规范数据库将被拒绝。绝不能将 `path` 指向 session-persistence 数据库。在具有 POSIX mode 的文件系统上，缺失的目录和数据库会以仅所有者可访问的方式创建（进程 umask 前为 `0700` 和 `0600`），SQLite sidecar 继承数据库 mode；现有 mode 保持不变。每个派生索引路径在一个进程中只能由一个服务拥有；不支持外部写入者或第二个进程，因为世代和 TEMP 遮蔽状态由连接持有。
 
 ## 配置
 
 | 键 | 默认值 | 契约 |
 |---|---:|---|
-| `path` | required | 专用派生索引 SQLite 路径；支持 `:memory:`。在 POSIX 文件系统上，缺失的文件系统路径会以仅所有者可访问的方式创建。 |
+| `path` | 必填 | 专用派生索引 SQLite 路径；支持 `:memory:`。在 POSIX 文件系统上，缺失的文件系统路径会以仅所有者可访问的方式创建。 |
 | `journalMode` | `wal` | `wal`、`delete`、`truncate` 或 `persist`。 |
 | `defaultLimit` | `20` | 请求省略 `limit` 时的分页大小；最多为 `Number.MAX_SAFE_INTEGER - 1`。 |
 | `maxLimit` | `100` | 接受的最大请求分页大小；最多为 `Number.MAX_SAFE_INTEGER - 1`。 |
@@ -32,23 +32,23 @@
 | `readWindowMax` | `50` | `before` 或 `after` 的最大原始事件数，用于继承的 `readEvent()`。 |
 | `persistedInspectConcurrency` | `4` | 继承批量读取的最大并发持久化日志检查数；必须是正安全整数。 |
 
-## Tokenizer 与限制
+## 分词器与限制
 
 该索引使用 FTS5 `unicode61`。在实现实验中，它支持双字符查询 `AI`，产生的索引比 trigram 备选方案小约 2.1 倍。取舍是 token/短语召回而非任意子字符串召回：`AI` 不匹配 token `BRAID`。需要执行字面的空白弹性子字符串扫描时，使用 `ctx.sessionQuery.filterEvents()` 并传入 `text` 子句。查询会拒绝 NUL；文档中的保留高亮标记和 NUL 会在索引前被规范化，使展示标记无法与源文本冲突。
 
-中止信号会停止已排队工作，并原样流经快照列表和非变更检查。来源工作一旦开始，串行化状态机会自行等待该后端 promise，即使后端忽略取消，之后也会在启动任何其他列表、检查、对账或查询工作前检查信号。因此，调用方只会在已启动后端工作完全停稳后观察到取消，而后续搜索在该清理尚未完成时无法进入 serializer。Node 的同步 `DatabaseSync` API 无法中断已在 JavaScript 线程上执行的元数据或 MATCH 语句；系统会在这些不可抢占调用前后立即检查信号。
+中止信号会停止已排队工作，并原样流经快照枚举和非修改式检查。来源工作一旦开始，串行化状态机会自行等待该后端 promise，即使后端忽略取消，之后也会在启动任何进一步的枚举、检查、对账或查询工作前检查信号。因此，调用方只会在已启动后端工作完全停稳后观察到取消，而后续搜索在该清理尚未完成时无法进入 serializer。Node 的同步 `DatabaseSync` API 无法中断已在 JavaScript 线程上执行的元数据或 MATCH 语句；系统会在这些不可抢占调用前后立即检查信号。
 
 ## 模型体验
 
 无。该可信搜索后端只向调用方返回命中，不注册面向模型的提示词、schema、工具或消息。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 无；该包既不组装也不发送提供方请求。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **无调用方授权**：这是上下文范围内的可信服务；模型工具或 UI 必须强制执行自己的访问策略。
 - **同步查询执行**：`DatabaseSync` 在 MATCH 执行期间会阻塞 JavaScript 线程，且无法中断已运行的语句。
 - **Token 召回，而非任意子字符串**：`unicode61` tokenizer 不会匹配更大 token 中的子字符串；对字面扫描使用 `filterEvents()`。
-- **单 owner 派生索引**：每个索引路径必须由一个进程中的一个服务拥有；不支持外部写入者和多进程共享。
+- **单一所有者的派生索引**：每个索引路径必须仅归一个进程中的一个服务所有；不支持外部写入者和多进程共享。
diff --git a/packages/session-query/session-query/README.i18n.yaml b/packages/session-query/session-query/README.i18n.yaml
index 1db5958d5f..3a122bbf43 100644
--- a/packages/session-query/session-query/README.i18n.yaml
+++ b/packages/session-query/session-query/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-query/session-query/README.md
 README.md: ebc577975f874a1a60c84061f9148282742bfdf2
-README.zh.md: c4d0c27c846bad6b9b621b6db391be1d4ee69fed
+README.zh.md: df504e8ae5172fa783769a6fa39830c446688e29
diff --git a/packages/session-query/session-query/README.zh.md b/packages/session-query/session-query/README.zh.md
index c4d0c27c84..df504e8ae5 100644
--- a/packages/session-query/session-query/README.zh.md
+++ b/packages/session-query/session-query/README.zh.md
@@ -10,47 +10,47 @@
 - `readSession(sessionId)` 在执行与恢复相同的核心回放验证后，返回一份完整、脱离存储的原始日志；它绝不会将该会话放入实时存储。
 - `filterSessions(filters, signal?)` 对同一份克隆逻辑语料库应用与提供方无关的会话元数据和可用性谓词。
 - `filterEvents(sessionId, filters)` 提取第一方语义文档，并按 seq 升序应用与提供方无关的元数据和字面文本谓词。
-- `readTitleSnapshots(sessionIds, signal?)` 从一次实时优先的语料库观察中解析唯一 id，将取消传递给持久化列表和检查，并按顺序返回每个会话的结算结果，使某个缺失或格式错误的标题来源不会丢弃其他来源。每个实时来源直接 fold，每个持久化 worker fold 为脱离存储的 header/标题结果，并在出队下一个 id 前释放完整日志。取消会拒绝整个批次。`readTitleSnapshot(sessionId, signal?)` 是单次观察视图；`readTitle(sessionId, signal?)` 只返回其可选的 folded `session/title`。
-- `listEvents(sessionId)` 加载实时优先的原始日志，将每个事件分类为 `current`、`shadowed` 或 `log-only`；该分类使用共享 `dsh-session` 接口 fold。
-- `readSurface(sessionId)` 返回一个克隆 header、原始日志捕获边界，以及按模型历史顺序排列的完整 folded 当前接口。实时会话优先于持久化；压缩只会在其替换追加之前或之后被观察，绝不会出现合成混合。
+- `readTitleSnapshots(sessionIds, signal?)` 从一次实时优先的语料库观察中解析唯一 id，将取消信号传递给持久化列表查询和检查，并按顺序返回每个会话的结算结果，使某个缺失或格式错误的标题来源不会丢弃其他来源。每个实时来源直接 fold，每个持久化 worker fold 为脱离存储的 header/标题结果，并在出队下一个 id 前释放完整日志。取消会拒绝整个批次。`readTitleSnapshot(sessionId, signal?)` 是单次观察视图；`readTitle(sessionId, signal?)` 只返回其可选的 folded `session/title`。
+- `listEvents(sessionId)` 加载实时优先的原始日志，将每个事件分类为 `current`、`shadowed` 或 `log-only`；该分类使用共享 `dsh-session` 表层 fold。
+- `readSurface(sessionId)` 返回一个克隆 header、原始日志捕获边界，以及按模型历史顺序排列的完整折叠后当前表层。实时会话优先于持久化；压缩（compaction）只会在其替换追加之前或之后被观察，绝不会出现合成混合。
 - `readEvent(request, signal?)` 返回一个克隆 header、完整目标事件和有界的原始 seq 窗口。`before` 和 `after` 默认为 0，且不得超过 `readWindowMax`。
-- `traceSession(sessionId, signal?)` 只读取一次语料库，返回从直接到向外的祖先，以及确定性的递归后代树。`complete: false` 标识第一个缺失父级；与目标相连的循环会以 `SESSION_QUERY_INVALID_LINEAGE` 失败。
+- `traceSession(sessionId, signal?)` 只读取一次语料库，返回从直接父级向外的祖先，以及确定性的递归后代树。`complete: false` 标识第一个缺失父级；与目标相连的循环会以 `SESSION_QUERY_INVALID_LINEAGE` 失败。
 - `traceEvent(request, signal?)` 只加载一次逻辑日志，返回其克隆源 header、直接位置替换和直接已记录来源信息。`replacementChain` 沿位置替换者跟踪到最终替换；来源链接仍不传递。
 
-持久化是可选的，可动态挂载或卸载。已挂载持久化无法读取时，跨语料库列表和血缘跟踪以 `SESSION_QUERY_PERSISTENCE_FAILED` 失败。针对已知实时会话的标题读取、事件跟踪或事件读取不会查询持久化，因此持久后端的健康状态无法使当前内存状态变得不可读。持久化标题和事件操作在加载前列表，并在元数据不匹配时拒绝，而不会组合不一致的观察。血缘跟踪取消传递到持久化列表；事件跟踪和事件读取取消传递到持久化列表和检查。每项操作都会等待已启动的后端调用结算，然后使用信号的精确原因拒绝，即使后端忽略了该信号。预先中止的已知实时标题读取、事件跟踪或事件读取会在 fold 或快照之前拒绝，且不查询持久化。批量标题观察执行一次元数据列表，使用最多 `persistedInspectConcurrency` 个 worker 检查唯一持久化 id，并保留每个标题自己观察到的 header，供下游授权使用。取消不会启动已排队检查，且只在已启动 worker 结算后拒绝。`listSessions()` 仍保持轻量，不加载日志或索引标题。
+持久化是可选的，可动态挂载或卸载。已挂载持久化无法读取时，跨语料库列表和血缘跟踪以 `SESSION_QUERY_PERSISTENCE_FAILED` 失败。针对已知实时会话的标题读取、事件跟踪或事件读取不会查询持久化，因此持久化后端的健康状态无法使当前内存状态变得不可读。持久化标题和事件操作在加载前先执行列表查询，并在元数据不匹配时拒绝，而不会组合不一致的观察。血缘跟踪的取消信号会传递给持久化列表查询；事件跟踪和事件读取的取消信号会传递给持久化列表查询和检查。每项操作都会等待已启动的后端调用结算，然后使用信号的精确原因拒绝，即使后端忽略了该信号。针对已知实时会话且预先中止的标题读取、事件跟踪或事件读取会在 fold 或快照之前拒绝，且不查询持久化。批量标题观察执行一次元数据列表查询，使用最多 `persistedInspectConcurrency` 个 worker 检查唯一持久化 id，并保留每个标题自己观察到的 header，供下游授权使用。取消不会启动已排队检查，且只在已启动 worker 结算后拒绝。`listSessions()` 仍保持轻量，不加载日志或索引标题。
 
 ## 过滤与提取
 
-`SessionResultFilter` 覆盖 id、可空 cwd、创建时间范围、可空父级和来源可用性。`SessionEventResultFilter` 覆盖 seq/时间范围、事件类型、接口和语义文本。过滤器数组使用 AND；同一列表子句内的值使用 OR。空列表值不匹配任何内容，范围包含端点，而格式错误的范围或封闭联合值以 `SESSION_QUERY_INVALID_FILTER` 失败。
+`SessionResultFilter` 覆盖 id、可空 cwd、创建时间范围、可空父级和来源可用性。`SessionEventResultFilter` 覆盖 seq/时间范围、事件类型、表层和语义文本。过滤器数组使用 AND；同一列表子句内的值使用 OR。空列表值不匹配任何内容，范围包含端点，而格式错误的范围或封闭联合值以 `SESSION_QUERY_INVALID_FILTER` 失败。
 
-文本子句刻意与 FTS 提供方无关：调用方文本会被转义为不区分大小写的 Unicode 正则表达式，每个空白运行匹配一个或多个空白字符。它是字面语义文本扫描，而非全文查询。`extractSessionEventText()` 和 `buildSessionEventSearchDocuments()` 定义共享的第一方文档投影；结构边界、流分片、请求 header 和未知声明合并变体不产生文档。
+文本子句刻意与 FTS 提供方无关：调用方文本会被转义为不区分大小写的 Unicode 正则表达式，每段连续空白匹配一个或多个空白字符。它是字面语义文本扫描，而非全文查询。`extractSessionEventText()` 和 `buildSessionEventSearchDocuments()` 定义共享的第一方文档投影；结构边界、流分片、请求 header 和未知声明合并变体不产生文档。
 
 ## 全文方法
 
-`SessionQueryService.searchSessions(request, exec?)` 按匹配最强的事件对逻辑语料库分组；`searchEvents(request, exec?)` 搜索一个逻辑会话。这两个是服务仅有的抽象方法。两者都返回分页结果，其延续信息是自有的品牌化 `SessionSearchCursor`；接受可选取消，并在不使用提供方专用数值分数的情况下公开 snippet。事件搜索分页结果还携带来自与命中相同索引世代的克隆目标 header，使授权消费方可将策略绑定到 payload 观察。搜索请求只接受元数据事件过滤器，因为字面文本过滤使用上文所述扫描路径。
+`SessionQueryService.searchSessions(request, exec?)` 按匹配最强的事件对逻辑语料库分组；`searchEvents(request, exec?)` 搜索一个逻辑会话。这两个是服务仅有的抽象方法。两者都返回分页结果，其延续信息是由服务持有的带品牌 `SessionSearchCursor`；接受可选取消，并在不使用提供方专用数值分数的情况下提供摘录。事件搜索分页结果还携带来自与命中相同索引世代的克隆目标 header，使授权消费方可将策略绑定到此次载荷观察。搜索请求只接受事件元数据过滤器，因为字面文本过滤使用上文所述扫描路径。
 
-该包没有提供方协调器、回退实现或独立具体插件。具体服务后端继承已实现的读取、过滤和跟踪，同时负责全文观察、对账、排名、游标世代和查询执行；第一个实现是 [`@deepseek-ai/dsh-session-query-sqlite`](../session-query-sqlite/README.md)。
+该包（package）没有提供方协调器、回退实现或独立具体插件。具体服务后端继承已实现的读取、过滤和跟踪，同时负责全文观察、对账、排名、游标世代和查询执行；第一个实现是 [`@deepseek-ai/dsh-session-query-sqlite`](../session-query-sqlite/README.md)。
 
-`SessionQueryError.code` 是一个封闭联合，覆盖请求验证、缺失目标、格式错误的接口、来源冲突、持久化/索引失败、取消，以及无效或陈旧游标；精确字面值在 [`src/config.ts`](src/config.ts) 中定义。
+`SessionQueryError.code` 是一个封闭联合，覆盖请求验证、缺失目标、格式错误的表层、来源冲突、持久化/索引失败、取消，以及无效或陈旧游标；精确字面值在 [`src/config.ts`](src/config.ts) 中定义。
 
-`listEvents()`、`readSurface()` 和 `traceEvent()` 执行同一个单遍 `dsh-session` 接口 fold。只有当事件 seq 从零开始且连续、接口标记符合事件类型资格、来源数组非空且无重复、引用指向较早事件，且每个位置替换都命名并引用它移除的每个接口节点时，加载的日志才有效；任何违规都以 `SESSION_QUERY_INVALID_SURFACE` 失败。
+`listEvents()`、`readSurface()` 和 `traceEvent()` 执行同一个单遍 `dsh-session` 表层 fold。只有当事件 seq 从零开始且连续、表层标记符合事件类型的适用性要求、来源数组非空且无重复、引用指向较早事件，且每个位置替换都命名并引用它移除的每个表层节点时，加载的日志才有效；任何违规都以 `SESSION_QUERY_INVALID_SURFACE` 失败。
 
 ## 配置
 
 | 键 | 默认值 | 契约 |
 |---|---:|---|
 | `readWindowMax` | `50` | `before` 或 `after` 的最大原始事件数。 |
-| `persistedInspectConcurrency` | `4` | 一次批量读取中的最大并发持久化日志检查数；必须是正安全整数。 |
+| `persistedInspectConcurrency` | `4` | 一次批量读取中的最大并发持久化日志检查数；必须是正的安全整数。 |
 
 ## 模型体验
 
 无。该可信查询服务只向调用方返回克隆会话记录，不注册面向模型的提示词、schema、工具或消息。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 无；该包既不组装也不发送提供方请求。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **无调用方授权**：这是上下文范围内的可信基础设施；未来的模型工具或 UI 必须限制调用方可检查的会话。
 - **无注册表或面向模型工具**：尚未提供提取器和搜索提供方注册表、递归事件来源遍历以及面向模型的工具。[跟踪决策](../../../.agents/notes/implemented/feature/2026-07-13-session-query-tracing.md) 负责关系语义；SQLite 归属和 tokenizer 决策位于[已实现搜索记录](../../../.agents/notes/implemented/feature/2026-07-10-sqlite-session-query-provider.md)。
diff --git a/packages/session-query/tool-session-query/README.i18n.yaml b/packages/session-query/tool-session-query/README.i18n.yaml
index 15156a8eeb..e86449af9c 100644
--- a/packages/session-query/tool-session-query/README.i18n.yaml
+++ b/packages/session-query/tool-session-query/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-query/tool-session-query/README.md
 README.md: d973daf1124c4be05f7335b18661d431d45be39f
-README.zh.md: 94cbd4e36c7146c7759b49421cf651abb327b5f4
+README.zh.md: b27d79a905a029d3750f24573e3c32785a314015
diff --git a/packages/session-query/tool-session-query/README.zh.md b/packages/session-query/tool-session-query/README.zh.md
index 94cbd4e36c..b27d79a905 100644
--- a/packages/session-query/tool-session-query/README.zh.md
+++ b/packages/session-query/tool-session-query/README.zh.md
@@ -2,20 +2,20 @@
 
 [English](README.md) | 中文
 
-位于 `ctx.sessionQuery` 之上、经工作区授权的模型工具。该 opt-in 包只依赖统一接口，并注册 `session_search`、`session_event_search`、`session_trace`、`session_event_trace` 和 `session_event_read`；已发布的宿主组合默认不挂载它。
+位于 `ctx.sessionQuery` 之上、经工作区授权的模型工具。该 opt-in 包（package）只依赖统一接口，并注册 `session_search`、`session_event_search`、`session_trace`、`session_event_trace` 和 `session_event_read`；已发布的宿主组合默认不挂载它。
 
 ## 配置
 
 | 键 | 默认值 | 含义 |
 |---|---:|---|
 | `maxSearchResults` | `100` | 在内部提供方分页中收集的最大已授权非自身命中数 |
-| `searchTimeoutMs` | `30000` | 连接到两个全文搜索工具的协作式 deadline |
+| `searchTimeoutMs` | `30000` | 附加到两个全文搜索工具的协作式截止时间 |
 
 调用方只能来自 `ToolExecution.exec.agent`。跨会话访问要求目标和调用方会话的 `cwd` 值严格相等；没有 `cwd` 的调用方只能检查自己。搜索绝不公开提供方游标、偏移、分页大小或模型可控上限。由于一次搜索会在内部消费与世代绑定的提供方游标，两个搜索工具都与同级工具调用排他执行；三个精确跟踪/读取工具选择并行执行。每个精确执行器都将未更改的执行信号传递给授权和服务跟踪/读取，因此取消会等待协作式持久化清理，并保留信号的精确原因。工具边界上的时间戳要求显式 `Z` 或数字偏移，并转换为包含端点的 epoch 毫秒过滤器。
 
 `session_search` 始终省略调用方会话。请求的父 id 会被去重，并在 FTS 前根据调用方工作区权限检查；只有已授权 id 会到达提供方，而缺失猜测和跨工作区猜测的行为完全相同，root 标记仍独立使用 OR。当前会话中的 `session_event_search` 会在调用它的步骤之前立即停止，因此当前 assistant 输出和已记录工具调用无法匹配自身。直接目标在跟踪、事件或标题读取前完成授权。血缘输出会用不含隐藏会话 id 的标记替换未授权祖先和后代边界。
 
-每个可信 `ctx.sessionQuery` 调用都会经过一个模型边界净化器。首先检查调用方取消，并精确保留。可用语料库和提供方诊断（包括可安全检查的嵌套原因）会尽力记录到内部日志；不可打印的失败使用固定日志占位符。诊断格式化和错误分类各自独立受保护，因此不可打印的原因无法逃逸，也无法阻止已安全分类的外层错误；不安全的分类或日志记录则回退到固定 `SESSION_QUERY_TOOL_FAILED` 代码和消息。本地参数验证和授权错误保留精确的工具自有消息。
+每个可信 `ctx.sessionQuery` 调用都会经过一个模型边界净化器。首先检查调用方取消，并精确保留。可获取的语料库诊断信息和提供方诊断信息（包括可安全检查的嵌套原因）会尽力记录到内部日志；不可打印的失败使用固定日志占位符。诊断格式化和错误分类各自独立受保护，因此不可打印的原因无法逃逸，也无法阻止已安全分类的外层错误；不安全的分类或日志记录则回退到固定 `SESSION_QUERY_TOOL_FAILED` 代码和消息。本地参数验证和授权错误保留精确的工具自有消息。
 
 该包刻意不执行字节或字符截断，也不导入 spill 后端。需要限制内联输出的部署应挂载 `@deepseek-ai/dsh-spill-policy`，它可在执行后替换已渲染文本，同时保留完整结果。
 
@@ -23,7 +23,7 @@
 
 ### 系统提示词
 
-#### 模型所见
+#### 模型看到的内容
 
 模型会收到一个固定的既往历史指引章节。
 
@@ -37,13 +37,13 @@ Use session_search to find relevant work from prior sessions, or session_event_s
 
 插件挂载期间，每次请求都存在一个固定精简章节。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 插件和指引文本不变时，前缀稳定。
 
 ### 工具 schema
 
-#### 模型所见
+#### 模型看到的内容
 
 模型会看到生成的 [`session_search`、`session_event_search`、`session_trace`、`session_event_trace` 和 `session_event_read` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-session-query)。搜索过滤器会增加固定 schema token，而游标、工作区路径、输出分页和模型可控结果上限仍不存在。
 
@@ -51,26 +51,26 @@ Use session_search to find relevant work from prior sessions, or session_event_s
 
 可见期间，每次请求都会发送 5 个固定只读 schema。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 工具可见性和定义不变时，前缀稳定。
 
 ### 工具结果
 
-#### 模型所见
+#### 模型看到的内容
 
-每次成功调用都会发出一个纯文本块。搜索结果包含标题和最佳匹配摘要；跟踪包含全部已授权关系；事件读取包含未缩写的目标 JSON。通用 spill 策略可以将过大的内联文本替换为预览、不透明定位信息和取回指引。
+每次成功调用都会发出一个纯文本块。搜索结果包含标题和最佳匹配摘录；跟踪包含全部已授权关系；事件读取包含未经删节的目标 JSON。通用 spill 策略可以将过大的内联文本替换为预览、不透明定位信息和取回指引。
 
 #### Token 影响
 
-结果取决于数据，并保留在已记录工具历史中直到压缩；`maxSearchResults` 限制搜索命中数。
+结果取决于数据，并保留在已记录工具历史中直到压缩（compaction）；`maxSearchResults` 限制搜索命中数。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 仅追加的结果文本位于可重用请求前缀之后，不会使较早的缓存条目失效。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - 搜索最多返回部署上限，匹配更多时会请模型缩小查询；不提供延续 token。
 - 工作区身份使用保守的字符串精确 `cwd` 相等性，因此符号链接等价的路径不共享权限。
-- 未挂载通用 spill 策略的自定义组合会在内联位置接受完整跟踪和事件 payload。
+- 未挂载通用 spill 策略的自定义组合会以内联方式接收完整跟踪和事件载荷。
diff --git a/packages/session-title/README.i18n.yaml b/packages/session-title/README.i18n.yaml
index 1609fda4ae..9cdb34dc49 100644
--- a/packages/session-title/README.i18n.yaml
+++ b/packages/session-title/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-title/README.md
 README.md: 64bca8153e566e1590776af511662a873198106e
-README.zh.md: b6779f3e798d695a3c22634bb298f2a7681edbfb
+README.zh.md: 5ff52211ac975e31845285f99618503cd775eded
diff --git a/packages/session-title/README.zh.md b/packages/session-title/README.zh.md
index b6779f3e79..5ff52211ac 100644
--- a/packages/session-title/README.zh.md
+++ b/packages/session-title/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-持久会话标题状态、一个可选异步提供方 seam，以及两个可选启用的模型后端实现。内置首消息回退属于服务本身，因此任何组合都能在不调用辅助模型的情况下为会话生成标题。
+持久化的会话标题状态、一个可选异步提供方 seam，以及两个由模型支持、可选启用的实现。内置首消息回退属于服务本身，因此任何组合都能在不调用辅助模型的情况下为会话生成标题。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
diff --git a/packages/session-title/session-title-all-messages-llm/README.i18n.yaml b/packages/session-title/session-title-all-messages-llm/README.i18n.yaml
index 352cfe4f68..54909b457e 100644
--- a/packages/session-title/session-title-all-messages-llm/README.i18n.yaml
+++ b/packages/session-title/session-title-all-messages-llm/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-title/session-title-all-messages-llm/README.md
 README.md: 25ec92432b5c2d18624f2f4d851552ee2b3bf5d0
-README.zh.md: 685046a2a4d0859e6eca86cfba67690b8271afaf
+README.zh.md: 947aa614741b2b0bd50e70f1d273a625a5624cc6
diff --git a/packages/session-title/session-title-all-messages-llm/README.zh.md b/packages/session-title/session-title-all-messages-llm/README.zh.md
index 685046a2a4..947aa61474 100644
--- a/packages/session-title/session-title-all-messages-llm/README.zh.md
+++ b/packages/session-title/session-title-all-messages-llm/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-可选的 `ctx.sessionTitle` 提供方，通过 `ctx.llm` 总结所有符合条件的用户消息。它注册 `all-user-messages` 节奏，并在每条新用户提示词后启动新 revision，同时使用 seed 历史与子会话提示词。较新的 revision 会中止并取代旧工作；即使提供方忽略取消，也无法提交陈旧输出。
+可选的 `ctx.sessionTitle` 提供方，通过 `ctx.llm` 总结所有符合条件的用户消息。它注册 `all-user-messages` 节奏，并在每条新用户提示词后启动新 revision，同时使用预置历史与子会话提示词。较新的 revision 会中止并取代旧工作；即使提供方忽略取消，也无法提交陈旧输出。
 
-该插件使用完整且必填的[共享 LLM 配置](../session-title-llm/README.md#configuration)。同时省略 `provider` 与 `model` 时，会继承每个当前已记录主请求的确切路由；也可以同时设置二者，使标题生成使用独立路由。如果最终封装的聚合提示词超过 `maxInputBytes`，请求会失败而不是截断历史；自动使用时会发出警告并保留先前标题。
+该插件使用完整且必填的[共享大语言模型（LLM）配置](../session-title-llm/README.md#configuration)。同时省略 `provider` 与 `model` 时，会继承每个当前已记录主请求的确切路由；也可以同时设置二者，使标题生成使用独立路由。如果最终封装的聚合提示词超过 `maxInputBytes`，请求会失败而不是截断历史；自动使用时会发出警告并保留先前标题。
 
 ## 模型体验
 
@@ -12,17 +12,17 @@
 
 #### 模型看到的内容
 
-标题模型会收到共享标题指令，以及一个 JSON 数组，其中按日志顺序包含截至当前 revision 的所有符合条件用户消息和确切 seq。Seed 历史也包含在内。
+标题模型会收到共享标题指令，以及一个 JSON 数组，其中按日志顺序包含截至当前 revision 的所有符合条件用户消息和确切 seq。预置历史也包含在内。
 
 #### Token 影响
 
-每条符合条件的新提示词之后都可能发送一项辅助请求，每次请求受 `maxInputBytes` 和 `maxOutputTokens` 约束；显式刷新可能增加调用。主 agent 请求不会增加 token。
+每条符合条件的新提示词之后都可能发送一次辅助请求，每次请求受 `maxInputBytes` 和 `maxOutputTokens` 约束；显式刷新可能增加调用。主 agent（智能体）请求不会增加 token。
 
 #### KV Cache 影响
 
-不会使主请求失效。每条提示词后，辅助输入都会增长或变化，因此提供方专用缓存复用会在第一个变化的 JSON token 处结束。
+不会使主请求的 KV Cache 失效。每条提示词后，辅助输入都会增长或变化，因此提供方专用缓存复用会在第一个变化的 JSON token 处结束。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- 输入溢出时保留先前标题；对于很长的会话，此提供方没有摘要再摘要机制或保留策略。
+- 输入溢出时保留先前标题；对于很长的会话，此提供方没有基于摘要继续生成摘要的机制或保留策略。
 - 它平等对待所有符合条件的用户消息，不提供权重、过滤或手动标题优先级。
diff --git a/packages/session-title/session-title-first-message-llm/README.i18n.yaml b/packages/session-title/session-title-first-message-llm/README.i18n.yaml
index ed6f3e7fe8..1353b0b4e6 100644
--- a/packages/session-title/session-title-first-message-llm/README.i18n.yaml
+++ b/packages/session-title/session-title-first-message-llm/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/session-title/session-title-first-message-llm/README.md
 README.md: c24d4de1fa945d11b5a868cad88a94ed14ad3891
-README.zh.md: a3adc5366c841ed6aa08220209ed372db1fc8666
+README.zh.md: 1f864b16c473654d0a52716cddbe26fe087860dc
diff --git a/packages/session-title/session-title-first-message-llm/README.zh.md b/packages/session-title/session-title-first-message-llm/README.zh.md
index a3adc5366c..1f864b16c4 100644
--- a/packages/session-title/session-title-first-message-llm/README.zh.md
+++ b/packages/session-title/session-title-first-message-llm/README.zh.md
@@ -4,7 +4,7 @@
 
 可选的 `ctx.sessionTitle` 提供方，通过 `ctx.llm` 总结第一条符合条件的用户消息。它注册 `first-message` 节奏，只在全新非 fork 会话首次创建回退时自动运行，并将结果归因于该消息的确切 seq。自动失败会保留回退，之后只能通过 `ctx.sessionTitle.refresh()` 重试。
 
-该插件使用完整且必填的[共享 LLM 配置](../session-title-llm/README.md#configuration)。同时省略 `provider` 与 `model` 时，会继承当前已记录主请求的确切路由；也可以同时设置二者，使标题生成使用独立路由。
+该插件使用完整且必填的[共享大语言模型（LLM）配置](../session-title-llm/README.md#configuration)。同时省略 `provider` 与 `model` 时，会继承当前已记录主请求的确切路由；也可以同时设置二者，使标题生成使用独立路由。
 
 ## 模型体验
 
@@ -12,17 +12,17 @@
 
 #### 模型看到的内容
 
-标题模型会收到共享标题指令，以及一个只包含第一条符合条件用户消息的 JSON 数组。后续提示词与继承的 fork 历史不会触发另一项自动调用。
+标题模型会收到共享标题指令，以及一个只包含第一条符合条件用户消息的 JSON 数组。后续提示词与继承的 fork 历史不会触发再次自动调用。
 
 #### Token 影响
 
-全新会话最多自动发出一项辅助请求，并受 `maxInputBytes` 和 `maxOutputTokens` 约束；显式刷新可能发出额外调用。主 agent 请求不会增加 token。
+全新会话最多自动发出一次辅助请求，并受 `maxInputBytes` 和 `maxOutputTokens` 约束；显式刷新可能发出额外调用。主 agent（智能体）请求不会增加 token。
 
 #### KV Cache 影响
 
-不会使主请求失效。辅助请求使用已配置或已记录路由，其缓存行为由提供方决定。
+不会使主请求的 KV Cache 失效。辅助请求使用已配置或已记录路由，其缓存行为由提供方决定。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - 对于长期会话，第一条消息可能不再具有代表性；如果后续提示词应触发重新生成标题，请使用全消息提供方。
-- Fork 会保留继承的标题，绝不会自动运行此提供方，即使 seed 中的首消息来自父会话。
+- fork 会保留继承的标题，绝不会自动运行此提供方，即使其预置的首消息来自父会话。
diff --git a/packages/session-title/session-title-llm/README.i18n.yaml b/packages/session-title/session-title-llm/README.i18n.yaml
index 2343a6cb3a..2ae140e9a5 100644
--- a/packages/session-title/session-title-llm/README.i18n.yaml
+++ b/packages/session-title/session-title-llm/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/session-title/session-title-llm/README.md
 README.md: 342687b5aa0cd35a70abf8cc66b3fe80342bce0b
-README.zh.md: 2bb48b52ae6298c5c895ef99ff54cf5f1d7a6d8c
+README.zh.md: d86366b589ba016793ca63daa7279504761744b3
diff --git a/packages/session-title/session-title-llm/README.zh.md b/packages/session-title/session-title-llm/README.zh.md
index 2bb48b52ae..d86366b589 100644
--- a/packages/session-title/session-title-llm/README.zh.md
+++ b/packages/session-title/session-title-llm/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-模型后端会话标题提供方的共享实现策略。它解析辅助路由，将精确选中的用户消息封装为 JSON，记录可分发的确切请求，应用语言感知的标题指令，强制执行输入和输出预算，组合超时与调用方取消，组装流，并返回带有确切来源 seq 和模型 provenance 的规范化文本。
+由模型支持的会话标题提供方的共享实现策略。它解析辅助路由，将精确选中的用户消息封装为 JSON，记录可分发的确切请求，应用语言感知的标题指令，强制执行输入和输出预算，组合超时与调用方取消，组装流，并返回带有确切来源 seq 和模型来源信息的规范化文本。
 
-此包是普通库，不是 Cordis 插件。提供方插件调用 `registerSessionTitleLlmProvider()`，传入各自节奏与消息选择器；该函数验证共享配置，并将每个 revision 委派给 `generateSessionTitleWithLlm()`，使两者的注册、路由、提示词、取消与验证行为不会漂移。
+此包（package）是普通库，不是 Cordis 插件。提供方插件调用 `registerSessionTitleLlmProvider()`，传入各自节奏与消息选择器；该函数验证共享配置，并将每次修订委派给 `generateSessionTitleWithLlm()`，使各插件的注册、路由、提示词、取消与验证行为不会漂移。
 
 ## 路由与失败契约
 
-`provider` 和 `model` 覆盖项都是可选的，但必须同时作为非空字符串提供。如果没有这一对取值，辅助模块会使用当前会话已记录 `request/header` 中捕获的确切提供方／模型路由；因此，在任何路由出现前显式刷新时必须提供覆盖项。辅助模块在记录或分发前，以 `maxInputBytes` 测量最终 JSON 封装的用户提示词，包括 seq 字段、包装层与 JSON 转义，而不是将其截断。消费流期间和流完成后都会重新检查超时与调用方取消，因此即使 interceptor 或适配器忽略 abort，也不能接受迟到的成功结果。格式错误或空输出、工具调用和非 stop 结束原因同样会 reject；会话标题服务决定该 reject 属于自动警告还是显式调用方失败。
+`provider` 和 `model` 覆盖项都是可选的，但必须同时作为非空字符串提供。如果没有这一对取值，辅助模块会使用当前会话已记录 `request/header` 中捕获的确切提供方／模型路由；因此，在任何路由出现前显式刷新时必须提供覆盖项。辅助模块在记录或分发前，依据 `maxInputBytes` 检查最终 JSON 封装用户提示词的大小，包括 seq 字段、包装层与 JSON 转义，而不是将其截断。消费流期间和流完成后都会重新检查超时与调用方取消，因此即使 interceptor 或适配器忽略 abort，也不能接受迟到的成功结果。格式错误或空输出、工具调用和非 stop 结束原因同样会导致调用被拒绝；会话标题服务决定该拒绝属于自动警告还是显式调用方失败。
 
-路由与输入验证完成后，辅助模块会在模型分发前直接通过 `Session` 追加仅写入日志的 `session/title-llm-request` 事件。它包含标题提供方 id、确切来源 seq、路由、系统提示词、消息列表，以及该调用使用的输出 token 上限。持久化会尽快观察该记录；追加不需要标题专属标记、类型断言、结算队列或 flush。分发的 envelope 会深度冻结，携带 `purpose: 'session-title'`，且有意不包含 dsh-agent-loop 的进程本地请求身份。Interceptor 会与记录保持一致，而循环专用重建观察者不会把它与对话 header 比较。DeepSeek 适配器会将该 purpose 映射为关闭 thinking，使少量输出预算全部用于可见标题文本；其他适配器负责自身 purpose 专用行为。后续模型失败会保留请求记录；从未成为可分发请求的验证失败不会创建记录。该事件始终位于派生模型历史之外。
+路由与输入验证完成后，辅助模块会在模型分发前直接通过 `Session` 追加仅写入日志的 `session/title-llm-request` 事件。它包含标题提供方 id、确切来源 seq、路由、系统提示词、消息列表，以及该调用使用的输出 token 上限。持久化会立即观察到该记录；追加不需要标题专属标记、类型断言、结算队列或刷写。分发的请求封套会深度冻结，携带 `purpose: 'session-title'`，且有意不包含 dsh-agent-loop 的进程本地请求身份。拦截器会与记录保持一致，而循环专用重建观察者不会把它与对话请求头比较。DeepSeek 适配器会根据该用途禁用思考，使少量输出预算全部用于可见标题文本；其他适配器负责自身用途专用行为。后续模型失败会保留请求记录；从未成为可分发请求的验证失败不会创建记录。该事件始终位于派生模型历史之外。
 
 ## 配置
 
@@ -22,7 +22,7 @@
 | `targetCjkCharacters` | 中文、日文或韩文标题的正整数目标字符数。 |
 | `maxInputBytes` | 最终 JSON 封装用户提示词的正整数 UTF-8 字节上限。 |
 | `maxOutputTokens` | 辅助生成的正整数 token 上限。 |
-| `timeoutMs` | 运行时定时器限制内的端到端正数 deadline。 |
+| `timeoutMs` | 运行时定时器限制内的正数端到端时限。 |
 | `provider`, `model` | 可选显式路由；二者同时提供或同时省略。 |
 
 ## 模型体验
@@ -31,17 +31,17 @@
 
 #### 模型看到的内容
 
-标题模型会收到固定系统指令，要求以输入语言返回一个简洁且无装饰的标题，其中包含所配置的词数与 CJK 字符数目标。它唯一的用户消息包含一个 JSON 数组，其中是精确选中的用户消息及其 seq。
+标题模型会收到固定系统指令，要求以输入语言返回一个简洁且无装饰的标题；该指令包含所配置的词数与 CJK 字符数目标。它唯一的用户消息包含一个 JSON 数组，其中是精确选中的用户消息及其 seq。
 
 #### Token 影响
 
-辅助请求根据所选输入大小和 `maxOutputTokens` 消耗 token。它与主 agent 请求相互独立，不会向 agent 历史增加标题文本或封装内容。DeepSeek 标题调用会关闭 thinking；主对话保留自身配置的 thinking 模式。
+辅助请求根据所选输入大小和 `maxOutputTokens` 消耗 token。它与主 agent（智能体）请求相互独立，不会向 agent 历史增加标题文本或封装内容。DeepSeek 标题调用会关闭思考；主对话保留自身配置的思考模式。
 
 #### KV Cache 影响
 
-不会使主请求失效。辅助缓存复用由提供方决定；固定指令可复用，而 JSON 消息数组会随每个 revision 变化。
+不会使主请求的 KV Cache 失效。辅助缓存复用由提供方决定；固定指令可复用，而 JSON 消息数组会随每次修订变化。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
 - 辅助模块只接受文本输出，并拒绝工具调用；不公开结构化输出适配器或提供方专用提示词变体。
 - 它对整个封装用户提示词强制执行字节上限，不会剪裁单条消息或应用保留策略。
diff --git a/packages/session-title/session-title/README.i18n.yaml b/packages/session-title/session-title/README.i18n.yaml
index 1e430b8974..5760b076a4 100644
--- a/packages/session-title/session-title/README.i18n.yaml
+++ b/packages/session-title/session-title/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/session-title/session-title/README.md
-README.md: 1939d00f7e78834ec19e2d6b4590cf12af297a30
-README.zh.md: d373c212a193a82567686a634bad79726185e832
+README.md: 9a5ec27c36f3411add37ebe231262eb5d205bc9e
+README.zh.md: 38fc9f82e3fcd94233cad31e291b8258c97d0975
diff --git a/packages/session-title/session-title/README.md b/packages/session-title/session-title/README.md
index 1939d00f7e..9a5ec27c36 100644
--- a/packages/session-title/session-title/README.md
+++ b/packages/session-title/session-title/README.md
@@ -10,6 +10,7 @@ Only text blocks from human `user/message` events are eligible. The first eligib
 
 - `get(session)` folds the latest accepted title from a live or replayed log.
 - `refresh(session, signal?)` materializes the fallback when needed, then explicitly runs the registered provider over the current eligible messages. Provider errors and caller cancellation reject; cancellation does not roll back an already accepted fallback event.
+- `rename(session, title)` accepts an explicit user title synchronously: it normalizes the text, supersedes in-flight automatic work, and appends a `session/title` event with the `user` source. A user-sourced latest title pins the session — later user messages schedule no automatic revision; an explicit `refresh` remains the deliberate unpin.
 - `register(provider)` installs the sole optional provider and returns its awaitable Cordis effect disposer. A second registration throws immediately; disposal aborts pending and active calls, waits for their settlement, and only then permits another provider to register.
 
 Automatic work never delays the main agent response. A provider starts only after a marked loop-built request's exact route matches the current logged `request/header`, including when the unchanged header needs no new snapshot. Its late completion appends a standalone log-only event directly through `Session` without opening a turn. Persistence observes that event eagerly and drains on ordinary lifecycle checkpoints; title publication itself does not force a flush. Automatic failures warn and retain the latest title. New all-message revisions, provider disposal, session disposal, and explicit refresh abort older work, and a stale completion cannot append. Concurrent explicit refreshes reserve their revision before provider work, while overlapping automatic and explicit fallback requests share one session-local in-flight append. The service and bundled model provider each append their own literal event type, so no generic title-write marker, cast, or settlement queue is needed. Service teardown cancels queued work and drains calls that ignore cancellation before unloading completes.
@@ -50,5 +51,5 @@ None for the main request; title events do not change its reconstructed content
 
 ## Known Limitations and Deferred Work
 
-- Manual rename, title deletion, generated-versus-user precedence, search, and list indexing are outside this service.
+- Title deletion (unpinning back to automatic titles without an explicit `refresh`), search, and list indexing are outside this service.
 - The provider registry deliberately accepts at most one implementation, so a deployment cannot compose competing title strategies without writing one provider that owns their precedence.
diff --git a/packages/session-title/session-title/README.zh.md b/packages/session-title/session-title/README.zh.md
index d373c212a1..38fc9f82e3 100644
--- a/packages/session-title/session-title/README.zh.md
+++ b/packages/session-title/session-title/README.zh.md
@@ -2,19 +2,20 @@
 
 [English](README.md) | 中文
 
-由日志支持的会话标题，提供即时确定性回退与一个可选异步提供方。每个已接受 revision 都是仅写入日志的 `session/title` 事件；`foldSessionTitle()` 与 `ctx.sessionTitle.get()` 会选择最新事件，并返回其事件 seq 和时间戳。
+由日志支持的会话标题，提供即时确定性回退与一个可选异步提供方。每次已接受的修订都是仅写入日志的 `session/title` 事件；`foldSessionTitle()` 与 `ctx.sessionTitle.get()` 会选择最新事件，并返回其事件 seq 和时间戳。
 
-只有用户 `user/message` 事件中的文本块符合条件。第一条符合条件的提示词会安排回退，从其开头若干词生成标题，并受所配置 UTF-8 字节上限约束。系统会规范化空白、移除终端控制序列，且截断绝不会切断 code point。空提示词和非文本提示词会等待后续符合条件的输入。
+只有用户 `user/message` 事件中的文本块符合条件。第一条符合条件的提示词会安排回退，从其开头若干词生成标题，并受所配置 UTF-8 字节上限约束。系统会规范化空白、移除终端控制序列，且截断绝不会切断码点。空提示词和非文本提示词会等待后续符合条件的输入。
 
 ## 服务：`SessionTitleService`（ctx 键：`sessionTitle`）
 
 - `get(session)` 从活跃或回放日志折叠最新已接受标题。
-- `refresh(session, signal?)` 在需要时物化回退，然后显式运行已注册提供方，处理当前符合条件的消息。提供方错误与调用方取消会 reject；取消不会回滚已接受的回退事件。
-- `register(provider)` 安装唯一可选提供方，并返回可等待的 Cordis effect disposer。第二次注册会立即抛出；资源释放会中止待处理和活跃调用，等待其结算，之后才允许注册另一个提供方。
+- `refresh(session, signal?)` 在需要时物化回退，然后显式运行已注册提供方，处理当前符合条件的消息。提供方错误或调用方取消都会导致返回的 Promise 被拒绝；取消不会回滚已接受的回退事件。
+- `rename(session, title)` 同步接受用户显式标题：规范化文本、取代在途自动工作，并追加一条 `user` 来源的 `session/title` 事件。最新标题来源为 user 即钉住该会话——后续用户消息不再安排自动 revision；显式 `refresh` 仍是有意的解钉手段。
+- `register(provider)` 安装唯一可选提供方，并返回可等待的 Cordis effect disposer。第二次注册会立即抛出；对提供方执行 dispose（资源释放）会中止待处理和活跃调用，等待其结算，之后才允许注册另一个提供方。
 
-自动工作绝不会延迟主 agent（智能体）响应。只有当带标记、由循环构建的请求，其确切路由与当前已记录的 `request/header` 匹配时，提供方才会启动；即使 header 未变而无需新快照，也适用此规则。延迟完成会直接通过 `Session` 追加一个独立的纯日志事件，而不打开轮次。持久化会尽快观察该事件，并在常规生命周期检查点排空；标题发布本身不会强制 flush。自动失败会发出警告并保留最新标题。新的全消息 revision、提供方资源释放、会话资源释放和显式刷新都会中止旧工作，陈旧完成值无法追加。并发显式刷新会在提供方工作之前预留修订号；重叠的自动／显式回退请求共享一个会话本地进行中追加。服务与随附模型提供方各自追加自己的字面量事件类型，因此不需要通用标题写入标记、类型断言或结算队列。服务 teardown 会取消排队工作，并在卸载完成前排空忽略取消的调用。
+自动工作绝不会延迟主 agent（智能体）响应。只有当带标记、由循环构建的请求，其确切路由与当前已记录的 `request/header` 匹配时，提供方才会启动；即使请求头未变而无需新快照，也适用此规则。延迟完成会直接通过 `Session` 追加一个独立的纯日志事件，而不打开轮次。持久化会立即观察到该事件，并在常规生命周期检查点完成刷写；标题发布本身不会强制刷写。自动失败会发出警告并保留最新标题。新的全消息修订、提供方 dispose、会话 dispose 和显式刷新都会中止旧工作，陈旧的完成结果无法追加。并发显式刷新会在提供方工作之前预留修订号；重叠的自动／显式回退请求共享一个会话本地正在进行的追加操作。服务与内置模型提供方各自追加自己的字面事件类型，因此不需要通用标题写入标记、类型断言或结算队列。服务拆卸会取消排队工作，并在卸载完成前等待不响应取消的调用结算完成。
 
-Fork 会原样继承 seed 中的标题事件。首消息节奏不会自动为子会话重新生成标题；全消息节奏可以在子会话收到后续用户提示词后追加新 revision。
+Fork 出的会话会原样继承种子中的标题事件。首消息节奏不会自动为子会话重新生成标题；全消息节奏可以在子会话收到后续用户提示词后追加新修订。
 
 ## 配置
 
@@ -28,7 +29,7 @@ Fork 会原样继承 seed 中的标题事件。首消息节奏不会自动为子
 
 ## 提供方契约
 
-提供方会提供品牌化稳定 id、自动模式（`first-message` 或 `all-user-messages`）和 `generate(request)`。请求携带活跃会话、截至一个固定 revision 的所有符合条件消息、可用时当前已记录的主请求路由，以及取消信号。结果包含非空标题、该请求中唯一且有序的来源消息 seq，以及可选模型 provenance。服务会在结果持久化前进行规范化和验证。
+提供方会提供带品牌类型的稳定 id、自动模式（`first-message` 或 `all-user-messages`）和 `generate(request)`。请求携带活跃会话、截至一次固定修订的所有符合条件消息、可用时当前已记录的主请求路由，以及取消信号。结果包含非空标题、该请求中唯一且有序的来源消息 seq，以及可选的模型来源信息。服务会在结果持久保存前进行规范化和验证。
 
 参见[会话标题数据结构](../../../docs/core-data-structures/session-title.md)与[已实现决策](../../../.agents/notes/implemented/feature/2026-07-21-log-backed-session-titles.md)。
 
@@ -42,13 +43,13 @@ Fork 会原样继承 seed 中的标题事件。首消息节奏不会自动为子
 
 #### Token 影响
 
-回退与已接受的提供方 revision 不会向主 agent 请求增加 token。可选提供方的独立辅助请求由对应提供方包记录。
+回退与已接受的提供方修订不会向主 agent 请求增加 token。可选提供方的独立辅助请求由对应提供方包（package）的文档说明。
 
 #### KV Cache 影响
 
 不影响主请求；标题事件不会改变重建内容或缓存键。
 
-## 已知限制与暂缓工作
+## 已知限制与暂缓事项
 
-- 手动重命名、删除标题、生成标题与用户标题的优先级、搜索和列表索引都不属于此服务。
+- 删除标题（不经显式 `refresh` 就解钉回自动标题）、搜索和列表索引不属于此服务。
 - 提供方注册表有意最多接受一个实现，因此部署若要组合相互竞争的标题策略，必须编写一个自行负责优先级的提供方。
diff --git a/packages/session-title/session-title/src/index.ts b/packages/session-title/session-title/src/index.ts
index 995b216086..12da431c68 100644
--- a/packages/session-title/session-title/src/index.ts
+++ b/packages/session-title/session-title/src/index.ts
@@ -7,7 +7,7 @@ import { Context, FiberState, Service, type Fiber } from 'cordis'
 import z from 'schemastery'
 import { z as zod } from 'zod'
 import type { Branded } from '@deepseek-ai/dsh-brand'
-import { deepFreeze, isAgentLoopRequest } from '@deepseek-ai/dsh-llm'
+import { assertNever, deepFreeze, isAgentLoopRequest } from '@deepseek-ai/dsh-llm'
 import type { GenerateOptions } from '@deepseek-ai/dsh-llm'
 import type {
   Session,
@@ -52,14 +52,18 @@ export type SessionTitleSource =
     readonly provider: SessionTitleProviderId
     readonly model?: SessionTitleModelProvenance
   }
+  | {
+    /** Explicit user rename: pins the title — automatic generation stops scheduling. */
+    readonly kind: 'user'
+  }
 
 /** Payload of the log-only `session/title` event. */
 export interface SessionTitleEventData {
   /** Normalized non-empty title text. */
   readonly title: string
-  /** Exact human `user/message` seqs used to derive this title. */
+  /** Exact human `user/message` seqs used to derive this title; empty for an explicit user rename. */
   readonly messageSeqs: number[]
-  /** Built-in fallback or registered-provider provenance. */
+  /** Built-in fallback, registered-provider, or explicit-user provenance. */
   readonly source: SessionTitleSource
 }
 
@@ -97,6 +101,16 @@ declare module '@deepseek-ai/dsh-session' {
   }
 }
 
+/**
+ * Rejection of an explicit user title whose text normalizes to empty — the
+ * one {@link SessionTitleService.rename} failure that blames the input.
+ * Callers translating rename failures onto a wire (`title-invalid`) narrow on
+ * this class; liveness and disposal failures stay plain `Error`s.
+ */
+export class SessionTitleInvalidError extends Error {
+  override readonly name = 'SessionTitleInvalidError'
+}
+
 /** One eligible human text message exposed to title providers. */
 export interface SessionTitleUserMessage {
   /** Source `user/message` event seq. */
@@ -180,20 +194,27 @@ export function foldSessionTitle(events: readonly SessionEvent[]): SessionTitleS
   return deepFreeze({
     title: event.data.title,
     messageSeqs: [...event.data.messageSeqs],
-    source: event.data.source.kind === 'fallback'
-      ? { kind: 'fallback' }
-      : {
-        kind: 'provider',
-        provider: event.data.source.provider,
-        ...(event.data.source.model === undefined
-          ? {}
-          : { model: { ...event.data.source.model } }),
-      },
+    source: copySessionTitleSource(event.data.source),
     eventSeq: event.seq,
     updatedAt: event.time,
   })
 }
 
+/** Defensive copy of a logged title source (the snapshot must not alias log-owned objects). */
+function copySessionTitleSource(source: SessionTitleSource): SessionTitleSource {
+  switch (source.kind) {
+    case 'fallback': return { kind: 'fallback' }
+    case 'provider': return {
+      kind: 'provider',
+      provider: source.provider,
+      ...(source.model === undefined ? {} : { model: { ...source.model } }),
+    }
+    case 'user': return { kind: 'user' }
+    /* v8 ignore next -- closed-union exhaustiveness guard */
+    default: return assertNever(source, 'SessionTitleSource')
+  }
+}
+
 /** Service-owned resolved limits. */
 interface ResolvedConfig {
   readonly fallbackMaxWords: number
@@ -328,6 +349,39 @@ export class SessionTitleService extends Service {
     return foldSessionTitle(session.events)
   }
 
+  /**
+   * Accept an explicit user title. Appends a `session/title` event with the
+   * `user` source, which pins the title: in-flight automatic generation is
+   * superseded and later user messages schedule none (an explicit
+   * {@link SessionTitleService.refresh} remains the deliberate unpin).
+   * @param session - exact live session to rename.
+   * @param title - raw user input; normalized before acceptance.
+   * @returns the accepted title snapshot.
+   * @throws {SessionTitleInvalidError} when the title normalizes to empty.
+   * @throws {Error} when the session is not live or the service is disposed.
+   */
+  rename(session: Session, title: string): SessionTitleSnapshot {
+    this.assertServiceActive()
+    if (this.ctx.sessions.get(session.id) !== session) {
+      throw new Error(`session "${session.id}" is not live in this store`)
+    }
+    const normalized = normalizeSessionTitle(title, this.config.maxTitleBytes)
+    if (normalized.length === 0) {
+      throw new SessionTitleInvalidError('session title must contain visible characters')
+    }
+    const state = this.stateFor(session)
+    this.supersede(state, 'user rename superseded automatic title generation')
+    session.append('session/title', {
+      title: normalized,
+      messageSeqs: [],
+      source: { kind: 'user' },
+    })
+    const snapshot = this.get(session)
+    /* v8 ignore next -- unreachable: the append above just committed a session/title event. */
+    if (snapshot === undefined) throw new Error('renamed title failed to fold')
+    return snapshot
+  }
+
   /**
    * Explicitly retry the registered provider, or materialize the built-in
    * fallback when no provider is registered.
@@ -345,6 +399,16 @@ export class SessionTitleService extends Service {
     const messages = collectSessionTitleMessages(session.events)
     const latest = messages.at(-1)
     if (registration === undefined || registration.closing || latest === undefined) {
+      // Explicit refresh is the unpin even without a provider: a standing
+      // user title must not short-circuit ensureFallback into a no-op, so
+      // re-derive and append the fallback over it when one is derivable.
+      const current = this.get(session)
+      const [first] = messages
+      if (current?.source.kind === 'user' && first !== undefined) {
+        this.appendFallback(session, first)
+        signal?.throwIfAborted()
+        return this.get(session)
+      }
       const fallback = await this.ensureFallback(session)
       signal?.throwIfAborted()
       return fallback
@@ -398,6 +462,8 @@ export class SessionTitleService extends Service {
   private onUserMessage(session: Session, event: Extract<SessionEvent, { type: 'user/message' }>): void {
     if (!this.serviceActive()) return
     if (event.data.source.kind !== 'user' || collectSessionTitleMessages([event]).length === 0) return
+    // A user rename pins the title: no automatic revision may override it.
+    if (this.get(session)?.source.kind === 'user') return
     const registration = this.registration
     if (registration !== undefined && !registration.closing) {
       const messages = collectSessionTitleMessages(session.events, event.seq)
@@ -668,6 +734,23 @@ export class SessionTitleService extends Service {
     }
   }
 
+  /**
+   * Derive and append the deterministic fallback title over whatever stands
+   * (the refresh unpin path: overwriting a pinned user title is the point).
+   * Synchronous on purpose — no await may separate derivation from append, so
+   * it needs neither ensureFallback's in-flight dedup nor its liveness
+   * re-check. An underivable fallback (empty after the caps) appends nothing.
+   */
+  private appendFallback(session: Session, first: SessionTitleUserMessage): void {
+    const title = fallbackSessionTitle(first.text, this.config.fallbackMaxWords, this.config.fallbackMaxBytes)
+    if (title.length === 0) return
+    session.append('session/title', {
+      title,
+      messageSeqs: [first.seq],
+      source: { kind: 'fallback' },
+    })
+  }
+
   /** Create the first deterministic fallback if the session still lacks a title. */
   private async ensureFallback(session: Session): Promise<SessionTitleSnapshot | undefined> {
     this.assertServiceActive()
diff --git a/packages/session-title/session-title/src/invariant.ts b/packages/session-title/session-title/src/invariant.ts
index 9bae01a72f..25b00a10aa 100644
--- a/packages/session-title/session-title/src/invariant.ts
+++ b/packages/session-title/session-title/src/invariant.ts
@@ -5,7 +5,8 @@
 
 /* jscpd:ignore-start */
 import type { Context } from 'cordis'
-import type { InvariantInstaller } from '@deepseek-ai/dsh-invariants'
+import type { InvariantFailure, InvariantInstaller } from '@deepseek-ai/dsh-invariants'
+import type { SessionEvent } from '@deepseek-ai/dsh-session'
 
 const PACKAGE_NAME = '@deepseek-ai/dsh-session-title'
 
@@ -15,11 +16,26 @@ export const name = 'session-title-invariant'
 export const inject = ['invariants']
 
 /**
- * No runtime invariant: the service validates provider revisions before their
- * title append, and its remaining lifecycle state is process-local and covered
- * by package tests.
+ * Durable title-provenance invariant: an automatic title always cites at
+ * least one human `user/message` seq, and an explicit user rename cites none
+ * — `messageSeqs` is empty iff `source.kind` is `user`. Provider revisions
+ * are validated by the service before their append; this checks the durable
+ * relationship every appended `session/title` event must keep, whichever
+ * writer produced it.
  */
-const install: InvariantInstaller = () => {}
+const install: InvariantInstaller = Object.assign((ctx: Context, fail: InvariantFailure) => {
+  // internal/dispatch interception rejects the append before publication
+  // (the session/event listener would only observe the already-committed log).
+  ctx.on('internal/dispatch', (_mode, eventName, args) => {
+    if (eventName !== 'session/event') return
+    const [, event] = args as [unknown, SessionEvent]
+    if (event.type !== 'session/title') return
+    const { source, messageSeqs } = event.data
+    if ((messageSeqs.length === 0) !== (source.kind === 'user')) {
+      fail(`session/title event ${String(event.seq)} breaks provenance: source "${source.kind}" with ${String(messageSeqs.length)} cited message seq(s)`)
+    }
+  }, { global: true })
+}, { inject: ['sessions'] })
 
 /**
  * Register this package's invariant companion.
diff --git a/packages/session-title/session-title/tests/invariant.spec.ts b/packages/session-title/session-title/tests/invariant.spec.ts
new file mode 100644
index 0000000000..5b316448da
--- /dev/null
+++ b/packages/session-title/session-title/tests/invariant.spec.ts
@@ -0,0 +1,44 @@
+// Title-provenance invariant: messageSeqs is empty iff source.kind is 'user'
+// — the durable relationship every appended session/title event must keep.
+import { describe, expect, it } from 'vitest'
+import { Context } from 'cordis'
+import * as SessionTitleInvariantCompanion from '@deepseek-ai/dsh-session-title/invariant'
+import InvariantService, { InvariantError } from '@deepseek-ai/dsh-invariants'
+import SessionStore, { SessionId } from '@deepseek-ai/dsh-session'
+
+async function setup(): Promise<Context> {
+  const ctx = new Context()
+  await ctx.plugin(SessionStore)
+  await ctx.plugin(InvariantService, { enabled: true })
+  await ctx.plugin(SessionTitleInvariantCompanion)
+  return ctx
+}
+
+describe('session-title provenance invariant', () => {
+  it('accepts cited automatic titles and citation-free user renames', async () => {
+    const ctx = await setup()
+    const session = ctx.sessions.create(SessionId('title-invariant-valid'))
+    expect(() => {
+      session.append('session/title', { title: 'auto', messageSeqs: [1], source: { kind: 'fallback' } })
+      session.append('session/title', { title: 'named', messageSeqs: [], source: { kind: 'user' } })
+    }).not.toThrow()
+  })
+
+  it('rejects a citation-free automatic title and a user rename that cites messages', async () => {
+    const ctx = await setup()
+    const session = ctx.sessions.create(SessionId('title-invariant-invalid'))
+    expect(() => {
+      session.append('session/title', { title: 'auto', messageSeqs: [], source: { kind: 'fallback' } })
+    }).toThrow(expect.objectContaining<Partial<InvariantError>>({
+      code: 'INVARIANT',
+      packageName: '@deepseek-ai/dsh-session-title',
+    }))
+    expect(() => {
+      session.append('session/title', { title: 'named', messageSeqs: [1], source: { kind: 'user' } })
+    }).toThrow(expect.objectContaining<Partial<InvariantError>>({
+      code: 'INVARIANT',
+      packageName: '@deepseek-ai/dsh-session-title',
+    }))
+    expect(session.seq).toBe(0)
+  })
+})
diff --git a/packages/session-title/session-title/tests/rename.spec.ts b/packages/session-title/session-title/tests/rename.spec.ts
new file mode 100644
index 0000000000..01d613ee3e
--- /dev/null
+++ b/packages/session-title/session-title/tests/rename.spec.ts
@@ -0,0 +1,181 @@
+// SessionTitleService.rename: user-source acceptance, normalization/rejection
+// boundaries, and the pin (a user-sourced latest title schedules no automatic
+// revision; explicit refresh stays the unpin).
+import { Context } from 'cordis'
+import { describe, expect, it, vi } from 'vitest'
+import { createUserMessage } from '@deepseek-ai/dsh-llm'
+import SessionStore, { Session, SessionId } from '@deepseek-ai/dsh-session'
+import SessionTitleService, {
+  SessionTitleProviderId,
+  foldSessionTitle,
+  type SessionTitleProviderRequest,
+} from '@deepseek-ai/dsh-session-title'
+
+const CONFIG = {
+  fallbackMaxWords: 5,
+  fallbackMaxBytes: 40,
+  maxTitleBytes: 40,
+} as const
+
+async function settle(): Promise<void> {
+  await new Promise(resolve => setTimeout(resolve, 0))
+}
+
+function appendHumanPrompt(session: ReturnType<Context['sessions']['create']>, text: string) {
+  return session.append('user/message', createUserMessage({
+    content: [{ type: 'text', text }],
+    source: { kind: 'user' },
+  }), { surfaceOp: 'append' })
+}
+
+describe('SessionTitleService.rename', () => {
+  it('appends a normalized user-source title', async () => {
+    const ctx = new Context()
+    await ctx.plugin(SessionStore)
+    await ctx.plugin(SessionTitleService, CONFIG)
+    const session = ctx.sessions.create(SessionId('rename-accept'))
+    session.append('turn/start', { turn: 1, trigger: { kind: 'message', source: { kind: 'user' } } })
+    appendHumanPrompt(session, 'Original prompt text')
+    await settle()
+
+    const accepted = ctx.sessionTitle.rename(session, '  Hand\tpicked   name  ')
+    expect(accepted).toMatchObject({
+      title: 'Hand picked name',
+      messageSeqs: [],
+      source: { kind: 'user' },
+    })
+    const event = session.events.findLast(item => item.type === 'session/title')
+    expect(event?.data).toEqual({
+      title: 'Hand picked name',
+      messageSeqs: [],
+      source: { kind: 'user' },
+    })
+    // foldSessionTitle round-trips the third source kind.
+    expect(foldSessionTitle(session.events)?.source).toEqual({ kind: 'user' })
+  })
+
+  it('rejects titles that normalize to empty and dead sessions', async () => {
+    const ctx = new Context()
+    await ctx.plugin(SessionStore)
+    await ctx.plugin(SessionTitleService, CONFIG)
+    const session = ctx.sessions.create(SessionId('rename-reject'))
+    expect(() => ctx.sessionTitle.rename(session, '    ')).toThrow(/visible characters/)
+
+    expect(() => ctx.sessionTitle.rename(new Session(SessionId('detached')), 'name'))
+      .toThrow(/not live in this store/)
+  })
+
+  it('pins the title: later user messages schedule no automatic revision; refresh unpins', async () => {
+    const ctx = new Context()
+    await ctx.plugin(SessionStore)
+    await ctx.plugin(SessionTitleService, CONFIG)
+    const generate = vi.fn(async (request: SessionTitleProviderRequest) => ({
+      title: 'Provider title',
+      messageSeqs: request.messages.map(message => message.seq),
+    }))
+    ctx.sessionTitle.register({
+      id: SessionTitleProviderId('pin-provider'),
+      automatic: 'all-user-messages',
+      generate,
+    })
+    const session = ctx.sessions.create(SessionId('rename-pin'))
+    session.append('turn/start', { turn: 1, trigger: { kind: 'message', source: { kind: 'user' } } })
+    appendHumanPrompt(session, 'First prompt')
+    await settle()
+    ctx.sessionTitle.rename(session, 'Pinned by hand')
+
+    // A later eligible prompt must schedule nothing while the pin stands.
+    appendHumanPrompt(session, 'Second prompt after the pin')
+    await settle()
+    session.append('request/header', {
+      header: { config: { provider: 'main-route', model: 'chat-model' } },
+      reason: 'change',
+    })
+    await settle()
+    expect(generate).not.toHaveBeenCalled()
+    expect(ctx.sessionTitle.get(session)?.title).toBe('Pinned by hand')
+
+    // Explicit refresh remains the deliberate unpin.
+    const refreshed = await ctx.sessionTitle.refresh(session)
+    expect(generate).toHaveBeenCalledOnce()
+    expect(refreshed?.title).toBe('Provider title')
+    expect(ctx.sessionTitle.get(session)?.source.kind).toBe('provider')
+  })
+
+  it('fallback-only refresh also unpins: the user title yields to a re-derived fallback', async () => {
+    const ctx = new Context()
+    await ctx.plugin(SessionStore)
+    await ctx.plugin(SessionTitleService, CONFIG)
+    const session = ctx.sessions.create(SessionId('rename-unpin-fallback'))
+    session.append('turn/start', { turn: 1, trigger: { kind: 'message', source: { kind: 'user' } } })
+    appendHumanPrompt(session, 'Derivable prompt words')
+    await settle()
+    ctx.sessionTitle.rename(session, 'Pinned without provider')
+    expect(ctx.sessionTitle.get(session)?.source.kind).toBe('user')
+
+    const refreshed = await ctx.sessionTitle.refresh(session)
+    expect(refreshed).toMatchObject({
+      title: 'Derivable prompt words',
+      source: { kind: 'fallback' },
+    })
+    // The pin is gone: the latest title is fallback-sourced, so the
+    // onUserMessage pin check no longer skips scheduling.
+    expect(ctx.sessionTitle.get(session)?.source.kind).toBe('fallback')
+  })
+
+  it('supersedes in-flight automatic generation: a late provider result cannot override the user title', async () => {
+    const ctx = new Context()
+    await ctx.plugin(SessionStore)
+    await ctx.plugin(SessionTitleService, CONFIG)
+    // The provider parks on a test-held deferred so rename lands while its
+    // generation is ACTIVE (not merely scheduled).
+    let releaseProvider: (() => void) | undefined
+    const gate = new Promise<void>((resolve) => { releaseProvider = resolve })
+    let aborted = false
+    const generate = vi.fn(async (request: SessionTitleProviderRequest) => {
+      request.signal.addEventListener('abort', () => { aborted = true })
+      await gate
+      return { title: 'Late provider title', messageSeqs: request.messages.map(message => message.seq) }
+    })
+    ctx.sessionTitle.register({
+      id: SessionTitleProviderId('deferred-provider'),
+      automatic: 'all-user-messages',
+      generate,
+    })
+    const session = ctx.sessions.create(SessionId('rename-supersede'))
+    session.append('turn/start', { turn: 1, trigger: { kind: 'message', source: { kind: 'user' } } })
+    appendHumanPrompt(session, 'Prompt that triggers generation')
+    session.append('request/header', {
+      header: { config: { provider: 'main-route', model: 'chat-model' } },
+      reason: 'change',
+    })
+    await settle()
+    expect(generate).toHaveBeenCalledOnce()
+
+    ctx.sessionTitle.rename(session, 'User wins')
+    expect(aborted).toBe(true)
+    releaseProvider?.()
+    await settle()
+    // The released provider result must not append over the user title, and
+    // the swallowed abort must not surface as an unhandled rejection.
+    const latest = session.events.findLast(item => item.type === 'session/title')
+    expect(latest?.data).toMatchObject({ title: 'User wins', source: { kind: 'user' } })
+  })
+
+  it('fallback-only refresh keeps the user title when no fallback is derivable', async () => {
+    const ctx = new Context()
+    await ctx.plugin(SessionStore)
+    // A 3-byte fallback cap cannot hold the 4-byte emoji prompt: the
+    // re-derived fallback is empty, so the pinned title survives the refresh.
+    await ctx.plugin(SessionTitleService, { ...CONFIG, fallbackMaxBytes: 3 })
+    const session = ctx.sessions.create(SessionId('rename-unpin-empty'))
+    session.append('turn/start', { turn: 1, trigger: { kind: 'message', source: { kind: 'user' } } })
+    appendHumanPrompt(session, '😀😀')
+    await settle()
+    ctx.sessionTitle.rename(session, 'Sticky emoji pin')
+
+    const refreshed = await ctx.sessionTitle.refresh(session)
+    expect(refreshed?.title).toBe('Sticky emoji pin')
+    expect(ctx.sessionTitle.get(session)?.source.kind).toBe('user')
+  })
+})
diff --git a/packages/skill/README.i18n.yaml b/packages/skill/README.i18n.yaml
index 5843c4fa86..ead6d96d42 100644
--- a/packages/skill/README.i18n.yaml
+++ b/packages/skill/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/skill/README.md
 README.md: 5c75661de17826e7ea4763e90b494e9e7a0a7c0f
-README.zh.md: d219f710c0185298af89ba2e074d9e3b2e093896
+README.zh.md: 0db8933ecff2a8ab4767b29bc5319453ddfb87fa
diff --git a/packages/skill/README.zh.md b/packages/skill/README.zh.md
index d219f710c0..0db8933ecf 100644
--- a/packages/skill/README.zh.md
+++ b/packages/skill/README.zh.md
@@ -1,8 +1,8 @@
-# skill/ - skill 功能家族
+# skill/ - skill（技能）能力家族
 
 [English](README.md) | 中文
 
-可复用 agent 指令的规范三包功能 seam：提供方注册表、本地实现，以及面向模型的目录/加载器消费方。全部都是**产品** 包。
+可复用 agent（智能体）指令的规范能力 seam 由三个包（package）组成：提供方注册表、本地实现，以及面向模型的目录／loader 消费方。全部均为**产品**包。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
@@ -10,4 +10,4 @@
 | `skill-local/` | 项目/自定义/用户文件系统提供方 | （注册到 `ctx.skills`） |
 | `tool-skill/` | 会话前缀目录和面向模型的 `skill` 加载器 | （注册到 `ctx.tools`） |
 
-接口位于 `skill/skill/`。提供方同步注册，并通过 `ctx.skills` 执行异步发现；`tool-skill` 只消费该接口，因此嵌入式或远程提供方可替换或补充 `skill-local`，无需改变面向模型的契约。`agent-core` 默认加载该家族，但它仍然是核心控制主干之外的功能，与 [`bash/`](../bash/README.md)、[`fs/`](../fs/README.md)、[`web/`](../web/README.md) 和 [`subagent/`](../subagent/README.md) 并列。
+接口位于 `skill/skill/`。提供方同步注册，并通过 `ctx.skills` 执行异步发现；`tool-skill` 只消费该接口，因此嵌入式或远程提供方可替换或补充 `skill-local`，无需改变面向模型的契约。`agent-core` 默认加载该家族，但它仍然是核心控制主干之外的能力，与 [`bash/`](../bash/README.md)、[`fs/`](../fs/README.md)、[`web/`](../web/README.md) 和 [`subagent/`](../subagent/README.md) 并列。
diff --git a/packages/skill/skill-local/README.i18n.yaml b/packages/skill/skill-local/README.i18n.yaml
index 603c1e0c90..ac773ab9f0 100644
--- a/packages/skill/skill-local/README.i18n.yaml
+++ b/packages/skill/skill-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/skill/skill-local/README.md
 README.md: c488fdc4b1d97b5aa1113e41a470484063526ded
-README.zh.md: 796c814a3c4064d545a966465caf6f99e9dd8601
+README.zh.md: c6e32a3ddbc2279428e23051e5636cf12c558ba9
diff --git a/packages/skill/skill-local/README.zh.md b/packages/skill/skill-local/README.zh.md
index 796c814a3c..c6e32a3ddb 100644
--- a/packages/skill/skill-local/README.zh.md
+++ b/packages/skill/skill-local/README.zh.md
@@ -4,19 +4,19 @@
 
 `ctx.skills` 注册表的本地文件系统提供方。
 
-该包实现一个 skill 来源。它扫描本地项目、自定义和用户 skill 根，解析 `SKILL.md` 或平铺 Markdown skill 文件，并将提供方注册到 `ctx.skills`。注册表仍位于 `@deepseek-ai/dsh-skill`；会话前缀目录和面向模型的加载器工具仍位于 `@deepseek-ai/dsh-tool-skill`。
+该包（package）实现一个 skill（技能）来源。它扫描本地项目、自定义和用户 skill 根目录，解析 `SKILL.md` 或平铺 Markdown skill 文件，并将提供方注册到 `ctx.skills`。注册表仍位于 `@deepseek-ai/dsh-skill`；会话前缀目录和面向模型的 loader 工具仍位于 `@deepseek-ai/dsh-tool-skill`。
 
 ## 插件
 
-需要 `ctx.skills` （`inject: ['skills']`）。
+需要 `ctx.skills`（`inject: ['skills']`）。
 
 ### 配置
 
 | 字段 | 默认值 | 含义 |
 |---|---|---|
-| `dshHome` | `$DSH_HOME` or `~/.dsh` | 由 [`@deepseek-ai/dsh-paths`](../../util/paths/README.md) 解析的 DeepSeek Harness 配置根；扫描该目录下的 `skills`。 |
-| `agentsHome` | `$DSH_AGENTS_HOME` or `~/.agents` | 为兼容 skill 扫描的共享 agent 配置根。 |
-| `customSkillDirs` | `[]` | 在项目根之后、用户根之前扫描的其他本地 skill 根。 |
+| `dshHome` | `$DSH_HOME` 或 `~/.dsh` | 由 [`@deepseek-ai/dsh-paths`](../../util/paths/README.md) 解析的 DeepSeek Harness 配置根目录；扫描该目录下的 `skills`。 |
+| `agentsHome` | `$DSH_AGENTS_HOME` 或 `~/.agents` | 为兼容 skill 扫描的共享 agent（智能体）配置根目录。 |
+| `customSkillDirs` | `[]` | 在项目根目录之后、用户根目录之前扫描的其他本地 skill 根目录。 |
 
 ## 发现
 
@@ -30,25 +30,25 @@
 | 400 | `user-dsh` | `<dshHome>/skills` |
 | 500 | `user-agents` | `<agentsHome>/skills` |
 
-项目根是包含 `.git` 的最近祖先；如果不存在，则使用当前 cwd。用户 DSH 根会跳过其 `.system` 子级，因此系统所有目录不会被当作普通用户 skill。该提供方提供项目和用户 skill；其他提供方可提供内置系统 skill。
+项目根目录是包含 `.git` 的最近祖先目录；如果不存在，则使用当前 cwd。用户 DSH 根目录会跳过其 `.system` 子目录，因此归系统所有的目录不会被当作普通用户 skill。该提供方提供项目和用户 skill；其他提供方可提供内置系统 skill。
 
 当 `ctx.fs` 可用时，发现通过 `ctx.fs.listDir` 列出根，通过 `ctx.fs.readText` 读取 skill 文件，并通过文件系统服务探测 `.git`。完整 skill 加载会将查找中止信号转发给文件系统元数据和内容读取。如果没有文件系统服务，提供方回退到可中止的 Node 文件系统 I/O，使最小本地上下文仍能加载 skill。缺失、不可读或格式错误的 skill 文件会警告并跳过，而不会使整个请求失败。
 
 ## Skill 格式
 
-Skill 可以是单层目录 bundle（`<name>/SKILL.md`），也可以是平铺 Markdown 文件（`<name>.md`）。v1 刻意不包含嵌套 `**/SKILL.md` 发现。Frontmatter 使用 `yaml` 包解析为 YAML；它要求 `name` 和 `description`，而 `whenToUse`、`disableModelInvocation` 和 `metadata` 可选。名称必须使用 kebab-case。
+Skill 可以是单层目录 bundle（`<name>/SKILL.md`），也可以是平铺 Markdown 文件（`<name>.md`）。v1 刻意不支持发现嵌套的 `**/SKILL.md`。Frontmatter 使用 `yaml` 包解析为 YAML；它要求 `name` 和 `description`，而 `whenToUse`、`disableModelInvocation` 和 `metadata` 可选。名称必须使用 kebab-case。
 
 ## 模型体验
 
-通过 `dsh-tool-skill` 间接影响模型。它将该提供方的可调用名称和有上限描述渲染到会话前缀目录中，并将所选指令正文与资源基底指引渲染到已保留工具历史中；路径、提供方 rank 和已禁用 skill 仍被隐藏。
+通过 `dsh-tool-skill` 间接影响模型。它将该提供方的可调用名称和有长度上限的描述渲染到会话前缀目录中，并将所选指令正文与资源基底指引渲染到保留的工具历史中；路径、提供方 rank 和已禁用 skill 仍被隐藏。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-不直接导致失效；指定的消费方负责其引起的任何请求前缀变更。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **发现深度为一层**：只识别 `<root>/<name>/SKILL.md` 和 `<root>/<name>.md`；忽略嵌套 skill 树和包 manifest。
+- **发现深度为一层**：只识别 `<root>/<name>/SKILL.md` 和 `<root>/<name>.md`；忽略嵌套 skill 树和包 manifest（元数据清单）。
 - **项目范围为最近 `.git` 祖先**：没有该标记的工作区回退到提供的 cwd，不支持其他项目根标记或 monorepo 子项目选择。
 - **不可读或格式错误的条目会随警告消失**：模型目录不会收到每个 skill 的诊断，无法区分缺失的 skill 与被跳过的 skill。
-- **无文件系统 watcher**：先前已收集 cwd 重新发现之前，编辑操作依赖注册表缓存被驱逐，或因提供方重新加载而失效。
+- **无文件系统监听**：在重新发现先前已收集的 cwd 之前，编辑内容能否生效取决于注册表缓存是否被淘汰，或是否因提供方重新加载而失效。
diff --git a/packages/skill/skill/README.i18n.yaml b/packages/skill/skill/README.i18n.yaml
index d9e7e42df7..2872a79b2d 100644
--- a/packages/skill/skill/README.i18n.yaml
+++ b/packages/skill/skill/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/skill/skill/README.md
 README.md: 639616d0b75f960e9ccd48546d44db841372bbe2
-README.zh.md: 3afdd415397927ebf107d6f862422c711a51888b
+README.zh.md: d2191d54502121c68976ead2541a0588a0530160
diff --git a/packages/skill/skill/README.zh.md b/packages/skill/skill/README.zh.md
index 3afdd41539..d2191d5450 100644
--- a/packages/skill/skill/README.zh.md
+++ b/packages/skill/skill/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-纯 agent skill 提供方注册表。
+纯 agent skill（智能体技能）提供方注册表。
 
-该包负责 `ctx.skills` 接口。它不知道 skill 来自本地文件、嵌入式插件数据、HTTP 还是其他后端；提供方通过 `ctx.skills.registerProvider(...)` 注册这些来源。已发布的本地实现是 [`@deepseek-ai/dsh-skill-local`](../skill-local)。
+该包（package）负责 `ctx.skills` 接口。它不知道 skill 来自本地文件、嵌入式插件数据、HTTP 还是其他后端；提供方通过 `ctx.skills.registerProvider(...)` 注册这些来源。已发布的本地实现是 [`@deepseek-ai/dsh-skill-local`](../skill-local)。
 
 ## 服务：`SkillService`（ctx 键：`skills`）
 
 ### 公开 API
 
-- `ctx.skills.registerProvider(provider): () => void` 使用唯一 `provider.name` 注册只读提供方。重复提供方名称会抛错，`runtime` 保留给 `ctx.skills.register(...)`。注册表借用提供方对象，并直接调用其方法。注册作用域绑定到 effect，可安全用于 HMR；精确的 Cordis disposer 支持有序组合拆卸。
+- `ctx.skills.registerProvider(provider): () => void` 使用唯一 `provider.name` 注册只读提供方。重复提供方名称会抛错，`runtime` 保留给 `ctx.skills.register(...)`。注册表借用提供方对象，并直接调用其方法。注册作用域绑定到 effect，可安全用于 HMR（热模块替换）；Cordis 返回的原始 disposer 支持有序组合拆卸。
 - `ctx.skills.list({ cwd?, signal? })` 借用只读查找选项，然后返回当前工作区中模型可调用的摘要；这些摘要跨提供方合并，并按名称排序。
 - `ctx.skills.get(name, { cwd?, signal? })` 在发现和加载中使用同一组只读选项和胜出候选项；在发现或缓存命中后重新检查取消，让提供方加载与信号竞速，验证已加载定义，然后将其返回，包括已对模型禁用的 skill。
 - `ctx.skills.register(skill): () => void` 注册只读运行时嵌入式 skill，省略时添加 `provider: "runtime"`。同名运行时注册使用先到先得：重复项会记录警告，并获得无操作 disposer。成功注册会返回精确的 Cordis disposer，以供有序组合拆卸。
@@ -23,15 +23,15 @@
 
 ## 提供方契约
 
-提供方同步注册，并在已等待的 `list(options)` 调用中执行远程设置、身份验证和发现。提供方对象、查找选项、候选项和定义都以只读方式借用，而不是克隆或重新绑定。提供方应遵守 `options.signal`；取消后，注册表也会停止等待不协作的发现或加载。
+提供方同步注册，并在可等待的 `list(options)` 调用中执行远程设置、身份验证和发现。提供方对象、查找选项、候选项和定义都以只读方式借用，而不是克隆或重新绑定。提供方应遵守 `options.signal`；取消后，注册表也会停止等待不协作的发现或加载。
 
 注册表在缓存前验证候选项，在返回前验证定义。胜出提供方会收到同一候选项和不透明 `locator`，两者都是它从 `list()` 返回的内容，从而支持后端专用文件、URL、id 或版本句柄。调用方和提供方必须保持只读契约。
 
-契约违反会快速失败。被拒绝的 `list()` 视为瞬时来源失败：系统记录它、跳过它，并且不缓存。只缓存已完成目录；提供方或运行时修订变更会丢弃正在进行的结果并重试。重复名称按 rank、提供方注册顺序，然后按提供方本地顺序解析。摘要按 skill 名称排序。
+违反契约时会快速失败。`list()` 返回的 Promise 被拒绝会被视为瞬时来源失败：系统记录并跳过该失败，且不缓存结果。只缓存已完成的目录；提供方或运行时修订发生变化时，会丢弃正在进行的结果并重试。重复名称依次按 rank、提供方注册顺序和提供方本地顺序解决冲突。摘要按 skill 名称排序。
 
-## 运行时 Skill
+## 运行时 skill
 
-`ctx.skills.register(...)` 是嵌入式运行时 skill 的便利接口。运行时 skill 使用 rank `250`：项目提供方可覆盖它们，它们则覆盖已发布本地提供方的自定义根和用户根。运行时定义和嵌套资源元数据均以只读方式借用；服务只实体化提供默认 `provider` 所需的顶层定义。运行时贡献内的注册使用先到先得，因此重复贡献无法通过其 disposer 移除活动项。
+`ctx.skills.register(...)` 是嵌入式运行时 skill 的便利接口。运行时 skill 使用 rank `250`：项目提供方可覆盖它们，它们则覆盖已发布本地提供方的自定义根目录和用户根目录。运行时定义和嵌套资源元数据均以只读方式借用；服务只物化提供默认 `provider` 所需的顶层定义。运行时贡献内的注册使用先到先得，因此重复贡献无法通过其 disposer 移除当前生效的贡献。
 
 ## 消费方边界
 
@@ -41,13 +41,13 @@
 
 通过 `dsh-tool-skill` 间接影响模型；该包将提供方摘要渲染到会话前缀中，并将已加载指令渲染到已保留工具结果中。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-不直接导致失效；指定的消费方负责其引起的任何请求前缀变更。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **已完成目录没有 TTL 或 watcher 失效机制**：提供方的底层文件或远程数据可在注册修订不变的情况下更改，因此已缓存 cwd 会保持陈旧，直到被驱逐或重新加载提供方/运行时。
-- **提供方依次查询**：一个缓慢的协作提供方会延迟之后注册的所有提供方；取消会停止调用方等待，但无法终止不协作提供方持续运行的工作。
-- **提供方列表失败会移除该请求的整个来源**：注册表会记录并跳过它，不提供模型可见诊断或部分目录恢复契约。
+- **已完成的目录没有 TTL 或监听失效机制**：提供方的底层文件或远程数据可在注册修订不变的情况下更改，因此已缓存的 cwd 会保持陈旧，直到缓存条目被淘汰或提供方／运行时重新加载。
+- **提供方依次查询**：一个响应取消但速度缓慢的提供方会延迟之后注册的所有提供方；取消会停止调用方等待，但无法终止不响应取消的提供方持续运行的工作。
+- **提供方列表失败会使该请求无法使用整个来源**：注册表会记录并跳过该来源，不提供模型可见诊断或部分目录恢复契约。
 - **重复解析使用先到先得**：系统会记录并隐藏较晚出现的低优先级候选项；不提供检查全部被遮蔽定义的 API。
diff --git a/packages/skill/tool-skill/README.i18n.yaml b/packages/skill/tool-skill/README.i18n.yaml
index 7e48e8619d..6cb393022f 100644
--- a/packages/skill/tool-skill/README.i18n.yaml
+++ b/packages/skill/tool-skill/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/skill/tool-skill/README.md
 README.md: 6b0af04a15bfda985be3e04868f18b9af702eae7
-README.zh.md: 4ca0ccd7734848d5774e92910e916bdf71c88c13
+README.zh.md: b2f073491a7c6d3c1cb0d6020d619ae725fed305
diff --git a/packages/skill/tool-skill/README.zh.md b/packages/skill/tool-skill/README.zh.md
index 4ca0ccd773..b2f073491a 100644
--- a/packages/skill/tool-skill/README.zh.md
+++ b/packages/skill/tool-skill/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-面向模型的 skill 目录和 `skill` 工具。
+面向模型的 skill（技能）目录和 `skill` 工具。
 
-需要 `ctx.tools` 和 `ctx.skills` （`inject: ['tools', 'skills']`）。
+需要 `ctx.tools` 和 `ctx.skills`（`inject: ['tools', 'skills']`）。
 
 ## 会话目录
 
-该插件在实时会话的第一个 `agent/step` 注入一条持久的用户角色 `<system-reminder>` 目录。它为调用会话的 cwd 解析 skill，将步骤中止信号转发到发现，并只列出已排序的 `name` 和 `description` 条目；skill 正文、路径、来源、提供方和 `whenToUse` 提示仍位于目录之外。如果没有模型可调用 skill，则省略目录；如果该 agent 的工具视图排除已发布的 `skill` 工具，或解析出一个同名作用域遮蔽，也会省略目录。这项精确定义检查使提示词指引、模型可见 schema 和可执行分派保持对齐。
+该插件在活动会话的第一个 `agent/step` 注入一条持久的用户角色 `<system-reminder>` 目录。它为调用会话的 cwd 解析 skill，将步骤中止信号转发到发现流程，并只列出已排序的 `name` 和 `description` 条目；skill 正文、路径、来源、提供方和 `whenToUse` 提示仍位于目录之外。如果没有模型可调用 skill，则省略目录；如果该 agent（智能体）的工具视图排除了随附的 `skill` 工具，或解析出同名的作用域内遮蔽项，也会省略目录。这项对工具定义的精确匹配检查使提示词指引、模型可见 schema 和可执行分派保持对齐。
 
 `catalogDescriptionMaxLength` 控制规范化且经 XML 转义的目录描述。其默认值是 `500`，且必须是不小于 `3` 的整数，以便为截断省略号保留空间。目录是一条带来源的 `user/message`，在第一个请求前注入，并保留在普通会话历史中。
 
@@ -18,9 +18,9 @@
 |---|---|---|
 | `name` | string（必填） | 可用 skill 列表中精确的 kebab-case skill 名称。 |
 
-执行使用调用 agent 的 `session.header.cwd`，使工作区敏感提供方解析胜出 skill。成功调用返回规范 `{ name, provider, resourceBase?, content }`，排除目录 rank 和提供方内部机制；其 Native 渲染器产生一个文本结果，其中包含 `<skill_content name="...">`、`<skill_resources>` 和 `<skill_instructions>`。
+执行使用调用 agent 的 `session.header.cwd`，使结果随工作区变化的提供方能够解析出胜出的 skill。成功调用返回规范形式的 `{ name, provider, resourceBase?, content }`，其中不包含目录排名和提供方内部机制；其 Native 渲染器会生成一个文本结果，其中包含 `<skill_content name="...">`、`<skill_resources>` 和 `<skill_instructions>`。
 
-资源指引只会根据 `resourceBase` 解析指令显式引用的路径或 URL；脚本、参考资料和产物按需加载，结果不会列举 skill 目录。本地提供方可以提供目录，而远程或嵌入式提供方可以提供 URL 或不透明加载指引。
+资源指引只会根据 `resourceBase` 解析指令显式引用的路径或 URL；脚本、参考资料和资源文件按需加载，结果不会列举 skill 目录。本地提供方可以提供目录，而远程或嵌入式提供方可以提供 URL 或不透明加载指引。
 
 无法解析的名称会报告 skill 未知或已不可用。无效名称和 `disableModelInvocation: true` skill 产生不同的错误结果。
 
@@ -30,9 +30,9 @@
 
 ### 会话目录
 
-#### 模型所见
+#### 模型看到的内容
 
-如果存在模型可调用 skill，且该精确 `skill` 工具可见，agent 会收到下方目录模板，其中包含每个已排序 skill 的一条数据依赖条目。该目录是一条持久的用户角色消息。
+如果存在模型可调用 skill，且可见的正是这个 `skill` 工具，agent 会收到下方目录模板，其中包含每个已排序 skill 的一条随数据而定的条目。该目录是一条持久的用户角色消息。
 
 ##### Skill 目录模板
 
@@ -52,29 +52,29 @@ If the user names a skill, or the task clearly matches a skill's description, ca
 
 重复输入成本随 skill 数量和 `catalogDescriptionMaxLength` 增长；当列表为空或工具被隐藏或遮蔽时，不会发送目录 token。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 仅追加，位于现有可重用前缀之后。如果新建或恢复的实例具有不同提供方、skill、描述、可见性或目录上限，则可能从新追加的目录位置起影响缓存重用。
 
 ### 工具 schema
 
-#### 模型所见
+#### 模型看到的内容
 
 模型会看到生成的 [`skill` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-skill)。
 
 #### Token 影响
 
-工具可见时，每次请求都有固定 schema 成本。
+工具可见时，每次请求都有固定的 schema token 开销。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 工具定义和可见性不变时，前缀稳定。遮蔽、限制或插件生命周期变更可能从该 schema 起使重用失效。
 
 ### 工具结果
 
-#### 模型所见
+#### 模型看到的内容
 
-成功调用使用下方结果模板，以及由提供方管理、目录、URL 或不透明的资源指引。
+成功调用使用下方结果模板，以及提供方管理的资源指引、目录资源指引、URL 资源指引或不透明资源指引。
 
 ##### Skill 结果模板
 
@@ -120,15 +120,15 @@ Load referenced resources only as needed.
 
 #### Token 影响
 
-已加载指令是取决于数据的工具结果 token，并在后续步骤中重新发送，直到压缩；不会制作重复的 `agent.inject()` 副本。
+已加载指令是取决于数据的工具结果 token，并在后续步骤中重新发送，直到压缩（compaction）；不会制作重复的 `agent.inject()` 副本。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-仅追加；新可见内容位于可重用请求前缀之后，不会使现有 KV 缓存条目失效。
+仅追加；新可见内容位于可重用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 工具错误
 
-#### 模型所见
+#### 模型看到的内容
 
 无效或陈旧选择会精确返回 `Error: invalid skill name "<name>"`、`Error: skill "<name>" is unknown or no longer available` 或 `Error: skill "<name>" is not available for model invocation`。提供方抛出的查找文本取决于数据，并接收同一个 `Error: <message>` 包装层。
 
@@ -136,13 +136,13 @@ Load referenced resources only as needed.
 
 只有失败调用会添加这些已保留 token。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-仅追加；新可见内容位于可重用请求前缀之后，不会使现有 KV 缓存条目失效。
+仅追加；新可见内容位于可重用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **目录省略 `whenToUse`、来源和提供方元数据**：路由只基于名称和有上限描述；`whenToUse` 仍是提供方元数据，加载后的包装层也不渲染它。
+- **目录省略 `whenToUse`、来源和提供方元数据**：路由只基于名称和有长度上限的描述；`whenToUse` 仍是提供方元数据，加载后的包装层也不渲染它。
 - **已加载指令正文没有大小上限**：提供方可返回足以占用大量下一步上下文的 skill；只有目录描述会被截断。
 - **资源是指引，而非附件**：工具报告基础目录/URL/不透明提示，但既不列举也不为模型获取引用文件。
-- **加载是一次性文本**：远程提供方缓慢或 skill 正文很大时，不提供部分、流式或缓存内容句柄。
+- **加载是一次性文本**：远程提供方缓慢或 skill 正文很大时，不提供部分内容、流式输出或缓存内容句柄。
diff --git a/packages/spill/README.i18n.yaml b/packages/spill/README.i18n.yaml
index 6b2792801c..6268366048 100644
--- a/packages/spill/README.i18n.yaml
+++ b/packages/spill/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/spill/README.md
 README.md: cb6f1e9af94ee5b0d0c06a8c4293e37e2d1eb856
-README.zh.md: 96de589f5aaf15a03d5d3d1b906dd274eea79440
+README.zh.md: bf51d0d1d88db38fa01122c3e4c78b4e1e608fee
diff --git a/packages/spill/README.zh.md b/packages/spill/README.zh.md
index 96de589f5a..bf51d0d1d8 100644
--- a/packages/spill/README.zh.md
+++ b/packages/spill/README.zh.md
@@ -1,15 +1,15 @@
-# spill/ - spill 存储功能家族
+# spill/ - spill 存储能力家族
 
 [English](README.md) | 中文
 
-工具输出 spill 的功能 seam：一个抽象存储接口、一个本地文件系统实现，以及一个使用该实现的工具结果策略。全部都是**产品** 包。
+工具输出 spill 的能力 seam：一个抽象存储接口、一个本地文件系统实现，以及一个使用该实现的工具结果策略。全部均为**产品**包（package）。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `spill/` | 抽象 spill 存储 seam（`saveText`：持久化过大的工具文本，返回定位信息与取回指引） | `ctx.spillStore` |
-| `spill-local/` | 本地文件系统后端：使用防路径遍历名称的私有会话级文件 | （注册到 `ctx.spillStore`） |
+| `spill-local/` | 本地文件系统后端：名称可防止路径遍历的私有会话级文件 | （注册到 `ctx.spillStore`） |
 | `spill-policy/` | `tools/post-execute` 策略：将过大的纯文本结果替换为预览和 spill 定位信息 | （无服务接口） |
 
 接口位于 `spill/spill/`。这种拆分方式与 bash/fs 相同：seam 只负责存储，`spill-local` 负责文件系统机制，`spill-policy` 负责决定何时 spill 以及面向模型的通知。预览机制位于 [`util/retention`](../util/README.md)；策略只组合两者，不会让任何一方承担对方的职责。
 
-设计原理见[工具输出 spill Agent Note](../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)，其中说明了为什么最终结果 spill 要与工具自行提前 spill（bash 流、subagent rollout）分离，以及为什么创建操作应由运行时 spill seam 而非面向模型的 `write` 工具承担。
+设计原理见[工具输出 spill Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)，其中说明了为什么最终结果 spill 要与工具自行提前 spill（bash 流、subagent rollout）分离，以及为什么创建操作应由运行时 spill seam 而非面向模型的 `write` 工具承担。
diff --git a/packages/spill/spill-local/README.i18n.yaml b/packages/spill/spill-local/README.i18n.yaml
index 03b817e00f..bc84d5a611 100644
--- a/packages/spill/spill-local/README.i18n.yaml
+++ b/packages/spill/spill-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/spill/spill-local/README.md
 README.md: 2270a65d9270e1549a9e48d6a36b821e48c29070
-README.zh.md: 2c5cd901d5a35e2129b7908082b6bc5e2b53be64
+README.zh.md: 0af8d81951cd3eaddc95724b73a84da271b35591
diff --git a/packages/spill/spill-local/README.zh.md b/packages/spill/spill-local/README.zh.md
index 2c5cd901d5..0af8d81951 100644
--- a/packages/spill/spill-local/README.zh.md
+++ b/packages/spill/spill-local/README.zh.md
@@ -2,33 +2,33 @@
 
 [English](README.md) | 中文
 
-[`@deepseek-ai/dsh-spill`](../spill) 存储 seam 的**本地文件系统** 实现。它注册为 `ctx.spillStore`，将工具过大的文本持久化到私有的会话级文件；定位信息是文件路径，取回指引会告诉模型对该路径使用 `read` 或 `grep`。
+[`@deepseek-ai/dsh-spill`](../spill) 存储 seam 的**本地文件系统**实现。它注册为 `ctx.spillStore`，将工具产生的过大文本持久化到私有的会话级文件；定位信息是文件路径，取回指引会告诉模型对该路径使用 `read` 或 `grep`。
 
 ## 存储布局
 
 文件存放在 `<root>/session-<hash>/​<random>-<safeName>`：
 
-- **`root`**：使用配置中的 `root`（解析为绝对路径）；如果省略，则在操作系统临时目录下延迟创建每进程私有（0700）目录。可预测且全球可读的根目录会让其他本地用户读取 spill 工具输出，或在其中预置符号链接。
-- **`session-<hash>`**：短 `sha256(sessionId)` 前缀，用于将一个会话的 spill 文件归组，以便未来的清理操作可按会话删除。
-- **`<random>-<safeName>`**：不可预测的十六进制前缀（防止在共享根目录中预置符号链接），加上经过清理的调用方 `suggestedName`，使其成为单个安全路径段（防路径遍历；与 JSONL 持久化后端的 `encodeSegment` 一致）。写入操作为排他且仅所有者可读写（`open(path, 'wx', 0o600)`）：如果路径已经存在，无论是否为符号链接，操作都会失败，因此预置的目标无法重定向写入。
+- **`root`**：使用配置中的 `root`（解析为绝对路径）；如果省略，则在操作系统临时目录下延迟创建每进程私有（0700）目录。可预测且任何用户均可读取的根目录会让其他本地用户读取 spill 工具输出，或在其中预置符号链接。
+- **`session-<hash>`**：截短的 `sha256(sessionId)` 前缀，用于将同一会话的 spill 文件归在一起，以便未来的清理操作可按会话删除。
+- **`<random>-<safeName>`**：不可预测的十六进制前缀（防止在共享根目录中预置符号链接），加上经过清理的调用方 `suggestedName`，使其成为单个安全路径段（防路径遍历；与 JSONL 持久化后端的 `encodeSegment` 一致）。写入操作采用排他方式，且权限仅限所有者（`open(path, 'wx', 0o600)`）：如果路径已经存在，无论是否为符号链接，操作都会失败，因此预置的目标无法重定向写入。
 
 ## 配置
 
 | 键 | 默认值 | 含义 |
 |---|---|---|
-| `root` | 私有 0700 临时目录 | spill 文件的根目录。进行设置可将它们保存在已知位置。 |
+| `root` | 私有 0700 临时目录 | spill 文件的根目录。设置后可将这些文件保存在已知位置。 |
 
-`saveText` 在发生真实存储故障（权限、ENOSPC）时拒绝；spill 策略会将该拒绝作为尽力而为的失败，并保留内联结果。词汇见 seam README，设计见[工具输出 spill Agent Note](../../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)。
+`saveText` 在发生真实存储故障（权限、ENOSPC）时返回拒绝；spill 策略会按尽力而为原则处理该拒绝，并保留内联结果。词汇见 seam README，设计见[工具输出 spill Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)。
 
 ## 模型体验
 
 通过渲染本地路径以及 `read`/`grep` 取回指引的 spill 消费方间接影响模型。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-不直接导致失效；指定的消费方负责其引起的任何请求前缀变更。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **本地 spill 文件会持续存在，直到外部清理为止**：该后端不提供会话生命周期删除或按时间保留的策略，因为已持久化、已恢复和 fork 后的会话可能仍在引用某个路径。
 - **定位信息需要与其位于同一文件系统的消费方**：远程或虚拟部署需要另一个 `SpillStore` 后端，其定位信息和取回指引在该环境中有明确含义。
diff --git a/packages/spill/spill-policy/README.i18n.yaml b/packages/spill/spill-policy/README.i18n.yaml
index e5cc04e53c..8d4746fc4c 100644
--- a/packages/spill/spill-policy/README.i18n.yaml
+++ b/packages/spill/spill-policy/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/spill/spill-policy/README.md
 README.md: 7638f62c0426964d83d7a635ebf73d0f117abd78
-README.zh.md: bf6d52a8c6b2b2e56052075af4a2150fba357f84
+README.zh.md: 522a0dafba621a808bc9e8c007b01274edf11dd3
diff --git a/packages/spill/spill-policy/README.zh.md b/packages/spill/spill-policy/README.zh.md
index bf6d52a8c6..522a0dafba 100644
--- a/packages/spill/spill-policy/README.zh.md
+++ b/packages/spill/spill-policy/README.zh.md
@@ -4,18 +4,18 @@
 
 **工具结果 spill 策略**：一个 `tools/post-execute` 转换器，用于防止过大的纯文本工具结果进入模型上下文。当最终结果超过 `maxInlineBytes` 时，它会通过 [`ctx.spillStore`](../spill) 保存完整文本，并将面向模型的结果替换为有界的首尾预览、后端定位信息与取回指引。
 
-该插件**不注册任何服务**，也不负责存储或预览机制：预览由 [`@deepseek-ai/dsh-retention`](../../util/retention) （`TextRetainer`）负责，存储由 `ctx.spillStore` 负责。它只决定何时 spill，并组合通知。
+该插件**不注册任何服务**，也不负责存储或预览机制：预览由 [`@deepseek-ai/dsh-retention`](../../util/retention)（`TextRetainer`）负责，存储由 `ctx.spillStore` 负责。它只决定何时 spill，并组合通知。
 
 ## 配置
 
 | 键 | 默认值 | 含义 |
 |---|---|---|
-| `maxInlineBytes` | *（省略）* | 面向模型的纯文本结果上下文上限，以 UTF-8 字节数计（在加载时验证为非负整数）。**省略时完全禁用该策略**（插件不注册任何内容）。设置后，较大的结果会被 spill，并替换为从同一预算派生的预览（首尾拆分）。 |
+| `maxInlineBytes` | *（省略）* | 面向模型的纯文本结果上下文上限，以 UTF-8 字节数计（在加载时验证为非负整数）。**省略时完全禁用该策略**（插件不注册任何内容）。设置后，超过该上限的结果会被 spill，并替换为从同一预算派生的预览（首尾拆分）。 |
 
 ## 行为
 
-1. 允许工具运行（通过 `next()` 委托，因此可以限制任何下游钩子接受的内容）。
-2. 跳过嵌套执行（存在 `exec.parent`——其持久副本由下方的 dispatch-log 分支设界）、已接受的值替换（注册表必须重新验证并渲染它们）、`read`（避免 `read → spill → read again` 循环）以及任何非 `accept` 决定（`block` 的纠正反馈会原样通过）。
+1. 允许工具运行（通过 `next()` 委托，因此可以限制任何下游钩子接受的结果）。
+2. 跳过嵌套执行（存在 `exec.parent`——其持久化副本由下方的 dispatch-log 分支设界）、已接受的值替换（注册表必须重新验证并重新渲染它们）、`read`（避免 `read → spill → read again` 循环）以及任何非 `accept` 决策（`block` 的纠正反馈会原样通过）。
 3. 仅在已接受的内容为**纯文本**（全部都是 `text` 块）时才将其展平；包含任何非文本块的结果都保持不变。
 4. 如果 UTF-8 大小为 `≤ maxInlineBytes`，则保持不变。
 5. 否则，保存完整文本，并将结果替换为预览和以下通知。系统会调整大小，使整个替换内容（预览、空行和通知）不超过 `maxInlineBytes`：先从预算中保留通知所需字节，再缩小预览以适配剩余空间，因此面向模型的结果绝不会超过上限：
@@ -28,31 +28,31 @@
 
    当通知本身已占满预算时（上限极小或定位信息很长），预览为空，只返回通知。如果仅通知的替换内容仍会超过 `maxInlineBytes`，策略将保留内联结果；它绝不会发出超过上限的替换内容（而且上限内的替换内容总比原结果更小，因此这也意味着 spill 绝不会增加字节数）。
 
-**尽力而为**：没有会话 owner、没有 `ctx.spillStore` 后端，或 `saveText` 拒绝 ⇒ 策略记录警告并返回原始结果。spill 失败绝不会将成功调用变为 `isError`，也不会隐藏内联结果。成功替换时只会更改 `content`；规范程序值保持不变。
+**尽力而为**：没有会话所有者、没有 `ctx.spillStore` 后端，或 `saveText` 返回拒绝 ⇒ 策略记录警告并返回原始结果。spill 失败绝不会将成功调用变为 `isError`，也不会隐藏内联结果。成功替换时只会更改 `content`；规范的程序化值保持不变。
 
-**dispatch-log 分支：**注册在 `tools/code-dispatch-log` 上的第二个监听器，把同一套上限、替换流水线与尽力而为的回退应用到每个 `run_code` 子调用结果的持久副本上（工件标签为 `dispatch`，按子调用 id 归档）。程序取得的值不受影响——它早已完整跨过 worker 边界；`read` 子调用同样设界：日志副本不是模型上下文，因此不会发生 read-again 循环，而 `read` 恰恰是最容易产生巨型日志的工具（[原理](../../../.agents/notes/implemented/feature/2026-07-26-code-dispatch-log-spill.md)）。
+**dispatch-log 分支：**注册在 `tools/code-dispatch-log` 上的第二个监听器，把同一套上限、替换流水线与尽力而为的回退应用到每个 `run_code` 子调用结果的持久化副本上（产物标签为 `dispatch`，按子调用 id 归档）。程序的值不受影响，因为它早已完整跨过 worker 边界；`read` 子调用同样设界：日志副本不是模型上下文，因此不会发生 read-again 循环，而 `read` 恰恰是最容易产生巨型日志的工具（[原理](../../../.agents/notes/implemented/feature/2026-07-26-code-dispatch-log-spill.md)）。
 
 ## 范围
 
-该策略只能看到最终格式化接口结果，看不到工具的内部资源或规范值。如果提供方已经截断内容（例如 `web-fetch-local.maxBodyChars`），spill 产物保存的是工具返回的完整格式化结果，而非完整原始源。提供方/资源上限仍必须存在，并且与该策略分离。`glob`/`grep` 负责对项级接口结果执行 spill，因为渲染前仍然存在完整的已获取值；bash 流负责在获取时 spill。通用策略预先注册自己的 waterfall 监听器，然后再委托，因此无论插件加载顺序如何，普通工具拥有的异步投影都会在通用字节限制之前完成。详见[工具输出 spill Agent Note](../../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)。
+该策略只能看到最终格式化的呈现结果，看不到工具的内部资源或规范值。如果提供方已经截断内容（例如 `web-fetch-local.maxBodyChars`），spill 产物保存的是工具返回的完整格式化结果，而非完整原始源。提供方／资源上限仍然是必需的，并且与该策略相互独立。`glob`/`grep` 负责对项级呈现结果执行 spill，因为渲染前仍然存在完整的已获取值；bash 流负责在获取时 spill。通用策略预先注册自己的 waterfall（瀑布式事件）监听器，然后再委托，因此无论插件加载顺序如何，普通工具自身的异步投影都会在通用字节限制之前完成。详见[工具输出 spill Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)。
 
 ## 模型体验
 
 ### 过大的纯文本结果
 
-#### 模型所见
+#### 模型看到的内容
 
-大小不超过 `maxInlineBytes` 的结果、嵌套结果、`read` 结果、已阻止的决定和包含非文本块的结果都保持不变。过大的纯文本接口结果会变为有界的首尾预览，后面附加 `(Omitted <bytes> bytes. Full formatted result stored at: <locator>. <retrievalHint>)`；存储或 owner 失败时，原始结果仍然可见。
+大小不超过 `maxInlineBytes` 的结果、嵌套结果、`read` 结果、被阻止的决策和包含非文本块的结果都保持不变。过大的纯文本呈现结果会变为有界的首尾预览，后面附加 `(Omitted <bytes> bytes. Full formatted result stored at: <locator>. <retrievalHint>)`；存储失败或没有会话所有者时，原始结果仍然可见。
 
 #### Token 影响
 
-成功替换后的内容最多为 `maxInlineBytes` 个 UTF-8 字节，并会保留在历史中直到压缩；完整 spill 文本不会重新发送给模型。
+成功替换后的内容最多为 `maxInlineBytes` 个 UTF-8 字节，并会保留在历史中直到压缩（compaction）；完整 spill 文本不会重新发送给模型。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-仅追加；新可见内容位于可重用请求前缀之后，不会使现有 KV 缓存条目失效。
+仅追加；新可见内容位于可重用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **只能对最终纯文本结果执行 spill**：混合内容结果、阻止反馈和 `read` 会原样通过；无法在此恢复先前已经发生的提供方截断或工具自有保留。
+- **只能对最终纯文本结果执行 spill**：混合内容结果、阻止反馈和 `read` 会原样通过；无法在此恢复先前已经发生的提供方截断或工具自身执行的保留处理。
 - **通知无法容纳时，该次调用的替换功能会禁用**：当上限极小或定位信息很长时，后端已经保存了无引用的 spill，但过大的原始结果仍会保留在内联位置。
diff --git a/packages/spill/spill/README.i18n.yaml b/packages/spill/spill/README.i18n.yaml
index b4117377bb..6a78a428f2 100644
--- a/packages/spill/spill/README.i18n.yaml
+++ b/packages/spill/spill/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/spill/spill/README.md
 README.md: c3434f90a8f7ab30baa6b237beab6dd9763f9bff
-README.zh.md: 4ceb4b1f4298d945f1f673403ad47a379263a674
+README.zh.md: 80a4431e9e57827e906fdfa4992a5a9af5b92aa9
diff --git a/packages/spill/spill/README.zh.md b/packages/spill/spill/README.zh.md
index 4ceb4b1f42..80a4431e9e 100644
--- a/packages/spill/spill/README.zh.md
+++ b/packages/spill/spill/README.zh.md
@@ -4,11 +4,11 @@
 
 **spill 存储 seam**：抽象的 `SpillStore` 服务（`ctx.spillStore`）定义 spill 后端做什么，即持久化某个工具过大的文本，并返回面向模型的定位信息与取回指引；它不规定如何实现。
 
-该包是 spill 功能的三个组成部分之一。拆分后，各项关注点可独立演进和替换：
+该包（package）是 spill 能力的三个组成部分之一。拆分后，各项关注点可独立演进和替换：
 
 | 包 | 职责 |
 |---|---|
-| `@deepseek-ai/dsh-spill` （本包） | 接口：抽象服务与词汇类型 |
+| `@deepseek-ai/dsh-spill`（本包） | 接口：抽象服务与词汇类型 |
 | `@deepseek-ai/dsh-spill-local` | 实现：位于宿主文件系统中的私有会话级文件 |
 | `@deepseek-ai/dsh-spill-policy` | 对过大最终结果执行 spill 的工具结果策略 |
 
@@ -18,25 +18,25 @@
 
 | 成员 | 语义 |
 |---|---|
-| `saveText(input)` | 逐字保存 `input.content`；解析并返回 `SpillRef`（不透明定位信息、写入的精确字节数和取回指引）。如果出现真实存储故障（权限、ENOSPC、后端不可用），则**拒绝**；由调用方决定如何降级。 |
+| `saveText(input)` | 逐字保存 `input.content`；成功时返回 `SpillRef`（不透明定位信息、写入的精确字节数和取回指引）。**如果出现真实存储故障，则返回拒绝**（权限、ENOSPC、后端不可用）；由调用方决定如何降级。 |
 
 存储操作以请求的 `owner` 会话作为保存时命名空间进行分组；后端自行选择私有表示，并可以从调用方的 `suggestedName` 派生名称，但绝不能将其当作可信路径。该 seam 只负责存储：不提供保留策略（由 [`@deepseek-ai/dsh-retention`](../../util/retention) 负责），不替换工具结果（由 `@deepseek-ai/dsh-spill-policy` 负责），也不提供取回/搜索 API（后端的 `retrievalHint` 会告诉模型如何使用定位信息）。
 
 ## 词汇
 
-`SaveTextSpill` （owner、source、suggestedName、content）是请求；`SpillRef` （locator、bytes、retrievalHint）是结果。`SpillLocator` 已经[品牌化](../../util/brand)，并以不透明字符串的形式呈现给模型；对 `dsh-spill-local` 而言它是本地路径，但未来的后端可以返回 URI、键或命令 token，无需修改策略/工具消费方。`SpillOwner.sessionId` 是保存时存储命名空间：fork 后的会话会从种子日志继承现有定位信息，无需复制文件或更改其归属；fork 后新产生的 spill 使用子会话 id。`SpillSource` （toolName、callId、label）是供后端命名和检查使用的描述性来源信息，而非访问控制信息。完整契约见 `src/types.ts`。
+`SaveTextSpill`（owner、source、suggestedName、content）是请求；`SpillRef`（locator、bytes、retrievalHint）是结果。`SpillLocator` 是[带品牌类型](../../util/brand)的值，并以不透明字符串的形式呈现给模型；对 `dsh-spill-local` 而言它是本地路径，但未来的后端可以返回 URI、键或命令 token，无需修改策略／工具消费方。`SpillOwner.sessionId` 是保存时存储命名空间：fork 后的会话会从种子日志继承现有定位信息，无需复制文件或更改其归属；fork 后新产生的 spill 使用子会话 id。`SpillSource`（toolName、callId、label）是供后端命名和检查使用的描述性来源信息，而非访问控制信息。完整契约见 `src/types.ts`。
 
-设计原理见[工具输出 spill Agent Note](../../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)，其中说明了为什么创建操作应由运行时 spill seam 而非面向模型的 `write` 工具承担。
+设计原理见[工具输出 spill Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-08-tool-output-spill-files.md)，其中说明了为什么创建操作应由运行时 spill seam 而非面向模型的 `write` 工具承担。
 
 ## 模型体验
 
 通过渲染后端定位信息和取回指引的 spill 消费方间接影响模型。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-不直接导致失效；指定的消费方负责其引起的任何请求前缀变更。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **该 seam 没有取回或删除 API**：消费方只能渲染后端的定位信息与指引；生命周期和访问语义仍由后端自行决定。
 - **存储不等于访问控制**：`SpillOwner` 会区分写入命名空间，但不会授予定位信息的读取权限；每个后端和取回消费方都必须自行强制执行访问边界。
diff --git a/packages/storage/README.i18n.yaml b/packages/storage/README.i18n.yaml
index 01831e2e2b..e4e025bc82 100644
--- a/packages/storage/README.i18n.yaml
+++ b/packages/storage/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/storage/README.md
 README.md: c9eca7355fd63e1023e9723486c86673e4c3574b
-README.zh.md: 4e848f16ea670eb4832a45eb7b7d28be3cb2da90
+README.zh.md: 0af55728a717b3725febca7ce812ba9d737d2a1a
diff --git a/packages/storage/README.zh.md b/packages/storage/README.zh.md
index 4e848f16ea..0af55728a7 100644
--- a/packages/storage/README.zh.md
+++ b/packages/storage/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-存储家族持久化会话事件日志之外的一切数据：命名后端与类型化数据形式在一个中心相接。设计记录：[领域 KV 存储 Agent Note](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
+存储家族持久化会话事件日志之外的一切数据：命名后端与类型化数据形式在一个中心相接。设计记录：[领域 KV 存储 Agent Note（agent 决策记录）](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
 
-| 包 | 职责 | ctx key |
+| 包（package） | 职责 | ctx key |
 |---|---|---|
-| `storage/` | 中心：命名后端注册表 + 可合并扩展的数据形式挂载、后端 facet 词汇、共享一致性测试套件 | `ctx.storage` |
+| `storage/` | 中心：命名后端注册表 + 可通过合并扩展的数据形式挂载、后端分面词汇、共享一致性测试套件 | `ctx.storage` |
 | `storage-json/` | JSON 后端：每个单元一个人类可读文件，以原子方式重写整个文件 | 注册后端 `json` |
 | `storage-sqlite/` | SQLite 后端：一个数据库承载所有已路由单元，每行一个文档 | 注册后端 `sqlite` |
 | `domain/` | 领域数据形式：经 zod 验证的记录、逐领域写入链、`domain/changed` 事件、按配置路由后端 | `ctx.storageDomain` + `ctx.storage.domain` |
 
-每个后端拥有一种介质，并公开数据形状 **facet**（目前为 `kv`；为未来的会话后端迁移预留 append-log facet）。每个后端插件都会在注册后发布内部生命周期服务；领域插件在公开自身服务前注入每个已配置的后端 key，因此配置树中的行序不携带启动语义。消费方绝不直接接触后端，而是注入 `storageDomain` 并通过它打开已声明的领域。
+每个后端拥有一种介质，并公开数据形状**分面**（目前为 `kv`；为未来的会话后端迁移预留追加日志分面）。每个后端插件都会在注册后发布内部生命周期服务；领域插件在公开自身服务前注入每个已配置的后端 key，因此配置树中的条目顺序不影响启动顺序。消费方绝不直接接触后端，而是注入 `storageDomain` 并通过它打开已声明的领域。
diff --git a/packages/storage/storage-domain/README.i18n.yaml b/packages/storage/storage-domain/README.i18n.yaml
index dc2e052c11..4f7fe3b7c7 100644
--- a/packages/storage/storage-domain/README.i18n.yaml
+++ b/packages/storage/storage-domain/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/storage/storage-domain/README.md
 README.md: 41e777c3d9a870530593a414839f905514a27134
-README.zh.md: cbef75e4ec445afae0a218259fe717fef6e23ae2
+README.zh.md: 97c55bb6d798366726f181608eccbbdf536e52f1
diff --git a/packages/storage/storage-domain/README.zh.md b/packages/storage/storage-domain/README.zh.md
index cbef75e4ec..97c55bb6d7 100644
--- a/packages/storage/storage-domain/README.zh.md
+++ b/packages/storage/storage-domain/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-DeepSeek Harness 存储中心的领域数据形式：在每个已配置后端注册后，公开可注入的 `ctx.storageDomain` 服务及对应的 `ctx.storage.domain` 投影。一个领域通过 `defineDomain`（zod 记录 schema、从 `z.infer` 派生的类型）声明一次，通过 `DomainFacility.open` 打开，并由具有最终决定权的内存状态提供服务：读取同步执行；写入在一条逐领域链上串行化，先在已路由后端达到持久状态，再更新内存并发出 `domain/changed`。打开消费方拥有 handle 的生命周期，并通过 `Domain.close()` 释放它（幂等；通常作为其自身的 `ctx.effect` disposer）；插件卸载时，facility 会关闭仍处于打开状态的领域。
+DeepSeek Harness 存储中心的领域数据形式：在所有已配置的后端注册后，公开可注入的 `ctx.storageDomain` 服务及对应的 `ctx.storage.domain` 投影。一个领域通过 `defineDomain`（zod 记录 schema、从 `z.infer` 派生的类型）声明一次，通过 `DomainFacility.open` 打开，并由具有最终决定权的内存状态提供服务：读取同步执行；写入在每个领域各自的一条链上串行化，先在已路由后端达到持久状态，再更新内存并发出 `domain/changed`。打开领域的消费方负责管理句柄的生命周期，并通过 `Domain.close()` 释放它（幂等；通常作为其自身的 `ctx.effect` 资源释放函数）；插件卸载时，该设施会关闭仍处于打开状态的领域。
 
-设计原理、打开语义和存储／领域分层见 [Agent Note](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
+设计原理、打开语义和存储／领域分层见 [Agent Note（agent 决策记录）](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
 
 ## 配置
 
 | key | 含义 |
 | --- | --- |
-| `backend` | 每个领域的默认后端名称（必填；不存在普遍正确的介质）。 |
+| `backend` | 每个领域的默认后端名称（必填；不存在普遍适用的存储介质）。 |
 | `routes` | 逐领域覆盖：领域名称 → 后端名称。 |
 
 ## 模型体验
@@ -19,7 +19,7 @@ DeepSeek Harness 存储中心的领域数据形式：在每个已配置后端注
 
 #### 模型看到的内容
 
-无。该包不注册工具、不注入提示词，也不追加会话事件；它在 `ctx.storageDomain` 后面存储非会话数据（Workspace 记录、未来的会话伴随元数据），只发出进程内 `domain/changed` 事件。只有消费方包通过自身已记录的表层渲染该事件时，它才会到达模型。
+无。该包（package）不注册工具、不注入提示词，也不追加会话事件；它在 `ctx.storageDomain` 后面存储非会话数据（工作区记录、未来的会话伴随数据），只发出进程内 `domain/changed` 事件。只有消费方包通过自身有文档说明的接口呈现该事件时，它才会到达模型。
 
 #### Token 影响
 
@@ -27,9 +27,9 @@ DeepSeek Harness 存储中心的领域数据形式：在每个已配置后端注
 
 #### KV Cache 影响
 
-相互独立：领域读写绝不触碰请求前缀，因此这里没有任何内容能使提供方 cache 复用失效。
+相互独立：领域读写绝不触碰请求前缀，因此这里没有任何内容能使提供方缓存复用失效。
 
 ## 已知限制与暂缓事项
 
-- **变更只在单进程内可见**：`domain/changed` 是进程内事件；在 Agent Note 暂缓的跨进程 revision 模式落地前，第二个主机进程或重新连接的 GUI 无法观察变更。
-- **没有跨表事务、二级索引或多 segment key**：每次写入只触碰一条记录；这些扩展的 trigger 和返工点列在 Agent Note 的暂缓工作清单中。
+- **变更只在单进程内可见**：`domain/changed` 是进程内事件；在 Agent Note 暂缓的跨进程修订模式落地前，第二个主机进程或重新连接的 GUI 无法观察变更。
+- **没有跨表事务、二级索引或多段键**：每次写入只触碰一条记录；这些扩展的触发点和返工点列在 Agent Note 的暂缓工作清单中。
diff --git a/packages/storage/storage-json/README.i18n.yaml b/packages/storage/storage-json/README.i18n.yaml
index efc252e025..25b54978b6 100644
--- a/packages/storage/storage-json/README.i18n.yaml
+++ b/packages/storage/storage-json/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/storage/storage-json/README.md
 README.md: c3417846a70f8a227a19a5abf834156d530e3597
-README.zh.md: 6a566aa1c2cdeac18760ec6e5964a623a3643928
+README.zh.md: 6a6ca9477fc0628516c97ff4d123beeea370d9be
diff --git a/packages/storage/storage-json/README.zh.md b/packages/storage/storage-json/README.zh.md
index 6a566aa1c2..6a6ca9477f 100644
--- a/packages/storage/storage-json/README.zh.md
+++ b/packages/storage/storage-json/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-[存储中心](../storage/README.md)的 JSON 后端：配置根目录下每个单元使用一个人类可读的 `<unit>.json` 文件，注册为后端 `json`。设计见[领域 KV 存储 Agent Note](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
+[存储中心](../storage/README.md)的 JSON 后端：配置根目录下每个单元使用一个人类可读的 `<unit>.json` 文件，注册为后端 `json`。设计见[领域 KV 存储 Agent Note（agent 决策记录）](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
 
 ## 模型
 
-- 内存中的单元状态具有最终决定权；每个写入原语都会通过临时写入 + fsync + 原子 `rename()` 替换重新发布整个文件。单元文件始终是完整的当前净状态：可读性是该后端存在的理由，规模问题则属于 SQLite 后端。
-- 缺失文件会作为空单元打开，并在第一次写入时物化。外部或无法解析的文件以 `malformed-medium` 拒绝；已存版本与 descriptor 不同时以 `version-mismatch` 拒绝（预发布立场，不迁移）。
-- 跨调用的写入顺序属于调用方（领域层的写入链）；每个单独调用具备原子性，并在 resolve 后持久。
+- 内存中的单元状态具有最终决定权；每个写入原语都会通过临时文件写入 + fsync + 原子 `rename()` 替换重新发布整个文件。单元文件始终是完整的当前状态：可读性是该后端存在的理由，规模问题则属于 SQLite 后端。
+- 缺失文件会作为空单元打开，并在第一次写入时物化。外来或无法解析的文件以 `malformed-medium` 拒绝；已存版本与描述符不同时以 `version-mismatch` 拒绝（预发布立场，不迁移）。
+- 跨调用的写入顺序属于调用方（领域层的写入链）；每次调用都具备原子性，并在完成时已达到持久状态。
 
 ## 配置
 
@@ -22,7 +22,7 @@
 
 #### 模型看到的内容
 
-无。该后端不贡献提示词、工具或 schema；它在 `ctx.storage` 后面持久化非会话领域数据，只供主机侧消费方使用。
+无。该后端不贡献提示词、工具或 schema；它在 `ctx.storage` 后面持久化非会话领域数据，只供宿主侧消费方使用。
 
 #### Token 影响
 
@@ -34,5 +34,5 @@
 
 ## 已知限制与暂缓事项
 
-- Windows 持久性依赖 libuv 的 `rename()`（使用替换的 `MoveFileExW`），没有显式 write-through 标志；append-log facet 落地时，计划把会话日志后端更严格的 Win32 write-through 发布辅助函数下移到此处（见 Agent Note 的迁移章节）。
-- 没有跨进程写锁：两个进程写入同一根目录时，可能交错执行整文件替换（最后写入者胜出）。当前消费方采用单主机进程部署；多进程方案按 Agent Note 的范围外表格暂缓。
+- Windows 持久性依赖 libuv 的 `rename()`（调用 `MoveFileExW` 并启用替换），没有显式 write-through 标志；追加日志分面落地时，计划把会话日志后端更严格的 Win32 write-through 发布辅助函数下移到此处（见 Agent Note 的迁移章节）。
+- 没有跨进程写锁：两个进程写入同一根目录时，可能交错执行整文件替换（最后写入者胜出）。当前消费方采用单一宿主进程部署；多进程方案按 Agent Note 的范围外事项表暂缓。
diff --git a/packages/storage/storage-sqlite/README.i18n.yaml b/packages/storage/storage-sqlite/README.i18n.yaml
index 389f4f99f4..ea968a8cf5 100644
--- a/packages/storage/storage-sqlite/README.i18n.yaml
+++ b/packages/storage/storage-sqlite/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/storage/storage-sqlite/README.md
 README.md: efc1b3ca54181a43c067594cc339ccc3a8fea510
-README.zh.md: bcf2dd44c5334ec0c2b6623d22244b3b6b77f18e
+README.zh.md: 0d680af2c2197e5e2d8773163a59927618b0e6cd
diff --git a/packages/storage/storage-sqlite/README.zh.md b/packages/storage/storage-sqlite/README.zh.md
index bcf2dd44c5..0d680af2c2 100644
--- a/packages/storage/storage-sqlite/README.zh.md
+++ b/packages/storage/storage-sqlite/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-[存储中心](../storage/README.md)的 SQLite 后端：注册为后端 `sqlite`，通过一个数据库文件提供 `kv` facet；该文件使用 `node:sqlite`（也可以是 `:memory:`）。设计与取舍见[领域 KV 存储 Agent Note](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
+[存储中心](../storage/README.md)的 SQLite 后端：注册为后端 `sqlite`，通过一个数据库文件提供 `kv` facet；该文件使用 `node:sqlite`（也可以是 `:memory:`）。设计与取舍见[领域 KV 存储 Agent Note（agent 决策记录）](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
 
 ## 存储模型
 
-每行一个文档：每个单元表都会成为一个物理 STRICT 表 `"u_<unit>_<table>" (key TEXT PRIMARY KEY, value TEXT)`，其中 `value` 是记录的 JSON 文本，因此一个 key 只更新一行（高频变更领域路由到这里而非 JSON 后端的原因）。单元标识位于两个元数据表中：`units` 在单元首次打开时标记其格式版本，descriptor 不同时以 `version-mismatch` 拒绝；`unit_globals` 保存每个单元的全局 singleton 行。物理布局版本位于 `PRAGMA user_version`；其他任何标记值都会被拒绝（未发布格式，不迁移）。单元名和表名在进入 DDL 之前依据中心的 `UNIT_NAME_RE` 接受验证，因此不会把外部输入插值到 SQL 标识符中。
+每行一个文档：每个单元表都会成为一个物理 STRICT 表 `"u_<unit>_<table>" (key TEXT PRIMARY KEY, value TEXT)`，其中 `value` 是记录的 JSON 文本，因此一个 key 只更新一行（高频变更领域路由到这里而非 JSON 后端的原因）。单元标识位于两个元数据表中：`units` 在单元首次打开时标记其格式版本，描述符不同时以 `version-mismatch` 拒绝；`unit_globals` 保存每个单元的全局单例行。物理布局版本位于 `PRAGMA user_version`；其他任何标记值都会被拒绝（未发布格式，不迁移）。单元名和表名在进入 DDL 之前依据中心的 `UNIT_NAME_RE` 进行验证，因此不会把外部输入插值到 SQL 标识符中。
 
-每个写入原语都是一条 prepared statement：SQLite 的逐语句原子性无需显式事务即可满足 KV 契约，写入顺序仍由调用方负责（领域层写入链）。缺失目录和数据库文件会以仅 owner 可访问的权限创建（`0o700`／`0o600`），与 session-persistence SQLite 后端一致；在计划的介质层提取完成前，该包逐字复用了后者的打开顺序。
+每个写入原语都是一条预处理语句：SQLite 的逐语句原子性无需显式事务即可满足 KV 契约，写入顺序仍由调用方负责（领域层写入链）。缺失目录和数据库文件会以仅所有者可访问的权限创建（`0o700`／`0o600`），与 session-persistence SQLite 后端一致；在计划的介质层提取完成前，该包（package）逐字复用了后者的打开顺序。
 
 ## 配置（schemastery）
 
@@ -25,7 +25,7 @@ interface Config {
 
 #### 模型看到的内容
 
-无。该后端不贡献提示词、工具或 schema；它在 `ctx.storage` 后面持久化非会话领域数据（Workspace 记录、未来的会话伴随元数据），只供主机侧消费方使用。
+无。该后端不贡献提示词、工具或 schema；它在 `ctx.storage` 后面持久化非会话领域数据（工作区记录、未来的会话伴随元数据），只供主机侧消费方使用。
 
 #### Token 影响
 
@@ -37,7 +37,7 @@ interface Config {
 
 ## 已知限制与暂缓事项
 
-- **`DatabaseSync` 是同步的**：每次写入会在其持续时间内阻塞事件循环（一条语句）；在领域数据规模下可以接受。
-- **没有 busy-wait 或重试策略**：另一个连接持有写事务时，该操作会立即被拒绝；多进程写入保护列在设计的未来工作清单中。
+- **`DatabaseSync` 是同步的**：每次写入都会在单条语句执行期间阻塞事件循环；在领域数据规模下可以接受。
+- **没有忙等待或重试策略**：另一个连接持有写事务时，该操作会立即被拒绝；多进程写入保护列在设计文档的未来工作清单中。
 - **只打开当前的 `STORAGE_SQLITE_SCHEMA_VERSION`**：其他任何已标记版本都会被拒绝而不是迁移（预发布立场）。
 - **`openDatabase` 重复了 session-persistence SQLite 打开顺序**：提取到共享介质层的工作暂缓至计划的会话后端迁移（见 Agent Note 的复用审计）。
diff --git a/packages/storage/storage/README.i18n.yaml b/packages/storage/storage/README.i18n.yaml
index 584d89f953..d84666fcb7 100644
--- a/packages/storage/storage/README.i18n.yaml
+++ b/packages/storage/storage/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/storage/storage/README.md
 README.md: 994285842925539e73f8f11780f19495f71d0280
-README.zh.md: 43cf542ceec7284228495e563dfa69b5c9af3157
+README.zh.md: de0f802c167146d53942ce6cb3768625602d9253
diff --git a/packages/storage/storage/README.zh.md b/packages/storage/storage/README.zh.md
index 43cf542cee..de0f802c16 100644
--- a/packages/storage/storage/README.zh.md
+++ b/packages/storage/storage/README.zh.md
@@ -2,17 +2,17 @@
 
 [English](README.md) | 中文
 
-非会话数据的存储中心（`ctx.storage`）：命名后端注册表加已挂载的数据形式 facility。中心自身不执行 IO：后端拥有介质，数据形式拥有语义。设计与取舍见[领域 KV 存储 Agent Note](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
+非会话数据的存储中心（`ctx.storage`）：命名后端注册表加已挂载的数据形式设施。中心自身不执行 IO：后端拥有介质，数据形式拥有语义。设计与取舍见[领域 KV 存储 Agent Note（agent 决策记录）](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
 
-## 形状
+## 结构
 
-- `ctx.storage.backend`：名称 → 后端表。多个后端并排保持挂载（`json`、`sqlite`）；为消费方提供服务的后端由该消费方自身的配置决定（领域层的路由表），绝非中心的全局选择。`register()` 返回 disposer；重复名称和未知 lookup 会高声失败。
-- `ctx.storage.mount(form, facility)`／`ctx.storage.form(form)`：数据形式挂载。`StorageForms` 可合并扩展；领域层合并 `domain`，并通过 `ctx.storage.domain` 访问。
-- 后端拥有一种介质（文件树根、数据库文件），并公开可选的数据形状 **facet**：目前为 `kv`；为未来的会话后端迁移预留 append-log facet。`src/backend.ts` 是规范契约文本；`tests/contract.ts` 导出每个后端都会运行的共享一致性测试套件。
+- `ctx.storage.backend`：名称 → 后端表。多个后端并排保持挂载（`json`、`sqlite`）；为消费方提供服务的后端由该消费方自身的配置决定（领域层的路由表），绝非中心的全局选择。`register()` 返回资源释放函数；注册重复名称或查找未知名称时都会明确报错。
+- `ctx.storage.mount(form, facility)`／`ctx.storage.form(form)`：数据形式挂载。`StorageForms` 可通过合并扩展；领域层合并 `domain`，并通过 `ctx.storage.domain` 访问。
+- 后端拥有一种介质（文件树根、数据库文件），并公开可选的数据形状**分面**：目前为 `kv`；为未来的会话后端迁移预留追加日志分面。`src/backend.ts` 是规范契约文本；`tests/contract.ts` 导出每个后端都会运行的共享一致性测试套件。
 
 ## 该分组中的包
 
-| 包 | 职责 |
+| 包（package） | 职责 |
 | --- | --- |
 | `dsh-storage` | 中心服务 + 后端词汇 + 共享一致性测试套件 |
 | `dsh-storage-json` | JSON 后端：每个单元一个人类可读文件，以原子方式重写整个文件 |
@@ -29,13 +29,13 @@
 
 #### Token 影响
 
-每次请求的直接 token 为零。
+每次请求都不会直接增加 token。
 
 #### KV Cache 影响
 
-与实时请求相互独立：中心绝不触碰请求前缀，因此无法使提供方 cache 复用失效。
+与实时请求相互独立：中心绝不触碰请求前缀，因此无法使提供方缓存复用失效。
 
 ## 已知限制与暂缓事项
 
 - **`kv` 是唯一的数据形状**：设计记录为未来的会话后端迁移预留了 append-log facet，但尚未定义；后端目前恰好只有一个 facet 需要实现。
-- **形式惰性解析**：在领域插件挂载前读取 `ctx.storage.domain` 会抛出 `form-not-mounted`；组装会按相应顺序排列插件（错误配置会高声失败，而不是静默等待）。
+- **数据形式按需解析**：在领域插件挂载前读取 `ctx.storage.domain` 会抛出 `form-not-mounted`；组装会按相应顺序排列插件（错误配置会明确报错，而不是静默推迟处理）。
diff --git a/packages/subagent/README.i18n.yaml b/packages/subagent/README.i18n.yaml
index 379c0fb498..a875878075 100644
--- a/packages/subagent/README.i18n.yaml
+++ b/packages/subagent/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/subagent/README.md
 README.md: fed0c3d6b252f5eeb8355c3b544066765999120a
-README.zh.md: 45f3c83f57613c16ba00063c9da9ac720a57727e
+README.zh.md: 9bc187aa972bc92385d32fe787e2fd65b6ce8361
diff --git a/packages/subagent/README.zh.md b/packages/subagent/README.zh.md
index 45f3c83f57..9bc187aa97 100644
--- a/packages/subagent/README.zh.md
+++ b/packages/subagent/README.zh.md
@@ -1,19 +1,19 @@
-# subagent/：subagent 能力族
+# subagent/：subagent 能力家族
 
 [English](README.md) | 中文
 
-subagent seam 允许 agent（智能体）把工作委派给子 agent。与 [bash](../bash/README.md) 和 [llm](../llm/README.md) 能力族一样，这也是一种能力 seam（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)），但有一个关键差异：**多个提供方实现在同一上下文中共存，并按名称注册**，而不是采用 bash 的单实现形态。该注册表仿照 LLM（大语言模型）适配器注册表。
+subagent（子 agent）seam 允许 agent（智能体）把工作委派给子 agent。与 [bash](../bash/README.md) 和 [llm](../llm/README.md) 能力家族一样，这也是一种能力 seam（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)），但有一个关键差异：**多个提供方实现在同一上下文中共存，并按名称注册**，而不是采用 bash 的单实现形态。该注册表仿照大语言模型（LLM）适配器注册表。
 
-| 包 | 角色 | ctx 键 |
+| 包（package） | 角色 | ctx 键 |
 |---|---|---|
 | `subagent/` | 抽象 subagent seam：具名提供方注册表与词汇 | `ctx.subagents` |
-| `subagent-inprocess/` | 共享进程内运行驱动器（不提供提供方；每次运行使用一个清理 effect） | 无 |
+| `subagent-inprocess/` | 共享进程内运行驱动器（不含提供方；每次运行使用一个清理 effect） | 无 |
 | `subagent-spawn/` | 进程内后端：全新的子 agent | （注册到 `ctx.subagents`） |
 | `subagent-fork/` | 进程内后端：以父 agent 已完成轮次的前缀作为初始内容的子 agent | （注册到 `ctx.subagents`） |
-| `subagent-acp/` | 进程外后端：在派生子进程中运行并通过 ACP（Agent Client Protocol）驱动的子 agent | （注册到 `ctx.subagents`） |
-| `subagent-dsh-sdk/` | 进程外后端：在派生子进程中运行的子 harness 运行时，经 TypeScript SDK 客户端走 stdio JSON-RPC 驱动 | （注册到 `ctx.subagents`） |
+| `subagent-acp/` | 进程外后端：在 spawn 的子进程中运行并通过 ACP（Agent Client Protocol）驱动的子 agent | （注册到 `ctx.subagents`） |
+| `subagent-dsh-sdk/` | 进程外后端：在 spawn 的子进程中运行的子 harness 运行时，经 TypeScript SDK 客户端走 stdio JSON-RPC 驱动 | （注册到 `ctx.subagents`） |
 | `tool-subagent/` | 面向模型的 `subagent` 委派工具，基于 `ctx.subagents` | （注册到 `ctx.tools`） |
 
-接口位于 `subagent/subagent/`。进程内 `subagent-spawn` / `subagent-fork` 后端共享 `subagent-inprocess` 驱动器（一个自身不提供提供方的库：两者都依赖它，彼此不依赖），进程外 `subagent-acp` / `subagent-dsh-sdk` 后端则经由 [`subprocess/`](../subprocess/README.md) seam spawn 其子进程（共享的凭据清除、以进程树为范围的拆卸、dispose（资源释放）阶梯）。测试只用包内 fixture（测试前置数据）替换子 agent 边界。
+接口位于 `subagent/subagent/`。进程内 `subagent-spawn` / `subagent-fork` 后端共享 `subagent-inprocess` 驱动器（一个自身不含提供方的库：两者都依赖它，彼此不依赖），进程外 `subagent-acp` / `subagent-dsh-sdk` 后端则经由 [`subprocess/`](../subprocess/README.md) seam spawn 其子进程（共享的凭据清除、以进程树为范围的拆卸、dispose（资源释放）阶梯）。测试只用包内 fixture（测试前置数据）替换子 agent 边界。
 
 提案与设计理由见 [.agents/notes/implemented/feature/2026-06-21-subagent-capability-seam.md](../../.agents/notes/implemented/feature/2026-06-21-subagent-capability-seam.md)。
diff --git a/packages/subagent/subagent-acp/README.i18n.yaml b/packages/subagent/subagent-acp/README.i18n.yaml
index 998a8e7396..e24b5f2f07 100644
--- a/packages/subagent/subagent-acp/README.i18n.yaml
+++ b/packages/subagent/subagent-acp/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/subagent/subagent-acp/README.md
 README.md: efcc77c442714a83631d009efb712fa7b8f5dfa0
-README.zh.md: 41723398ffa34995282814011cd03b2fc1cc5a4b
+README.zh.md: 216d66ed5d259c1bf2ea2383356df97be3e5ced1
diff --git a/packages/subagent/subagent-acp/README.zh.md b/packages/subagent/subagent-acp/README.zh.md
index 41723398ff..216d66ed5d 100644
--- a/packages/subagent/subagent-acp/README.zh.md
+++ b/packages/subagent/subagent-acp/README.zh.md
@@ -6,15 +6,15 @@ ACP（Agent Client Protocol）提供方会在全新的子进程中运行每个 s
 
 ## 启动与所有权
 
-`start(request)` 先解析子 agent 的工作目录，再依次执行 `spawn` → ACP `initialize` → `newSession`，然后才兑现。因此，兑现表示远程会话已就绪，所有权也已转移给调用方。派生、初始化、新建会话或发布前取消失败时，只有在子进程已回收后才会拒绝；工作目录解析失败则会在派生任何内容前拒绝。
+`start(request)` 先解析子 agent 的工作目录，再依次执行 `spawn` → ACP `initialize` → `newSession`，然后才兑现。因此，兑现表示远程会话已就绪，所有权也已转移给调用方。spawn、初始化、新建会话或发布前取消失败时，只有在子进程已回收后才会拒绝；工作目录解析失败则会在尚未 spawn 任何内容时拒绝。
 
 工作目录优先使用已配置的 `cwd` 覆盖值，否则使用执行委派的父会话 cwd，绝不使用服务器进程自身的 cwd，因为同一个服务器进程会服务来自多个工作区的会话。从父级取得的值必须是绝对路径，指向 harness 可以进入的目录（具备搜索权限，这是子进程 cwd 的要求）；解析后的同一路径同时作为子进程 cwd 和 ACP `session/new` 工作区。
 
 返回的运行 id 在父级命名空间中生成。子服务器的会话 id 只用于 ACP 协议调用，因为 ACP 只保证它在该全新子进程中唯一；若将其用作父级生命周期 id，可能与另一个远程运行或本地 agent 冲突。
 
-发布后，提供方发送提示词，并把流式 `agent_message_chunk` 文本收集到 `SubagentResult.output`。提示词/传输失败会以 `stopReason: 'error'` 兑现；如果必需的请求信号或 dispose 请求了取消，则以 `aborted` 兑现。
+发布后，提供方发送提示词，并把流式 `agent_message_chunk` 文本收集到 `SubagentResult.output`。提示词/传输失败会以 `stopReason: 'error'` 兑现；如果必需的请求信号或 dispose（资源释放）请求了取消，则以 `aborted` 兑现。
 
-`dispose()` 是幂等的。它会移除信号监听器，在可行时请求 ACP 取消，然后经由该 seam 的动词运行本后端自有的拆卸阶梯（`disposeAcpChild`）：先关闭 stdin 并等待 `disposeEofGraceMs` 让子进程协作停稳，再触发句柄的 `terminate()` 升级（SIGTERM、spawn 宽限期、SIGKILL——Windows 直接强制终止），最后进行有界的整树退出等待；若仍有存活进程，则拒绝。每次运行都使用全新进程；尚未实现进程池。
+`dispose()` 是幂等的。它会移除信号监听器，在可行时请求 ACP 取消，然后经由该 seam 的动词运行本后端自有的拆卸阶梯（`disposeAcpChild`）：先关闭 stdin 并等待 `disposeEofGraceMs` 让子进程协作式完全停稳，再触发句柄的 `terminate()` 升级（SIGTERM、spawn 宽限期、SIGKILL——Windows 直接强制终止），最后进行有界的整树退出等待；若仍有存活进程，则拒绝。每次运行都使用全新进程；尚未实现进程池。
 
 ## 能力与上下文
 
@@ -25,7 +25,7 @@ ACP 不声明任何启动时能力，因为当前进程无法强制执行远程
 | 键 | 默认值 | 含义 |
 |---|---|---|
 | `providerName` | `acp` | `ctx.subagents` 上的注册表名称。 |
-| `command` | 必填 | 每次运行时派生的可执行文件。 |
+| `command` | 必填 | 每次运行时 spawn 的可执行文件。 |
 | `args` | `[]` | 命令参数。 |
 | `cwd` | 父会话 cwd | 子进程及其 ACP 会话的工作目录覆盖值；不得为空。相对值会在加载时以 harness 启动目录为基准解析，结果必须指向 harness 可以进入的目录。 |
 | `permission` | `reject` | 自动回答权限请求：拒绝，或选择第一个允许形态的选项。 |
@@ -57,9 +57,9 @@ ACP 不声明任何启动时能力，因为当前进程无法强制执行远程
 
 ## 进程边界
 
-子进程经由 [`dsh-subprocess`](../../subprocess/subprocess/README.md) seam spawn：共享的凭据清除先移除名称形似凭据的环境变量和环境中已有的 `DSH_*` 名称，显式 `config.env` 值在清除之后合并（有意转发的 `DEEPSEEK_API_KEY` 会保留下来，`DSH_PERMISSION_MODE` 这类 `DSH_*` 部署事实也以同样的方式到达子进程——清除只丢弃其陈旧的同名环境值），stderr 以 inherit 方式直通父进程自身的流，dispose 则以本插件配置的宽限期运行该 seam 的协作式 stdin EOF→SIGTERM→SIGKILL 阶梯。ACP 协议是真正的序列化边界；同进程 subagent 值不会为防御目的而克隆。
+子进程经由 [`dsh-subprocess`](../../subprocess/subprocess/README.md) seam spawn：共享的凭据清除先移除疑似凭据的环境变量和环境中已有的 `DSH_*` 名称，显式 `config.env` 值在清除之后合并（有意转发的 `DEEPSEEK_API_KEY` 会保留下来，`DSH_PERMISSION_MODE` 这类 `DSH_*` 部署事实也以同样的方式到达子进程——清除只丢弃其陈旧的同名环境值），stderr 会继承到父进程自身的流，dispose 则以本插件配置的宽限期运行该 seam 的协作式 stdin EOF→SIGTERM→SIGKILL 阶梯。ACP 协议格式（wire format）是真正的序列化边界；同进程 subagent 值不会为防御目的而克隆。
 
-本包没有默认导出。否则 Cordis loader 的解包会隐藏具名 `inject` 元数据；见[事故复盘 0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)。
+本包（package）没有默认导出。否则 Cordis loader 的解包会隐藏具名 `inject` 元数据；见[事故复盘（postmortem）0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)。
 
 无密钥测试通过真实 stdio 驱动脚本化 ACP 子进程，其中包括一个由 Loader 组合的 stdio 应用，用于端到端证明父会话 cwd 继承。带密钥 e2e 会驱动仓库中的真实 ACP agent；没有 `DEEPSEEK_API_KEY` 时自行跳过。
 
@@ -87,17 +87,17 @@ ACP 不声明任何启动时能力，因为当前进程无法强制执行远程
 
 #### Token 影响
 
-父级输入只增加最终结果或错误，其内容依赖数据，并保留到上下文压缩为止。该提供方自身不会添加父级 schema。
+父级输入只增加最终结果或错误，其内容依赖数据，并保留到压缩（compaction）为止。该提供方自身不会添加父级 schema。
 
 #### KV Cache 影响
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **每次运行使用全新进程**：持久进程池属于后续优化（见 [seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-21-subagent-capability-seam.md)）。
+- **每次运行使用全新进程**：持久进程池属于后续优化（见 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-06-21-subagent-capability-seam.md)）。
 - **仅支持本地工作区**：解析后的 cwd 是交给同一台机器上子进程的本地路径；远程 ACP agent 的工作区映射需要独立的后端能力，本包尚未设计。
 - **不支持可选启动时能力**：该提供方无法在远程进程内应用本地 harness 的 `outputSchema`、深度上限、工具过滤器或 persona，因此不会声明这些能力；服务会拒绝需要它们的请求。
-- **只收集已提交的 `agent_message_chunk` 文本**：自动化服务器把推理、工具活动、计划和其他 trace 数据保留在子 agent 会话日志中，不通过 ACP 发出。
+- **只收集已提交的 `agent_message_chunk` 文本**：自动化服务器把推理（reasoning）、工具活动、计划和其他 trace 数据保留在子 agent 会话日志中，不通过 ACP 发出。
 - **权限提示自动回答**（`permission: allow | reject`）：当前版本不会把子 agent 的 `session/request_permission` 呈现给人。
 - **没有快照层回放覆盖率**（`TODO(acp-subagent-replay)`）：ACP 子 agent 拥有独立进程和独立回放形态，该工作延期处理。
diff --git a/packages/subagent/subagent-dsh-sdk/README.i18n.yaml b/packages/subagent/subagent-dsh-sdk/README.i18n.yaml
index 54cf6ef79d..296617f484 100644
--- a/packages/subagent/subagent-dsh-sdk/README.i18n.yaml
+++ b/packages/subagent/subagent-dsh-sdk/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/subagent/subagent-dsh-sdk/README.md
 README.md: e904ce3c09a1b44f8f5a0072b9ca85812898e74f
-README.zh.md: b11cb9c8e0bf2577be71230292d73878b22896a0
+README.zh.md: cec89ad9a65c68163fc136fe7b04cb135c57fb51
diff --git a/packages/subagent/subagent-dsh-sdk/README.zh.md b/packages/subagent/subagent-dsh-sdk/README.zh.md
index b11cb9c8e0..cec89ad9a6 100644
--- a/packages/subagent/subagent-dsh-sdk/README.zh.md
+++ b/packages/subagent/subagent-dsh-sdk/README.zh.md
@@ -2,21 +2,21 @@
 
 [English](README.md) | 中文
 
-SDK provider 把每个子代理作为一个完整的 DeepSeek Harness 运行时跑在全新子进程里，经由 [TypeScript SDK 客户端](../../sdk/sdk-client/README.md)走 stdio JSON-RPC 驱动。它是 [`subagent-acp`](../subagent-acp/README.md) 之外的第二个进程外后端，差异在线协议与子进程契约：ACP 后端能驱动任何 Agent Client Protocol 代理；本后端专门驱动 harness SDK 运行时（`dsh-jsonrpc-agent` bin 或打包可执行文件），因此子进程是一个完整的对等 harness——自有 `cordis.yml` 决定的组成、会话持久化、模型路由与工具。
+SDK 提供方会在全新的子进程中把每个 subagent 作为完整的 DeepSeek Harness 运行时运行，并经由 [TypeScript SDK 客户端](../../sdk/sdk-client/README.md) 通过 stdio JSON-RPC 驱动。它是 [`subagent-acp`](../subagent-acp/README.md) 之外的第二个进程外后端，差异在协议格式（wire format）和子进程契约：ACP（Agent Client Protocol）后端能驱动任何 Agent Client Protocol agent（智能体）；本后端专门驱动 harness SDK 运行时（`dsh-jsonrpc-agent` bin 或打包后的可执行文件），因此子进程是一个完整的对等 harness，拥有由 `cordis.yml` 决定的组合、会话持久化、模型路由和工具。
 
 ## 启动与所有权
 
-`start(request)` 先解析子进程工作目录，经 `DeepSeekHarness` 生成运行时，并在履行前完成 `initialize` 握手（携带配置的 `provider`/`model` 路由及可选的 `maxTokens` 输出上限）。因此履行意味着子运行时已就绪、所有权已移交调用方。生成、握手或发布前取消的失败只在子进程被收割之后拒绝；工作目录解析失败在生成任何东西之前拒绝。
+`start(request)` 先解析子进程工作目录，通过 `DeepSeekHarness` spawn 运行时，并在履行前完成 `initialize` 握手（携带配置的 `provider`/`model` 路由及可选的 `maxTokens` 输出上限）。因此，履行意味着子运行时已就绪、所有权已移交给调用方。spawn、握手或发布前取消失败时，只会在子进程被回收后拒绝；工作目录解析失败则会在尚未 spawn 任何内容时拒绝。
 
-工作目录的解析与 ACP 后端完全一致，经由接缝共享的进程外助手（[`dsh-subagent`](../subagent/README.md)）：设置了 `cwd` 覆盖则用之（加载时校验一次），否则用发起委托的父会话 cwd——绝不用服务器进程自己的 cwd。解析出的路径同时成为子进程 cwd 与其 SDK 会话的工作区 cwd。
+工作目录的解析与 ACP 后端完全一致，并使用 seam 共享的进程外辅助工具（[`dsh-subagent`](../subagent/README.md)）：设置了 `cwd` 覆盖值时使用该值（加载时校验一次），否则使用发起委派的父会话 cwd，绝不使用服务器进程自身的 cwd。解析出的路径同时成为子进程 cwd 和其 SDK 会话的工作区 cwd。
 
-返回的 run id 铸造于父命名空间；子运行时的会话 id 只存在于子进程内部。发布之后，provider 跑一个 SDK 回合，并从子会话事件中读取答案：最后一条完整 `assistant/message`，或回合被截断时已累积的 `text-delta` 流——部分答案在取消与错误路径上都得以保留。
+返回的 run id 在父级命名空间中生成；子运行时的会话 id 只存在于子进程内部。发布后，提供方运行一个 SDK 轮次，并从子会话事件中读取答案：最后一条完整的 `assistant/message`，或轮次被截断时已累积的 `text-delta` 流；部分答案在取消和错误路径上都得以保留。
 
-`dispose()` 幂等：先把结果就地定格为 `aborted`（线上没有 prompt 取消方法），再关闭运行时——一次有界的协议 `shutdown` 请求，随后是共享的 stdin-EOF → SIGTERM → SIGKILL 阶梯直到真正退出。
+`dispose()`（资源释放）是幂等的：先在本地把结果确定为 `aborted`（协议层面没有提示词取消机制），再关闭运行时，即先发出一次有界的协议 `shutdown` 请求，随后通过共享的 stdin-EOF → SIGTERM → SIGKILL 阶梯使进程实际退出。
 
 ## 停止原因映射
 
-子进程在 `session.finished` 上以结构化 `TurnEndReason` 报告回合结局；provider 把它映射进接缝词汇表。`completed` → `completed`，`max-tokens` → `max-tokens`，`aborted` → `aborted`；其余一切——`error`、`interrupted`、`disposed`、未来变体、或根本没跑回合——映射为 `error`，不洁终止绝不报告为成功。发布后的传输层失败经 `onError` 诊断汇（接到 `ctx.logger.warn`）压平为 `stopReason: 'error'`；接缝契约禁止 `result` 拒绝。
+子进程在 `session.finished` 上以结构化 `TurnEndReason` 报告轮次结果；提供方将其映射为 seam 词汇。`completed` → `completed`，`max-tokens` → `max-tokens`，`aborted` → `aborted`；其余情况，包括 `error`、`interrupted`、`disposed`、未来变体或根本未运行轮次，均映射为 `error`，因此非正常停止绝不会报告为成功。发布后的传输层失败会通过 `onError` 诊断接收器（连接到 `ctx.logger.warn`）压平为 `stopReason: 'error'`；seam 契约禁止 `result` 被拒绝。
 
 ## 能力与上下文
 
@@ -27,14 +27,14 @@ Provider 不宣告任何启动期能力（`outputSchema`/`depthLimit`/`toolFilte
 | 键 | 默认 | 含义 |
 |---|---|---|
 | `providerName` | `dsh-sdk` | `ctx.subagents` 上的注册名。 |
-| `command` | 必填 | 每次 run 生成的可执行文件（子运行时 bin 或打包 exe）。 |
+| `command` | 必填 | 每次运行时 spawn 的可执行文件（子运行时 bin 或打包后的可执行文件）。 |
 | `args` | `[]` | 命令参数（通常是子进程的 `cordis.yml` 路径）。 |
 | `cwd` | 父会话 cwd | 工作目录覆盖；校验规则与 [`subagent-acp`](../subagent-acp/README.md) 相同。 |
-| `provider` | `deepseek` | 写入子进程 `initialize` 的 provider 路由。 |
+| `provider` | `deepseek` | 写入子进程 `initialize` 的提供方路由。 |
 | `model` | `deepseek-v4-flash` | 写入子进程 `initialize` 的模型。 |
-| `maxTokens` | provider 默认值 | 写入子进程 `initialize` 的单次请求输出 token 上限；对子根 Agent 及其进程内后代生效。 |
+| `maxTokens` | 提供方默认值 | 写入子进程 `initialize` 的单次请求输出 token 上限；对子运行时的根 agent 及其进程内后代生效。 |
 | `env` | `{}` | 在凭据擦除后的父环境之上叠加的显式子环境（例如子进程自己的 `DEEPSEEK_API_KEY`，或 `DSH_CORDIS_CONFIG`）。 |
-| `shutdownTimeoutMs` | `1000` | 处置期间协议 `shutdown` 交换的时限。 |
+| `shutdownTimeoutMs` | `1000` | dispose 期间协议 `shutdown` 交换的时限。 |
 | `disposeEofGraceMs` | `6000` | stdin EOF 之后、平台终止之前的宽限。 |
 | `disposeGraceMs` | `3000` | 终止后的退出确认窗口；POSIX 在 SIGTERM 之后、SIGKILL 之前也等待同样时长。 |
 
@@ -55,45 +55,45 @@ Provider 不宣告任何启动期能力（`outputSchema`/`depthLimit`/`toolFilte
 
 ## 进程边界
 
-子环境以 [`dsh-subprocess`](../../subprocess/README.md) 接缝的 `scrubbedParentEnv()` 为基底——移除形似凭据与 `DSH_*` 的环境变量——再在擦除之后合并显式 `config.env` 值。子进程由 SDK 客户端生成而非经 `ctx.subprocess`（subprocess README 记载的 SDK 托管传输例外），因此本后端自行应用该擦除。JSON-RPC 线就是真实的序列化边界。
+子进程环境以 [`dsh-subprocess`](../../subprocess/README.md) seam 的 `scrubbedParentEnv()` 为基础，先移除疑似凭据和名称为 `DSH_*` 的环境变量，再合并显式 `config.env` 值。子进程由 SDK 客户端 spawn，而不是经由 `ctx.subprocess` spawn（这是 subprocess README 中记录的 SDK 托管传输例外），因此本后端会自行执行环境清理。JSON-RPC 协议格式才是真正的序列化边界。
 
-本包没有默认导出。否则 Cordis loader 解包会隐藏具名 `inject` 元数据；见[事后分析 0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)。
+本包（package）没有默认导出。否则 Cordis loader 解包会隐藏具名 `inject` 元数据；见[事故复盘（postmortem）0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)。
 
 免密钥测试通过真实 stdio 驱动 SDK 客户端包的脚本化伪运行时，还包括一个 Loader 组合 e2e：子进程是真实的第二个 harness 运行时，端到端证明父会话 cwd 继承（`tests/loader-composition.e2e.ts`）。
 
-## Model Experience
+## 模型体验
 
-### Child-agent request
+### 子 agent 请求
 
-#### What the model sees
+#### 模型看到的内容
 
-子运行时的模型收到独立任务作为其用户消息，加上该运行时自己配置的系统提示、工具与全新会话。它收不到任何父对话。本 provider 不宣告可选启动期能力，因此本地服务会拒绝需要 persona、工具过滤、深度强制或结构化输出的请求，而不是静默省略。
+子运行时的模型会收到作为用户消息的独立任务，以及该运行时自身配置的系统提示词、工具和全新会话。它不会收到父级对话。本提供方不声明可选的启动时能力，因此本地服务会拒绝要求 persona、工具过滤、深度强制或结构化输出的请求，而不是静默省略这些要求。
 
-#### Token effect
+#### Token 影响
 
-子进程支付一份独立的完整上下文与自己的多步历史。这些 token 绝不进入父上下文。
+子运行时会为独立的完整上下文及其多步骤历史消耗 token。这些 token 绝不会进入父级上下文。
 
-#### KV Cache effect
+#### KV Cache 影响
 
-独立于父请求缓存。每个 SDK 子进程只能复用在其自身 provider、模型、组成与历史下完全相同的前缀；子步骤在此之外只增不改。
+与父级请求缓存相互独立。每个 SDK 子进程只能复用其自身提供方、模型、组合和历史均相同时的前缀；除此之外，子 agent 的步骤仅追加增长。
 
-### Parent tool result, indirectly
+### 父级工具结果（间接）
 
-#### What the model sees
+#### 模型看到的内容
 
-经由 `dsh-tool-subagent`，父方只收到子进程的最终助手文本（或累积的部分文本），或该消费者精确的停止原因错误——收不到中间消息与工具流量。
+经由 `dsh-tool-subagent`，父级只会收到子运行时最终的 assistant 文本（或累积的部分文本），或该消费方给出的精确停止原因错误；不会收到中间消息或工具流量。
 
-#### Token effect
+#### Token 影响
 
-父输入只增长最终结果或错误，其大小依数据而定，保留至压缩。本 provider 自身不给父方增加任何 schema。
+父级输入只增加最终结果或错误，其大小取决于数据，并保留到压缩（compaction）为止。本提供方自身不会向父级添加任何 schema。
 
-#### KV Cache effect
+#### KV Cache 影响
 
-只追加；新可见内容跟在可复用请求前缀之后，不使既有 KV 缓存条目失效。
+仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## Known Limitations and Deferred Work
+## 已知限制与暂缓事项
 
-- **每次 run 一个全新运行时进程** —— 无池化；harness 运行时要启动完整插件树，单次生成成本高于 ACP 后端的典型子进程。
-- **无可选启动期能力** —— 父方无法在子进程内强制 `outputSchema`、深度、工具过滤或 persona；请改为配置子进程自己的 `cordis.yml`。
-- **子进程的转录留在其自己的会话根** —— 父日志只记录委托工具调用/结果（接缝的子隔离规则）；流式 `session.event` 通道只用于提取输出，不桥接进父日志。
-- **仅限本地子进程** —— 解析出的 cwd 是本地路径；远程运行时需要自己的后端。
+- **每次运行都使用全新的运行时进程**：不使用进程池；harness 运行时需要启动完整的插件树，因此每次运行的 spawn 成本高于 ACP 后端通常使用的子进程。
+- **不支持可选的启动时能力**：父级无法在子进程内强制执行 `outputSchema`、深度限制、工具过滤或 persona；应改为配置子进程自身的 `cordis.yml`。
+- **子进程的 transcript（文本记录）保留在其自身的会话根目录中**：父级日志只记录委派工具调用／结果（seam 的子级隔离规则）；流式 `session.event` 通道只用于提取输出，不会桥接到父级日志中。
+- **仅支持本地子进程**：解析出的 cwd 是本地路径；远程运行时需要独立的后端。
diff --git a/packages/subagent/subagent-fork/README.i18n.yaml b/packages/subagent/subagent-fork/README.i18n.yaml
index 73456dc604..14b40beb88 100644
--- a/packages/subagent/subagent-fork/README.i18n.yaml
+++ b/packages/subagent/subagent-fork/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/subagent/subagent-fork/README.md
 README.md: b448dc309bff07c744443530a648c7c30e4d20d9
-README.zh.md: be6730ebe23510a017ffa80563016133dc87cae3
+README.zh.md: 3e14206d5637fded9edb4e173608c55e3341f8fc
diff --git a/packages/subagent/subagent-fork/README.zh.md b/packages/subagent/subagent-fork/README.zh.md
index be6730ebe2..3e14206d56 100644
--- a/packages/subagent/subagent-fork/README.zh.md
+++ b/packages/subagent/subagent-fork/README.zh.md
@@ -31,7 +31,7 @@ fork 声明 `{ outputSchema: true, depthLimit: true, toolFilter: true, persona:
 
 #### 模型看到的内容
 
-子 agent 先接收父 agent 平衡的已完成轮次界面前缀，再逐字接收新的任务内容。配置的 persona 会在子 agent 的全新作用域中遮蔽提示词文本；工具限制会过滤其全局协议 schema、可执行工具查找和 Code Mode SDK 绑定，但不影响独立注册的指导内容。父 agent 的工具视图与权限不会被继承。可选的结构化输出请求会添加仅属于子 agent 的契约。父 agent 当前进行中的轮次会被排除。
+子 agent 先接收父 agent 已配平的完整轮次表层前缀，再逐字接收新的任务内容。配置的 persona 会在子 agent 的全新作用域中遮蔽提示词文本；工具限制会过滤其全局协议 schema、可执行工具查找和 Code Mode SDK 绑定，但不影响独立的指导内容。父 agent 的工具视图与权限不会被继承。可选的结构化输出请求会添加仅属于子 agent 的契约。父 agent 当前进行中的轮次会被排除。
 
 #### Token 影响
 
@@ -49,13 +49,13 @@ fork 会把保留的已完成历史复制到独立的子 agent 请求中；随
 
 #### Token 影响
 
-父 agent 输入增加一个依赖数据的最终结果，并保留到上下文压缩（compaction）为止。
+父 agent 输入会增加一个取决于数据的最终结果，并保留到压缩（compaction）为止。
 
 #### KV Cache 影响
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **运行不公开 `sendMessage`/`resume`**：进程内运行不具备这些可选运行时能力。
 - **初始内容是一次性快照**：子 agent 只能看到 fork 时父 agent 已完成的轮次，看不到父 agent 此后记录的任何内容；不会实时共享上下文。
diff --git a/packages/subagent/subagent-spawn/README.i18n.yaml b/packages/subagent/subagent-spawn/README.i18n.yaml
index 68d37783c7..970a148213 100644
--- a/packages/subagent/subagent-spawn/README.i18n.yaml
+++ b/packages/subagent/subagent-spawn/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/subagent/subagent-spawn/README.md
 README.md: 868f829edbcfe2eb4d66ccd0ff9988924c70298b
-README.zh.md: 823291baae0ca09b86565516af1843aff033f354
+README.zh.md: 99cfdf0e633345d1152c59cbe5ce7a029eb6ec9d
diff --git a/packages/subagent/subagent-spawn/README.zh.md b/packages/subagent/subagent-spawn/README.zh.md
index 823291baae..99cfdf0e63 100644
--- a/packages/subagent/subagent-spawn/README.zh.md
+++ b/packages/subagent/subagent-spawn/README.zh.md
@@ -6,9 +6,9 @@ spawn 提供方会在当前进程中创建一个全新的子 `Agent`。子 agent
 
 ## 行为
 
-`start(request)` 不提供初始内容，直接委托给 [`startInProcessRun`](../subagent-inprocess/README.md)，并在子 agent 发布后才返回。子 agent 获得父 agent 的工作目录/会话谱系，并默认继承父 agent 模型（除非覆盖），但以空对话开始运行。
+`start(request)` 不传入 seed，直接委托给 [`startInProcessRun`](../subagent-inprocess/README.md)，并在子 agent 发布后才返回。子 agent 获得父 agent 的工作目录/会话谱系，并默认继承父 agent 模型（除非覆盖），但以空对话开始运行。
 
-共享驱动器负责深度检查、persona 与工具过滤器设置、结构化输出、必需信号取消、单次执行、结果读取和完全停稳后的 dispose（资源释放）。启动失败不会留下已发布的子 agent；提供方插件在完成后卸载，也不会撤销由持有方拥有的运行。
+共享驱动器负责深度检查、persona 与工具过滤器设置、结构化输出、通过必需的信号执行取消、单次执行、结果读取和完全停稳后的 dispose（资源释放）。启动遭拒不会留下已发布的子 agent；启动调用兑现后卸载提供方，也不会撤销由持有方拥有的运行。
 
 ## 能力
 
@@ -30,7 +30,7 @@ spawn 声明 `{ outputSchema: true, depthLimit: true, toolFilter: true, persona:
 
 #### Token 影响
 
-子 agent 为全新的独立上下文和历史支付 token 成本；不会复制父 agent 历史 token。persona 会改变该子 agent 的重复提示词成本，过滤则会改变其 schema 或生成 SDK 的成本。
+子 agent 会为全新的独立上下文和历史消耗 token；不会复制父 agent 历史的 token。persona 会改变该子 agent 反复使用的提示词成本，过滤则会改变其 schema 或生成 SDK 的成本。
 
 #### KV Cache 影响
 
@@ -50,7 +50,7 @@ spawn 声明 `{ outputSchema: true, depthLimit: true, toolFilter: true, persona:
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **运行不公开 `sendMessage`/`resume`**：进程内运行不具备这些可选运行时能力。
-- **全新表示不含父 agent transcript**：子 agent 会继承 cwd、谱系、模型及显式配置的 persona/工具限制，但不继承父 agent 的任何对话；需要已完成轮次上下文时，请使用 fork 提供方。
+- **全新表示不含父 agent transcript（文本记录）**：子 agent 会继承 cwd、谱系、模型及显式配置的 persona/工具限制，但不继承父 agent 的任何对话；需要已完成轮次上下文时，请使用 fork 提供方。
diff --git a/packages/subagent/subagent/README.i18n.yaml b/packages/subagent/subagent/README.i18n.yaml
index e5d5d48e33..e2e1b999ab 100644
--- a/packages/subagent/subagent/README.i18n.yaml
+++ b/packages/subagent/subagent/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/subagent/subagent/README.md
 README.md: 3d5d5e7498b1700c07486cc6e894e72fed681bec
-README.zh.md: 3eb3d2bb379ec1a39035f42ffbda08bed6698823
+README.zh.md: eb26c79665d387a1e779050dad476d8672f67642
diff --git a/packages/subagent/subagent/README.zh.md b/packages/subagent/subagent/README.zh.md
index 3eb3d2bb37..eb26c79665 100644
--- a/packages/subagent/subagent/README.zh.md
+++ b/packages/subagent/subagent/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-subagent seam 允许一个 agent（智能体）通过具名提供方把工作委派给子 agent。调用方使用统一的服务 API（`ctx.subagents`）；提供方决定子 agent 在当前进程、另一进程还是未来的传输之上运行。
+subagent seam 允许一个 agent（智能体）通过具名提供方把工作委派给子 agent。调用方使用统一的服务 API（`ctx.subagents`）；提供方决定子 agent 在当前进程中、另一进程中，还是通过未来的传输机制运行。
 
-## 包角色
+## 包（package）的角色
 
-该能力族把稳定接口与实现、面向模型的工具分开：
+该系列包把稳定接口与实现、面向模型的工具分开：
 
 | 包 | 角色 |
 |---|---|
@@ -24,12 +24,12 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委
 
 | 成员 | 含义 |
 |---|---|
-| `registerProvider(provider)` | 按名称注册一个可信的同进程实现。注册受 effect 作用域约束；移除注册会阻止新的启动，但不会撤销已返回给调用方的运行。重复名称会立即失败。 |
+| `registerProvider(provider)` | 按名称注册一个可信的同进程实现。注册受 effect 作用域约束；移除注册会阻止新的启动，但不会撤销已返回给调用方的运行。重复名称会明确报错。 |
 | `getProvider(name)` | 返回提供方；不存在时返回 `undefined`。 |
 | `list()` | 按插入顺序返回提供方名称。 |
-| `start(name, request)` | 校验请求的能力和语义值，然后等待提供方，直到真实子 agent 就绪。兑现时返回由持有方拥有的 `SubagentRun`；拒绝表示提供方已清理所有局部启动资源。 |
+| `start(name, request)` | 校验请求的能力和语义值，然后等待提供方，直到真实子 agent 就绪。兑现时返回由持有方拥有的 `SubagentRun`；拒绝表示提供方已清理启动过程中取得的全部资源。 |
 
-`SubagentStartRequest.signal` 是必填项，也是规范取消通道。发布前中止会使 `start()` 在回滚后拒绝；发布后中止会取消实时子 agent。请求还可以选择模型、要求结构化输出、限制委派深度、约束子 agent 工具或设置子 agent persona。
+`SubagentStartRequest.signal` 是必填项，也是规范取消通道。发布前中止会使 `start()` 在回滚后拒绝；发布后中止会取消正在运行的子 agent。请求还可以选择模型、要求结构化输出、限制委派深度、约束子 agent 工具或设置子 agent persona。
 
 同进程请求、描述符、结果和事件 payload 都是以不可变方式借用的可信类型值。服务不会克隆或冻结它们；序列化和不可信输入校验属于真实的进程、worker、持久化和模型边界。
 
@@ -46,17 +46,17 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委
 
 该 seam 拥有实现和消费方共享的深度词汇：`AgentOptions.subagentDepth` 声明、`assertSubagentMaxDepth` 和 `delegationDepthOf(agent)`。持久化的 `SessionHeader.delegationDepth` 具有权威性且单调：运行时选项可以加深计数，但绝不能降低它，因此恢复后的子 agent 不会被重新计为顶层。
 
-运行时功能是 `SubagentRun` 上的可选方法：`sendMessage?` 会引导实时子 agent，`resume?` 则异步创建延续运行。方法是否存在就是能力检查。
+运行时功能是 `SubagentRun` 上的可选方法：`sendMessage?` 可对正在运行的子 agent 进行 steering（中途引导），`resume?` 则异步创建延续运行。方法是否存在就是能力检查。
 
 `inheritsParentContext` 只用于描述，不能强制执行。它仅说明子 agent 是否能看到父级已完成的对话历史（`fork` 可以；`spawn` 和 ACP 不可以），不表示是否继承工具、服务或权限。
 
 ## 所有权与生命周期
 
-`provider.start(request): Promise<SubagentRun>` 是所有权转移边界。兑现前，提供方拥有设置过程，并且每次失败时都必须取消、回滚并使局部资源完全停稳。兑现后，调用方拥有该运行，并且必须在每条路径上调用 `dispose()`。
+`provider.start(request): Promise<SubagentRun>` 是所有权转移边界。兑现前，提供方拥有设置过程，并且每次失败时都必须取消、回滚并使启动过程中已取得的资源完全停稳。兑现后，调用方拥有该运行，并且必须在每条路径上调用 `dispose()`。
 
 `SubagentRun.result` 兑现为 `{ output, structured?, stopReason }`。子 agent 级失败会以非 `completed` 原因兑现；只有 seam 无法表示的基础设施故障才可以拒绝。`dispose()` 是幂等的，会取消剩余工作，并等待子 agent 资源完全停稳。
 
-本地运行会在 `start()` 兑现前发布普通的子 agent/会话，把该共享会话 id 作为 `SubagentRun.id` 返回，以 `SubagentRun.localAgent` 公开准确的子 agent，并把 `request.parent.session.id` 记录到子 agent 的 `parentSession` header。远程提供方则生成父级作用域的生命周期 id，并返回 `localAgent: undefined`。
+本地运行会在 `start()` 兑现前发布普通的子 agent/会话，把该共享会话 id 作为 `SubagentRun.id` 返回，以 `SubagentRun.localAgent` 公开该子 agent 本身，并把 `request.parent.session.id` 记录到子 agent 的 `parentSession` header。远程提供方则生成父级作用域的生命周期 id，并返回 `localAgent: undefined`。
 
 服务只会发出 `subagent/start`，而且是在 `start()` 兑现后。它在同步通知前附加结果观察器，因此即使子 agent 已经结算，也仍会先产生 `subagent/start`，再产生 `subagent/end`。这对事件共享服务生成的 `runId`；其 `local` 标志取自提供方准确 `localAgent` 的快照，因此观察器绝不会从可复用的提供方/会话名称推断运行身份或本地性。
 
@@ -66,7 +66,7 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委
 
 ## 收集模型
 
-面向模型的工具默认同步收集：先等待子 agent 结果，再 dispose 运行，然后才返回。后台委派不会改变该 seam；消费方把启动过程和最终运行注册到通用 `ctx.tasks` 运行时，随后使用共享任务工具进行收集和取消。完整契约见[后台 subagent 任务 Agent Note](../../../.agents/notes/implemented/feature/2026-07-08-background-subagent-tasks.md)、[能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-21-subagent-capability-seam.md)和 `src/types.ts`。
+面向模型的工具默认同步收集：先等待子 agent 结果，再对运行执行 dispose（资源释放），然后才返回。后台委派不会改变该 seam；消费方把启动过程和最终运行注册到通用 `ctx.tasks` 运行时，随后使用共享任务工具进行收集和取消。完整契约见[后台 subagent 任务 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-08-background-subagent-tasks.md)、[能力 seam Agent Note](../../../.agents/notes/implemented/feature/2026-06-21-subagent-capability-seam.md)和 `src/types.ts`。
 
 ## 模型体验
 
@@ -76,7 +76,7 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委
 
 不会直接使缓存失效；具名消费方负责请求前缀的任何变化。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **运行时引导和延续只是 seam 能力**：当前工具中没有消费 `sendMessage` 和 `resume` 的面向模型消费方。
+- **运行时 steering 和延续只是 seam 能力**：当前工具中没有消费 `sendMessage` 和 `resume` 的面向模型消费方。
 - **生命周期事件只供观察**：影响运行的 `subagent/end` 延续或决策接口仍需等待具体消费方。
diff --git a/packages/subagent/tool-subagent/README.i18n.yaml b/packages/subagent/tool-subagent/README.i18n.yaml
index f021a7266e..9c8e235669 100644
--- a/packages/subagent/tool-subagent/README.i18n.yaml
+++ b/packages/subagent/tool-subagent/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/subagent/tool-subagent/README.md
 README.md: 20bb6b9c59a13f23301368faefe18849ccc0b1e9
-README.zh.md: 9435ff638b0d021507af487a54fabac3018ca400
+README.zh.md: 8da57896359f5ac47d0ec076c3395d2e7fb1e02a
diff --git a/packages/subagent/tool-subagent/README.zh.md b/packages/subagent/tool-subagent/README.zh.md
index 9435ff638b..8da5789635 100644
--- a/packages/subagent/tool-subagent/README.zh.md
+++ b/packages/subagent/tool-subagent/README.zh.md
@@ -10,7 +10,7 @@
 
 前台调用会让执行信号贯穿启动和执行，等待 `run.result`，并且在返回前总会等待 `run.dispose()`。只有 `completed` 会返回规范值 `{ kind: 'foreground', runId, output: JsonValue[] }`，并渲染为相同的最终文本；中止、拒绝、token 上限和其他失败都会变成出错的工具结果，不包含局部输出。
 
-设置 `run_in_background: true` 后，工具会在启动提供方前注册父级拥有的任务，并返回规范值 `{ kind: 'background', taskId }`，渲染为 `started background subagent task <id>`。任务拥有的信号覆盖待处理的启动阶段，以及启动调用返回后的子 agent。`task_kill` 和所有者 dispose（资源释放）会中止它。结算会等待启动回滚或子 agent dispose，然后把完成的最终文本映射为完成、中止映射为 `killed`、其他失败映射为 `failed`。任务不提供增量读取；通用任务工具负责后续状态、收集、取消和通知。见[后台 subagent Agent Note](../../../.agents/notes/implemented/feature/2026-07-08-background-subagent-tasks.md)。
+设置 `run_in_background: true` 后，工具会在启动提供方前注册父级拥有的任务，并返回规范值 `{ kind: 'background', taskId }`，渲染为 `started background subagent task <id>`。任务拥有的信号覆盖待处理的启动阶段，以及启动调用返回后的子 agent。`task_kill` 和所有者 dispose（资源释放）会中止它。结算会等待启动回滚或子 agent dispose，然后把完成的最终文本映射为完成、中止映射为 `killed`、其他失败映射为 `failed`。任务不提供增量读取；通用任务工具负责后续状态、收集、取消和通知。见[后台 subagent Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-08-background-subagent-tasks.md)。
 
 `toolFilter` 会改变子 agent 的全局工具层，但不是从父级派生的权限上限。见 [agent 作用域的安全非目标](../../../.agents/notes/implemented/architecture/2026-07-08-agent-scope-contexts.md#security-and-authority-are-non-goals)。
 
@@ -21,14 +21,14 @@
 | `provider`（必填） | 提供方名称（`spawn`、`fork`、`acp` 等）。 |
 | `toolName` | 面向模型的名称，默认 `subagent`；每个已加载实例必须不同。 |
 | `enableRunInBackground` | 公开后台模式，默认 `true`；禁用时也会拒绝强制后台调用。 |
-| `agentOptions` | 传给具体 provider 的子 agent `provider`、`model` 和正整数 `maxTokens`；进程内 provider 会用显式值覆盖继承的父级选项。 |
+| `agentOptions` | 传给具体提供方的子 agent `provider`、`model` 和正整数 `maxTokens`；进程内提供方会用显式值覆盖继承的父级选项。 |
 | `persona` | 每个子 agent 独立的 persona；要求提供方具备 `persona` 能力。 |
 | `toolFilter` | 每个子 agent 独立的全局工具限制；要求提供方具备 `toolFilter` 能力。 |
 | `maxDepth` | 绝对委派深度上限，默认 `3`（`0` 禁止委派）；数值上限要求 `depthLimit` 能力，缺失时挂载失败。对于预算由子 harness 拥有的进程外提供方，`'provider-managed'` 不发送上限。工具在达到上限时仍然可见；每次尝试启动都会检查调用 agent 的当前深度，被拒绝时返回出错的工具结果。 |
 
 ## 并发
 
-前台调用与后台调用互斥。子 agent 可能共享父级工作区或外部资源，一元分类器无法证明同级委派的效果彼此不相交。见[并行工具调用 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md)。
+前台调用和后台调用均互斥。子 agent 可能共享父级工作区或外部资源，一元分类器无法证明同级委派的效果彼此不相交。见[并行工具调用 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-parallel-tool-call-execution.md)。
 
 ## 模型体验
 
@@ -40,7 +40,7 @@
 
 #### Token 影响
 
-每个父级请求支付固定 schema 成本；每个提供方实例增加一个 schema。
+每个父级请求都会产生固定的 schema token 开销；每个提供方实例增加一个 schema。
 
 #### KV Cache 影响
 
@@ -64,7 +64,7 @@
 
 #### 模型看到的内容
 
-启动时精确返回 `started background subagent task <id>`。通用任务接口提供后续状态、最终输出、取消响应和通知。
+启动时原样返回 `started background subagent task <id>`。通用任务接口提供后续状态、最终输出、取消响应和通知。
 
 #### Token 影响
 
@@ -74,7 +74,7 @@
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **后台运行只公开最终输出**：子 agent 中间步骤留在子 agent 会话中。
 - **等待中实例的重复名称发现较晚**（`TODO(subagent-dup-toolname)`）：若要阻止提供方注册回滚，需要一份预期名称注册表。
diff --git a/packages/subprocess/README.i18n.yaml b/packages/subprocess/README.i18n.yaml
index f536302eea..5f310fc9b3 100644
--- a/packages/subprocess/README.i18n.yaml
+++ b/packages/subprocess/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/subprocess/README.md
 README.md: 64e4740c7ac2706e45bb3517891504bf31a6109b
-README.zh.md: e30b6c7f51ed0dffba15a6d1dff632e4ff4c6402
+README.zh.md: 615b492fa7da5b12a7d0cecfe98fc4f1eb94b504
diff --git a/packages/subprocess/README.zh.md b/packages/subprocess/README.zh.md
index e30b6c7f51..615b492fa7 100644
--- a/packages/subprocess/README.zh.md
+++ b/packages/subprocess/README.zh.md
@@ -1,12 +1,12 @@
-# subprocess/：进程管理能力家族
+# subprocess/：子进程能力家族
 
 [English](README.md) | 中文
 
-spawn 受管子进程树的共用归属位置：完全显式的 spawn spec，其 stdio 处置方式（disposition）为 Node 形状、按流划分（原始管道、inherit、附带 spill 文件的有界尾部保留收集）；harness 中所有 spawn 调用方共用的那一份凭据清除；基于偏移量的增量读取；以进程树为范围、带 SIGTERM→宽限期→SIGKILL 升级的信号发送；以及协作式 dispose（资源释放）阶梯。命令默认值补全、shell 语义、deadline、协议分帧与呈现留在消费方：[bash 执行器](../bash/README.md)、[LSP 主机](../lsp/README.md)与 [ACP（Agent Client Protocol）subagent 后端](../subagent/README.md)。参见[进程管理器 seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-26-subprocess-seam.md)。
+这里集中提供受管子进程树的 spawn 能力：完整指定的 spawn spec，采用 Node 风格、按流划分的 stdio 处置方式（disposition），包括原始管道、inherit、附带 spill 文件的有界尾部保留收集；harness 中所有 spawn 调用方共用的凭据清除机制；基于偏移量的增量读取；以进程树为范围、带 SIGTERM→宽限期→SIGKILL 升级的信号发送；以及协作式 dispose（资源释放）阶梯。命令默认值补全、shell 语义、时限、协议分帧与呈现留在消费方：[bash 执行器](../bash/README.md)、[LSP 主机](../lsp/README.md)与 [ACP（Agent Client Protocol）subagent 后端](../subagent/README.md)。参见[subprocess seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-26-subprocess-seam.md)。
 
 | 包（package） | ctx 键 | 角色 |
 |---|---|---|
 | [`subprocess`](subprocess/README.md)（`@deepseek-ai/dsh-subprocess`） | `ctx.subprocess` | seam 本体：抽象的 `SubprocessService.spawn(spec)`、完全显式且带按流划分 stdio 处置方式的 `SubprocessSpawnSpec`、`SubprocessHandle`（流、基于偏移量的读取器、terminate/waitForExit/dispose），以及共享的凭据清除 + `DSH_*`/`CollectedOutput` 词汇 |
 | [`subprocess-local`](subprocess-local/README.md)（`@deepseek-ai/dsh-subprocess-local`） | 无 | 本地实现：detached 进程树、按处置方式接线的流、附带有界私有 spill 文件的尾部保留截断、`DSH_*` 合并次序、带升级的进程树信号发送、dispose 阶梯，以及先终止再等待退出的 dispose |
 
-服务拥有跨消费方重载的进程存续期；消费方拥有一个进程的含义（一条 bash 命令、未来的非 shell 运行器）以及塑造它的每一项默认值。
+即使消费方重载，进程生命周期仍由服务负责管理；消费方负责定义进程的含义（一条 bash 命令、未来的非 shell 运行器），以及决定塑造该进程的每一项默认值。
diff --git a/packages/subprocess/subprocess-local/README.i18n.yaml b/packages/subprocess/subprocess-local/README.i18n.yaml
index b778e094f4..64daa32c81 100644
--- a/packages/subprocess/subprocess-local/README.i18n.yaml
+++ b/packages/subprocess/subprocess-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: a4e20b5c64afd14cb9dce9f8cc46fa96a1f5f19d
-README.zh.md: 2bd7ebe7c3407a4ad6a86c520594465f7f1e710c
+#   pnpm run verify-translation-pairing --write packages/subprocess/subprocess-local/README.md
+README.md: 202fc57080400afcbf5a65c17ea6bc8ac758c96f
+README.zh.md: c3a00cfd327c6865eda57cc63b87fd5f40e2b24f
diff --git a/packages/subprocess/subprocess-local/README.md b/packages/subprocess/subprocess-local/README.md
index a4e20b5c64..202fc57080 100644
--- a/packages/subprocess/subprocess-local/README.md
+++ b/packages/subprocess/subprocess-local/README.md
@@ -8,7 +8,7 @@ Local implementation of the [`@deepseek-ai/dsh-subprocess`](../subprocess/README
 
 - **Detached process trees with platform-correct signalling** — POSIX children are spawned `detached` (own process group) and signalled by negative pgid with a direct-child fallback; Windows terminates the tree via `taskkill /PID <pid> /T /F` (injectable for tests). `terminate()` — the handle's only termination verb — sends SIGTERM then SIGKILL after the spec's grace (OpenCode's escalation; pipelines and subshells die with the parent) and is a no-op once the tree is gone; `waitForExit()` polls whole-tree liveness so consumer teardown confirms real quiescence. After the leader exits, still-open pipes receive the same bounded drain grace so a surviving descendant cannot hold the outcome open indefinitely. ESRCH is tolerated; daemons that re-parent away from the group can still survive — the same caveat as the surveyed tools.
 - **Per-stream dispositions** — `'pipe'` hands the raw stream to the caller untouched (protocol framing stays consumer-owned); `'inherit'` passes the parent descriptor through; collect mode keeps the in-memory TAIL beyond its cap (errors and results cluster at the end — pi/OpenCode rationale) while the FULL stream is appended to a private temp file when a spill cap is configured — omitting `spill` keeps only the tail, the diagnostic shape. A stream larger than the spill cap discards its now-incomplete spill and returns only the marked truncated tail; spill fds are sealed at settlement, and a failed final close withholds the path rather than advertising an incomplete file. Spill files are `0600` with random names under a lazily-created `0700` per-process directory.
-- **Credential scrub + explicit merge** — `process.env` minus credential-shaped vars (`*KEY*`/`*SECRET*`/`*TOKEN*`) and all ambient `DSH_*` names; the spec's explicit `env` merges after that scrub with no namespace validation, so a deliberately supplied credential or current `DSH_*` fact wins while stale nested-harness identity cannot leak in ambiently. Supplied stdin is written and closed; otherwise fd 0 is `/dev/null`. See the [stdin/env Agent Note](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md) and [managed environment Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md).
+- **Credential scrub + explicit merge** — `process.env` minus credential-shaped vars (`*KEY*`/`*PASSWORD*`/`*SECRET*`/`*TOKEN*`) and all ambient `DSH_*` names; the spec's explicit `env` merges after that scrub with no namespace validation, so a deliberately supplied credential or current `DSH_*` fact wins while stale nested-harness identity cannot leak in ambiently. Supplied stdin is written and closed; otherwise fd 0 is `/dev/null`. See the [stdin/env Agent Note](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md) and [managed environment Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md).
 - **Offset-based reads** — collect-mode readers return deltas in whole-stream byte coordinates; the service never holds a cursor, so consumer-owned cursors (the bash background read path) and full-stream re-reads coexist, before and after settlement.
 - **Terminate-and-join disposal** — the service retains live handles only so its own disposal can escalate every running tree and await its exit; settled and spawn-failed handles leave the live set on settlement.
 
@@ -23,7 +23,7 @@ No direct invalidation; the named consumers own any request-prefix changes.
 ## Known Limitations and Deferred Work
 
 - **Windows tree support is best-effort and untested in CI** — termination routes through `taskkill /PID <pid> /T /F` with all outcomes contained (absent tree, races, missing binary), and liveness falls back to the direct-child boundary; the suites cover the routing through an injected runner only, and `packages/subprocess/*` is excluded from the Windows test matrix.
-- **The credential scrub is a name heuristic** — `*KEY*`/`*SECRET*`/`*TOKEN*` only; differently-named secrets (e.g. `*PASSWORD*`) pass through, and a whitelist for over-scrubbed vars is noted future work.
+- **The credential scrub is a name heuristic** — `*KEY*`/`*PASSWORD*`/`*SECRET*`/`*TOKEN*` only; differently-named secrets (e.g. `*PASSPHRASE*`) pass through, and a whitelist for over-scrubbed vars is noted future work.
 - **Completed spill files are not deleted** — bounded full-output recovery files (and the private per-process spill dir) accumulate under the OS tmpdir until something external cleans them; oversize incomplete spills are discarded and deletion is attempted immediately, but a cleanup failure can leave a bounded file behind.
 
 The raw process handling lives in `src/spawn.ts`; `src/index.ts` is the service wiring.
diff --git a/packages/subprocess/subprocess-local/README.zh.md b/packages/subprocess/subprocess-local/README.zh.md
index 2bd7ebe7c3..c3a00cfd32 100644
--- a/packages/subprocess/subprocess-local/README.zh.md
+++ b/packages/subprocess/subprocess-local/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-[`@deepseek-ai/dsh-subprocess`](../subprocess/README.md) seam 的本地实现：`LocalSubprocessService` 把每个 spec 的 argv 作为 detached 进程树 spawn，依照 spec 中按流划分的 stdio 处置方式（disposition）完成接线（原始管道、inherit、附带可选 spill 文件的有界尾部保留收集），并以进程树为范围、按 SIGTERM→SIGKILL 升级发送信号。该实现没有任何配置：每项处置方式、限制与目录都随 spawn spec 到达，因此随部署变化的旋钮留在各调用方 seam 的配置里（[`dsh-bash-local`](../../bash/bash-local/README.md)、[`dsh-lsp-local`](../../lsp/lsp-local/README.md)、[`dsh-subagent-acp`](../../subagent/subagent-acp/README.md)）。
+[`@deepseek-ai/dsh-subprocess`](../subprocess/README.md) seam 的本地实现：`LocalSubprocessService` 将每个 spec 的 argv spawn 为 detached 进程树，依照 spec 中按流划分的 stdio 处置方式（disposition）完成接线（原始管道、inherit、附带可选 spill 文件的有界尾部保留收集），并以进程树为范围发送信号，按 SIGTERM→SIGKILL 逐级升级。该实现没有任何配置：每项处置方式、限制与目录都随 spawn spec 传入，因此随部署变化的可调参数留在各调用方 seam 的配置里（[`dsh-bash-local`](../../bash/bash-local/README.md)、[`dsh-lsp-local`](../../lsp/lsp-local/README.md)、[`dsh-subagent-acp`](../../subagent/subagent-acp/README.md)）。
 
 ## 行为（以及设计来源）
 
-- **带平台正确信号发送的 detached 进程树**：POSIX 子进程使用 `detached` spawn（拥有独立进程组），信号以负 pgid 发送并以直接子进程作为回退；Windows 通过 `taskkill /PID <pid> /T /F` 终止进程树（可为测试注入）。`terminate()`（句柄唯一的终止动词）先发送 SIGTERM，经过 spec 的宽限期后再发送 SIGKILL（沿用 OpenCode 的升级策略；管道与子 shell 会随父进程一起结束），进程树消亡后为空操作；`waitForExit()` 轮询整棵进程树的存活状态，使消费方的拆卸能确认真正的完全停稳。组长进程退出后，仍然打开的管道也只获得同样有界的排空宽限期，因此存活的后代进程无法无限期地拖住结果不结算。系统会容忍 ESRCH；脱离该组重新挂载的 daemon 仍可能存活，这与调研工具的局限相同。
-- **按流划分的处置方式**：`'pipe'` 把原始流原样交给调用方（协议分帧仍归消费方所有）；`'inherit'` 直通父进程的描述符；收集模式（collect）在输出超过上限后于内存中保留尾部（错误与结果通常聚集在末尾，沿用 pi/OpenCode 的理由），并在配置了 spill 上限时把完整流追加到一个私有临时文件；省略 `spill` 则只保留尾部，即诊断尾部的形状。某条流大于 spill 上限时，会丢弃已不完整的 spill，仅返回带截断标记的尾部；spill 文件描述符在结算时封存，最终关闭失败时则不公布路径，以免声称存在不完整的文件。spill 文件权限为 `0600`、名称随机，位于按需延迟创建的 `0700` 每进程目录之下。
-- **凭据清除 + 显式合并**：以 `process.env` 为基础，移除形似凭据的变量（`*KEY*`／`*SECRET*`／`*TOKEN*`）和所有环境中已有的 `DSH_*` 名称；spec 的显式 `env` 在该清除之后合并且不做命名空间校验，因此有意提供的凭据或当前 `DSH_*` 事实会胜出，而陈旧的嵌套 harness 身份无法从环境中隐式漏入。提供的 stdin 会被写入后关闭；否则 fd 0 指向 `/dev/null`。参见 [stdin/env Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md)与[受管环境 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md)。
-- **基于偏移量的读取**：收集模式的读取器以全流字节坐标返回增量；服务自身从不持有游标，因此消费方自有的游标（bash 的后台读取路径）与完整流重读可以共存，结算前后皆然。
-- **先终止再等待退出的 dispose**：服务保留存活句柄，只为让自身的 dispose 能对每个仍在运行的进程树执行升级并等待其退出；已结算与 spawn 失败的句柄在结算时即离开存活集合。
+- **以适合平台的方式发送信号的 detached 进程树**：POSIX 子进程使用 `detached` spawn（拥有独立进程组），信号以负 pgid 发送并以直接子进程作为回退；Windows 通过 `taskkill /PID <pid> /T /F` 终止进程树（可为测试注入）。`terminate()`（句柄唯一的终止操作）先发送 SIGTERM，经过 spec 的宽限期后再发送 SIGKILL（沿用 OpenCode 的升级策略；流水线与子 shell 会随父进程一起结束），进程树消亡后为空操作；`waitForExit()` 轮询整棵进程树的存活状态，使消费方的拆卸能确认真正的完全停稳。组长进程退出后，仍然打开的管道也只获得同样有界的排空宽限期，因此存活的后代进程无法无限期地拖住结果不结算。系统会容忍 ESRCH；重新指定父进程并脱离该组的 daemon 仍可能存活，这与所调研工具的局限相同。
+- **按流划分的处置方式**：`'pipe'` 把原始流原样交给调用方（协议分帧仍归消费方所有）；`'inherit'` 直通父进程的描述符；收集模式（collect）在输出超过上限后于内存中保留尾部（错误与结果通常聚集在末尾，沿用 pi/OpenCode 的理由），并在配置了 spill 上限时把完整流追加到一个私有临时文件；省略 `spill` 则只保留用于诊断的尾部。某条流大于 spill 上限时，会丢弃已不完整的 spill，仅返回带截断标记的尾部；spill 文件描述符在结算时封存，最终关闭失败时则不公布路径，以免声称存在不完整的文件。spill 文件权限为 `0600`、名称随机，位于按需创建、权限为 `0700` 的每进程目录之下。
+- **凭据清除 + 显式合并**：以 `process.env` 为基础，移除形似凭据的变量（`*KEY*`／`*PASSWORD*`／`*SECRET*`／`*TOKEN*`）和所有环境中已有的 `DSH_*` 名称；spec 的显式 `env` 在该清除之后合并且不做命名空间校验，因此有意提供的凭据或当前 `DSH_*` 事实会胜出，而陈旧的嵌套 harness 身份无法从环境中隐式漏入。提供的 stdin 会被写入后关闭；否则 fd 0 指向 `/dev/null`。参见 [stdin/env Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md)与[受管环境 Agent Note](../../../.agents/notes/implemented/feature/2026-07-10-agent-session-identity-and-log-location.md)。
+- **基于偏移量的读取**：收集模式的读取器按完整流的字节坐标返回增量；服务自身从不持有游标，因此消费方自有的游标（bash 的后台读取路径）与完整流重读可以共存，结算前后皆然。
+- **先终止再等待退出的 dispose（资源释放）**：服务保留存活句柄，只为让自身的 dispose 能对每个仍在运行的进程树执行升级并等待其退出；已结算与 spawn 失败的句柄在结算时即离开存活集合。
 
 ## 模型体验
 
@@ -18,12 +18,12 @@
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由具名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
 - **Windows 进程树支持仅为尽力而为，且未经 CI 测试**：终止经由 `taskkill /PID <pid> /T /F` 完成，所有结果都被就地吸收，不向外抛出（进程树已不存在、竞态、二进制缺失），存活探测则回退到直接子进程边界；测试套件只通过注入的运行器覆盖这条路由，且 `packages/subprocess/*` 被排除在 Windows 测试矩阵之外。
-- **凭据清除依赖名称启发式规则**：只匹配 `*KEY*`／`*SECRET*`／`*TOKEN*`；名称不同的 secret（例如 `*PASSWORD*`）会继续传递，对误删变量引入白名单属于已记录的后续工作。
+- **凭据清除依赖名称启发式规则**：只匹配 `*KEY*`／`*PASSWORD*`／`*SECRET*`／`*TOKEN*`；名称不同的 secret（例如 `*PASSPHRASE*`）会继续传递，对误删变量引入白名单属于已记录的后续工作。
 - **不会删除已完成的 spill 文件**：有界的完整输出恢复文件（以及每个进程的私有 spill 目录）会在 OS tmpdir 下累积，直到外部机制进行清理；超大的不完整 spill 会被丢弃并立即尝试删除，但清理失败可能留下一个有界文件。
 
 原始进程处理位于 `src/spawn.ts`；`src/index.ts` 负责服务接线。
diff --git a/packages/subprocess/subprocess-local/tests/spawn.spec.ts b/packages/subprocess/subprocess-local/tests/spawn.spec.ts
index c02f6b0d55..491756f01f 100644
--- a/packages/subprocess/subprocess-local/tests/spawn.spec.ts
+++ b/packages/subprocess/subprocess-local/tests/spawn.spec.ts
@@ -255,10 +255,10 @@ describe('stdin and extra env (set by in-process plugins)', () => {
   })
 
   it('an explicit extra env entry overrides the credential scrub', async () => {
-    // EXPLICIT_OVERRIDE_KEY matches the credential scrub pattern, yet an explicit
+    // EXPLICIT_OVERRIDE_PASSWORD matches the credential scrub pattern, yet an explicit
     // entry is still honored — the scrub only drops AMBIENT process.env creds.
-    const result = await finish(spawnSubprocess(spec('echo "$EXPLICIT_OVERRIDE_KEY"', {
-      env: { EXPLICIT_OVERRIDE_KEY: 'explicit-wins' },
+    const result = await finish(spawnSubprocess(spec('echo "$EXPLICIT_OVERRIDE_PASSWORD"', {
+      env: { EXPLICIT_OVERRIDE_PASSWORD: 'explicit-wins' },
     })))
     expect(result.stdout.text).toBe('explicit-wins\n')
   })
@@ -748,13 +748,17 @@ describe('environment and spill-file hardening', () => {
   it('scrubs credential-shaped and ambient DSH env vars from child processes', async () => {
     process.env.DSH_TEST_API_KEY = 'super-secret'
     process.env.DSH_TEST_TOKEN = 'also-secret'
+    process.env.SUBPROCESS_TEST_PASSWORD = 'password-secret'
     process.env.DSH_TEST_PLAIN = 'visible'
     try {
-      const result = await finish(spawnSubprocess(spec('echo "[${DSH_TEST_API_KEY:-absent}|${DSH_TEST_TOKEN:-absent}|${DSH_TEST_PLAIN:-absent}]"')))
-      expect(result.stdout.text.trim()).toBe('[absent|absent|absent]')
+      const result = await finish(spawnSubprocess(spec(
+        'echo "[${DSH_TEST_API_KEY:-absent}|${DSH_TEST_TOKEN:-absent}|${SUBPROCESS_TEST_PASSWORD:-absent}|${DSH_TEST_PLAIN:-absent}]"',
+      )))
+      expect(result.stdout.text.trim()).toBe('[absent|absent|absent|absent]')
     } finally {
       delete process.env.DSH_TEST_API_KEY
       delete process.env.DSH_TEST_TOKEN
+      delete process.env.SUBPROCESS_TEST_PASSWORD
       delete process.env.DSH_TEST_PLAIN
     }
   })
diff --git a/packages/subprocess/subprocess/README.i18n.yaml b/packages/subprocess/subprocess/README.i18n.yaml
index 121d2fec86..64f64d65ed 100644
--- a/packages/subprocess/subprocess/README.i18n.yaml
+++ b/packages/subprocess/subprocess/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/subprocess/subprocess/README.md
 README.md: c360437bf2b2b95734f55f6aec46b0cecffb9260
-README.zh.md: 888b8863cc1dfb0828a6d0c8c59a48790cb1522e
+README.zh.md: dac459a6ed1b92c2354bf0a2cc4e0c23e824154f
diff --git a/packages/subprocess/subprocess/README.zh.md b/packages/subprocess/subprocess/README.zh.md
index 888b8863cc..dac459a6ed 100644
--- a/packages/subprocess/subprocess/README.zh.md
+++ b/packages/subprocess/subprocess/README.zh.md
@@ -2,28 +2,28 @@
 
 [English](README.md) | 中文
 
-进程管理器 seam（`ctx.subprocess`）。抽象的 `SubprocessService` 只暴露一个方法：`spawn(spec): SubprocessHandle`，外加所有消费方共享的词汇：完全显式的 `SubprocessSpawnSpec`、携带基于偏移量的非消费式输出读取器的 `SubprocessHandle`、`SubprocessOutcome`、`CollectedOutput`，以及受管的 `DSH_*` 环境命名空间（`DSH_ENV_PREFIX`、`DshEnvironment`）。本地实现位于 [`dsh-subprocess-local`](../subprocess-local/README.md)。
+子进程 seam（`ctx.subprocess`）。抽象的 `SubprocessService` 只暴露一个方法：`spawn(spec): SubprocessHandle`，外加所有消费方共享的词汇：完全显式的 `SubprocessSpawnSpec`、携带基于偏移量的非消费式输出读取器的 `SubprocessHandle`、`SubprocessOutcome`、`CollectedOutput`，以及受管的 `DSH_*` 环境命名空间（`DSH_ENV_PREFIX`、`DshEnvironment`）。本地实现位于 [`dsh-subprocess-local`](../subprocess-local/README.md)。
 
 ## 契约
 
-- `spawn(spec)` 立即返回一个实时句柄；`done` 在进程关闭时以退出事实 resolve（`SubprocessOutcome` 不携带输出，也不携带原因分类），仅在 spawn 层面失败时 reject。
-- spec 完全显式（argv、cwd、按流划分的 stdio 处置方式（disposition）、宽限期），因为随部署变化的默认值属于调用方 seam 的配置，而不属于某个隐藏的进程管理器默认值（`dsh-bash` 的 request/spec 拆分是这条规则的所属模板）。`argv` 绝不经过 shell 解释；需要 shell 的消费方自行传入 `['bash', '-c', command]`。
-- stdio 按流采用 Node 形状：`'pipe'` 把原始流交给调用方做自己的协议分帧（LSP 的 JSON-RPC、ACP（Agent Client Protocol）的 ndjson），`'inherit'` 直通父进程描述符以承载诊断输出，收集模式（collect）`{ maxBytes, spill? }` 则缓冲一段有界尾部，外加可选的完整流 spill 文件。收集模式的读取器接受全流字节偏移量且从不消费，因此独立的读取器不会抢走彼此的增量；偏移量滑出内存尾部窗口的读取标记为 `lossy`，并在 spill 文件存在时指向它。收集到的输出在结算后仍可读取。
+- `spawn(spec)` 立即返回一个活动句柄；`done` 在进程关闭时以退出事实 resolve（`SubprocessOutcome` 不携带输出，也不携带原因分类），仅在 spawn 层面失败时 reject。
+- spec 完全显式（argv、cwd、按流划分的 stdio 处置方式（disposition）、宽限期），因为随部署变化的默认值属于调用方 seam 的配置，而不属于某个隐藏的子进程默认值（`dsh-bash` 的 request/spec 拆分是这条规则的所属模板）。`argv` 绝不经过 shell 解释；需要 shell 的消费方自行传入 `['bash', '-c', command]`。
+- stdio 按流采用 Node 风格：`'pipe'` 把原始流交给调用方做自己的协议分帧（LSP 的 JSON-RPC、ACP（Agent Client Protocol）的 ndjson），`'inherit'` 直通父进程描述符以承载诊断输出，收集模式（collect）`{ maxBytes, spill? }` 则缓冲一段有界尾部，外加可选的完整流 spill 文件。收集模式的读取器接受全流字节偏移量且从不消费，因此独立的读取器不会抢走彼此的增量；偏移量滑出内存尾部窗口的读取标记为 `lossy`，并在 spill 文件存在时指向它。收集到的输出在结算后仍可读取。
 - 终止在每个平台上都以进程树为范围（POSIX 用 detached 进程组并以直接子进程回退；Windows 用 `taskkill /T`）：`terminate()`（唯一的终止动词）执行 SIGTERM→宽限期→SIGKILL 升级（幂等，也由 spec 的 abort 信号驱动，进程树消亡后为空操作）；`waitForExit(signal?)` 观察整棵进程树的存活状态，使消费方自有的拆卸阶梯能在真正完全停稳后才进入下一层。管理器只响应中止，但绝不判定原因（deadline、拆卸阶梯与原因分类归调用方所有）。
-- `scrubbedParentEnv()` / `SENSITIVE_ENV_PATTERN` 是唯一一份共享的凭据清除定义：环境中形似凭据的名称与 `DSH_*` 名称都会被丢弃，spec 的显式 `env` 在清除之后合并且不做命名空间校验——有意转发的凭据或当前 `DSH_*` 事实之所以能保留下来，正因为它是调用方的显式选择，而陈旧的同名环境值永远到不了子进程。无法把 spawn 路由到该服务的调用点（node-pty 后端、由 SDK 管理的传输层）改为导入凭据清除函数。
+- `scrubbedParentEnv()` / `SENSITIVE_ENV_PATTERN` 是唯一一份共享的环境清理定义：环境中形似凭据的名称与 `DSH_*` 名称都会被丢弃，spec 的显式 `env` 在清除之后合并且不做命名空间校验——有意转发的凭据或当前 `DSH_*` 事实之所以能保留下来，正因为它是调用方的显式选择，而陈旧的同名环境值永远到不了子进程。无法把 spawn 路由到该服务的进程启动方（node-pty 后端、由 SDK 管理的传输层）改为导入环境清理函数。
 - 服务自身的 dispose（资源释放）会终止所有仍在运行的受管进程并等待其退出。
 
-参见[进程管理器数据结构目录](../../../docs/core-data-structures/subprocess.md)与 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-26-subprocess-seam.md)。
+参见[子进程数据结构目录](../../../docs/core-data-structures/subprocess.md)与[seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-26-subprocess-seam.md)。
 
 ## 模型体验
 
-通过消费方 seam 间接影响（目前是 `dsh-tool-bash` 背后的 bash 执行器家族）；进程输出与生命周期面向模型的全部渲染归消费方所有。
+通过消费方 seam 间接影响（目前是 `dsh-tool-bash` 背后的 bash 执行器家族）；进程输出和生命周期的全部面向模型渲染均由消费方负责。
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由具名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **node-pty 与由 SDK 管理的 spawn 只共享凭据清除**：PTY 后端的终端 fork 与 MCP SDK 自己的 stdio 传输层无法把 spawn 路由到这道 seam（fork/spawn 调用归库所有）；它们改为导入 `scrubbedParentEnv`，使环境策略保持单一来源。
-- **拆卸阶梯归消费方所有**：该 seam 只提供信号动词与进程树存活等待，不提供现成的停稳序列；每个进程外消费方自行编码其子进程的配合形状（ACP 后端以 stdin EOF 打头的阶梯是仓库内模板）。
+- **node-pty 与由 SDK 管理的 spawn 只共享环境清理**：PTY 后端的终端 fork 与 MCP SDK 自己的 stdio 传输层无法把 spawn 路由到这道 seam（fork/spawn 调用归库所有）；它们改为导入 `scrubbedParentEnv`，使环境策略保持单一来源。
+- **拆卸阶梯归消费方所有**：该 seam 只提供信号动词与进程树存活等待，不提供现成的停稳序列；每个进程外消费方自行编码其子进程的配合方式（ACP 后端以 stdin EOF 打头的阶梯是仓库内模板）。
diff --git a/packages/subprocess/subprocess/src/index.ts b/packages/subprocess/subprocess/src/index.ts
index d52433175b..5d0bbe2c76 100644
--- a/packages/subprocess/subprocess/src/index.ts
+++ b/packages/subprocess/subprocess/src/index.ts
@@ -37,7 +37,7 @@ export type {
  * deliberately supplied entry survives because explicit env layers merge
  * after the scrub.
  */
-export const SENSITIVE_ENV_PATTERN = /KEY|SECRET|TOKEN/i
+export const SENSITIVE_ENV_PATTERN = /KEY|PASSWORD|SECRET|TOKEN/i
 
 /**
  * The ambient parent environment minus credential-shaped names and minus all
diff --git a/packages/subprocess/subprocess/tests/service.spec.ts b/packages/subprocess/subprocess/tests/service.spec.ts
index 46cddaba63..28a1dce299 100644
--- a/packages/subprocess/subprocess/tests/service.spec.ts
+++ b/packages/subprocess/subprocess/tests/service.spec.ts
@@ -55,16 +55,19 @@ describe('SubprocessService seam', () => {
   it('scrubbedParentEnv drops credential-shaped and DSH_ names but keeps PATH', () => {
     process.env.DSH_SCRUB_PROBE = 'stale'
     process.env.SCRUB_PROBE_TOKEN = 'secret'
+    process.env.SCRUB_PROBE_PASSWORD = 'secret'
     process.env.SCRUB_PROBE_PLAIN = 'visible'
     try {
       const env = scrubbedParentEnv()
       expect(env.DSH_SCRUB_PROBE).toBeUndefined()
       expect(env.SCRUB_PROBE_TOKEN).toBeUndefined()
+      expect(env.SCRUB_PROBE_PASSWORD).toBeUndefined()
       expect(env.SCRUB_PROBE_PLAIN).toBe('visible')
       expect(env.PATH).toBeDefined()
     } finally {
       delete process.env.DSH_SCRUB_PROBE
       delete process.env.SCRUB_PROBE_TOKEN
+      delete process.env.SCRUB_PROBE_PASSWORD
       delete process.env.SCRUB_PROBE_PLAIN
     }
   })
diff --git a/packages/support/README.i18n.yaml b/packages/support/README.i18n.yaml
index 4870773e06..6adfa91c58 100644
--- a/packages/support/README.i18n.yaml
+++ b/packages/support/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/support/README.md
 README.md: b9550fd54feb36448227faae8485fe8b6dbf4fb0
-README.zh.md: dc017c5f6544e29533f14600deb5b9261e351536
+README.zh.md: 8eec9b96f621724e1ddc9a70abed069f11cb7fdc
diff --git a/packages/support/README.zh.md b/packages/support/README.zh.md
index dc017c5f65..8eec9b96f6 100644
--- a/packages/support/README.zh.md
+++ b/packages/support/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-这些包用于开发、测试和示例，而非作为产品 API 发布。它们是真实工作区包（有类型、经过测试、受覆盖率门禁约束），但具有**较低的兼容性预期**：当其背后的开发需求变化时，它们可以改变或被移除，无需像产品包那样谨慎执行弃用流程。
+这些包（package）用于开发、测试和示例，而非作为产品 API 发布。它们是实际的工作区包（具备类型、经过测试，并受覆盖率门禁约束），但具有**较低的兼容性预期**：当其背后的开发需求变化时，它们可以改变或被移除，无需像产品包那样谨慎执行弃用流程。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
-| `acp-snapshot/` | ACP 测试工具包：共享子进程/客户端启动器、快照 harness、规范化器和套件工厂 | （库：由 ACP e2e 和 `*.snapshot.ts` 套件导入） |
-| `agent-loop-testkit/` | 为测试具体 agent loop 的测试挂载共享先决条件 | （库：由 AgentLoop 集成测试导入） |
+| `acp-snapshot/` | ACP（Agent Client Protocol）测试工具包：共享子进程/客户端启动器、快照 harness、规范化器和套件工厂 | （库：由 ACP e2e 和 `*.snapshot.ts` 套件导入） |
+| `agent-loop-testkit/` | 为验证具体 agent loop（智能体循环）的测试挂载共享先决条件 | （库：由 AgentLoop 集成测试导入） |
 | `invariants/` | 用于开发诊断的运行时事件契约断言 | （监听 `session/*`、`agent/*`） |
 | `loader-smoke/` | 共享的真实 Loader 子进程 harness，用于无密钥示例冒烟测试 | （库：由示例 e2e 套件导入） |
-| `llm-mock-server/` | 可编脚本的 OpenAI 兼容 HTTP/SSE 故障服务器与 CLI，用于 LLM 恢复测试 | （独立服务器和测试库） |
-| `llm-replay/` | 录制/回放适配器：从已记录的会话 JSONL 短路 `llm/stream`（无密钥快照测试） | （监听 `llm/stream`） |
+| `llm-mock-server/` | 可编程的 OpenAI 兼容 HTTP/SSE（Server-Sent Events）故障服务器与 CLI（命令行界面），用于 LLM（大语言模型）恢复测试 | （独立服务器和测试库） |
+| `llm-replay/` | 录制/回放适配器：通过已记录的会话 JSONL 对 `llm/stream` 进行短路处理（无密钥快照测试） | （监听 `llm/stream`） |
 
-`invariants` 是开发支持，但没有环境保护：无论在何处注册，它都会运行；默认 `dsh-agent-spine-demo` bundle 无条件挂载它。`agent-loop-testkit` 为手工构建的 AgentLoop 测试集中管理必需服务主干，而不负责其 loop 或场景。`llm-replay` 支撑演示和受每文件覆盖率门禁约束的快照测试层，`llm-mock-server` 则通过确定性 HTTP/SSE 故障驱动真实提供方适配器。`acp-snapshot` 包含 ACP 子进程/客户端边界以及快照 harness、规范化器和套件机制，`loader-smoke` 负责无密钥示例 e2e 套件使用的并列真实 Loader 启动边界。只有当某个包获得已记录的产品消费方时，它才会从 `support/` 升级到产品分组。
+`invariants` 是开发支持，但没有环境条件限制：无论在何处注册，它都会运行；默认 `dsh-agent-spine-demo` bundle 无条件挂载它。`agent-loop-testkit` 为手工构建的 AgentLoop 测试集中管理必需服务主干，而不负责其 agent loop 或场景。`llm-replay` 支撑演示和受逐文件覆盖率门禁约束的快照测试层，`llm-mock-server` 则通过确定性 HTTP/SSE 故障驱动真实提供方适配器。`acp-snapshot` 包含 ACP 子进程/客户端边界以及快照 harness、规范化器和套件机制，`loader-smoke` 负责无密钥示例 e2e 套件使用的并列真实 Loader 启动边界。只有当某个包获得文档记载的产品消费方时，它才会从 `support/` 转入产品分组。
diff --git a/packages/support/agent-loop-testkit/README.i18n.yaml b/packages/support/agent-loop-testkit/README.i18n.yaml
index d919a89cf5..2236b6aa8e 100644
--- a/packages/support/agent-loop-testkit/README.i18n.yaml
+++ b/packages/support/agent-loop-testkit/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/support/agent-loop-testkit/README.md
 README.md: 18c46069d3cfd402c83b5ecab68458667738163b
-README.zh.md: 93af58a279558007c252ef0734cc95b2ad79a5a5
+README.zh.md: 2151591154899094fe33f9c25c9b9144e6d02a22
diff --git a/packages/support/agent-loop-testkit/README.zh.md b/packages/support/agent-loop-testkit/README.zh.md
index 93af58a279..2151591154 100644
--- a/packages/support/agent-loop-testkit/README.zh.md
+++ b/packages/support/agent-loop-testkit/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-为测试具体 `AgentLoop` 的测试挂载共享先决条件。`mountAgentLoopTestDependencies(ctx, options?)` 按依赖顺序安装 LLM、会话、系统提示词、工具和 agent 服务，然后在 loop 挂载前返回。
+为运行具体 `AgentLoop` 的测试共享挂载先决依赖。`mountAgentLoopTestDependencies(ctx, options?)` 按依赖顺序安装 LLM（大语言模型）、会话、系统提示词、工具和 agent（智能体）服务，然后在 agent loop 挂载前返回。
 
-调用方注册适配器和可选插件，使用待测配置挂载 `AgentLoop`，并 dispose 自己的 Context。系统提示词和工具注册表配置可通过 `options` 转发；辅助工具不提供超出服务自有默认值的测试默认值。插件加载失败会拒绝辅助工具调用，而顺序中较早激活的服务仍归调用方的 Context 所有。
+调用方注册适配器和可选插件，使用待测配置挂载 `AgentLoop`，并 dispose（资源释放）自己的 Context。系统提示词和工具注册表配置可通过 `options` 转发；该辅助函数不提供超出服务自有默认值的测试默认值。插件加载失败会使辅助函数调用被拒绝，而顺序中较早激活的服务仍归调用方的 Context 所有。
 
 ```ts
 import { Context } from 'cordis'
@@ -18,16 +18,16 @@ await mountAgentLoopTestDependencies(ctx)
 await ctx.plugin(AgentLoop, { agents: [] })
 ```
 
-针对注入失败、部分拓扑、服务加载顺序或服务拆卸的测试会直接挂载其依赖，而不使用此辅助工具。
+针对注入失败、部分拓扑、服务加载顺序或服务清理的测试会直接挂载其依赖，而不使用此辅助函数。
 
 ## 模型体验
 
 无。该测试专用组合辅助工具既不驱动也不修改模型请求。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 无；该包既不组装也不发送提供方请求。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **只共享必需的先决主干**：适配器、可选插件、`AgentLoop`、agent 和 Context 拆卸仍由调用方负责，以使场景专用顺序保持可见。
+- **只共享必需的先决主干**：适配器、可选插件、`AgentLoop`、agent 和 Context 清理仍由调用方负责，以使特定场景的挂载顺序清晰可见。
diff --git a/packages/support/invariants/README.i18n.yaml b/packages/support/invariants/README.i18n.yaml
index 4cf92d6542..b51a6af3ee 100644
--- a/packages/support/invariants/README.i18n.yaml
+++ b/packages/support/invariants/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/support/invariants/README.md
 README.md: 203dbd5ad09f5b1378061fbf9adcff885889eae2
-README.zh.md: e101a30046c6b1ac18870c268ea6a2c960cb5d90
+README.zh.md: 841e7a32f8e2605c25aadef35862a0223046933f
diff --git a/packages/support/invariants/README.zh.md b/packages/support/invariants/README.zh.md
index e101a30046..841e7a32f8 100644
--- a/packages/support/invariants/README.zh.md
+++ b/packages/support/invariants/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-用于包自有运行时不变量检查的可配置注册表服务。根插件注册 `ctx.invariants`；它不包含产品检查或产品包导入。每个工作区包都发布一个 `./invariant` 配套入口，用于注册其精确 NPM 包名。
+用于包（package）自有运行时不变量检查的可配置注册表服务。根插件注册 `ctx.invariants`；它不包含产品检查或产品包导入。每个工作区包都发布一个 `./invariant` 配套入口，用于注册其精确 NPM 包名。
 
 ## 服务：`InvariantService`（`ctx.invariants`）
 
@@ -16,19 +16,19 @@ interface Config {
 
 默认值为 `enabled: true`、`package_allowlist: []` 和 `package_blocklist: []`。只有在服务启用、allowlist 为空或至少一个 allowlist pattern 匹配完整 NPM 名称，且没有 blocklist pattern 匹配时，包才被选中。因此，blocklist 匹配优先于 allowlist 匹配。
 
-每个条目都是区分大小写的 JavaScript 正则表达式源，使用 `new RegExp(pattern)` 编译。除非源提供 `^` 和 `$`，否则匹配不锚定；不解析 `/pattern/flags` 语法。同一列表中的空白、带前后空白、无效或重复条目会使服务启动失败。有效 pattern 可以不匹配任何当前已加载包，以使后续加载和 HMR 保持确定性。
+每个条目都是区分大小写的 JavaScript 正则表达式源，使用 `new RegExp(pattern)` 编译。除非源提供 `^` 和 `$`，否则匹配不锚定；不解析 `/pattern/flags` 语法。同一列表中的空白、带前后空白、无效或重复条目会使服务启动失败。有效 pattern 可以不匹配任何当前已加载包，以使后续加载和 HMR（热模块替换）保持确定性。
 
-`ctx.invariants.register(packageName, installer)` 为完整 NPM 包名保留一个活动注册，即使过滤器使其 installer 保持非活动，并返回 disposer。已启用贡献在专用子 Cordis fiber 中运行。Installer 可以通过 `installer.inject` 声明所需服务接口，并收到 `fail(message)`；后者抛出绑定到注册包的 `InvariantError`。在注册成功前，系统会等待同步或异步 installer 完成；失败会 dispose 子级，并原子释放归属。
+`ctx.invariants.register(packageName, installer)` 为完整 NPM 包名保留一个活动注册，即使过滤器使其 installer 保持非活动，并返回 disposer。已启用贡献在专用子 Cordis fiber 中运行。installer 可以通过 `installer.inject` 声明所需服务接口，并收到 `fail(message)`；后者抛出绑定到注册包的 `InvariantError`。在注册成功前，系统会等待同步或异步 installer 完成；失败会原子地 dispose（资源释放）子级并释放归属。
 
 服务拥有每个注册 fiber，返回的 disposer 同时属于配套 fiber。卸载任一侧都会移除监听器、跟踪状态和保留。因此，配套入口可以重新加载并注册同一包名，而不保留旧状态。由会话支撑的配套入口从持久事件重建 baseline；仅实时配套入口观察重新加载后开始的操作。
 
 `InvariantError` 扩展 `Error`，携带稳定 `code: 'INVARIANT'`，并公开所属 `packageName`，而不向服务添加产品依赖。
 
-在每个组合中，Session 自身负责不可变且接口有效的日志存储：它对每个候选项制作一份无损 JSON 快照，验证完整来源和位置替换，将 `tool/result` 替换限制为一个当前结果的 `content`，深度冻结已接受记录，并通过不可变数组快照公开日志。`dsh-session` 不变量配套入口检查 Session 不负责的其余跨记录规则。
+在每个组合中，Session 自身负责不可变且通过对外接口校验的日志存储：它对每个候选项制作一份无损 JSON 快照，验证完整来源和位置替换，将 `tool/result` 替换限制为一个当前结果的 `content`，深度冻结已接受记录，并通过不可变数组快照公开日志。`dsh-session` 不变量配套入口检查 Session 不负责的其余跨记录规则。
 
 ## 包配套入口
 
-发布和注册覆盖全部包；运行时断言刻意不使用合成内容。只有当包拥有可观察事件关系或相关可变数据关系时，配套入口才安装检查。确认必需方法、插件名称、注入、effect 或固定纯函数结果属于类型、加载或单元测试关注点，而非运行时不变量。
+发布和注册覆盖全部包；但不会为了覆盖全部包而人为编造运行时断言。只有当包拥有可观察事件关系或相关可变数据关系时，配套入口才安装检查。确认必需方法、插件名称、注入、effect 或固定纯函数结果属于类型、加载或单元测试关注点，而非运行时不变量。
 
 如果不存在合理的运行时关系，配套入口使用空 installer，并以包专用的前置 `No runtime invariant:` 注释说明原因。纯工具、行为已通过 seam 观察的薄实现、仅组合包、二进制程序、契约需要崩溃/往返测试的持久化适配器和测试支持包通常属于此类。当 owner 获得可变状态或事件协议时，必须重新审视该说明。
 
@@ -36,18 +36,18 @@ interface Config {
 
 | 配套入口 | 检查 |
 |---|---|
-| `dsh-session`, `dsh-agent`, `dsh-scope`, `dsh-agent-loop` | 会话包含关系和调用/结果跟踪、agent 状态转换、inbox FIFO 守恒、作用域 subject 和模型请求重建。 |
+| `dsh-session`, `dsh-agent`, `dsh-scope`, `dsh-agent-loop` | 会话包含关系和调用/结果跟踪、agent（智能体）状态转换、inbox FIFO 守恒、作用域 subject 和模型请求重建。 |
 | `dsh-llm`, `dsh-llm-retry`, `dsh-tools`, `dsh-system-prompt` | 流语法、持久重试位置和边界、工具流水线阶段与冻结结果，以及权威提示词组装数据。 |
-| `dsh-compact`, `dsh-hook-protocol`, `dsh-sandbox-policy` | 持久压缩与钩子配对、压缩元数据和沙箱 mode 词汇。 |
+| `dsh-compact`, `dsh-hook-protocol`, `dsh-sandbox-policy` | 持久压缩（compaction）与钩子配对、压缩元数据和沙箱 mode 词汇。 |
 | `dsh-fs`, `dsh-subagent`, `dsh-workflow` | 文件系统事件身份、提供方/子级配对和工作流/agent 生命周期身份。 |
-| `dsh-goal`, `dsh-goal-session` | 持久 goal 来源/内容一致性、修订和生命周期转换、时间戳、顺序接纳 Round 和重建的继续提示词。 |
+| `dsh-goal`, `dsh-goal-session` | 持久 goal 来源/内容一致性、修订和生命周期转换、时间戳、依次获准的 Round 和重建的继续提示词。 |
 | `dsh-permission`, `dsh-user-approval` | 活动 preset 引用和审批询问/决定审计配对。 |
 | `dsh-tasks`, `dsh-tool-todo` | 任务快照生命周期/归属字段和持久整表 todo 结构。 |
-| `dsh-time-context` | 持久时钟读数与会话开放轮次、下一个步骤前位置和已用 baseline 一致；渲染时间可解析，且不晚于其事件。 |
+| `dsh-time-context` | 持久化时钟读数与会话中正在进行的轮次、下一步骤开始前的位置及已用时间 baseline 一致；渲染时间可解析，且不晚于其事件。 |
 
-每个 owner 的根入口仍与诊断独立。单独加载服务不会安装产品检查；在没有服务时加载配套入口，会等待其声明的 `invariants` 注入。
+每个 owner 的根入口仍独立于诊断。单独加载服务不会安装产品检查；在没有服务时加载配套入口，会等待其声明的 `invariants` 注入。
 
-`pnpm run verify-package-invariants` 发现全部工作区包。它拒绝生成标记、未说明的空 installer、省略或忽略 reporter 的非空 installer、错误注册名称，以及不完整的导出、发布、依赖、TypeScript 引用或 bundle 接线。该源规则是最低归属检查；聚焦测试证明每个可执行配套入口的语义。
+`pnpm run verify-package-invariants` 发现全部工作区包。它拒绝生成标记、未说明的空 installer、省略或忽略 reporter 的非空 installer、错误注册名称，以及不完整的导出、发布、依赖、TypeScript 引用或 bundle 接线。该源码规则是最低归属检查；聚焦测试证明每个可执行配套入口的语义。
 
 ## 组合
 
@@ -68,18 +68,18 @@ ctx.plugin(SessionInvariant)
 
 标准 agent 主干挂载服务和 4 个核心有状态配套入口。自定义组合为希望检查其契约的其他已加载包显式添加配套入口；过滤器可以在不改变包入口的情况下禁用或选择注册。
 
-每个普通 Vitest 拓扑都挂载显式启用的服务和当前测试包的配套入口。聚焦套件覆盖可执行配套入口的有效和无效观察，一个穷尽拓扑则挂载全部配套入口，以证明注册和 dispose 接线。
+每个普通 Vitest 拓扑都挂载显式启用的服务和当前测试包的配套入口。聚焦套件覆盖可执行配套入口的合法与违规观测，一个穷尽拓扑则挂载全部配套入口，以证明注册和 dispose 接线。
 
 ## 模型体验
 
 无。服务和配套入口观察运行时事件和可变快照，不会更改提示词、消息、schema、流或工具结果。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 无；不变量检查不组装或发送提供方请求。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- 请求重建覆盖 loop 在冻结前显式标记的请求；直接一次性 LLM 调用即使由调用方冻结或附加会话 id，仍不在该标记契约内。
+- 请求重建覆盖 loop 在冻结前显式标记的请求；直接一次性 LLM（大语言模型）调用即使由调用方冻结或附加会话 id，仍不在该标记契约内。
 - 仅实时生命周期配套入口无法重建自身重新加载前开始的操作。标准组合和测试组合会在相应操作开始前挂载它们。
 - 正则表达式过滤器在服务生命周期内固定；更改它们需要执行普通 Cordis 插件重新加载。
diff --git a/packages/support/llm-replay/README.i18n.yaml b/packages/support/llm-replay/README.i18n.yaml
index d0c335ee24..cfd0fb0b63 100644
--- a/packages/support/llm-replay/README.i18n.yaml
+++ b/packages/support/llm-replay/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/support/llm-replay/README.md
 README.md: 6c934e01a5f13b94724d0e5435524d9d463adb2f
-README.zh.md: 03309931d01957ef60aa65d5d9ba3c2a844b0a16
+README.zh.md: d13d84ad95a6b63177ee0e2aa83965be19fdeeb8
diff --git a/packages/support/llm-replay/README.zh.md b/packages/support/llm-replay/README.zh.md
index 03309931d0..d13d84ad95 100644
--- a/packages/support/llm-replay/README.zh.md
+++ b/packages/support/llm-replay/README.zh.md
@@ -2,21 +2,21 @@
 
 [English](README.md) | 中文
 
-用于无密钥快照测试的回放 LLM 插件。它从已记录的**会话 JSONL** fixture 重建模型流，使测试可以在无 API 密钥的情况下使用固定模型 transcript 启动真实 agent。配置 `providers` 后，它会注册仅回放适配器，其目录可供测试模型发现的场景使用；没有 `providers` 时，它会安装不需要发现的测试所用 catch-all `llm/stream` waterfall。
+用于无密钥快照测试的大语言模型（LLM）回放插件。它根据已记录的**会话 JSONL** fixture（测试前置数据）重建模型流，使测试无需 API 密钥即可针对固定的模型 transcript（文本记录）启动真实 agent（智能体）。配置 `providers` 后，它会注册仅用于回放的适配器，其模型目录可供测试模型发现功能的场景使用；未配置 `providers` 时，它会安装无需模型发现功能的测试所用 catch-all `llm/stream` waterfall（瀑布式事件）。
 
-其消费方是 ACP、headless `stream-json` 和 TUI 快照套件，以及 web 浏览器 e2e lane。Loader 驱动套件使用此插件替换真实 LLM 适配器；web lane 直接安装它，以保留拆卸消费句柄。将派生和回放逻辑保留在此处，可使其受 `packages/*/src` 上每文件 100% 覆盖率门禁约束。
+其消费方包括 ACP（Agent Client Protocol）、headless `stream-json` 和 TUI 快照套件，以及 Web 浏览器 e2e 流水线。Loader 驱动的套件使用此插件替代真实 LLM 适配器；Web 流水线直接安装它，以保留清理阶段的消费检查句柄。将派生和回放逻辑放在此处，可使其受 `packages/*/src` 的逐文件 100% 覆盖率门禁约束。
 
-## Fixture 的工作方式
+## fixture 的工作方式
 
-Fixture 就是持久化会话日志（`<scenario>/session.jsonl`）。其 `assistant/chunk` 事件携带每个 `StreamChunk`，因此按 `(turn, step)` 对其分组可重建每次 `stream()` 调用的分片序列（每个 loop 步骤一次模型调用）。因此，录制操作是「运行一次真实 agent 并收集 `.jsonl`」，由快照 harness 完成；该插件不执行录制。Fixture 的 `request/header` 内容可能被 token 化为 `{{system}}`/`{{tools}}`（harness 在一个场景中固定该内容，并擦除其余场景）；回放对此并不关心，因为派生只读取 `assistant/chunk` 事件和第 0 行会话 header。
+fixture 就是持久化的会话日志（`<scenario>/session.jsonl`）。其 `assistant/chunk` 事件包含每个 `StreamChunk`，因此按 `(turn, step)` 分组即可重建每次 `stream()` 调用的分片序列（每个循环步骤调用一次模型）。因此，录制就是「运行一次真实 agent 并收集 `.jsonl`」，由快照 harness 完成；该插件本身不录制。fixture 的 `request/header` 内容可能被标记化为 `{{system}}`/`{{tools}}`（harness 会在一个场景中固定该内容，并清除其余场景中的内容）；回放不受影响，因为派生过程只读取 `assistant/chunk` 事件和第 0 行的会话 header。
 
-有两种失败 mode 无法仅从 `assistant/chunk` 重建：在任何分片前纯抛出（例如 HTTP 401，日志只包含 `turn/end {error}` 而没有分片），以及 cancel/hang（是时序，而非分片内容）。需要这些的场景提供可选 sidecar（`<scenario>/replay.override.json`），它要么替换派生脚本（裸 `ReplayEntry[]`），要么增补派生脚本（`{ patches: [{ at, entry }] }`：保留全部由 JSONL 派生的调用，仅在点名的调用索引处换入，索引从 0 计；`at` 等于派生长度时为追加，正是注入的瞬态抛出之后那次重试尝试所占的槽位）。Patch 索引必须互不重复。覆写文档、每个 patch 与每个条目，以及每个分片的判别字段都会在文件加载时接受校验。`hang` 条目可以指定 `readyFile`；在其前缀分片到达 loop 后、等待取消前，回放会写入该空标记，使外部驱动器可以在不观察展示更新的情况下确定性取消。
+有两种失败模式无法仅根据 `assistant/chunk` 重建：在产生任何分片前直接抛出异常（例如 HTTP 401，此时日志只有 `turn/end {error}` 而没有分片），以及取消或挂起（差异在时序，而非分片内容）。需要这些行为的场景可提供伴随文件（`<scenario>/replay.override.json`）：它可以替换派生脚本（裸 `ReplayEntry[]`），也可以增补派生脚本（`{ patches: [{ at, entry }] }`：保留所有从 JSONL 派生的调用，只替换指定的从 0 开始计数的调用索引；当 `at` 等于派生长度时，则在注入瞬态异常后的重试位置追加一次调用）。补丁索引不得重复。文件加载时会校验覆写文档、每个补丁和条目，以及每个分片的判别标签。`hang` 条目可以指定 `readyFile`；当前缀分片到达循环后、开始等待取消前，回放会写入这个空标记，使外部驱动程序无需观察展示层更新即可确定性地取消。
 
 ## 嵌套 agent：每会话键控
 
-父 agent 委托给进程内 subagent 的场景会记录多个日志：父级（`session.jsonl`）和每个子级各一个（`session.1.jsonl`等）。每个 agent 在同一上下文中作为自己的 `Session` 运行，因此回放必须为每个 agent 提供自己的脚本。
+父 agent 委托给进程内 subagent（子 agent）的场景会记录多个日志：父会话使用 `session.jsonl`，每个子会话各使用一个日志（`session.1.jsonl` 等）。每个 agent 都在同一上下文中作为独立的 `Session` 运行，因此回放必须为每个 agent 提供各自的脚本。
 
-回放按调用会话 id 为每次调用建键（由 agent loop 标记的 `GenerateOptions.sessionId`）。实时会话 id 在每次运行中都是新的随机值，绝不等于已记录值，因此实时会话通过**首次调用顺序** 绑定到已记录脚本：脚本按 header `createdAt` 排序（父级在前，因为它必须先进行流式输出才能委托）；第一个进行任何调用的实时会话领取第一个脚本，下一个新会话领取下一个，以此类推。然后，每个会话推进自己的游标。没有 `sessionId` 的调用是绑定到主脚本的单一匿名会话，因此单会话场景与以前完全相同。实时会话数超过已记录脚本数时快速失败。
+回放根据发起调用的会话 id 为每次调用建立键（`GenerateOptions.sessionId` 由 agent loop 写入）。实时会话 id 每次运行时都会重新随机生成，绝不会等于记录中的 id，因此实时会话按**首次调用顺序**绑定到已记录脚本：脚本按 header 中的 `createdAt` 排序（父会话在前，因为它必须先开始流式输出才能委托）；第一个发起调用的实时会话取得第一个脚本，下一个新会话取得下一个脚本，以此类推。此后每个会话分别推进自己的游标。没有 `sessionId` 的调用视为一个绑定主脚本的匿名会话，因此单会话场景的行为与以前完全相同。不同实时会话的数量超过已记录脚本数时会明确报错。
 
 ## 配置
 
@@ -24,9 +24,9 @@ Fixture 就是持久化会话日志（`<scenario>/session.jsonl`）。其 `assis
 |---|---|---|---|
 | `file` | string | `$DSH_SNAPSHOT_FILE` | 主（父）`session.jsonl` fixture 的路径。必需（配置或 env）。 |
 | `overrideFile` | string | `$DSH_SNAPSHOT_OVERRIDE` | 主会话的可选 `ReplayOverrideDoc` sidecar：裸 `ReplayEntry[]` 替换其派生脚本，`{ patches }` 则按调用索引增补该脚本。 |
-| `childFiles` | string[] | `$DSH_SNAPSHOT_CHILD_FILES` (path-delimited) | 嵌套场景中已记录的 subagent 子会话日志；单会话场景为空。 |
+| `childFiles` | string[] | `$DSH_SNAPSHOT_CHILD_FILES`（以路径分隔符分隔） | 嵌套场景中已记录的 subagent 子会话日志；单会话场景为空。 |
 | `providers` | `ReplayProviderConfig[]` | 无 | 可选的仅回放提供方和模型目录。每个提供方可以设置 `retryPolicy`，每个模型可以发布 `contextWindow`；已配置路由通过回放适配器分派，绝不执行提供方 I/O。 |
-| `paceMs` | number | 无（突发） | 可选的每分片毫秒延迟，使下游传输（例如真实浏览器观察的 web SSE mux）看到真正的增量传递。它只是仿真开关，测试不得依赖它保证正确性。值必须是非负整数；pace 等待期间中止会迅速取消流。 |
+| `paceMs` | number | 无（突发） | 可选的每分片毫秒延迟，使下游传输（例如真实浏览器观察到的 Web SSE（Server-Sent Events）多路复用器）看到真正的增量传递。它只是仿真开关，测试不得依赖它保证正确性。值必须是非负整数；pace 等待期间中止会迅速取消流。 |
 
 ```yaml
 - id: llm-replay
@@ -52,10 +52,10 @@ Fixture 就是持久化会话日志（`<scenario>/session.jsonl`）。其 `assis
 
 ## 导出项
 
-- `installLlmReplay(ctx, config)`：安装已配置回放适配器或 catch-all `llm/stream` 监听器；返回 `ReplayHandle`（包含用于 HMR 安全的 `dispose()`，以及 `assertConsumed()` 拆卸检查；后者确保每个已记录脚本都绑定到实时会话，且每个已绑定游标都已耗尽，从而将场景静默驱动的模型调用少于记录数转换为明确诊断）。在测试中使用它，可以不通过 Loader 或 env var 驱动回放。
-- `loadSessionScripts(config)`：解析场景的有序 `SessionScript[]` （主级 + 子级），准备按首次调用顺序绑定到实时会话。
-- `loadReplayScript(config)`：只解析主会话的 `ReplayEntry[]` （如果存在则使用经校验的 sidecar 替换或 patch，否则从 JSONL 派生；fixture 缺失时快速失败）。
-- `deriveReplayScript(events)` / `parseSessionLog(text)` / `parseSessionHeader(text)`：将已记录会话日志转换为脚本并读取其 header `id`/`createdAt` 的纯辅助工具。派生分组必须以 `finish` 分片结束；没有该分片的分组是已抛出 `stream()` 的指纹，必须改用 override sidecar 表达。
+- `installLlmReplay(ctx, config)`：安装已配置回放适配器或 catch-all `llm/stream` 监听器；返回 `ReplayHandle`（包含用于保证 HMR（热模块替换）安全的 `dispose()`，以及清理阶段执行的 `assertConsumed()` 检查；后者确保每个已记录脚本都绑定到实时会话，且每个已绑定游标都已耗尽，从而将场景静默驱动的模型调用少于记录数转换为明确诊断）。在测试中使用它，可以不通过 Loader 或 env var 驱动回放。
+- `loadSessionScripts(config)`：解析场景的有序的 `SessionScript[]`（主会话 + 子会话），准备按首次调用顺序绑定到实时会话。
+- `loadReplayScript(config)`：只解析主会话的 `ReplayEntry[]`（如果伴随文件存在，则使用经校验的替换或补丁；否则从 JSONL 派生；fixture 缺失时明确报错）。
+- `deriveReplayScript(events)` / `parseSessionLog(text)` / `parseSessionHeader(text)`：将已记录会话日志转换为脚本并读取其 header `id`/`createdAt` 的纯辅助工具。派生分组必须以 `finish` 分片结束；没有该分片的分组是 `stream()` 抛出异常的指纹，必须改用 override sidecar 表达。
 - 类型 `ReplayEntry` / `ReplayOverrideDoc` / `ReplayOverridePatch` / `SessionScript` / `ReplayConfig` / `ReplayProviderConfig` / `ReplayModelConfig` / `ReplayHandle` / `Config`。
 
 ## 插件导出形态
@@ -66,11 +66,11 @@ Fixture 就是持久化会话日志（`<scenario>/session.jsonl`）。其 `assis
 
 无。该无密钥测试适配器不向提供方模型发送请求，只将已记录 assistant 分片回放到测试 loop 中。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-无；该包既不组装也不发送提供方请求。
+无；该包（package）既不组装也不发送提供方请求。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **首次调用顺序脚本绑定假设串行委托**：并发运行同级 subagent 的 cut（或运行中落地的压缩摘要调用）会非确定性地将实时会话绑定到已记录脚本；在这种场景出现前暂不实现更强的键控（`XXX(concurrent-subagents)`）。
-- **只有生产分片的调用可派生**：纯分片前抛出或 cancel/hang 场景需要 `replay.override.json` sidecar。替换和 patch 两种形式都只影响主会话；子会话脚本仍从各自日志派生。
+- **首次调用顺序脚本绑定假设串行委托**：并发运行同级 subagent 的 cut（或运行中发生的上下文压缩（context compaction）摘要调用）会非确定性地将实时会话绑定到已记录脚本；在这种场景出现前暂不实现更强的键控（`XXX(concurrent-subagents)`）。
+- **只有会产生分片的调用才能派生**：在产生分片前直接抛出异常或取消/挂起的场景需要 `replay.override.json` 伴随文件。替换和补丁两种形式都只影响主会话；子会话脚本仍从各自日志派生。
diff --git a/packages/support/loader-smoke/README.i18n.yaml b/packages/support/loader-smoke/README.i18n.yaml
index 4794ac72cb..2c99378710 100644
--- a/packages/support/loader-smoke/README.i18n.yaml
+++ b/packages/support/loader-smoke/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/support/loader-smoke/README.md
 README.md: 73610ce50ebac4c6fc7bb9135f7b41b347c60685
-README.zh.md: 17f8481220136e8edf9fccd23fabfca5ccf41dfc
+README.zh.md: 9254cd1592d4bf3f119d350f04165228db87b1cf
diff --git a/packages/support/loader-smoke/README.zh.md b/packages/support/loader-smoke/README.zh.md
index 17f8481220..9254cd1592 100644
--- a/packages/support/loader-smoke/README.zh.md
+++ b/packages/support/loader-smoke/README.zh.md
@@ -4,20 +4,20 @@
 
 用于测试通过 Cordis Loader 启动应用和 `cordis.yml` 的共享子进程 harness。`resolveExampleLaunch` 选择本地 `src` mode（tsx 和根 tsconfig 路径）或 CI `lib` mode（普通 Node 和包导出）；选择依据为显式 mode 或 `DSH_EXAMPLE_MODE`。
 
-`runLoaderSmoke` 接受 bin 和配置路径、可选的完整 bin 参数、环境覆盖、stdin、运行前设置和清理前检查。它负责隔离 cwd、DSH 主目录、诊断、deadline、终止、EOF 和清理；在零退出后返回两个流，失败时拒绝并携带两个流。
+`runLoaderSmoke` 接受可执行文件路径和配置路径、可选的完整可执行文件参数、环境变量覆盖、标准输入、运行前准备和清理前检查。它负责隔离工作目录、DSH 主目录、诊断、截止时间、终止、EOF 和清理；进程以零状态退出后返回两个流，失败时则返回拒绝并附带两个流。
 
 这是支持层测试基础设施，而非产品 API。
 
 ## 模型体验
 
-无。该测试专用 harness 启动示例进程并检查它们的流，不会改变已组装模型请求。
+无。该测试专用 harness 启动示例进程并检查它们的流，不会改变组装后的模型请求。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
 无；该包既不组装也不发送提供方请求。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
-- **构建 mode 需要事先构建**：配置还必须能够通过 `examples/node_modules` 向上解析每个命名包。
+- **构建模式需要事先构建**：配置还必须能够通过 `examples/node_modules` 向上解析每个命名包。
 - **捕获的 stdout 和 stderr 仅受 execa 默认 100 MB `maxBuffer` 约束**：失控子进程会在该上限处被终止，而不是在冒烟测试自选的预算处。
-- **超时只终止直接子进程**：故障 fixture 生成的进程树可以比冒烟测试存活更久，需要外部清理。
+- **超时只终止直接子进程**：有故障的 fixture（测试前置数据）spawn 的进程树可能比冒烟测试存活更久，需要外部清理。
diff --git a/packages/tasks/README.i18n.yaml b/packages/tasks/README.i18n.yaml
index 79f5e7b8e2..18c81fdaed 100644
--- a/packages/tasks/README.i18n.yaml
+++ b/packages/tasks/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/tasks/README.md
 README.md: 9bafe5633bb7e57a5404ffb41fad04b621832b6d
-README.zh.md: 73c87a2c95ccebf70558a2051149eca4ba41f60e
+README.zh.md: f07990f45a636314da358c4f39a72d70ce51db1a
diff --git a/packages/tasks/README.zh.md b/packages/tasks/README.zh.md
index 73c87a2c95..f07990f45a 100644
--- a/packages/tasks/README.zh.md
+++ b/packages/tasks/README.zh.md
@@ -1,13 +1,13 @@
-# tasks/：后台任务能力包族
+# tasks/：后台任务能力家族
 
 [English](README.md) | 中文
 
-后台 task id、拥有者隔离、读取、取消、等待和完成通知的共用归属位置。Bash、subagent 及未来的长时间运行工具共用一套面向模型的协议。参见[后台任务运行时 Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-06-20-generic-long-running-tool-runtime.md)和[任务注册表 seam Agent Note](../../.agents/notes/implemented/architecture/2026-07-26-task-registry-seam.md)。
+这是后台任务 id、所有者隔离、读取、取消、等待和完成通知的共用归属位置。Bash、subagent（子 agent）及未来的长时间运行工具共用一套面向模型的协议。参见[后台任务运行时 Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-06-20-generic-long-running-tool-runtime.md)和[任务注册表 seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-26-task-registry-seam.md)。
 
 | 包（package） | ctx 键 | 角色 |
 |---|---|---|
-| [`tasks`](tasks/README.md)（`@deepseek-ai/dsh-tasks`） | `ctx.tasks` | 注册表 seam：品牌化 `<kind>-N` id、按拥有者设防的 read／kill／wait／list 契约、快照词汇、防止 `attachSurface` 配置错误的防线，以及快照不变式配套插件 |
-| [`tasks-local`](tasks-local/README.md)（`@deepseek-ai/dsh-tasks-local`） | 无 | 进程局部的注册表实现：内存记录、首次结果优先的结算簿记，以及等待完成的拥有者清理与拆卸路径 |
+| [`tasks`](tasks/README.md)（`@deepseek-ai/dsh-tasks`） | `ctx.tasks` | 注册表 seam：带品牌类型的 `<kind>-N` id、按所有者隔离的读取／终止／等待／列出契约、快照词汇、`attachSurface` 配置错误防线，以及快照不变式配套项 |
+| [`tasks-local`](tasks-local/README.md)（`@deepseek-ai/dsh-tasks-local`） | 无 | 进程内注册表实现：内存记录、以首次结算为准的簿记，以及会等待执行完毕的所有者清理路径和销毁路径 |
 | [`tool-tasks`](tool-tasks/README.md)（`@deepseek-ai/dsh-tool-tasks`） | 无 | 面向模型的控制接口：`task_output`、`task_list`、`task_kill`、完成通知注入和后台工作习惯提示词段落 |
 
-注册表拥有跨生产方或接口重载的状态；工具包拥有呈现。生产方通过 `ctx.tasks.start` 注册执行钩子，并自行决定其配置是否公开 `run_in_background`。
+生产方或接口重载时，状态仍由注册表持有；工具包负责呈现。生产方通过 `ctx.tasks.start` 注册执行钩子，并自行决定其配置是否公开 `run_in_background`。
diff --git a/packages/tasks/tasks-local/README.i18n.yaml b/packages/tasks/tasks-local/README.i18n.yaml
index 532331c5be..7dbd6dd7e4 100644
--- a/packages/tasks/tasks-local/README.i18n.yaml
+++ b/packages/tasks/tasks-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/tasks/tasks-local/README.md
 README.md: 23ca6fca61ccb59c855e5d6da6b0a2e23e7cb632
-README.zh.md: c5553a76690278f5b6d5ec40a55d213ef7e1e2d9
+README.zh.md: 5b04caaf640042049e4ef9e6fa58a30adc6f8f72
diff --git a/packages/tasks/tasks-local/README.zh.md b/packages/tasks/tasks-local/README.zh.md
index c5553a7669..5b04caaf64 100644
--- a/packages/tasks/tasks-local/README.zh.md
+++ b/packages/tasks/tasks-local/README.zh.md
@@ -2,25 +2,25 @@
 
 [English](README.md) | 中文
 
-[`@deepseek-ai/dsh-tasks`](../tasks/README.md) 注册表 seam 的进程局部实现：`LocalTaskService` 把每条记录保存在内存中，按 kind 签发 `<kind>-N` id，并且只交出全新快照，从不交出实时状态。它没有配置；作为插件加载后即注册为 `ctx.tasks`。
+[`@deepseek-ai/dsh-tasks`](../tasks/README.md) 注册表 seam 的进程本地实现：`LocalTaskService` 把每条记录保存在内存中，按 kind 签发 `<kind>-N` id，并且只交出全新快照，从不交出实时状态。它没有配置；作为插件加载后即注册为 `ctx.tasks`。
 
 ## 生命周期
 
-任务属于其 owner 和后端，而不是生产方工具 fiber，因此重载生产方或表层不会停止任务。某个 owner 的第一个任务会把一个受等待的 effect 附加到精确的 `Agent` scope。owner 释放会取消该对象的任务，等待生产方完全停稳，并移除其快照；复用 agent 或 Session id 无法重定向旧清理。
+任务属于其所有者和后端，而不是生产方工具 fiber，因此重载生产方或接口不会停止任务。某个所有者的第一个任务会把一个会被等待的 effect 附加到对应 `Agent` 对象的 scope 上。所有者的 dispose（资源释放）会取消该对象的任务，等待生产方完全停稳，并移除其快照；复用的 agent（智能体）id 或会话 id 无法重定向旧的清理操作。
 
-服务释放会关闭监听器、取消所有存活任务、等待其记录，并从仍存活的 owner scope 分离 effect。如果拆卸取消抛出异常，服务会强制把记录标为失败，并警告工作可能遗留，而不会死锁。取消已返回但始终不终止 `done` 时，系统无法将其与缓慢停止区分开，拆卸可能因此停滞。
+服务 dispose 会关闭监听器、取消所有存活任务、等待其记录完成，并从仍存活的所有者 scope 中分离 effect。如果销毁期间的取消操作抛出异常，服务会强制将记录标为失败，并警告工作可能成为孤立工作，而不会死锁。取消操作已返回但 `done` 始终未结算时，系统无法将其与缓慢停止区分开，销毁过程可能因此停滞。
 
-结算遵循首次结果优先：最早出现的终止结果（生产方结算、被隔离为 `failed` 的 `done` 拒绝，或拆卸强制失败）只记录一次，只通知监听器一次并对每个监听器单独隔离故障，然后释放等待方。挂起的等待会在监听器运行前把任务标记为已报告，因此呈现完成情况的表层不会重复发出通知。
+结算遵循首次结算优先原则：最早出现的终止结果（生产方结算、作为 `failed` 隔离处理的 `done` 拒绝，或销毁时的强制失败）只记录一次，也只通知监听器一次；各监听器的故障会单独隔离，随后释放等待方。挂起的等待会在监听器运行前把任务标记为已报告，因此呈现完成情况的表层不会重复发出通知。
 
 ## 模型体验
 
-通过生产方插件和 [`dsh-tool-tasks`](../tool-tasks/README.md) 间接影响；它们会渲染 task id、输出、状态、取消和完成通知。
+通过生产方插件和 [`dsh-tool-tasks`](../tool-tasks/README.md) 间接影响；它们会呈现任务 id、输出、状态、取消和完成通知。
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **任务只存在于进程本地**：记录随 harness 进程一起消亡；持久或跨重启执行需要一个单独实现该 seam 的后端。
-- **静默无效的取消可能使拆卸停滞**：只有显式抛出异常才能安全地强制标为失败。
+- **任务只存在于进程本地**：记录会随 harness 进程终止而消失；持久或跨重启执行需要一个单独实现该 seam 的后端。
+- **静默无效的取消可能使销毁过程停滞**：只有显式抛出异常才能安全地强制标为失败。
diff --git a/packages/tasks/tasks/README.i18n.yaml b/packages/tasks/tasks/README.i18n.yaml
index b86c63e859..97dc186a8e 100644
--- a/packages/tasks/tasks/README.i18n.yaml
+++ b/packages/tasks/tasks/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/tasks/tasks/README.md
 README.md: 2f822bad139020f0ebae0165aa4e8893853f635d
-README.zh.md: 4adb249f31241d5c61c3f8cbee638e8243e4a92e
+README.zh.md: 19f77347896fc9da631c819ea4086cc24ad5f1a8
diff --git a/packages/tasks/tasks/README.zh.md b/packages/tasks/tasks/README.zh.md
index 4adb249f31..19f7734789 100644
--- a/packages/tasks/tasks/README.zh.md
+++ b/packages/tasks/tasks/README.zh.md
@@ -6,21 +6,21 @@
 
 ## 服务契约
 
-- `start(spec): TaskId` 验证控制表层、spec、精确的存活 owner，以及可选的正 `outputLimitBytes`，然后只调用生产方的 `run()` 一次。启动方抛出异常时不注册任何内容；成功返回会直接提交，不再执行其他可能失败的步骤。
+- `start(spec): TaskId` 验证控制表层、spec、确切且仍存活的 owner，以及可选的 `outputLimitBytes`（如提供则须为正数），然后只调用生产方的 `run()` 一次。启动方抛出异常时不注册任何内容；成功返回会直接提交，不再执行其他可能失败的步骤。
 - `get(id, caller?)` 和 `list(caller?)` 返回非消费式快照。列表只包含调用方拥有及无 owner 的任务。
 - `read(id, caller?)` 消费流任务的唯一游标；对于最终输出任务，则以幂等方式读取终止输出。
 - `kill(id, caller?, reason?)` 在更改状态前调用生产方取消。取消抛出异常时任务保持运行；成功则把状态改为 `stopping`，并将终止交付标记为已报告。
 - `wait(id, timeoutMs, caller?, signal?)` 返回终止快照，或在超时时返回存活快照。中止只会停止等待；一旦终止交付已向该等待方提交，终止结果优先。
-- `onTaskDone(listener)` 观察每条终止记录及其精确 owner。监听器抛出异常或拒绝会被封装；系统不会等待监听器工作。
+- `onTaskDone(listener)` 观察每条终止记录及其精确 owner。监听器抛出的异常和产生的拒绝都会被隔离；系统不会等待监听器工作。
 - `attachSurface(name)` 在其 effect 生命周期内声明控制表层。如果没有附加任何表层，`start()` 会在生产方执行前失败。
 
 有 owner 的访问会比较任务的 `SessionId` 与调用方。`bash-1` 等 id 可预测，因此这道隔离是安全边界。无 owner 的任务向调用方开放，并持续到服务释放。
 
 `outputLimitBytes` 是生产方拥有的模型呈现策略，会原样携带到快照中。控制表层在添加状态或通知元数据后应用它；注册表不会重写生产方输出，也不会为省略此字段的生产方虚构默认值。
 
-实现还必须兑现契约的生命周期语义：注册的存续期长于生产方与控制表层的 fiber，owner 释放和服务释放会取消存活工作并等待守约的生产方，结算遵循首次结果优先（一条终止记录、一轮故障隔离的监听器通知，然后释放等待方）。
+实现还必须兑现契约的生命周期语义：注册的存续期长于生产方 fiber 与控制表层 fiber，owner 释放和服务释放会取消仍在运行的工作并等待守约的生产方，结算遵循首次结果优先（一条终止记录、一轮异常受到隔离的监听器通知，然后释放等待方）。
 
-参见[任务类型目录](../../../docs/core-data-structures/tasks.md)、[运行时 Agent Note](../../../.agents/notes/implemented/architecture/2026-06-20-generic-long-running-tool-runtime.md)和 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-26-task-registry-seam.md)。
+参见[任务类型目录](../../../docs/core-data-structures/tasks.md)、[运行时 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-20-generic-long-running-tool-runtime.md)和 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-26-task-registry-seam.md)。
 
 ## 模型体验
 
@@ -28,10 +28,10 @@
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
 - **流输出只有一个消费游标**：独立观察者需要游标或快照 API。
-- **前台工作无法提升**：生产方在启动前选择前台或后台。
-- **契约是进程内的**：`TaskStart.run()` 传入回调和确切的 `Agent` 对象；持久或跨进程后端必须先重塑身份、重启、所有权与观察语义，才能实现此 seam。
+- **前台工作无法转为后台**：生产方在启动前选择前台或后台。
+- **契约是进程内的**：`TaskStart.run()` 传入回调和确切的 `Agent` 对象；持久化或跨进程后端必须先重塑身份、重启、所有权与观察语义，才能实现此 seam。
diff --git a/packages/tasks/tool-tasks/README.i18n.yaml b/packages/tasks/tool-tasks/README.i18n.yaml
index efa174ffae..646ac8ca6c 100644
--- a/packages/tasks/tool-tasks/README.i18n.yaml
+++ b/packages/tasks/tool-tasks/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/tasks/tool-tasks/README.md
 README.md: 276ca3f8284b366bee54e585297c3a033a65c8d1
-README.zh.md: f4d61b23c1b90e24591efd6773ffe9f9c46c2b4b
+README.zh.md: 823c25c3e0c7c208ffef9c5be5b2e4ef7e96bad8
diff --git a/packages/tasks/tool-tasks/README.zh.md b/packages/tasks/tool-tasks/README.zh.md
index f4d61b23c1..823c25c3e0 100644
--- a/packages/tasks/tool-tasks/README.zh.md
+++ b/packages/tasks/tool-tasks/README.zh.md
@@ -14,11 +14,11 @@
 
 它们的规范值依次为 `{ text, task }`、`PublicTaskSnapshot[]` 和 `{ outcome: 'cancellation-requested' | 'already-finished', task }`。公共快照携带 id、kind、label、status/detail 及开始／结束时间；它有意省略 `ownerSession` 和内部 `reported` 通知位。原生 renderer 保留上述状态与确认文本。
 
-当生产方提供 `outputLimitBytes` 时，`task_output`、终止 `task_kill` 和完成通知会在添加状态或通知文本后，对完整的原生 UTF-8 结果施加上限。只要能够容纳，读取就会保留输出尾部与控制后缀；有界完成通知则先为 `background task <id>` 和 `task_output` 收集指令预留空间，再把剩余字节用于可变的 kind、label、status、detail 与截断标记。一个前置 pre-execute 监听器会在策略运行前捕获调用方可见任务；每个任务控制定义的 final-content 回调会把其生产方上限应用到单文本拒绝、短路、规范化工具或流水线失败、替换和阻止；结构化多块策略结果保持自身形状。已有的生产方截断标记会复用，不会重复添加。省略该字段的生产方保留现有的无界控制表层行为。
+当生产方提供 `outputLimitBytes` 时，`task_output`、针对已终止任务的 `task_kill` 和完成通知会在添加状态或通知文本后，对完整的原生 UTF-8 结果施加上限。只要能够容纳，读取就会保留输出尾部与控制后缀；有界完成通知则先为 `background task <id>` 和 `task_output` 收集指令预留空间，再把剩余字节用于可变的 kind、label、status、detail 与截断标记。一个前置 pre-execute 监听器会在策略运行前捕获调用方可见任务；每个任务控制定义的 final-content 回调会把其生产方上限应用到单文本拒绝、短路、规范化工具或流水线失败、替换和阻止；结构化多块策略结果保持自身形状。已有的生产方截断标记会复用，不会重复添加。省略该字段的生产方保留现有的无界控制表层行为。
 
 ## 完成通知
 
-一项尚未报告的完成会向精确 owner 的会话注入 `background task <id> (<kind>: <label>) finished [status: ...]. Read its output with task_output.`。应用上限时，在 PTY 支持的 64 字节下限内，稳定 id 前缀和收集命令的优先级高于可变 label/detail，因此通知仍可操作。注入是下一次请求使用的持久上下文，并非唤醒。kill 或终止性 read/wait 会把交付标为已报告，并抑制重复通知。owner 释放竞态无需特殊处理：循环没有终结状态，teardown 期间注入的通知作为空闲上下文追加——会话仍挂接时随之持久化以供恢复，脱离后随无引用日志一并丢弃。
+一项尚未报告的完成会向精确 owner 的会话注入 `background task <id> (<kind>: <label>) finished [status: ...]. Read its output with task_output.`。应用上限时，即使采用 PTY 支持的 64 字节下限，稳定 id 前缀和收集命令的优先级也高于可变 label/detail，因此通知仍可操作。注入是下一次请求使用的持久上下文，并非唤醒。kill 或针对已终止任务的 read/wait 会把交付标为已报告，并抑制重复通知。owner 释放竞态无需特殊处理：循环没有终结状态，teardown 期间注入的通知作为空闲上下文追加——会话仍挂接时随之持久化以供恢复，之后随脱离挂接的日志一并丢弃。
 
 ## 配置
 
@@ -35,7 +35,7 @@
 
 #### 模型看到的内容
 
-该插件注册 scope 中的每次请求都包含以下指引。按 agent scope 过滤工具时，可能会隐藏工具，却不会移除独立注册的提示词区段。
+该插件注册 scope 中的每次请求都包含以下指引。按 agent（智能体）scope 过滤工具时，可能会隐藏工具，却不会移除独立注册的提示词区段。
 
 ##### 后台任务指引
 
@@ -45,7 +45,7 @@ Track every background task id you start. You are notified in-session when a tas
 
 #### Token 影响
 
-激活期间，每次请求承担少量固定输入成本。
+激活期间，每次请求都会产生少量固定的输入 token 开销。
 
 #### KV Cache 影响
 
@@ -59,7 +59,7 @@ Track every background task id you start. You are notified in-session when a tas
 
 #### Token 影响
 
-工具可见的每次请求承担固定 schema 成本。
+工具可见时，每次请求都会产生固定的 schema token 开销。
 
 #### KV Cache 影响
 
@@ -69,15 +69,15 @@ Track every background task id you start. You are notified in-session when a tas
 
 #### 模型看到的内容
 
-读取会返回输出或 `(no new output)`，随后是 `[status: <status>]` 和可选 detail。空列表返回 `(no background tasks)`。kill 返回 `requested cancellation of task <id>` 或现有终止状态。尚未报告且有 owner 的完成使用上述通知。
+读取会返回输出或 `(no new output)`，随后是 `[status: <status>]` 和可选 detail。空列表返回 `(no background tasks)`。kill 返回 `requested cancellation of task <id>` 或现有终止状态。尚未报告且有 owner 的任务完成时使用上述通知。
 
 #### Token 影响
 
-结果与通知在压缩前保留于父级历史。流读取不会重复已消费的输出；生产方提供的 `outputLimitBytes` 会限制每次完整读取或通知。
+结果与通知在压缩（compaction）前保留于父级历史。流读取不会重复已消费的输出；生产方提供的 `outputLimitBytes` 会限制每次完整读取或通知。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
diff --git a/packages/tasks/tool-tasks/package.json b/packages/tasks/tool-tasks/package.json
index 2fd0b464a4..c63e49ddc9 100644
--- a/packages/tasks/tool-tasks/package.json
+++ b/packages/tasks/tool-tasks/package.json
@@ -32,6 +32,7 @@
   "peerDependencies": {
     "@deepseek-ai/dsh-agent": "^0.0.1",
     "@deepseek-ai/dsh-invariants": "^0.0.1",
+    "@deepseek-ai/dsh-llm": "^0.0.1",
     "@deepseek-ai/dsh-retention": "^0.0.1",
     "@deepseek-ai/dsh-system-prompt": "^0.0.1",
     "@deepseek-ai/dsh-tasks": "^0.0.1",
diff --git a/packages/telemetry/README.i18n.yaml b/packages/telemetry/README.i18n.yaml
index b10509cb56..41f1bd956f 100644
--- a/packages/telemetry/README.i18n.yaml
+++ b/packages/telemetry/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/telemetry/README.md
 README.md: 944cb3f9bac6169feddf8b49bc481cfbe7c6fa9d
-README.zh.md: 7be8af93654565cd9fb4d8b4376e0960fd7eb72b
+README.zh.md: 795b20abb47e1bf791730cc7f3ebb0522549a271
diff --git a/packages/telemetry/README.zh.md b/packages/telemetry/README.zh.md
index 7be8af9365..795b20abb4 100644
--- a/packages/telemetry/README.zh.md
+++ b/packages/telemetry/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-面向外部的会话上报：遥测（telemetry）seam 及其 OpenTelemetry 后端。整套设计归档于[复活 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md)：边界公理（harness 的职责止于 `emit()`，投递由上报 SDK 负责）、`telemetry/record` waterfall（瀑布式事件；脱敏规则由部署方挂载，seam 自身不带任何规则）、固定分片投影、handoff 游标，以及运维记录通道。
+面向外部的会话上报：遥测（telemetry）seam 及其 OpenTelemetry 后端。整套设计固定在[复活 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md)中：边界公理（harness 的职责止于 `emit()`，投递由上报 SDK 负责）、`telemetry/record` waterfall（瀑布式事件；脱敏规则由部署方挂载，seam 自身不带任何规则）、固定分片投影、handoff 游标，以及运维记录通道。
 
-| 包 | 职责 |
+| 包（package） | 职责 |
 |---|---|
 | [`@deepseek-ai/dsh-session-telemetry`](session-telemetry/) | seam 本体：捕获点、投影、脱敏、handoff 游标、运维信号，以及最小后端契约（`emit`/`flush?`/`shutdown`）。 |
 | [`@deepseek-ai/dsh-session-telemetry-otel`](session-telemetry-otel/) | 部署方要加载的后端：OTel JS SDK 的日志流水线（`LoggerProvider` + `BatchLogRecordProcessor` + OTLP/HTTP 导出器），经透传（passthrough）原样配置。 |
diff --git a/packages/telemetry/session-telemetry-otel/README.i18n.yaml b/packages/telemetry/session-telemetry-otel/README.i18n.yaml
index e8cafdc470..b1a2052a3f 100644
--- a/packages/telemetry/session-telemetry-otel/README.i18n.yaml
+++ b/packages/telemetry/session-telemetry-otel/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/telemetry/session-telemetry-otel/README.md
 README.md: 9b208e291e77bee50d9d4fd14808268dca75f2db
-README.zh.md: f36cfe74146b779c4ddb1101227cb45a06f0968c
+README.zh.md: 76de1bf1ad58a0239907f3b63c672177874c7966
diff --git a/packages/telemetry/session-telemetry-otel/README.zh.md b/packages/telemetry/session-telemetry-otel/README.zh.md
index f36cfe7414..76de1bf1ad 100644
--- a/packages/telemetry/session-telemetry-otel/README.zh.md
+++ b/packages/telemetry/session-telemetry-otel/README.zh.md
@@ -17,7 +17,7 @@
     processor: {}            # optional; passed verbatim to BatchLogRecordProcessor
 ```
 
-`exporter.url` 是本包唯一自行校验的字段：必填、无默认值、必须能解析为 `http(s)`，因此缺失端点会在插件加载时失败（`processor.maxExportBatchSize` 不是正整数时同样如此：SDK 会接受该值，随后却在关闭时因它挂起）。其余全部是 SDK 自己的选项形态，由 SDK 拥有并在 SDK 文档中说明，两个配置块都整体透传（passthrough）：`OTLPExporterNodeConfigBase` 的每个字段（`headers`、`timeoutMillis`、`compression`、`keepAlive` 等）都会到达导出器；批处理、导出节奏（`scheduledDelayMillis`）、重试、队列上限，以及持续失败下的丢失策略，都是 SDK 的文档化行为，经 `processor` 透传调优。该后端刻意不实现 `flush()`：批处理器是进程内唯一执行 flush 的组件，`shutdown()` 的排空正因如此才是完整的。从 `cordis.yml` 中删除该配置块即为退出方式：无残留状态，也没有 `enabled` 开关。
+`exporter.url` 是本包（package）唯一自行校验的字段：必填、无默认值、必须能解析为 `http(s)`，因此缺失端点会在插件加载时失败（`processor.maxExportBatchSize` 不是正整数时同样如此：SDK 会接受该值，随后却在关闭时因它挂起）。其余全部是 SDK 自己的选项形态，由 SDK 拥有并在 SDK 文档中说明，两个配置块都整体透传（passthrough）：`OTLPExporterNodeConfigBase` 的每个字段（`headers`、`timeoutMillis`、`compression`、`keepAlive` 等）都会到达导出器；批处理、导出节奏（`scheduledDelayMillis`）、重试、队列上限，以及持续失败下的丢失策略，都是 SDK 的文档化行为，经 `processor` 透传调优。该后端刻意不实现 `flush()`：批处理器是进程内唯一执行 flush 的组件，`shutdown()` 的排空正因如此才是完整的。从 `cordis.yml` 中删除该配置块即为退出方式：无残留状态，也没有 `enabled` 开关。
 
 ## 哪些数据会离开本机
 
@@ -25,7 +25,7 @@
 
 ## 字段映射
 
-seam 记录 → SDK 日志记录：`time` → `timestamp`/`observedTimestamp`；`severity` → `severityNumber`/`severityText`（INFO 9 / WARN 13 / ERROR 17）；`body` → 结构化日志 body；`attributes` 原样照搬。接收端基于 `(session.id, event.seq)` 去重、按严重级别告警，并通过 `shutdown` 记录的缺失检测崩溃（一个曾有活动、没有 `shutdown` 运维记录、且已然陈旧的会话，就是未干净结束的会话）。该标记的含义是遥测干净地停止了对该会话的观察：它在会话自身 dispose（资源释放）时发出，对于届时仍在运行的会话，则在应用拆卸时发出；标记之后又出现该会话的更多事件，说明发生的是遥测重载，而不是会话重启。跨谱系（lineage）的流并不自足：恢复的会话在其自身 id 的流上从上一个进程停止之处继续；fork 出的会话，其流从继承边界开始，前缀位于父会话的流中，由接收端基于 `session.parent_id` + `session.seed_length` 拼接。继续而非回放的一个后果：流中一个开启后再未关闭的轮次，标志着上一个进程死在了该轮次之内。恢复时本地日志会以合成的关闭事件修复，但这些修复绝不导出：导出的流忠实于崩溃进程实际发出的内容，其后干净的 `shutdown` 标记也只证明恢复后进程自身的退出。
+seam 记录 → SDK 日志记录：`time` → `timestamp`/`observedTimestamp`；`severity` → `severityNumber`/`severityText`（INFO 9 / WARN 13 / ERROR 17）；`body` → 结构化日志 body；`attributes` 原样照搬。接收端基于 `(session.id, event.seq)` 去重、按严重级别告警，并通过 `shutdown` 记录的缺失检测崩溃（一个曾有活动、没有 `shutdown` 运维记录、且已然陈旧的会话，就是未干净结束的会话）。该标记的含义是遥测干净地停止了对该会话的观察：它在会话自身 dispose（资源释放）时发出，对于届时仍在运行的会话，则在应用关闭时发出；标记之后又出现该会话的更多事件，说明发生的是遥测重载，而不是会话重启。跨谱系（lineage）的流并不自足：恢复的会话在其自身 id 的流上从上一个进程停止之处继续；fork 出的会话，其流从继承边界开始，前缀位于父会话的流中，由接收端基于 `session.parent_id` + `session.seed_length` 拼接。继续而非回放的一个后果：流中一个开启后再未关闭的轮次，标志着上一个进程死在了该轮次之内。恢复时本地日志会以合成的关闭事件修复，但这些修复绝不导出：导出的流忠实于崩溃进程实际发出的内容，其后干净的 `shutdown` 标记也只证明恢复后进程自身的退出。
 
 ## 模型体验
 
@@ -35,7 +35,7 @@ seam 记录 → SDK 日志记录：`time` → `timestamp`/`observedTimestamp`；
 
 无；本包既不组装也不发送提供方请求。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **上游实验性源码树**：`@opentelemetry/sdk-logs` 仍从上游实验性（experimental）源码树发布；SDK API 的变动只会落在本包，也仅落在本包；seam 契约不动。
 - **无真实 collector 覆盖**：所有测试都导出到本地 mock collector；无密钥的 Loader 组合 e2e（`tests/loader-composition.e2e.ts`）在每次运行中都覆盖协议格式（wire format）形态，而面对真实 OTLP 部署的行为（认证、TLS、限流）属于 SDK 导出器文档的职责范围。
diff --git a/packages/telemetry/session-telemetry/README.i18n.yaml b/packages/telemetry/session-telemetry/README.i18n.yaml
index ae4c852011..18f6751424 100644
--- a/packages/telemetry/session-telemetry/README.i18n.yaml
+++ b/packages/telemetry/session-telemetry/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/telemetry/session-telemetry/README.md
 README.md: 272c9abe78849be3d2bba2c54cd7e25bcbe2d4c2
-README.zh.md: fdf35d423b0a650f4e6280e500436fe8bcf5a48c
+README.zh.md: e6f077c1d12d00e746147908560d05381fde11c3
diff --git a/packages/telemetry/session-telemetry/README.zh.md b/packages/telemetry/session-telemetry/README.zh.md
index fdf35d423b..e6f077c1d1 100644
--- a/packages/telemetry/session-telemetry/README.zh.md
+++ b/packages/telemetry/session-telemetry/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-遥测（telemetry）seam：会话事件上报的捕获侧，隔在一个后端契约之后，任何上报 SDK 都无需变形即可满足该契约。塑造本包一切设计的边界公理：**本包的职责止于 `emit()`**。批处理、重试、排队与丢失策略都属于后端自身的 SDK，本包既不为其立规，也不做包装。设计依据与被否决的替代方案见[复活 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md)。
+遥测（telemetry）seam：会话事件上报的捕获侧，隔在一个后端契约之后，任何上报 SDK 都无需变形即可满足该契约。塑造本包（package）一切设计的边界公理：**本包的职责止于 `emit()`**。批处理、重试、排队与丢失策略都属于后端自身的 SDK，本包既不为其立规，也不做包装。设计依据与被否决的替代方案见[复活 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md)。
 
 ## 后端契约
 
@@ -10,11 +10,11 @@
 
 ## 捕获点
 
-协调器的全部注册都经由组合方 fiber 的 effect 完成：`session/created`（收养：记录 header，并经投影从构造边界起回读日志；来自 fork 或恢复的构造函数种子绝不会在 firehose 上再次发出，也绝不会再次导出）、`session/event`（投影、深拷贝、脱敏、交接；零 I/O）、`session/flush`（转发可选的 `flush()` 提示并返回 void；循环所等待的并行任务绝不能等待遥测）、`session/disposed`（在会话自身的终止边缘发出该会话的 `shutdown` 运维记录，接收端正是在这个边缘锚定崩溃检测；随后将该会话退役，因此长生命周期的后端既不会保留已关闭的会话，也不会在卸载时再次标记它们）、`agent/error`（唯一的实时总线转发；会话事件词汇有意不包含运行错误记录）、一个 dispose effect（拆卸时先标记每个仍存活的会话，再等待后端的 `shutdown()`；失败只发出警告而不抛出），以及对 `ctx.sessions.list()` 的收养扫描（热重载不会重放 `session/created`）。
+协调器的全部注册都经由组合方 fiber 的 effect 完成：`session/created`（收养：记录 header，并经投影从构造边界起回读日志；来自 fork 或恢复的构造函数种子绝不会在 firehose 上再次发出，也绝不会再次导出）、`session/event`（投影、深拷贝、脱敏、交接；零 I/O）、`session/flush`（转发可选的 `flush()` 提示并返回 void；循环所等待的并行任务绝不能等待遥测）、`session/disposed`（在会话自身的终止边缘发出该会话的 `shutdown` 运维记录，接收端正是在这个边缘锚定崩溃检测；随后将该会话退役，因此长生命周期的后端既不会保留已关闭的会话，也不会在卸载时再次标记它们）、`agent/error`（唯一的实时总线转发；会话事件词汇有意不包含运维错误记录）、一个 dispose effect（拆卸时先标记每个仍存活的会话，再等待后端的 `shutdown()`；失败只发出警告而不抛出），以及对 `ctx.sessions.list()` 的收养扫描（热重载不会重放 `session/created`）。
 
-## 脱敏 waterfall
+## 脱敏 waterfall（瀑布式事件）
 
-每条记录在投影与 `emit()` 之间都要经过 `telemetry/record` waterfall（瀑布式事件），这是该 seam 的擦除扩展点。seam 自身不带任何规则：最内层的 `next()` 原样透传记录，因此未挂载监听器时，记录以捕获时的原样到达后端；导出数据能干净到什么程度，恰恰取决于部署方挂载了什么规则。监听器通过变换 `next()` 的返回值来堆叠；不调用 `next()` 就返回，即替换其下方的全部逻辑；抛出异常的监听器会在协调器的隔离范围内以 fail-closed 方式扣下这一条记录。脱敏只作用于导出副本；权威会话日志永不改写。
+每条记录在投影与 `emit()` 之间都要经过 `telemetry/record` waterfall，这是该 seam 的脱敏扩展点。seam 自身不带任何规则：最内层的 `next()` 原样透传记录，因此未挂载监听器时，记录以捕获时的原样到达后端；导出数据能干净到什么程度，恰恰取决于部署方挂载了什么规则。监听器通过变换 `next()` 的返回值来堆叠；不调用 `next()` 就返回，即替换其下方的全部逻辑；抛出异常的监听器会在协调器的隔离范围内以 fail-closed 方式拦下这一条记录。脱敏只作用于导出副本；权威会话日志永不改写。
 
 ## handoff 游标
 
@@ -36,7 +36,7 @@
 
 无；本包既不组装也不发送提供方请求。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **尽力而为的投递**：游标标记的是已交接而非已投递；在重载窗口内被拆除的会话无法重新收养；崩溃时留在后端队列中的内容会丢失。持久 outbox（spool、每 sink 游标、at-least-once）推迟到有部署方提出明确的崩溃丢失要求时再实现；见[复活 Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md)。
+- **尽力而为的投递**：游标标记的是已交接而非已投递；在重载窗口内被拆除的会话无法重新收养；崩溃时留在后端队列中的内容会丢失。持久化 outbox（spool、每 sink 游标、at-least-once）推迟到有部署方提出明确的崩溃丢失要求时再实现；见[复活 Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-session-telemetry-otel-revival.md)。
 - **不内置脱敏规则**：未挂载 `telemetry/record` 监听器时，记录以捕获时的原样离开进程，包括文件内容或命令输出中内嵌的任何凭据；向共享 collector 导出的部署方自行负责其规则集。
diff --git a/packages/timeout/README.i18n.yaml b/packages/timeout/README.i18n.yaml
index 85007025aa..c2637365b8 100644
--- a/packages/timeout/README.i18n.yaml
+++ b/packages/timeout/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/timeout/README.md
 README.md: 2a75e4d54aa1f518858d97e2f4c7d09a23a80a70
-README.zh.md: d0066fb3c1101b195f6da6798c6a1e0dad49a6c5
+README.zh.md: d0192f7441eb212afa3752cf703877d8d9b58e5c
diff --git a/packages/timeout/README.zh.md b/packages/timeout/README.zh.md
index d0066fb3c1..d0192f7441 100644
--- a/packages/timeout/README.zh.md
+++ b/packages/timeout/README.zh.md
@@ -2,10 +2,10 @@
 
 [English](README.md) | 中文
 
-工具调用超时策略插件。它是单一 **产品** 包（package）：它是 `tools/execute` 环绕分发 seam（由 [`dsh-tools`](../core/tools) 拥有）和纯 [`dsh-timeout`](../util/timeout) 库的部署策略消费方，而非带接口／实现拆分的可替换能力，因此无需 seam 三包组合。
+工具调用超时策略插件。它是单一 **产品**包（package）：它是 `tools/execute` 环绕分发 seam（由 [`dsh-tools`](../core/tools) 拥有）和纯 [`dsh-timeout`](../util/timeout) 库的部署策略消费方，而非带接口／实现拆分的可替换能力，因此无需 seam 三包组合。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
-| `timeout-policy/` | `tools/execute` 包装层：对每个已配置工具，它都在 `exec.signal` 上启动单次调用截止时间，并在截止时间先到时返回结构化 `TOOL_TIMEOUT` 结果 | （注册 `tools/execute` 监听器；不注入任何内容） |
+| `timeout-policy/` | `tools/execute` 包装层：对每个已配置工具，它都在 `exec.signal` 上设置单次调用截止时间，并在截止时间先到时返回结构化 `TOOL_TIMEOUT` 结果 | （注册 `tools/execute` 监听器；不注入任何内容） |
 
-超时被拆分为三层：[`dsh-timeout`](../util/timeout) 拥有纯计时／分类原语（`deadline`/`timeoutOf`）；每种能力拥有终止操作（bash 终止其进程组，fetch 提供方关闭其 socket）；本包则拥有 *作为部署策略的面向模型工具调用预算*：没有面向模型的超时参数，也没有全局默认值。它是[超时库 Agent Note](../../.agents/notes/implemented/architecture/2026-07-06-timeout-deadline-library.md) 所预见的中间件。`bash` 和钩子命令执行保留各自的 `BASH_TIMEOUT` 后端超时，不经过此策略。
+超时被拆分为三层：[`dsh-timeout`](../util/timeout) 拥有纯计时／分类原语（`deadline`/`timeoutOf`）；每种能力拥有终止操作（bash 终止其进程组，fetch 提供方关闭其 socket）；本包则拥有 *作为部署策略的面向模型工具调用预算*：没有面向模型的超时参数，也没有全局默认值。它是[超时库 Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-06-timeout-deadline-library.md) 所预见的中间件。`bash` 和钩子命令执行保留各自的 `BASH_TIMEOUT` 后端超时，不经过此策略。
diff --git a/packages/timeout/timeout-policy/README.i18n.yaml b/packages/timeout/timeout-policy/README.i18n.yaml
index 59b88721a8..94509e3669 100644
--- a/packages/timeout/timeout-policy/README.i18n.yaml
+++ b/packages/timeout/timeout-policy/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/timeout/timeout-policy/README.md
 README.md: 3e5769e2f8b95392e9d489659c6bae634ab3ff18
-README.zh.md: d9c14af1b609b3c6edf3472664007b8119f283db
+README.zh.md: 8d69e7d4aacb70beb5eaaa63c74a0759e0038ec5
diff --git a/packages/timeout/timeout-policy/README.zh.md b/packages/timeout/timeout-policy/README.zh.md
index d9c14af1b6..8d69e7d4aa 100644
--- a/packages/timeout/timeout-policy/README.zh.md
+++ b/packages/timeout/timeout-policy/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-工具调用超时强制执行器：一个 `tools/execute` 环绕分发监听器。它会在 `exec.signal` 上启动单次调用的协作式截止时间；适用条件是工具声明了 `timeoutMs`，且声明位于其 `ToolDefinition` 上。截止时间先到时，它返回结构化 `TOOL_TIMEOUT` 结果。预算从工具自身的声明中读取（`ToolDefinition.timeoutMs`，由拥有该工具的插件设置），因此此插件是 **零配置** 的。它是 `tools/execute` 包装层的参考实现，也是面向模型工具调用预算的强制执行归属地（超时库 Agent Note 所预见的中间件）。
+工具调用超时强制执行器：单个 `tools/execute` 环绕分发监听器，会在 `exec.signal` 上设置单次调用的协作式截止时间；适用于声明了 `timeoutMs` 且声明位于其 `ToolDefinition` 上的工具。该截止时间先到时，它返回结构化 `TOOL_TIMEOUT` 结果。预算从工具自身的声明中读取（`ToolDefinition.timeoutMs`，由拥有该工具的插件设置），因此此插件是**零配置**的。它是 `tools/execute` 包装层的参考实现，也是面向模型工具调用预算的强制执行归属地（超时库 Agent Note（agent 决策记录）所预见的中间件）。
 
 ## 插件（命名空间：`timeout-policy`）
 
-它是函数／命名空间插件（`name`/`inject`/`apply`），而非服务。它不注册工具，也不接受配置；它消费 `ctx.tools` 的 `tools/execute` waterfall（由 `dsh-tools` 注册表始终提供），并读取每个已分发工具声明的 `timeoutMs`；该声明来自注册表（`ctx.tools.get(exec.name)`）。
+它是函数／命名空间插件（`name`/`inject`/`apply`），而非服务。它不注册工具，也不接受配置；它消费 `ctx.tools` 的 `tools/execute` waterfall（瀑布式事件）（由 `dsh-tools` 注册表始终提供），并读取每个已分发工具声明的 `timeoutMs`；该声明来自注册表（`ctx.tools.get(exec.name)`）。
 
 ```yaml
 - id: timeout-policy
@@ -19,39 +19,39 @@
 
 对 **声明了 `timeoutMs` 的工具**，监听器会：
 
-1. 从注册表中的工具自身声明（`ctx.tools.get(exec.name)?.timeoutMs`）读取预算，并启动 `deadline(exec.signal, timeoutMs, 'TOOL_TIMEOUT')`：一个将调用方中止与此插件计时器融合的信号（`@deepseek-ai/dsh-timeout`）。
-2. 将该派生信号替换到 `exec` 上用于下游分发，然后恢复调用方自身的信号（cordis `next()` 忽略传入的参数，因此包装层会原地修改共享 `exec`；恢复可使 `tools/post-execute` 看到调用方的信号）。
-3. 分发后，如果 `timeoutOf(d.signal, 'TOOL_TIMEOUT')` 匹配，即此插件自身的计时器触发，则将结果替换为结构化 `TOOL_TIMEOUT` 工具结果：`{ isError: true, error: { message, info: { name: 'ToolTimeoutError', code: 'TOOL_TIMEOUT' } }, content: 'Error: tool call timed out after <ms>ms' }`。
+1. 从注册表中的工具自身声明（`ctx.tools.get(exec.name)?.timeoutMs`）读取预算，并设置 `deadline(exec.signal, timeoutMs, 'TOOL_TIMEOUT')`：一个将调用方中止与此插件计时器融合的信号（`@deepseek-ai/dsh-timeout`）。
+2. 将该派生信号替换到 `exec` 上用于下游分发，然后恢复调用方自身的信号（Cordis `next()` 忽略传入的参数，因此包装层会原地修改共享 `exec`；恢复可使 `tools/post-execute` 看到调用方的信号）。
+3. 分发后，如果 `timeoutOf(d.signal, 'TOOL_TIMEOUT')` 检测到此插件自身的计时器已触发，则将结果替换为结构化 `TOOL_TIMEOUT` 工具结果：`{ isError: true, error: { message, info: { name: 'ToolTimeoutError', code: 'TOOL_TIMEOUT' } }, content: 'Error: tool call timed out after <ms>ms' }`。
 
 **未声明预算的工具** 会原样委托（不启动截止时间）。
 
-基础 `next()` 是 `tools/execute` 在注册表中带规范化的分发 thunk，因此当超时信号到达抛出自身上游中止错误的提供方时，分发会先将其转换为普通错误结果，再由此包装层替换为 `TOOL_TIMEOUT`。这一顺序就是替换依据信号（`timeoutOf`）而非已分发结果形状的原因。
+基础 `next()` 是注册表为 `tools/execute` 提供的、带规范化处理的分发 thunk，因此当超时信号到达抛出自身上游中止错误的提供方时，分发会先将其转换为普通错误结果，再由此包装层替换为 `TOOL_TIMEOUT`。这一顺序就是替换依据信号（`timeoutOf`）而非已分发结果形状的原因。
 
 ### 协作式，而非硬终止
 
-派生信号只会 **通知**；终止仍属于工具以及它将 `exec.signal` 转发给的能力（`dsh-timeout` 库不拥有 kill）。**因此，声明 `timeoutMs` 意味着「与 `exec.signal` 协作」**：忽略该信号的工具不会在超时时停止。只有转发信号的工具才应声明该字段；已交付的 `web_fetch`/`web_search`（通过 `ctx.web` 转发给提供方）是参考实现。`TOOL_TIMEOUT` 无需会话事件以满足可重建性：它是最终面向模型的 `tool/result`，已由循环记录。
+派生信号只会 **通知**；是否终止仍取决于工具及其将 `exec.signal` 转发到的能力（`dsh-timeout` 库本身不负责硬终止）。**因此，声明 `timeoutMs` 意味着「与 `exec.signal` 协作」**：忽略该信号的工具不会在超时时停止。只有转发信号的工具才应声明该字段；已交付的 `web_fetch`/`web_search`（通过 `ctx.web` 转发给提供方）是参考实现。`TOOL_TIMEOUT` 无需会话事件以满足可重建性：它是最终面向模型的 `tool/result`，已由循环记录。
 
 ### 与其他 `tools/execute` 包装层组合
 
-多个 `tools/execute` 监听器按 cordis 注册顺序组合。与未来的重试／沙箱／指标包装层一起使用时，注册顺序决定语义：「超时覆盖整个重试操作」（超时注册在外层），或「超时覆盖每次尝试」（超时注册在内层）。
+多个 `tools/execute` 监听器按 Cordis 注册顺序组合。与未来的重试／沙箱／指标包装层一起使用时，注册顺序决定语义：「超时覆盖整个重试操作」（超时注册在外层），或「超时覆盖每次尝试」（超时注册在内层）。
 
 ## 模型体验
 
 ### 条件工具结果
 
-#### 模型所见内容
+#### 模型看到的内容
 
 此插件不添加提示词或 schema。如果已声明的截止时间先到，它会将提供方结果替换为 `Error: tool call timed out after <ms>ms` 与结构化 `TOOL_TIMEOUT`；否则原结果保持不变。
 
 #### Token 影响
 
-未超时调用为零 token。超时会添加一条短小且保留的错误结果，并可防止体积更大的延迟提供方结果进入上下文。
+未超时的调用不会增加 token。超时会添加一条会被保留的简短错误结果，并可防止体积更大、较晚返回的提供方结果进入上下文。
 
 #### KV Cache 影响
 
 仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - **协作式，绝不是硬终止**：截止时间只通过 `exec.signal` 通知；忽略该信号的工具不会在超时时停止（参见「协作式，而非硬终止」一节）。
 - **没有统一预算**：只有声明 `timeoutMs` 并将其放在 `ToolDefinition` 上的工具才会获得截止时间；未声明工具没有注册表级默认值（已交付的 `bash`/`read`/`write`/`edit` 有意不声明）。
diff --git a/packages/todo/README.i18n.yaml b/packages/todo/README.i18n.yaml
index d930492e76..a7b1f6f529 100644
--- a/packages/todo/README.i18n.yaml
+++ b/packages/todo/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/todo/README.md
 README.md: 1e5ae9a1583b9e9d3913fcd1dca7ef11a5f391fe
-README.zh.md: 0048940424f71c5709714d88346433fcfbf1554f
+README.zh.md: a77f788a41353ea547059864dc7cf73ac5025219
diff --git a/packages/todo/README.zh.md b/packages/todo/README.zh.md
index 0048940424..a77f788a41 100644
--- a/packages/todo/README.zh.md
+++ b/packages/todo/README.zh.md
@@ -1,11 +1,11 @@
-# todo/：todo／规划能力系列
+# todo/：todo／规划能力家族
 
 [English](README.md) | 中文
 
-面向模型的 todo 工具。它是单一 **产品** 包（package）：这里没有接口／实现 seam，因为该列表是由单一所有者管理的会话状态（每个 agent（智能体）会话拥有自己的列表），而非可替换能力。
+面向模型的 todo 工具。它是单一 **产品**包（package）：这里没有接口／实现 seam，因为该列表是由单一所有者管理的会话状态（每个 agent（智能体）会话拥有自己的列表），而非可替换能力。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `tool-todo/` | 面向模型的 `todo_write` 工具；将完整列表写入会话日志（`todo/write`） | （注册到 `ctx.tools`） |
 
-列表存在于事件溯源会话日志中（`SessionEventMap['todo/write']`，由 [`dsh-session`](../core/session) 拥有）；本包是追加快照的轻量消费方。[TUI 应用](../examples/tui-demo)等 UI 以及宿主／客户端运行时会根据会话事件渲染该持久列表。
+列表存在于事件溯源会话日志中（`SessionEventMap['todo/write']`，由 [`dsh-session`](../core/session) 拥有）；本包是追加快照的轻量消费方。[TUI 应用](../examples/tui-demo)等 UI 以及宿主／客户端运行时会根据会话事件渲染该持久化列表。
diff --git a/packages/todo/tool-todo/README.i18n.yaml b/packages/todo/tool-todo/README.i18n.yaml
index 3a84acde3b..603d2cfef0 100644
--- a/packages/todo/tool-todo/README.i18n.yaml
+++ b/packages/todo/tool-todo/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/todo/tool-todo/README.md
 README.md: b05ef43e7137dcf5678b1f1ad6d8c00b8a43baef
-README.zh.md: 5bc54c7cdf04f9bd77b385119bb88b2bb6ca43a3
+README.zh.md: 88a9f5d69ff52dcedbd8f27a4ace5fde3e0dc5a2
diff --git a/packages/todo/tool-todo/README.zh.md b/packages/todo/tool-todo/README.zh.md
index 5bc54c7cdf..88a9f5d69f 100644
--- a/packages/todo/tool-todo/README.zh.md
+++ b/packages/todo/tool-todo/README.zh.md
@@ -6,62 +6,62 @@
 
 ## 功能
 
-注册一个工具 `todo_write(todos: [{ content, status }])` 到 `ctx.tools`。模型每次调用都会发送完整列表，不存在部分更新或单项编辑。每次调用都会向调用 agent 的会话日志追加 `todo/write` 事件（完整列表快照），具体调用 `agent.session.append('todo/write', { todos })`；当前列表是最新的该类事件（回放时后写者胜）。
+注册一个工具 `todo_write(todos: [{ content, status }])` 到 `ctx.tools`。模型每次调用都会发送完整列表，不存在部分更新或单项编辑。每次调用都会向调用 agent 的会话日志追加 `todo/write` 事件（完整列表快照），具体调用 `agent.session.append('todo/write', { todos })`；当前列表是最新的该类事件（回放时后写覆盖先写）。
 
 `status` 是 `pending`、`in_progress` 或 `completed` 之一。
 
 ## 单一所有者
 
-该列表属于调用工具的唯一 agent 会话。不存在 subagent／共享／swarm scope：非 agent 调用方（没有 `exec.agent`）无处写入列表，因此会被拒绝。这是有意设置的 scope 限制，详见 Agent Note。
+该列表属于调用工具的唯一 agent 会话。不存在 subagent／共享／swarm scope：非 agent 调用方（没有 `exec.agent`）无处写入列表，因此会被拒绝。这是有意设置的 scope 限制，详见 Agent Note（agent 决策记录）。
 
 ## 验证
 
-除 schema 的类型／必填／枚举检查外，`execute` 还会拒绝空或重复的 `content`、同时存在多个 `in_progress` 任务的情况（连贯计划最多只有一个活跃任务），以及 `content`/`status` 之外的任何条目键——扩展条目形状（id、嵌套）会响亮失败而不是被静默压平，保证落日志的快照与模型自认为写入的内容一致。顺序与保持列表最新的纪律由模型根据工具描述负责。
+除 schema 的类型／必填／枚举检查外，`execute` 还会拒绝空或重复的 `content`、同时存在多个 `in_progress` 任务的情况（连贯计划最多只有一个活跃任务），以及 `content`/`status` 之外的任何条目键——扩展条目形状（id、嵌套）会明确报错而不是被静默压平，保证落日志的快照与模型自认为写入的内容一致。列表的顺序及及时更新由模型依照工具描述负责。
 
 ## 渲染
 
-规范结果为 `{ todos, counts: { pending, inProgress, completed } }`；其 Native 渲染器返回精简的更新确认。工具还会写入完整 `todo/write` 会话事件。UI 订阅事件流，并自行渲染该持久列表：[TUI 应用](../../examples/tui-demo)与 [web 客户端](../../client/ui-conversation)基于站立计划（其后没有更晚 `turn/start` 的最近一次 `todo/write`）显示计划条（web 另有专属工具行）（[展示](../../../.agents/notes/implemented/feature/2026-07-23-web-todo-display.md)、[生命周期](../../../.agents/notes/implemented/feature/2026-07-28-todo-plan-clears-on-next-turn.md)）。
+规范结果为 `{ todos, counts: { pending, inProgress, completed } }`；其 Native 渲染器返回精简的更新确认。工具还会写入完整 `todo/write` 会话事件。UI 订阅事件流，并自行渲染该持久化列表：[TUI 应用](../../examples/tui-demo)与 [web 客户端](../../client/ui-conversation)基于当前有效计划（其后没有更晚 `turn/start` 的最近一次 `todo/write`）显示计划条（web 另有专属工具行）（[展示](../../../.agents/notes/implemented/feature/2026-07-23-web-todo-display.md)、[生命周期](../../../.agents/notes/implemented/feature/2026-07-28-todo-plan-clears-on-next-turn.md)）。
 
 ## 会话投影
 
-当组合挂载了 `ctx.sessionProjections`（[`@deepseek-ai/dsh-session-projection`](../../session-projection/session-projection/README.md)）时，本包在一个注入式子插件下注册 `todos` 投影单元：`init` = `null`（尚无写入）、`apply` = 从每个 `todo/write` 取整表，并在每个 `turn/start` 清为 `null`（站立计划；`turn/end` 保留刚完成的清单；其余事件都返回同一个状态引用）、`view` = 恒等、`stateVersion` = 2。key 在本包合并进 `SessionProjectionMap`（经接口包的 `/types` 出口）；框架驱动该单元，载体在历史尾页与 `session/projection` 推送帧上供给该值。未装注册表的组合不受影响。生命周期理由见 [下一轮清空 todo 计划条](../../../.agents/notes/implemented/feature/2026-07-28-todo-plan-clears-on-next-turn.md)。
+当组合挂载了 `ctx.sessionProjections`（[`@deepseek-ai/dsh-session-projection`](../../session-projection/session-projection/README.md)）时，本包在一个注入的子插件中注册 `todos` 投影单元：`init` = `null`（尚无写入）、`apply` = 从每个 `todo/write` 取整表，并在每个 `turn/start` 清为 `null`（当前有效计划；`turn/end` 保留刚完成的清单；其余事件都返回同一个状态引用）、`view` = 恒等、`stateVersion` = 2。该键在本包中合并进 `SessionProjectionMap`（经接口包的 `/types` 出口）；框架驱动该单元，载体通过历史尾页与 `session/projection` 推送帧提供该值。未挂载注册表的组合不受影响。生命周期理由见 [在下一轮次清空 todo 计划](../../../.agents/notes/implemented/feature/2026-07-28-todo-plan-clears-on-next-turn.md)。
 
 ## 导出形状
 
-函数／命名空间插件：导出 `name`/`inject`/`apply`，不提供默认导出。意外的 `export default` 会通过 Loader 的 `unwrapExports` 折叠模块并丢弃 `inject`（参见 [docs/postmortem/0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)）。
+函数／命名空间插件：导出 `name`/`inject`/`apply`，不提供默认导出。意外的 `export default` 会被 Loader 的 `unwrapExports` 折叠为默认导出，并导致 `inject` 丢失（参见 [docs/postmortem/0001](../../../docs/postmortem/0001-acp-default-export-drops-inject.md)）。
 
 ## 模型体验
 
 ### 工具 schema
 
-#### 模型所见内容
+#### 模型看到的内容
 
 模型会看到生成的 [`todo_write` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-todo)。
 
 #### Token 影响
 
-工具可见的每个请求都有固定 schema 成本。
+工具可见的每个请求都有固定的 schema token 开销。
 
 #### KV Cache 影响
 
-只要定义和可见性不变，前缀就保持稳定。插件生命周期或 scope 限制可能会使此 schema 之后的复用失效。
+只要定义和可见性不变，前缀就保持稳定。插件生命周期或 scope 限制可能会使从此 schema 起的缓存复用失效。
 
 ### 工具调用历史与结果
 
-#### 模型所见内容
+#### 模型看到的内容
 
-每个 assistant 工具调用都会在参数中保留整个替换列表。成功时精确返回 `Updated todo list: <pending> pending, <inProgress> in progress, <completed> completed.`。稳定失败文本为 ``Error: invalid todo: `content` must be a non-empty string``、`Error: invalid todos: duplicate content "<content>"`、`Error: invalid todos: at most one task may be in_progress, got <count>` 和 `Error: todo_write requires an owning agent session`。完整 `todo/write` 会话事件是 UI 与回放状态，而非第二条模型消息。
+每个 assistant 工具调用都会在参数中保留整个替换列表。成功时原样返回 `Updated todo list: <pending> pending, <inProgress> in progress, <completed> completed.`。稳定失败文本为 ``Error: invalid todo: `content` must be a non-empty string``、`Error: invalid todos: duplicate content "<content>"`、`Error: invalid todos: at most one task may be in_progress, got <count>` 和 `Error: todo_write requires an owning agent session`。完整 `todo/write` 会话事件是 UI 与回放状态，而非第二条模型消息。
 
 #### Token 影响
 
-Token 增长与模型每次提交的完整列表成比例，且这些调用参数会保留到压缩（compaction）。结果本身很小，且形状固定。
+token 用量会随模型每次提交的完整列表增长，且这些调用参数会保留到压缩（compaction）。结果本身很小，且形状固定。
 
 #### KV Cache 影响
 
 仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
-- **仅单一所有者 scope**：列表属于唯一调用 agent 会话；subagent／共享／swarm scope 是有意裁减（参见「单一所有者」一节），非 agent 调用方会被拒绝。
-- **项目形状有意保持最小**：`content` 加三态 `status`；整表替换不需要稳定 id、优先级或 active-form 字段。
+- **仅单一所有者 scope**：列表属于唯一调用 agent 会话；subagent／共享／swarm scope 是有意设置的限制（参见「单一所有者」一节），非 agent 调用方会被拒绝。
+- **条目形状有意保持最小**：`content` 加三态 `status`；整表替换不需要稳定 id、优先级或 active-form 字段。
 - **整表替换是唯一操作**：没有部分更新，也没有回读工具；模型每次调用都必须重新发送完整列表。
diff --git a/packages/ui/README.i18n.yaml b/packages/ui/README.i18n.yaml
index 78f6fab567..7a6175f898 100644
--- a/packages/ui/README.i18n.yaml
+++ b/packages/ui/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/ui/README.md
 README.md: f08157d411a018141cdc21c487f81ae198f4de56
-README.zh.md: d321ee9a181eb27ccecb143d3733ee46e417b27f
+README.zh.md: ed4fbf576224a61e680fca337ac5e60829f8a90e
diff --git a/packages/ui/README.zh.md b/packages/ui/README.zh.md
index d321ee9a18..ed4fbf5762 100644
--- a/packages/ui/README.zh.md
+++ b/packages/ui/README.zh.md
@@ -2,21 +2,21 @@
 
 [English](README.md) | 中文
 
-面向用户的交互通道和进程外 SDK 服务器。这些是**产品** 包（package）：由用户或 SDK 客户端直接操作的真实接口。
+面向用户的交互通道和进程外 SDK 服务器。这些是**产品**包（package）：由用户或 SDK 客户端直接操作的真实接口。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `commands/` | 用户命令注册表：共享发现元数据、作用域遮蔽、取消以及 UI 直接分派 | `ctx.commands` |
 | `user-approval/` | 一次性用户审批机制、封闭的结果词汇、审计事件和逐会话审批策略 | `ctx.approval` |
-| `permission/` | 面向用户的权限预设（`workspace-write`/`danger-full-access`）：用一个产品级选择器组合沙箱模式与审批策略两个调节项，并写入各自的会话事件 | `ctx.permission` |
+| `permission/` | 面向用户的权限预设（`workspace-write`/`danger-full-access`）：通过一项产品级选择组合沙箱模式与审批策略两个可调参数，并写入各自的会话事件 | `ctx.permission` |
 | `user-interaction/` | UI 支持的确认工具所使用的抽象用户问答 seam | `ctx.userInteraction` |
 | `tool-ask-user/` | 模型侧 `ask_user_question` 工具，基于 `ctx.userInteraction` 实现 | （注册到 `ctx.tools`） |
 | `tui/` | 交互式 pi-tui 终端通道：渲染会话标题、事件和工具意图，响应 `ctx.userInteraction`，并托管由 effect 持有的插件浮层 | `ctx.tui`（驱动 `ctx.agents`） |
 | `jsonrpc/` | 面向进程外 SDK 客户端的 stdio JSON-RPC 服务器 | （驱动 `ctx.agents`） |
-| `app-boot/` | app bin 的共享启动粘合层：加载 `.env`、Loader 快速失败保护、感知快照的配置解析，以及等待整棵树停稳的启动序列 | （供各 bin 使用的库） |
+| `app-boot/` | app bin 的共享启动粘合层：加载 `.env`、会明确报错的 Loader 保护机制、感知快照的配置解析，以及等待整棵树停稳的启动序列 | （供各 bin 使用的库） |
 
-UI 集成属于客户端驱动插件，而非对循环的修改：它使用现有的 `agent/*` 事件分类和 `dsh-agent` 工厂。[`tui`](tui/README.md) 是交互式终端入口，并提供终端本地的 `ctx.tui` 扩展服务；[`jsonrpc`](jsonrpc/README.md) 为进程外 SDK 客户端提供服务，而非交互式单次任务使用 `cli-demo`。[`commands`](commands/README.md) 是 TUI 使用的纯用户发现与分派通道；命令输入和输出不会成为模型消息。
+UI 集成属于由客户端驱动的插件，而非对循环的修改：它使用现有的 `agent/*` 事件分类和 `dsh-agent` 工厂。[`tui`](tui/README.md) 是交互式终端入口，并提供终端本地的 `ctx.tui` 扩展服务；[`jsonrpc`](jsonrpc/README.md) 为进程外 SDK 客户端提供服务，非交互式的一次性任务则使用 `cli-demo`。[`commands`](commands/README.md) 是 TUI 使用的仅面向用户的发现与分派通道；命令输入和输出不会成为模型消息。
 
-`user-approval`、`user-interaction` 和 `tool-ask-user` 位于此处，因为向用户提问是由 UI 支持的产品功能，并不属于提供方无关的核心主干。`user-approval` 持有一次性的 `ctx.approval` 决策机制及其策略层级；应答方仍归拥有 agent（智能体）的通道或自动化传输层所有。`user-interaction` 保持提供方无关（`ctx.userInteraction`），`tool-ask-user` 是其模型侧消费方，而交互式 app 包提供具体实现。
+`user-approval`、`user-interaction` 和 `tool-ask-user` 位于此处，因为向用户提问是由 UI 支持的产品功能，并不属于无提供方的核心主干。`user-approval` 负责一次性的 `ctx.approval` 决策机制及其策略层级；应答逻辑仍由负责 agent（智能体）的通道或自动化传输层提供。`user-interaction` 保持提供方无关（`ctx.userInteraction`），`tool-ask-user` 是其模型侧消费方，而交互式 app 包提供具体的提供方。
 
-基于 [`agent-spine-demo`](../examples/agent-spine-demo/README.md) 组合的可运行 app bundle 位于 [`examples/`](../examples/README.md)（`tui-demo`、`acp-demo`、`jsonrpc-demo`）。`acp-demo` 和 `jsonrpc-demo` 持有启动 bin；`tui-demo` bundle 则由产品 [`dsh`](../../apps/cli/README.md) CLI 启动。`ui/` 保留可复用的用户／SDK 通道插件和共享 `app-boot` 粘合层；仅供自动化使用的 ACP 传输层位于 [`acp/`](../acp/README.md)。每个入口都持有自己的 stdout 策略，叶子 `cordis.yml` 则提供后端与可选工具。
+基于 [`agent-spine-demo`](../examples/agent-spine-demo/README.md) 组合的可运行 app bundle 位于 [`examples/`](../examples/README.md)（`tui-demo`、`acp-demo`、`jsonrpc-demo`）。`acp-demo` 和 `jsonrpc-demo` 各自提供启动 bin；`tui-demo` bundle 则由产品 [`dsh`](../../apps/cli/README.md) CLI（命令行界面）启动。`ui/` 保留可复用的用户／SDK 通道插件和共享 `app-boot` 粘合层；仅供自动化使用的 ACP（Agent Client Protocol）传输层位于 [`acp/`](../acp/README.md)。每个入口都负责自己的 stdout 策略，叶子 `cordis.yml` 则提供后端与可选工具。
diff --git a/packages/ui/commands/README.i18n.yaml b/packages/ui/commands/README.i18n.yaml
index 5c37ccbb16..400fef35bf 100644
--- a/packages/ui/commands/README.i18n.yaml
+++ b/packages/ui/commands/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/ui/commands/README.md
 README.md: 4ad72cf9e232c8d41e525f42eecde5637032a391
-README.zh.md: bace8f6346ac737a838d802dfc5c6ffe52c56edd
+README.zh.md: 9cc6a3f31e55da5d56c5b49ba78fbd66381ed680
diff --git a/packages/ui/commands/README.zh.md b/packages/ui/commands/README.zh.md
index bace8f6346..9cc6a3f31e 100644
--- a/packages/ui/commands/README.zh.md
+++ b/packages/ui/commands/README.zh.md
@@ -2,17 +2,17 @@
 
 [English](README.md) | 中文
 
-由插件拥有、供交互式 UI 适配器使用的面向用户命令注册表。[插件命令注册 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-plugin-command-registration.md)定义了其边界与分发契约。
+由插件负责、供交互式 UI 适配器使用的面向用户命令注册表。[插件命令注册 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-plugin-command-registration.md)定义了其边界与分发契约。
 
 ## 服务契约
 
-`ctx.commands.register(definition)` 注册一个小写命令名称、描述、可选的非结构化输入提示，以及可中止的处理器。每个已注册命令都可供所有已组合的命令适配器使用；与某项部署不兼容的插件不会在此注册。普通上下文中的注册全局生效。在 `agent.ctx` 下挂载的命令生产插件会声明自身的 `commands` 注入，并创建精确限定到该 agent 的定义；该定义会遮蔽同名的全局定义。这种子级注入形态保留了 agent 作用域，同时不会让核心 agent loop 依赖 UI 服务。同一层中的名称重复会在注册时失败。每个 disposer 都是 Cordis effect 返回的确切 disposer；注册或移除命令时，系统会通知每个 `commands/change` 观察者，使实时适配器能够刷新发现结果。观察者失败会写入日志，既不能否决注册表变更，也不能阻止后续观察者运行。
+`ctx.commands.register(definition)` 注册一个小写命令名称、描述、可选的非结构化输入提示，以及可中止的处理器。每个已注册命令都可供所有已组合的命令适配器使用；与某项部署不兼容的插件不会在此注册。普通上下文中的注册全局生效。在 `agent.ctx` 下挂载的命令生产插件会声明自身的 `commands` 注入，并创建精确限定到该 agent（智能体）的定义；该定义会遮蔽同名的全局定义。这种子级注入形态保留了 agent 作用域，同时不会让核心 agent loop（智能体循环）依赖 UI 服务。同一层中的名称重复会在注册时失败。每个 disposer 都是 Cordis effect 返回的确切 disposer；注册或移除命令时，系统会通知每个 `commands/change` 观察者，使运行中的适配器能够刷新发现结果。观察者失败会写入日志，既不能否决注册表变更，也不能阻止后续观察者运行。
 
-`list(agent)` 在应用作用域遮蔽后，返回按名称排序的不可变描述符。`find(agent, name)` 返回相应定义。`execute(agent, line, signal)` 使用 `parseCommand()`，且只运行已知命令，返回已结算的 `CommandExecution`（规范化结果加生命周期配对 `commandId`）；语法无效或名称未知时返回 `undefined`。已解析命令的生命周期会以 log-only 事件对的形式记录在接收 agent 的会话日志中：`command/run`（进入处理器前记录，携带铸造的 `commandId`、解析器的结构化 `name`/`args` 切分和发起方 `CommandSource`）与 `command/done`（结算时记录，携带结局种类与原样文本；处理器抛出或被中止时以 `kind: 'error'` 结算）。未通过准入的输入不记录任何事件。两者都是直接独立追加：没有轮次包裹它们，持久化在常规检查点与 teardown 时排空它们。
+`list(agent)` 在应用作用域遮蔽后，返回按名称排序的不可变描述符。`find(agent, name)` 返回相应定义。`execute(agent, line, signal)` 使用 `parseCommand()`，且只运行已知命令，返回已结算的 `CommandExecution`（规范化结果加生命周期配对 `commandId`）；语法无效或名称未知时返回 `undefined`。已解析命令的生命周期会以 log-only 事件对的形式记录在接收 agent 的会话日志中：`command/run`（进入处理器前记录，携带新生成的 `commandId`、解析器的结构化 `name`/`args` 切分和发起方 `CommandSource`）与 `command/done`（结算时记录，携带结果类型与原样文本；处理器抛出或被中止时以 `kind: 'error'` 结算）。未通过准入的输入不记录任何事件。两者都直接独立追加到接收 agent 的会话中：没有轮次包裹它们，持久化机制会在常规检查点和销毁期间排空这些事件。
 
-`parseCommand()` 识别位于字节零位置的斜杠、由小写字母、数字、`_` 或 `-` 构成的名称，以及名称后紧接输入末尾或空白的形式。它将名称后的每个字节作为 `rawInput` 返回，其中包括分隔空白；消费方拥有各命令专用的语法，只能执行该语法允许的规范化。
+`parseCommand()` 识别位于第 0 字节的斜杠、由小写字母、数字、`_` 或 `-` 构成的名称，以及名称后紧接输入末尾或空白的形式。它将名称后的每个字节作为 `rawInput` 返回，其中包括分隔空白；消费方负责各命令专用的语法，只能执行该语法允许的规范化。
 
-处理器返回 `success` 或 `error`，并可附带 UI 文本。适配器直接渲染结果，结果绝不进入模型历史。注册表绝不会隐式地把 `rawInput` 提交给 agent；命令生产方可以通过接收命令的 `Agent` 显式安排模型可见工作，此时该生产方拥有由此产生的消息契约。注册表会让处理器完成与所提供的中止信号竞速，但不协作的处理器可能在调用方停止等待后继续产生自身的外部副作用。
+处理器返回 `success` 或 `error`，并可附带 UI 文本。适配器直接渲染结果，结果绝不进入模型历史。注册表绝不会隐式地把 `rawInput` 提交给 agent；命令生产方可以通过接收命令的 `Agent` 显式安排模型可见工作，此时该生产方负责由此产生的消息契约。注册表会同时等待处理器完成和所提供的中止信号，以先发生者为准，但不响应中止的处理器可能在调用方停止等待后继续产生自身的外部副作用。
 
 ## 组合
 
@@ -32,9 +32,9 @@
 
 #### KV Cache 影响
 
-注册表元数据、命令输入和直接输出绝不会进入模型请求，也不会影响其缓存。被命令变更的领域拥有之后产生的所有缓存影响。
+注册表元数据、命令输入和直接输出绝不会进入模型请求，也不会影响其缓存。发生变更的领域负责之后产生的所有缓存影响。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **仅支持非结构化文本输入**：表单、补全 schema 和类型化参数仍由各命令自行解析。
 - **副作用采用协作式取消**：中止后，分发会停止等待；处理器必须遵循信号，才能停止已经进入外部系统的工作。
diff --git a/packages/ui/jsonrpc/README.i18n.yaml b/packages/ui/jsonrpc/README.i18n.yaml
index 3176b2ce40..324f8c4d99 100644
--- a/packages/ui/jsonrpc/README.i18n.yaml
+++ b/packages/ui/jsonrpc/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/ui/jsonrpc/README.md
 README.md: b1219ba10269fc7d046da22c280ff1b91424a5ae
-README.zh.md: 63615654769bf4ed7a69c09dc818af034c3a3c3c
+README.zh.md: 1c27f5edf2f1f172aa6303697b17e2e77a65842a
diff --git a/packages/ui/jsonrpc/README.zh.md b/packages/ui/jsonrpc/README.zh.md
index 6361565476..1c27f5edf2 100644
--- a/packages/ui/jsonrpc/README.zh.md
+++ b/packages/ui/jsonrpc/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-`jsonrpc` 插件通过 stdio 提供以换行符分隔的 JSON-RPC，使进程外 SDK 客户端能够驱动 harness agent（智能体）。[`HarnessSdkServer`](src/server.ts) 持有协议方法和通知；传输与具名线类型位于 [`dsh-sdk-protocol`](../../sdk/sdk-protocol/README.md)，与客户端 SDK 共享；[`jsonrpc-demo`](../../examples/jsonrpc-demo/README.md) 提供外围的 `cordis.yml` 应用。
+`jsonrpc` 插件通过 stdio 提供以换行符分隔的 JSON-RPC，使进程外 SDK 客户端能够驱动 harness agent（智能体）。[`HarnessSdkServer`](src/server.ts) 负责协议方法和通知；传输与具名协议类型位于 [`dsh-sdk-protocol`](../../sdk/sdk-protocol/README.md)，与客户端 SDK 共享；[`jsonrpc-demo`](../../examples/jsonrpc-demo/README.md) 提供外围的 `cordis.yml` 应用。
 
 ## 组装
 
-`inject: ['agents']`。服务器按 `sessionId` 获取或创建一个 agent。只有服务建立快照时的生命周期 `local` 标志为 true，服务器才会转发 subagent 完成事件；提供方名称、子级 id 和持久化谱系均不能证明本地性。已注册的适配器优先；未被持有的 `deepseek` 路由会挂载 `dsh-llm-deepseek`，任何其他未被持有的提供方都会导致初始化失败。其他功能由外围 `cordis.yml` 提供。
+`inject: ['agents']`。服务器按 `sessionId` 获取或创建一个 agent。只有服务对生命周期建立快照时记录的 `local` 标志为 true，服务器才会转发 subagent 完成事件；提供方名称、子级 id 和持久化谱系均不能证明本地性。已注册的适配器优先；尚无适配器负责的 `deepseek` 路由会挂载 `dsh-llm-deepseek`，任何其他尚无适配器负责的提供方都会导致初始化失败。其他功能由外围 `cordis.yml` 提供。
 
 ## 配置
 
@@ -40,7 +40,7 @@ Stdout 只承载 JSON-RPC 帧。部署不得组合 stdout logger；诊断应写
 
 仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **协议没有逐会话关闭或提示词取消方法**：SDK 创建的 agent 会一直存活到进程关闭；一条已接受的提示词必须运行到 agent 空闲，该会话才能接受下一条。
 - **stdout 纯净性由部署保证**：外围配置仍可能加载 stdout logger 并破坏 JSON-RPC 通道；此插件不会检查或否决同级 logger。
diff --git a/packages/ui/tool-ask-user/README.i18n.yaml b/packages/ui/tool-ask-user/README.i18n.yaml
index a03a7326fa..7b8e667b58 100644
--- a/packages/ui/tool-ask-user/README.i18n.yaml
+++ b/packages/ui/tool-ask-user/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/ui/tool-ask-user/README.md
 README.md: 8e779f4025c20cd200344efb7cb8cd6bc09ba64d
-README.zh.md: fe1dc5559882532c4f44e705cc6daa2c7f4f8905
+README.zh.md: acaffec0764404a0e0e842ffc2b4efdee8869c4f
diff --git a/packages/ui/tool-ask-user/README.zh.md b/packages/ui/tool-ask-user/README.zh.md
index fe1dc55598..acaffec076 100644
--- a/packages/ui/tool-ask-user/README.zh.md
+++ b/packages/ui/tool-ask-user/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-模型侧 `ask_user_question` 工具，基于 `ctx.userInteraction` 实现。当模型需要确认、选择或缺失信息才能继续时，它可以借此向用户提出简明问题。
+模型侧 `ask_user_question` 工具，基于 `ctx.userInteraction` 实现。当模型需要确认、选择结果或缺失的信息才能继续时，它可以借此向用户提出简明问题。
 
 ## 工具
 
@@ -15,7 +15,7 @@
 - `options`：可选选项，包含 `label` 和 `description`。如需推荐某个选项，请将其置于首位，并在该标签末尾追加 `(Recommended)`。
 - `multi_select`：该问题是否可以返回多个选中的选项。
 
-工具调用 `ctx.userInteraction.ask()`，并返回规范的 `{ answers: [{ id, selected, custom? }] }`。`selected` 包含选项标签；仅当用户自由填写回答时才会出现 `custom`，并覆盖选中的选项。Native renderer 会保留紧凑的 JSON 文本形式 `{ "answers": [{ "id": "...", "selected": ["..."], "custom": "..." }] }`。
+工具调用 `ctx.userInteraction.ask()`，并返回规范的 `{ answers: [{ id, selected, custom? }] }`。`selected` 包含选项标签；仅当用户自由填写回答时才会出现 `custom`，并覆盖选中的选项。Native 渲染器会保留紧凑的 JSON 文本形式 `{ "answers": [{ "id": "...", "selected": ["..."], "custom": "..." }] }`。
 
 ## 职责
 
@@ -31,17 +31,17 @@
 
 #### Token 影响
 
-工具可见的每个请求都会产生固定的 schema 开销。
+工具可见时，每个请求都会产生固定的 schema token 开销。
 
 #### KV Cache 影响
 
-只要定义和可见性保持不变，前缀即可稳定复用。插件生命周期变化或作用域限制可能从此 schema 开始使复用失效。
+只要定义和可见性保持不变，前缀即可稳定复用。插件生命周期变化或作用域限制可能会使从此 schema 起的缓存复用失效。
 
 ### 工具调用历史与结果
 
 #### 模型看到的内容
 
-模型提出的完整问题保留在 assistant 工具调用参数中。用户回答后，下一步骤会看到精确采用 `{"answers":[{"id":"<id>","selected":["<label>"],"custom":"<text>"}]}` 形式的紧凑 JSON；不使用 `custom` 时会省略该字段，`selected` 可以包含零个、一个或多个标签。调用等待期间的 UI 交互不属于模型上下文。
+模型提出的完整问题保留在 assistant 工具调用参数中。用户回答后，下一步会看到精确采用 `{"answers":[{"id":"<id>","selected":["<label>"],"custom":"<text>"}]}` 形式的紧凑 JSON；不使用 `custom` 时会省略该字段，`selected` 可以包含零个、一个或多个标签。调用等待期间的 UI 交互不属于模型上下文。
 
 #### Token 影响
 
@@ -51,7 +51,7 @@
 
 仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **待处理问题会阻塞工具调用，直至用户作答**：该工具未声明 `timeout-policy` 预算；取消仅沿用当前轮次的 `exec.signal`。
 - **Native 回答渲染为 JSON 文本**：规范值仍为结构化数据，但模型侧结果使用紧凑 JSON，而非更丰富的内容块词汇。
diff --git a/packages/ui/tui/package.json b/packages/ui/tui/package.json
index f9260e3231..ca12063262 100644
--- a/packages/ui/tui/package.json
+++ b/packages/ui/tui/package.json
@@ -45,6 +45,7 @@
     "@deepseek-ai/dsh-session-reference": "^0.0.1",
     "@deepseek-ai/dsh-session-title": "^0.0.1",
     "@deepseek-ai/dsh-skill": "^0.0.1",
+    "@deepseek-ai/dsh-subprocess": "^0.0.1",
     "@deepseek-ai/dsh-system-prompt": "^0.0.1",
     "@deepseek-ai/dsh-token-meter": "^0.0.1",
     "@deepseek-ai/dsh-tools": "^0.0.1",
@@ -82,6 +83,7 @@
     "@deepseek-ai/dsh-session-reference": "workspace:^",
     "@deepseek-ai/dsh-session-title": "workspace:^",
     "@deepseek-ai/dsh-skill": "workspace:^",
+    "@deepseek-ai/dsh-subprocess": "workspace:^",
     "@deepseek-ai/dsh-system-prompt": "workspace:^",
     "@deepseek-ai/dsh-token-meter": "workspace:^",
     "@deepseek-ai/dsh-tool-cordis": "workspace:^",
diff --git a/packages/ui/tui/src/chat/helpers.ts b/packages/ui/tui/src/chat/helpers.ts
index 58c2a71b21..d004500ba7 100644
--- a/packages/ui/tui/src/chat/helpers.ts
+++ b/packages/ui/tui/src/chat/helpers.ts
@@ -16,6 +16,7 @@ import {
   visibleWidth,
 } from '@earendil-works/pi-tui'
 import type { Session } from '@deepseek-ai/dsh-session'
+import { scrubbedParentEnv } from '@deepseek-ai/dsh-subprocess'
 
 /** Editor that shows a placeholder without making it editable content. */
 export class HintEditor extends Editor {
@@ -65,13 +66,10 @@ export function formatCwd(cwd: string | undefined): string {
  */
 export function gitBranch(cwd: string): string | undefined {
   try {
-    const env = Object.fromEntries(
-      Object.entries(process.env).filter(([name]) => !/(?:KEY|SECRET|TOKEN)/iu.test(name)),
-    )
     const branch = execFileSync('git', ['branch', '--show-current'], {
       cwd,
       encoding: 'utf8',
-      env,
+      env: scrubbedParentEnv(),
       stdio: ['ignore', 'pipe', 'ignore'],
       timeout: 1_000,
     }).trim()
diff --git a/packages/ui/tui/tests/chat-helpers.spec.ts b/packages/ui/tui/tests/chat-helpers.spec.ts
new file mode 100644
index 0000000000..17b8c29c0d
--- /dev/null
+++ b/packages/ui/tui/tests/chat-helpers.spec.ts
@@ -0,0 +1,29 @@
+import { execFileSync } from 'node:child_process'
+import { afterEach, describe, expect, it, vi } from 'vitest'
+import { gitBranch } from '../src/chat/helpers.ts'
+
+vi.mock('node:child_process', () => ({
+  execFileSync: vi.fn(() => 'main\n'),
+}))
+
+afterEach(() => {
+  vi.unstubAllEnvs()
+  vi.clearAllMocks()
+})
+
+describe('chat helpers', () => {
+  it('scrubs ambient credentials and DSH names from the Git child', () => {
+    vi.stubEnv('TUI_TEST_PASSWORD', 'ambient-password')
+    vi.stubEnv('DSH_TUI_TEST_FLAG', 'ambient-harness-state')
+    expect(gitBranch('/workspace')).toBe('main')
+    const call = vi.mocked(execFileSync).mock.calls[0] as unknown as [
+      string,
+      string[],
+      { env: NodeJS.ProcessEnv },
+    ]
+    expect(call[0]).toBe('git')
+    expect(call[1]).toEqual(['branch', '--show-current'])
+    expect(call[2].env).not.toHaveProperty('TUI_TEST_PASSWORD')
+    expect(call[2].env).not.toHaveProperty('DSH_TUI_TEST_FLAG')
+  })
+})
diff --git a/packages/ui/tui/tsconfig.json b/packages/ui/tui/tsconfig.json
index 3560d6bc9d..1cfc20fd88 100644
--- a/packages/ui/tui/tsconfig.json
+++ b/packages/ui/tui/tsconfig.json
@@ -56,6 +56,9 @@
     {
       "path": "../../skill/skill"
     },
+    {
+      "path": "../../subprocess/subprocess"
+    },
     {
       "path": "../user-interaction"
     },
diff --git a/packages/ui/user-approval/README.i18n.yaml b/packages/ui/user-approval/README.i18n.yaml
index 083bc84d20..453c761357 100644
--- a/packages/ui/user-approval/README.i18n.yaml
+++ b/packages/ui/user-approval/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/ui/user-approval/README.md
 README.md: 38bcfbfe81c3ff5f16d1835259bd4c35a06dcb64
-README.zh.md: 2a3a6d08d66c70a22b3a23a2341efc0452b8a782
+README.zh.md: 7f2678d8572b191ec88a326374420dde7deed3dc
diff --git a/packages/ui/user-approval/README.zh.md b/packages/ui/user-approval/README.zh.md
index 2a3a6d08d6..7f2678d857 100644
--- a/packages/ui/user-approval/README.zh.md
+++ b/packages/ui/user-approval/README.zh.md
@@ -4,9 +4,9 @@
 
 与通道无关的一次性审批 seam。`ctx.approval.request(req)` 返回 `allowed-once`、`rejected`、`cancelled` 或 `unavailable`；应答者缺失或失败时会以拒绝方式关闭，授权也只适用于所请求的操作。确切事件签名见生成的 [Cordis 目录](../../../docs/cordis-catalog/events.md)。
 
-每个请求都必须属于一个打开的 agent（智能体）轮次。服务会追加一对 `approval/asked` 与 `approval/decided` 审计记录，而模型只会看到由此产生且已写入日志的工具结果。已中止的请求会 resolve 为 `cancelled`；如果审计追加在提交前失败，请求会被拒绝，而不会返回一项未记录的决定。
+每个请求都必须属于一个尚未结束的 agent（智能体）轮次。服务会追加一对 `approval/asked` 与 `approval/decided` 审计记录，而模型只会看到由此产生且已写入日志的工具结果。已中止的请求会解析为 `cancelled`；如果审计记录的追加在提交前失败，Promise 会被拒绝，而不会返回一项未记录的决定。
 
-应答者是 `approval/request` waterfall（瀑布式事件）监听器。要回答所拥有 agent 的请求，请返回一个结果；否则调用 `next()` 委托。限定到 agent 的监听器只接收该 agent 的请求；每项部署应当组合一个终端应答者，因为同级监听器的顺序不是策略优先级机制。ACP（Agent Client Protocol）自动化桥接层为其拥有的会话提供一次性机器决定。
+应答者是 `approval/request` waterfall（瀑布式事件）监听器。要回答其负责的 agent 请求，请返回一个结果；否则调用 `next()` 委托。限定到 agent 的监听器只接收该 agent 的请求；每项部署应当组合一个最终应答者，因为同级监听器的顺序不是策略优先级机制。ACP（Agent Client Protocol）自动化桥接层为其负责的会话提供一次性机器决定。
 
 `ApprovalPolicy` 为 `'ask'` 或 `'never'`。实际值取最后一条 `approval/policy` 事件，并回退到配置；`setApprovalPolicy()` 是写入路径。`'never'` 会在交互式分发之前拒绝请求，也是提示词中唯一声明的策略。切换最多产生一条合并通知：如果覆盖发生在最后一个 `request/header` 之后，则归因于用户；否则归因于操作方／配置。
 
@@ -49,15 +49,15 @@ Approval prompts are disabled in this session: actions that require approval are
 
 #### Token 影响
 
-不会产生重复的审计 token。拒绝可能以一条少量且保留的错误替换正常工具结果，而允许会保留消费方的普通结果。
+不会产生重复的审计 token。拒绝可能以一条简短且会保留的错误信息替换正常工具结果，而允许会保留消费方的普通结果。
 
 #### KV Cache 影响
 
 仅追加；新出现的可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **请求只在打开的轮次内有效**：空闲时或轮次之间的调用方会在审计前抛出异常；持久化的轮次外审批工作流仍属延期事项。
-- **仅存在一次性授权**：结果词汇包含 `allowed-once`，但不含 `allow-always`、记忆规则、撤销或授权存储；会话策略只有 `ask`／`never`。
+- **请求只在尚未结束的轮次内有效**：在空闲时或轮次之间发起调用，会在审计前抛出异常；持久化的轮次外审批工作流仍属暂缓事项。
+- **仅存在一次性授权**：结果词汇包含 `allowed-once`，但不含 `allow-always`、已记住的规则、撤销或授权存储；会话策略只有 `ask`／`never`。
 - **请求不携带工具参数**：应答者会看到工具名称、原因和可选调用 id；ACP 机器通道要求调用 id，并会委托不含 id 的请求。
 - **没有内置应答者**：无头或组合不完整的部署会 resolve 为 `unavailable` 并以拒绝方式关闭；服务自身绝不会提示人类。
diff --git a/packages/ui/user-interaction/README.i18n.yaml b/packages/ui/user-interaction/README.i18n.yaml
index 2a3b525012..98e40aa0d3 100644
--- a/packages/ui/user-interaction/README.i18n.yaml
+++ b/packages/ui/user-interaction/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/ui/user-interaction/README.md
 README.md: d234d6677bdd772f1bbd2c979c0d41f90aef5c32
-README.zh.md: b70a61d6491e0bb0e52215cdeaeea3d728f7f153
+README.zh.md: c89210b6955a661313ca9e0e82e43da5a4d1db79
diff --git a/packages/ui/user-interaction/README.zh.md b/packages/ui/user-interaction/README.zh.md
index b70a61d649..c89210b695 100644
--- a/packages/ui/user-interaction/README.zh.md
+++ b/packages/ui/user-interaction/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-抽象用户交互 seam。它拥有 `ctx.userInteraction`：当面向模型的工具或权限插件需要暂停工作并询问人类决定时所使用的服务。
+抽象用户交互 seam。它定义 `ctx.userInteraction`，供面向模型的工具或权限插件在需要暂停工作并询问人类决定时使用。
 
 ## 服务：`UserInteractionService`（ctx 键：`userInteraction`）
 
@@ -19,7 +19,7 @@
 - `UserInteractionProvider`：包含 `ask(request)` 的 UI 实现。
 - `UserInteractionError`：`HarnessError` 的子类，包含 `EMPTY_QUESTIONS`、`NO_PROVIDER`、`DUPLICATE_PROVIDER` 和 `ASK_ABORTED` 等代码。
 
-当回答包含 `custom` 时，`selected` 为空；自定义文本会覆盖所选选项，而不是补充它们。UI 可以把跳过的条目保留为 `{ id, selected: [] }`，既维持现有回答形态，也保留该批次中的其他回答。
+当回答包含 `custom` 时，`selected` 为空；自定义文本是所选选项的替代，而不是补充。UI 可以把跳过的条目保留为 `{ id, selected: [] }`，既维持现有回答形态，也保留该批次中的其他回答。
 
 ## 职责
 
@@ -31,9 +31,9 @@
 
 #### KV Cache 影响
 
-不会直接使缓存失效；具名消费方拥有所有请求前缀变更。
+不会直接使 KV Cache 失效；请求前缀的任何变更均由上述消费方负责。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **每个上下文只能有一个提供方**：不支持路由或扇出到多个 UI；第二次注册会抛出 `DUPLICATE_PROVIDER`，未注册任何提供方时，`ask()` 会抛出 `NO_PROVIDER`，而不会降级。
-- **词汇仅包含问题表单形态**：可选选项加可选自定义文本；更丰富的交互形态（文件选择器、diff 预览确认）尚无 seam 词汇。
+- **词汇仅包含问题表单形态**：可供选择的选项加可选的自定义文本；更丰富的交互形态（文件选择器、diff 预览确认）尚无 seam 词汇。
diff --git a/packages/util/brand/README.i18n.yaml b/packages/util/brand/README.i18n.yaml
index b7f9ff0337..2c9d24d392 100644
--- a/packages/util/brand/README.i18n.yaml
+++ b/packages/util/brand/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/util/brand/README.md
 README.md: 68401d95a31ed2122386794ad5256a5cd93bb2a6
-README.zh.md: c8b773937426487e67001bc133cf7fae3aea2c6e
+README.zh.md: 98fcad0487af3ba1c2aaada4716744fb6088cdfe
diff --git a/packages/util/brand/README.zh.md b/packages/util/brand/README.zh.md
index c8b7739374..98fcad0487 100644
--- a/packages/util/brand/README.zh.md
+++ b/packages/util/brand/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-`Branded<B>` 名义类型原语：一个微小的**仅类型** 包（无运行时代码，无 harness 包依赖），由每个拥有跨边界 id 的包共享。
+`Branded<B>` 名义类型原语：一个微小的**仅类型**包（package），无运行时代码，也不依赖其他 harness 包；所有负责跨边界 id 的包都会共享它。
 
 ## `Branded` 是什么
 
@@ -19,10 +19,10 @@ export function SessionId(id: string): SessionId {
 }
 ```
 
-构造操作通过所属包中针对每个 id 的工厂完成。比较、日志记录、JSON 序列化和协议格式与普通字符串表现相同；品牌会在编译时被擦除。
+构造操作通过所属包中各 id 专用的工厂完成。比较、日志记录、JSON 序列化和协议格式（wire format）的行为与普通字符串相同；品牌信息会在编译时被擦除。
 
 ## 策略：为跨包边界的 id 添加品牌
 
 包为自己拥有的 id 添加品牌：`CallId` 位于 `dsh-llm`，共享的 agent/会话 `SessionId` 位于 `dsh-session`，`TaskId` 位于 `dsh-tasks`。为可能被混淆的跨包 id 添加品牌，但无需为每个字符串都添加。
 
-该包只负责原语。保持无依赖意味着，例如 `dsh-tasks` 可以为 `TaskId` 添加品牌，而无需仅为使用 `Branded` 而导入不相关的功能包。
+该包只负责这一原语。保持无依赖意味着，例如 `dsh-tasks` 可以为 `TaskId` 使用品牌类型，而无需仅为使用 `Branded` 而导入不相关的功能包。
diff --git a/packages/util/paths/README.i18n.yaml b/packages/util/paths/README.i18n.yaml
index 5fa3570408..a57d18b601 100644
--- a/packages/util/paths/README.i18n.yaml
+++ b/packages/util/paths/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/util/paths/README.md
 README.md: b28e684f3183d739c8e229a9b341801dbf345d86
-README.zh.md: 20cc033e4d0ee22d75c7ca315c462a7846146444
+README.zh.md: ab4e8123d19fd56749e3e7a0d59e8cd6bea0c3d0
diff --git a/packages/util/paths/README.zh.md b/packages/util/paths/README.zh.md
index 20cc033e4d..ab4e8123d1 100644
--- a/packages/util/paths/README.zh.md
+++ b/packages/util/paths/README.zh.md
@@ -16,9 +16,9 @@ DeepSeek Harness 用户数据的共享文件系统路径辅助工具。
 
 `expandHomePath()` 使用操作系统主目录展开 `~`、`~/...` 和 Windows 风格的 `~\...` 前缀。它会保留非波浪号路径和 `~user/...` 原样不变。
 
-该包刻意保持规模小且不依赖 harness，以便产品包共享用户数据路径约定，而不必彼此依赖。
+该包（package）刻意保持规模小且不依赖 harness，以便产品包共享用户数据路径约定，而不必彼此依赖。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **展开范围刻意保持狭窄**：只有单独的 `~`、`~/...` 和 `~\...` 使用当前操作系统主目录；`~alice/...` 等指定用户的形式、环境变量和 shell 表达式保持不变。
 - **辅助工具不会操作文件系统**：调用方仍负责目录创建、存在性检查、权限，以及对结果路径应用信任策略。
diff --git a/packages/util/retention/README.i18n.yaml b/packages/util/retention/README.i18n.yaml
index ad0fdc79f2..dc9529004b 100644
--- a/packages/util/retention/README.i18n.yaml
+++ b/packages/util/retention/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/util/retention/README.md
 README.md: d257075a67b35e92bce53a88fc6d002f4f4d5d9b
-README.zh.md: 433d512144612137fbfccf107cfe0987cec091fa
+README.zh.md: c97c1d5afe504c54851f35b86d7bb893b32bcea4
diff --git a/packages/util/retention/README.zh.md b/packages/util/retention/README.zh.md
index 433d512144..c97c1d5afe 100644
--- a/packages/util/retention/README.zh.md
+++ b/packages/util/retention/README.zh.md
@@ -2,11 +2,11 @@
 
 [English](README.md) | 中文
 
-一个轻依赖的**保留** 库：为必须限制返回上下文量的工具提供有界的面向模型输出。调用方将项或文本分片送入有界对象，然后取回保留的内容和精确的省略元数据。
+一个轻依赖的**保留**库：为必须限制返回上下文量的工具提供有界的面向模型输出。调用方将项或文本分片送入有界对象，然后取回保留的内容和精确的省略元数据。
 
-该库**只** 负责这个机制问题：*「我们保留了什么，又省略了什么？」*。工具专用代码保留其业务语义：文件分组、行号、退出码、提供方错误状态、每行预览截断、spill 文件以及面向模型的文案。这就是 [Agent Note](../../../.agents/notes/implemented/architecture/2026-07-06-tool-result-retention-library.md) 划定的边界。
+该库**只**负责这个机制问题：*「我们保留了什么，又省略了什么？」*。工具专用代码保留其业务语义：文件分组、行号、退出码、提供方错误状态、每行预览截断、spill 文件以及面向模型的文案。这就是 [Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-06-tool-result-retention-library.md) 划定的边界。
 
-它是**库，而非服务或插件**：没有 `ctx`，不注册任何内容，不发出任何事件。状态只存在于每个 retainer（一次累积）中，绝不跨调用。工具包直接导入它。
+它是**库，而非服务或插件**：没有 `ctx`，不注册任何内容，不发出任何事件。状态只存在于每个 retainer（一次累积）中，绝不跨调用。工具包（package）直接导入它。
 
 ## 对外接口
 
@@ -32,18 +32,18 @@ import type {
 
 ## 资源模式
 
-两个 retainer 使用独立名称，而不是同一个通用收集器，因为它们的**资源模型** 不同。
+两个 retainer 使用独立名称，而不是同一个通用收集器，因为它们的**资源模型**不同。
 
 - **`ItemRetainer` 限制有序逻辑单元**。搜索工具可收集完整结果集用于 spill 文件恢复，同时只为面向模型的预览保留前 `maxItems` 项。因为调用方会继续送入每个已观察到的项，所以省略数量是精确的。
 - **`TextRetainer` 限制面向字节的文本**。`head`、`tail` 和 `headTail` 在 `finish()` 时保留 UTF-8 边界；`headTail` 是 `dsh-spill-policy` 用于围绕 spill 文件通知构建有界预览的形态。
 
 ## `truncated` 是预算事实，绝不表示「不完整」
 
-`truncated` 表示*因为预算限制，retainer 省略了本可获得的内容*。它**不** 表示上游不完整。权限失败、跳过二进制文件、提供方部分失败、不可读候选项和无效 UTF-8 保留在工具领域字段中，绝不合并到 `truncated`。将两者混为一谈是该库命名最容易诱发的缺陷；务必保持分离。
+`truncated` 表示*因为预算限制，retainer 省略了本可获得的内容*。它**不**表示上游不完整。权限失败、跳过二进制文件、提供方部分失败、不可读候选项和无效 UTF-8 保留在工具领域字段中，绝不合并到 `truncated`。将两者混为一谈是该库命名最容易诱发的缺陷；务必保持分离。
 
 ## 字节，而非字符
 
-文本上限和 `omittedBytes` 按**字节** 计数，以保证进程/正文安全（子进程 pipe 和 HTTP 正文都是字节流）。跨越码点的分片会被正确处理：`finish()` 会修剪每个切割位置的不完整码点，使返回文本绝不在边界引入替换字符；首尾两侧会分开解码，因此绝不会跨越被省略的中间部分重建码点。按字符或行限制的预览预算属于独立的工具职责。
+文本上限和 `omittedBytes` 按**字节**计数，以保证进程/正文安全（子进程管道和 HTTP 正文都是字节流）。跨越码点的分片会被正确处理：`finish()` 会修剪每个切割位置的不完整码点，使返回文本绝不在边界引入替换字符；首尾两侧会分开解码，因此绝不会跨越被省略的中间部分重建码点。按字符或行限制的预览预算属于独立的工具职责。
 
 ## 工具映射
 
@@ -51,11 +51,11 @@ import type {
 
 | 工具 | Retainer 与策略 | 说明 |
 |---|---|---|
-| `glob` | `ItemRetainer<FsGlobEntry>`, `head` | 收集完整的已排序路径列表用于 spill 文件，同时在内联位置保留第一页。路径映射、已跳过候选项和 `incomplete` 保留在外部。 |
-| `grep` | `ItemRetainer<FlatGrepMatch>`, `head` | 收集匹配项用于 spill 文件，同时在内联位置保留第一页。每个匹配项的预览截断、分组、排序和 `incomplete` 保留在外部。 |
-| `bash` | `TextRetainer`, `tail` or `headTail` | 执行器仍负责 spill 文件、退出状态、信号、超时和后台任务。 |
-| `web_fetch` | `TextRetainer`, `head` or `headTail` | 提供方/资源上限保留为提供方事实；retainer 只提供保留文本和省略元数据。 |
-| `web_search` | `ItemRetainer<WebSearchSource>`, `head` | 当提供方返回的来源超过面向模型的结果应包含的数量时，标准化「来源已达上限」通知。 |
+| `glob` | `ItemRetainer<FsGlobEntry>`，`head` | 收集完整的已排序路径列表用于 spill 文件，同时在内联位置保留第一页。路径映射、已跳过候选项和 `incomplete` 保留在外部。 |
+| `grep` | `ItemRetainer<FlatGrepMatch>`，`head` | 收集匹配项用于 spill 文件，同时在内联位置保留第一页。每个匹配项的预览截断、分组、排序和 `incomplete` 保留在外部。 |
+| `bash` | `TextRetainer`，`tail` 或 `headTail` | 执行器仍负责 spill 文件、退出状态、信号、超时和后台任务。 |
+| `web_fetch` | `TextRetainer`，`head` 或 `headTail` | 提供方/资源上限保留为提供方事实；retainer 只提供保留文本和省略元数据。 |
+| `web_search` | `ItemRetainer<WebSearchSource>`，`head` | 当提供方返回的来源超过面向模型的结果应包含的数量时，标准化「来源已达上限」通知。 |
 
 `read` **刻意不在 v1 范围内**。其 `read-render` 辅助工具负责文件专用的分页契约：`offset`/`limit`、行号、`totalLines`、偏移越界错误、每行预览截断、针对已选窗口的字节上限。这是行窗口渲染器，而非通用保留机制。单个 `Omitted` 数量无法表示行窗口两侧。
 
@@ -87,11 +87,11 @@ const footer = formatRetentionNotice(
 
 通过渲染保留内容和省略元数据的工具消费方间接影响模型。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-不直接导致失效；指定的消费方负责其引起的任何请求前缀变更。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **项保留只支持 `head`**：tail、head/tail、分页、分组和提供方完整性语义仍由工具负责。
 - **文本保留面向字节**：`read` 分页等行窗口和字符窗口需要单独的渲染器；切割可能会丢弃部分 UTF-8 边界字节，以保持返回文本有效。
diff --git a/packages/util/timeout/README.i18n.yaml b/packages/util/timeout/README.i18n.yaml
index 89bf05b258..0436cc4d34 100644
--- a/packages/util/timeout/README.i18n.yaml
+++ b/packages/util/timeout/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/util/timeout/README.md
 README.md: 11c55a45a1255e14fb551e42ba3965453dbd94ae
-README.zh.md: 699fb62a4fa19f97d234f83b21f639f61a9f0777
+README.zh.md: 8b63f00595139f9af2e9a31e8ab3c3494088e0ff
diff --git a/packages/util/timeout/README.zh.md b/packages/util/timeout/README.zh.md
index 699fb62a4f..8b63f00595 100644
--- a/packages/util/timeout/README.zh.md
+++ b/packages/util/timeout/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-超时的**时序与分类** 部分：一个零依赖纯函数库（无运行时 harness 依赖），由每个需要限制调用方超时提示、启动 deadline，并在之后区分「已超时」与「已取消」的功能共享。
+超时的**时序与分类**部分：一个零依赖纯函数库（无运行时 harness 依赖），由每个需要限制调用方超时提示、启动 deadline，并在之后区分「已超时」与「已取消」的功能共享。
 
-它**不负责终止**。它发出的信号只会*通知*；真正停止工作仍由各功能负责，因为机制各不相同：bash 对操作系统进程组发送 SIGKILL，web 拆除 `fetch` socket，没有任何共享层能够承担全部终止机制。[Agent Note](../../../.agents/notes/implemented/architecture/2026-07-06-timeout-deadline-library.md) 将边界划定为：共享时序/分类，将强制终止保留在本地。
+它**不负责终止**。它发出的信号只会*通知*；真正停止工作仍由各功能负责，因为机制各不相同：bash 对操作系统进程组发送 SIGKILL，web 关闭 `fetch` 套接字，没有任何共享层能够承担全部终止机制。[Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-06-timeout-deadline-library.md) 将边界划定为：共享时序/分类，将强制终止保留在本地。
 
 它是**库，而非服务或插件**：没有 `ctx`，不注册任何内容，不持有状态，也不发出事件。「超时服务」必须了解如何停止每项功能的工作，这正是微内核要排除在共享层之外的知识。
 
@@ -16,16 +16,16 @@ import { clampTimeout, deadline, idleWatchdog, MAX_TIMER_DELAY_MS, timeoutOf, Ti
 
 | 导出项 | 职责 |
 |---|---|
-| `clampTimeout(requested, def, max, name?)` | 验证调用方可选的正有限提示，从 `def` 填充，并限制在 `max` 以内。如果提示不为正数或有限数，则抛出错误（包含 `name`）。 |
+| `clampTimeout(requested, def, max, name?)` | 验证调用方可选的、值为正且有限的提示，从 `def` 填充，并限制在 `max` 以内。如果提示为非正数或非有限数，则抛出错误（包含 `name`）。 |
 | `deadline(upstream, timeoutMs, code)` | 将 `upstream` 取消与超时融合为一个 `AbortSignal`（`AbortSignal.any`）；超时携带 `TimeoutReason`。`[Symbol.dispose]` 清除 timer。 |
-| `idleWatchdog(upstream, timeoutMs, code)` | 保持一个稳定的融合信号，并且只在受保护的异步迭代器 `next()` 尚未完成时启动。解析后取消启动；后续需求重新启动；dispose 清除；并发需求被拒绝。 |
-| `MAX_TIMER_DELAY_MS` | Node 在不将延迟限制为 1 毫秒时可调度的最大延迟（`2_147_483_647`）。拥有 timer 的配置不得超过该值。 |
+| `idleWatchdog(upstream, timeoutMs, code)` | 保持一个稳定的融合信号，并且只在受保护的异步迭代器 `next()` 尚未完成时启动 timer。完成后停止 timer；后续需求重新启动 timer；dispose（资源释放）时清除；并发需求被拒绝。 |
+| `MAX_TIMER_DELAY_MS` | Node 在不将延迟限制为 1 毫秒时可调度的最大延迟（`2_147_483_647`）。负责 timer 的配置不得超过该值。 |
 | `timeoutOf(signal \| { reason }, code?)` | 从已中止的信号/错误中恢复 `TimeoutReason`，否则返回 `undefined`，即超时与取消的分类器。传入 `code` 可仅匹配这个 deadline 的 timer（见下文的嵌套）。 |
-| `TimeoutReason` | 印在超时中止上的内部原因（`code` + `timeoutMs`）。它不是公开错误；提供方将其转换为自己的错误/字段。 |
+| `TimeoutReason` | 标记在超时中止上的内部原因（`code` + `timeoutMs`）。它不是公开错误；提供方将其转换为自己的错误/字段。 |
 
 ## `timeoutMs <= 0` 哨兵值
 
-`0` 是后端自有后台工作（bash `start()`）使用的「无超时」值，其可见范围为：**内部**。`deadline()` 不启动 timer，只转发 `upstream`；如果也没有 upstream，它将返回永不中止的信号和无操作 disposer，因此每个调用方都能保持同一种调用形态。外部请求提示会通过 `clampTimeout` 验证为**正有限数**，之后才进入 `deadline`，因此 `0` 绝不是面向模型/插件的「禁用超时」值。
+`0` 是后端自有后台工作（bash `start()`）使用的**内部**「无超时」值。`deadline()` 不启动 timer，只转发 `upstream`；如果也没有 upstream，它将返回永不中止的信号和无操作 disposer，因此每个调用方都能保持同一种调用形态。外部请求提示会通过 `clampTimeout` 验证为**正有限数**，之后才进入 `deadline`，因此 `0` 绝不是面向模型/插件的「禁用超时」值。
 
 ## 使用形态
 
@@ -44,27 +44,27 @@ export async function runWithDeadline(upstream: AbortSignal | undefined, timeout
 }
 ```
 
-该信号只会*通知*；调用方必须连接自己的终止机制（`d.signal.addEventListener('abort', kill)`，或将 `d.signal` 传给 `fetch`）。让 promise 与 timer 竞速，会在子进程或 socket 泄漏的情况下就解析工具调用；发出信号则会强制要求存在真正的终止路径。
+该信号只会*通知*；调用方必须接入自己的终止机制（`d.signal.addEventListener('abort', kill)`，或将 `d.signal` 传给 `fetch`）。让 promise 与 timer 竞速，会在子进程或套接字仍在泄漏时就让工具调用完成；发出信号则会强制要求存在真正的终止路径。
 
 将你自己的 `code` 传给 `timeoutOf`，以便分类可在嵌套中组合：当你收到的 `upstream` *本身*就是 deadline 信号时（未来启动每次调用 deadline 的 `tools/execute` 中间件），如果外层 timer 首先触发，`AbortSignal.any` 会保留外层 `TimeoutReason`。将范围限定为你的 `code`，可将外部超时视为普通 upstream 取消，这才是你所属功能视角下的正确分类，而不会在本地 timer 尚未到期时就声称自己超时。
 
-对于流式传输，创建一个 `idleWatchdog`，将其稳定的 `signal` 传入传输，并为每次提供方读取调用 `watchdog.next(iterator)`。间隔必须为正有限数，且不得超过 `MAX_TIMER_DELAY_MS`；否则 Node 会将其限制为 1 毫秒。它只测量尚未完成的需求，因此当下游代码进行渲染或在请求下一个分片前以其他方式等待时，timer 不会运行。该原语仍然只会通知，因此传输必须观察稳定信号；DeepSeek 和 pi-ai 适配器证明，超时会关闭它们的真实响应正文或 SDK 请求。
+对于流式传输，创建一个 `idleWatchdog`，将其稳定的 `signal` 传给传输层，并为提供方的每次读取调用 `watchdog.next(iterator)`。间隔必须为正有限数，且不得超过 `MAX_TIMER_DELAY_MS`；否则 Node 会将其限制为 1 毫秒。它只对尚未完成的读取请求计时，因此当下游代码进行渲染或在请求下一个分片前以其他方式等待时，timer 不会运行。该原语仍然只会通知，因此传输层必须观察稳定信号；DeepSeek 和 pi-ai 适配器证明，超时会关闭它们的真实响应正文或 SDK 请求。
 
 ## 哪些操作不设置超时
 
-本地文件 `read`/`write`/`edit` 不接受 `timeoutMs`：系统调用最多只能尽力中止，超时无法强制 `fsync`/`rename` 停止，而添加超时将成为违反显式优于隐式的默认值。详见 [`fs/`](../../fs/README.md)。
+本地文件 `read`/`write`/`edit` 不接受 `timeoutMs`：中止系统调用至多只能尽力而为，超时无法强制 `fsync`/`rename` 停止，而添加超时会引入一个违反显式优于隐式原则的隐式默认值。详见 [`fs/`](../../fs/README.md)。
 
 ## 模型体验
 
 通过 `dsh-timeout-policy` 等消费方间接影响模型；消费方可能会将提供方结果替换为已保留的超时错误，或抑制延迟结果。
 
-#### KV 缓存影响
+#### KV Cache 影响
 
-不直接导致失效；指定的消费方负责其引起的任何请求前缀变更。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
-## 已知限制与待完成工作
+## 已知限制与暂缓事项
 
 - **只发出通知**：deadline 无法停止忽略其信号的工作；每项功能仍需要自己的 socket/进程/任务终止路径。
 - **`timeoutMs <= 0` 是内部词汇**：只有在所属后端已解析策略后，它才会禁用本地 timer；绝不会作为面向模型/插件的公开开关。
 - **第一个中止原因决定分类**：当 upstream 取消早于本地 timer 发生时，即使自己的超时之后也会到期，该层也无法再报告。
-- **空闲 watchdog 不是总 deadline**：它针对每个尚未完成的迭代器需求重新启动，并刻意排除消费方的思考时间。
+- **空闲 watchdog 不是总 deadline**：它针对每个尚未完成的迭代器需求重新启动，并刻意排除消费方的处理时间。
diff --git a/packages/web/README.i18n.yaml b/packages/web/README.i18n.yaml
index 8d554a4573..636e5ddd63 100644
--- a/packages/web/README.i18n.yaml
+++ b/packages/web/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/web/README.md
 README.md: 8cd173b922ead32219ffab2b2d6b6428b3ce375c
-README.zh.md: b10849bbddf6d62701d72b064a58c728ffb1e1b7
+README.zh.md: 42f09e19d0faf709cc4f669ae5f7835b809a79a5
diff --git a/packages/web/README.zh.md b/packages/web/README.zh.md
index b10849bbdd..42f09e19d0 100644
--- a/packages/web/README.zh.md
+++ b/packages/web/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-web 访问能力 seam：抽象 web 接口、搜索／抓取提供方实现，以及面向模型的 web 工具。这些全是**产品** 包。
+web 访问能力 seam：抽象 web 接口、搜索／抓取提供方实现，以及面向模型的 web 工具。这些全是**产品**包（package）。
 
 | 包 | 职责 | ctx key |
 |---|---|---|
@@ -10,9 +10,9 @@ web 访问能力 seam：抽象 web 接口、搜索／抓取提供方实现，以
 | `web-search-exa/` | Exa 支持的 `WebSearchProvider` | （注册到 `ctx.web`） |
 | `web-search-perplexity/` | Perplexity 支持的 `WebSearchProvider` | （注册到 `ctx.web`） |
 | `web-search-deepseek/` | DeepSeek 支持的 `WebSearchProvider`，通过 Anthropic 兼容 API 使用原生 `web_search` | （注册到 `ctx.web`） |
-| `web-fetch-local/` | 匿名公共 HTTP(S) `WebFetchProvider` | （注册到 `ctx.web`） |
+| `web-fetch-local/` | 用于匿名访问公共 HTTP(S) 的 `WebFetchProvider` | （注册到 `ctx.web`） |
 | `tool-web/` | 面向模型的 `web_search`／`web_fetch` 工具 schema | （注册到 `ctx.tools`） |
 
-接口位于 `web/web/`。与 bash/fs 不同，该 seam 跨越**两种能力**（搜索和抓取），每种能力都可能有多个提供方：`ctx.web` 是单一的 web 访问中间层，拥有一项提供方选择策略、一套中止／错误词汇，以及一个面向产品的「该 harness 如何访问 web」配置表层。提供方注册的是**能力** 而非工具；`tool-web` 是面向模型名称、schema、提示词指引和呈现的唯一 owner。替换搜索提供方不会改变模型提出查询的方式，替换抓取实现也不会改变模型请求 URL 的方式。
+接口位于 `web/web/`。与 bash/fs 不同，该 seam 跨越**两种能力**（搜索和抓取），每种能力都可能有多个提供方：`ctx.web` 是单一的 web 访问中间层，拥有一项提供方选择策略、一套中止／错误词汇，以及一个面向产品的「该 harness 如何访问 web」配置接口。提供方注册的是**能力**而非工具；`tool-web` 是面向模型名称、schema、提示词指引和呈现的唯一负责方。替换搜索提供方不会改变模型提出查询的方式，替换抓取实现也不会改变模型请求 URL 的方式。
 
-设计原理见 [web 能力 seam Agent Note](../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)，其中也解释了搜索与抓取为何有意合并为一个 seam，以及为何暂缓实现 `web_fetch` 的 SSRF 防护。
+设计原理见 [web 能力 seam Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)，其中也解释了搜索与抓取为何有意合并为一个 seam，以及为何暂缓实现 `web_fetch` 的 SSRF 防护。
diff --git a/packages/web/tool-web/README.i18n.yaml b/packages/web/tool-web/README.i18n.yaml
index 58641384b9..d239e581fa 100644
--- a/packages/web/tool-web/README.i18n.yaml
+++ b/packages/web/tool-web/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/web/tool-web/README.md
 README.md: 9b78920b1b6c611118294421dec1e75e381ed5d6
-README.zh.md: 2152c40f1ccac2272fa0b2681514a712417c0ad3
+README.zh.md: d36258d3a5bd8af6716e1fd9c3384389e8395e23
diff --git a/packages/web/tool-web/README.zh.md b/packages/web/tool-web/README.zh.md
index 2152c40f1c..d36258d3a5 100644
--- a/packages/web/tool-web/README.zh.md
+++ b/packages/web/tool-web/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-面向模型的 web 工具套件 `web_search` 与 `web_fetch`，构建于 [web 能力 seam](../web/README.md)（`ctx.web`）之上。它只拥有面向模型的事项：工具名称、JSON schema、snake_case 参数名称、提示词区段、结果数量上限、结果格式、HTML→markdown 呈现，以及 `presentCall`。所有 web 访问都通过 `ctx.web`；该包绝不导入具体提供方。两个工具都不公开面向模型的超时：每个工具的协作式工具调用预算通过配置在此声明（`fetchTimeoutMs`／`searchTimeoutMs`，附加为 `ToolDefinition.timeoutMs`），由 [`@deepseek-ai/dsh-timeout-policy`](../../timeout/timeout-policy/README.md)（`tools/execute` 包装层）强制执行；每个工具只把 `exec.signal` 转发给 seam。
+面向模型的 web 工具套件 `web_search` 与 `web_fetch`，构建于 [web 能力 seam](../web/README.md)（`ctx.web`）之上。它只负责面向模型的事项：工具名称、JSON Schema、snake_case 参数名称、提示词区段、结果数量上限、结果格式、HTML→markdown 呈现，以及 `presentCall`。所有 web 访问都通过 `ctx.web`；该包（package）绝不导入具体提供方。两个工具都不公开面向模型的超时：每个工具的协作式工具调用超时预算通过配置在此声明（`fetchTimeoutMs`／`searchTimeoutMs`，附加为 `ToolDefinition.timeoutMs`），由 [`@deepseek-ai/dsh-timeout-policy`](../../timeout/timeout-policy/README.md)（`tools/execute` 包装层）强制执行；每个工具只把 `exec.signal` 转发给 seam。
 
 每个工具独立注册；只需要其中一个工具的产品可以通过配置禁用另一个（`{ search: false }`／`{ fetch: false }`）。
 
@@ -10,20 +10,20 @@
 
 | 工具 | 参数 | 行为 |
 |---|---|---|
-| `web_search` | `query`（string） | 发现。返回可选答案与源 URL。`max_results` **不** 面向模型：工具设置上限（`searchMaxResults` 配置，默认 8）并传给 seam。 |
+| `web_search` | `query`（string） | 用于发现信息。返回可选答案与来源 URL。`max_results` **不**面向模型：工具设置上限（`searchMaxResults` 配置，默认 8）并传给 seam。 |
 | `web_fetch` | `url`（string） | 获取特定 URL。HTML 主体渲染为 markdown（turndown，带 GFM 表格／删除线）；文本主体原样通过。非 2xx 状态会报告，而非报错。工具调用超时是部署策略（`dsh-timeout-policy`），不是模型参数。 |
 
-两个工具都选择并发调度，因为提供方读取会返回内容，不会修改父 agent 状态。
+两个工具都选择并发调度，因为提供方读取会返回内容，不会修改父 agent（智能体）的状态。
 
-规范化 seam 结果也是规范工具值：`WebSearchResult` 与 `WebFetchResult`。原生 renderer 保留下述答案／源与抓取主体文本；提供方搜索／主体上限仍是获取限制，而非仅呈现截断。
+规范化后的 seam 结果也是标准工具值：`WebSearchResult` 与 `WebFetchResult`。原生渲染器会保留下文所述的答案、来源和抓取正文文本；提供方对搜索结果数量和正文大小的上限仍属于获取限制，而非仅用于呈现的截断。
 
 ## 配置
 
-| Key | 默认值 | 含义 |
+| 配置键 | 默认值 | 含义 |
 |---|---|---|
 | `search` | `true` | 注册 `web_search`。 |
 | `fetch` | `true` | 注册 `web_fetch`。 |
-| `searchMaxResults` | `8` | 一次 `web_search` 调用返回的源数量上限（seam 截断更长的提供方列表并标记）。 |
+| `searchMaxResults` | `8` | 一次 `web_search` 调用返回的来源数量上限（seam 截断更长的提供方列表并标记）。 |
 | `fetchTimeoutMs` | `30000` | `web_fetch` 的协作式工具调用超时预算（ms）。 |
 | `searchTimeoutMs` | `30000` | `web_search` 的协作式工具调用超时预算（ms）。 |
 | `fetchMaxOutputChars` | `200000` | 同步转换的源字符数与单次完整 `web_fetch` 输出的上限（状态头、渲染后的主体与页脚合并计算）；主体被截断时，在能容纳的情况下附带截断提示。 |
@@ -37,9 +37,9 @@
 
 ## 稳定注册
 
-工具注册遵循产品**启用状态**，而非后端可用性。即使选中的提供方缺失、错误配置、存在歧义或暂时不可用，工具仍保持可见；seam 在执行时解析提供方，执行以结构化 `WebError`（例如 `WEB_PROVIDER_UNAVAILABLE`、`WEB_PROVIDER_AMBIGUOUS`）失败，`ToolRegistry.execute()` 会把它转为模型可读、hook／UI 可路由的错误工具结果。这样无需把插件加载顺序、credential 状态或 HMR 时机纳入面向模型契约，也能保持模型 schema 稳定。要彻底移除 web 工具，请在此处通过配置将其禁用。
+工具注册遵循产品**启用状态**，而非后端可用性。即使选中的提供方缺失、错误配置、存在歧义或暂时不可用，工具仍保持可见；seam 在执行时解析提供方，执行以结构化 `WebError`（例如 `WEB_PROVIDER_UNAVAILABLE`、`WEB_PROVIDER_AMBIGUOUS`）失败，`ToolRegistry.execute()` 会把它转为模型可读、钩子／UI 可路由的错误工具结果。这样无需把插件加载顺序、凭据状态或 HMR（热模块替换）时机纳入面向模型契约，也能保持模型 schema 稳定。要彻底移除 web 工具，请在此处通过配置将其禁用。
 
-工具绝不会调用提供方的 `available()`，也不会枚举提供方；唯一执行路径是 `ctx.web.search()`／`ctx.web.fetch()`，提供方不可用会作为选择机制在执行时抛出的结构化 `WebError` code 到达工具。提供方选择完全留在 seam 内，只有一个 owner。
+工具绝不会调用提供方的 `available()`，也不会枚举提供方；唯一执行路径是 `ctx.web.search()`／`ctx.web.fetch()`，提供方不可用时，选择机制会在执行阶段抛出结构化 `WebError`，其错误码由工具接收。提供方选择完全留在 seam 内，由单一主体负责。
 
 ## 模型体验
 
@@ -63,7 +63,7 @@ Use the web_fetch tool to retrieve the content of a specific HTTP(S) URL (for ex
 
 #### Token 影响
 
-每个通过配置启用的工具会为每次请求增加固定指引成本，即使限制隐藏了其 schema。
+每个通过配置启用的工具都会为每次请求增加固定的指引 token 开销，即使限制隐藏了其 schema。
 
 #### KV Cache 影响
 
@@ -77,7 +77,7 @@ Use the web_fetch tool to retrieve the content of a specific HTTP(S) URL (for ex
 
 #### Token 影响
 
-每次请求承担固定 schema 成本；通过配置禁用会同时移除 schema 与指引，scope 限制只移除 schema。
+每次请求都会产生固定的 schema token 开销；通过配置禁用会同时移除 schema 与指引，scope 限制只移除 schema。
 
 #### KV Cache 影响
 
@@ -87,21 +87,21 @@ Use the web_fetch tool to retrieve the content of a specific HTTP(S) URL (for ex
 
 #### 模型看到的内容
 
-可选的提供方答案之后是 `Sources:`，再跟随数据相关、形状精确为 `- [<title-or-url>](<url>)` 的行，并可添加后缀 ` — <snippet> (<publishedAt>)`。既无答案也无源时，结果显示 `No results found.`。列表达到上限时会添加 `(Showing the first <count> sources. Refine the query for more.)`；每项结果都以 `Cite the relevant URLs above as markdown links in your answer.` 结尾。
+可选的提供方答案之后是 `Sources:`，再跟随内容取决于数据且格式严格为 `- [<title-or-url>](<url>)` 的行，并可添加后缀 ` — <snippet> (<publishedAt>)`。既无答案也无来源时，结果显示 `No results found.`。列表达到上限时会添加 `(Showing the first <count> sources. Refine the query for more.)`；每个结果都以 `Cite the relevant URLs above as markdown links in your answer.` 结尾。
 
 #### Token 影响
 
-数据相关结果会重复发送直到压缩，源数量由 `searchMaxResults` 限制。
+数据相关结果会重复发送直到压缩（compaction），来源数量由 `searchMaxResults` 限制。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 抓取结果
 
 #### 模型看到的内容
 
-成功抓取的精确形状是 `Fetched <finalUrl> (HTTP <statusCode>)`、一个空行，以及提供方拥有的解码主体。发生截断时会再添加一个空行和 `(Content truncated. Fetch a more specific URL or section for the full text.)`；失败变为 `Error: <message>`。查询与 URL 保留在调用历史中。
+成功抓取的精确形状是 `Fetched <finalUrl> (HTTP <statusCode>)`、一个空行，以及由提供方返回的已解码正文。发生截断时会再添加一个空行和 `(Content truncated. Fetch a more specific URL or section for the full text.)`；失败变为 `Error: <message>`。查询与 URL 保留在调用历史中。
 
 #### Token 影响
 
@@ -109,7 +109,7 @@ Use the web_fetch tool to retrieve the content of a specific HTTP(S) URL (for ex
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ### 参数错误
 
@@ -123,10 +123,10 @@ Use the web_fetch tool to retrieve the content of a specific HTTP(S) URL (for ex
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
-- **HTML→markdown 转换会在 GFM 无法安全表示的输入上降级**：[turndown](https://github.com/mixmark-io/turndown)（带 GFM 表格／删除线）通过真实 DOM 转换至多 `fetchMaxOutputChars` 个源字符。保守的 512 层词法守卫会将深层或嵌套有歧义的主体作为原始 HTML 直接透传，转换异常也会如此处理；表格的 `colspan` 会被忽略，因为 GFM 无法表示跨列单元格。这些限制可避免阻塞事件循环，也避免不受信任的数值属性使输出膨胀（[决策记录](../../../.agents/notes/implemented/simplification/2026-07-26-turndown-for-tool-web-html-markdown.md)）。
-- **面向模型的表层有意保持最小，提升项暂缓**：`max_results` 保持为配置上限（不是模型参数），`web_fetch` 只接受 `url`（没有 `format`／`prompt`／LLM 摘要模式）；两项都列为 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md) 中的后续步骤。
-- **没有 web 专用权限策略**：两个工具都不会请求 `ctx.approval` 就直接执行；需要确认的部署必须添加 `tools/pre-execute` 策略，该包不定义持久 URL／domain 授权。
+- **HTML→markdown 转换会在 GFM 无法安全表示的输入上降级**：[turndown](https://github.com/mixmark-io/turndown)（带 GFM 表格／删除线）通过真实 DOM 转换至多 `fetchMaxOutputChars` 个源字符。保守的 512 层词法守卫会将深层或嵌套有歧义的主体作为原始 HTML 直接透传，转换异常也会如此处理；表格的 `colspan` 会被忽略，因为 GFM 无法表示跨列单元格。这些限制可避免阻塞事件循环，也避免不受信任的数值属性使输出膨胀（[Agent Note（agent 决策记录）](../../../.agents/notes/implemented/simplification/2026-07-26-turndown-for-tool-web-html-markdown.md)）。
+- **面向模型的接口有意保持精简，后续扩展暂缓**：`max_results` 保持为配置上限（不是模型参数），`web_fetch` 只接受 `url`（没有 `format`／`prompt`／LLM（大语言模型）摘要模式）；两项都列为 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md) 中的后续步骤。
+- **没有 web 专用权限策略**：两个工具都不会请求 `ctx.approval` 就直接执行；需要确认的部署必须添加 `tools/pre-execute` 策略，该包不定义持久化的 URL／域名授权。
diff --git a/packages/web/web-fetch-local/README.i18n.yaml b/packages/web/web-fetch-local/README.i18n.yaml
index 7da50f07d8..dc8929c47e 100644
--- a/packages/web/web-fetch-local/README.i18n.yaml
+++ b/packages/web/web-fetch-local/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/web/web-fetch-local/README.md
 README.md: 8cadba2de7a2708252ebc7143840825fd4fe4549
-README.zh.md: bd6257b41736153a608df93c5baa74eac38d50e3
+README.zh.md: e4ae567d9b763d0fc1192b59d89ac5304c1c34de
diff --git a/packages/web/web-fetch-local/README.zh.md b/packages/web/web-fetch-local/README.zh.md
index bd6257b417..e4ae567d9b 100644
--- a/packages/web/web-fetch-local/README.zh.md
+++ b/packages/web/web-fetch-local/README.zh.md
@@ -2,30 +2,30 @@
 
 [English](README.md) | 中文
 
-一个匿名公共 HTTP(S) `WebFetchProvider`，用于 harness [web 能力 seam](../web/README.md)（`ctx.web`）。它获取具体 URL，返回状态码与有界解码内容。
+一个匿名公共 HTTP(S) `WebFetchProvider`，用于 harness [web 能力 seam](../web/README.md)（`ctx.web`）。它获取具体 URL，返回状态码和长度受限的解码内容。
 
-这是一个**实现** 包：它向 `ctx.web` 注册提供方，不拥有该 key，也不注册面向模型的工具。它是函数／namespace 插件（`inject: ['web']`）。
+这是一个**实现**包（package）：它向 `ctx.web` 注册提供方，不拥有该键，也不注册面向模型的工具。它是函数／命名空间插件（`inject: ['web']`）。
 
 ## 职责拆分
 
 提供方拥有**安全资源获取**：URL 验证、HTTP 传输、重定向策略、资源兜底超时、中止传播、字节上限、charset 解码、内容类型分类与二进制拒绝。`@deepseek-ai/dsh-tool-web` 拥有**呈现**（HTML→markdown、截断格式）。非 2xx HTTP 响应是*结果*（状态码 + 解码主体），不是错误；`WebError` 只用于无法安全获取或表示资源的失败。
 
-提供方的 `timeoutMs` 是直接 `ctx.web.fetch()` 调用方与错误配置部署的资源兜底，不是面向模型的工具调用预算。[`dsh-timeout-policy`](../../timeout/timeout-policy/README.md) 拥有 `web_fetch` 工具调用预算，并通过武装 `exec.signal` 强制执行该预算。
+提供方的 `timeoutMs` 是直接 `ctx.web.fetch()` 调用方和配置有误的部署所用的资源兜底，不是面向模型的工具调用预算。[`dsh-timeout-policy`](../../timeout/timeout-policy/README.md) 拥有 `web_fetch` 工具调用预算，并让 `exec.signal` 在超时时触发，以强制执行该预算。
 
-已交付的 web 工具部署会把提供方兜底设为高于工具预算，因此模型调用通常返回 `TOOL_TIMEOUT`。如果外层 deadline 先到达提供方，提供方报告 `WEB_ABORTED`，外层策略再将其替换为 `TOOL_TIMEOUT`。因此，`WEB_FETCH_TIMEOUT` 标识提供方预算已经耗尽的直接 seam 调用方。
+已交付的 web 工具部署会把提供方兜底设为高于工具预算，因此模型调用通常返回 `TOOL_TIMEOUT`。如果外层截止期限先于提供方的兜底超时触发，提供方会报告 `WEB_ABORTED`，外层策略再将其替换为 `TOOL_TIMEOUT`。因此，`WEB_FETCH_TIMEOUT` 表明直接 seam 调用方的提供方预算已经耗尽。
 
 ## 传输卫生
 
-- 只接受 `http:` 和 `https:` URL；拒绝 URL 中的 credential（`WEB_BLOCKED_URL`）以及过长／格式错误的 URL（`WEB_INVALID_URL`）。
+- 只接受 `http:` 和 `https:` URL；拒绝 URL 中的凭据（`WEB_BLOCKED_URL`）以及过长／格式错误的 URL（`WEB_INVALID_URL`）。
 - 强制执行 URL 最大长度、响应字节上限（`WEB_FETCH_TOO_LARGE`）、解码主体字符上限、超时（`WEB_FETCH_TIMEOUT`）和重定向跳数上限。
 - 把调用方的中止信号（`WEB_ABORTED`）传播到网络请求与流式读取。
-- 只跟随**同源** 重定向；跨源重定向以 `WEB_REDIRECT_BLOCKED` 失败，要求发起新的工具调用（沿用 Claude Code 的 WebFetch 模型）。
+- 只跟随**同源**重定向；跨源重定向以 `WEB_REDIRECT_BLOCKED` 失败，要求发起新的工具调用（沿用 Claude Code 的 WebFetch 模式）。
 - 发送显式的产品 `User-Agent`，绝不伪装成浏览器。
 - 不受支持的内容类型（例如二进制）以 `WEB_UNSUPPORTED_CONTENT_TYPE` 拒绝。
 
 ## 配置
 
-| Key | 默认值 | 含义 |
+| 配置键 | 默认值 | 含义 |
 |---|---|---|
 | `maxUrlLength` | `2048` | 接受的请求 URL 最大长度。 |
 | `maxResponseBytes` | `5_000_000` | 响应主体最大字节数。 |
@@ -38,14 +38,14 @@
 
 ## 模型体验
 
-通过 [`dsh-tool-web`](../tool-web/README.md) 间接影响；该工具把此提供方经 `maxBodyChars` 限制的解码文本或 markdown 形状 HTML 置于抓取结果包装中，并保留提供方失败；重定向、标头与传输机制保持隐藏。
+通过 [`dsh-tool-web`](../tool-web/README.md) 间接影响；该工具把此提供方经 `maxBodyChars` 限制的解码文本或 markdown 形状 HTML 置于抓取结果包装层中，并保留提供方失败；重定向、标头与传输机制保持隐藏。
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **SSRF／私有网络防护暂缓**：不会阻止私有、loopback、link-local、multicast 或其他非公开目标，也不进行 DNS 解析后验证或逐跳重新验证（见 [web 能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。在此功能落地前，该提供方是 SSRF 原语；能够访问敏感内部网络目标的部署**禁止启用它**。
+- **SSRF／私有网络防护暂缓**：不会阻止私有、loopback、link-local、multicast 或其他非公开目标，也不进行 DNS 解析后验证或逐跳重新验证（见 [web 能力 seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。在此功能落地前，该提供方是 SSRF 原语；能够访问敏感内部网络目标的部署**禁止启用它**。
 - **只解码文本内容**：包括 html/xhtml 与 `text/*` 加 JSON/XML 家族；缺少 `Content-Type` 或任何二进制类型都会抛出 `WEB_UNSUPPORTED_CONTENT_TYPE`，可提取文本的 PDF 解码属于明确的暂缓工作。
-- **charset 只来自 `Content-Type` 标头**（默认为 UTF-8）：HTML `<meta charset>` 声明会被忽略；声明但无法识别的 charset label 会抛出异常，而非回退。
+- **charset 只来自 `Content-Type` 标头**（默认为 UTF-8）：HTML `<meta charset>` 声明会被忽略；声明但无法识别的 charset 标签会抛出异常，而非回退。
diff --git a/packages/web/web-search-deepseek/README.i18n.yaml b/packages/web/web-search-deepseek/README.i18n.yaml
index a65feb5185..83129cba21 100644
--- a/packages/web/web-search-deepseek/README.i18n.yaml
+++ b/packages/web/web-search-deepseek/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/web/web-search-deepseek/README.md
 README.md: 54eb7561b9d81a9e2da565e3870abe094dbe984d
-README.zh.md: 5ce5f46882efb94c7d12fd137b77b167a0088741
+README.zh.md: 8862b8a9c1ba69d247942bb6e41289214826c682
diff --git a/packages/web/web-search-deepseek/README.zh.md b/packages/web/web-search-deepseek/README.zh.md
index 5ce5f46882..8862b8a9c1 100644
--- a/packages/web/web-search-deepseek/README.zh.md
+++ b/packages/web/web-search-deepseek/README.zh.md
@@ -4,7 +4,7 @@
 
 由 [DeepSeek](https://deepseek.com) 支持的 `WebSearchProvider`，用于 harness [web 能力 seam](../web/README.md)（`ctx.web`）。它调用 DeepSeek 的 **Anthropic 兼容 Messages API**（`POST {baseURL}/messages`），启用原生 `web_search_20250305` 服务器工具，并把 DeepSeek 返回的结构化 `web_search_tool_result` 块映射为 seam 规范化的 `WebSearchResult`。
 
-这是一个**实现** 包：它向 `ctx.web` 注册提供方，不拥有该 key，也不注册面向模型的工具。与 `@deepseek-ai/dsh-llm-deepseek` 一样，它是函数／namespace 插件（`inject: ['web']`）。Anthropic 协议形状是提供方私有细节，并**不** 使该提供方依赖 `ctx.llm`。
+这是一个**实现**包（package）：它向 `ctx.web` 注册提供方，不拥有该键，也不注册面向模型的工具。与 `@deepseek-ai/dsh-llm-deepseek` 一样，它是函数／命名空间插件（`inject: ['web']`）。Anthropic 协议格式（wire format）是提供方私有细节，并**不**使该提供方依赖 `ctx.llm`。
 
 ## 与专用搜索端点的区别
 
@@ -12,13 +12,13 @@ Exa 和 Perplexity 提供专用搜索端点，DeepSeek 则没有。该提供方
 
 **严格模式**：如果响应不含 `web_search_tool_result` 块（未触发原生搜索），提供方会抛出 `WebError` `WEB_PROVIDER_ERROR`，而非降级为文本抓取；这种行为诚实且可诊断。
 
-它复用 `$DEEPSEEK_API_KEY`（不增加 secret），但**不会** 复用 `$DEEPSEEK_BASE_URL`：搜索端点使用 Anthropic 兼容基址（`https://api.deepseek.com/anthropic/v1`），不同于 LLM 适配器使用的 chat-completions 基址（`https://api.deepseek.com`）。
+它复用 `$DEEPSEEK_API_KEY`（不增加密钥），但**不会**复用 `$DEEPSEEK_BASE_URL`：搜索端点使用 Anthropic 兼容基址（`https://api.deepseek.com/anthropic/v1`），不同于大语言模型（LLM）适配器使用的 chat-completions 基址（`https://api.deepseek.com`）。
 
 ## 配置
 
-| Key | 默认值 | 含义 |
+| 配置键 | 默认值 | 含义 |
 |---|---|---|
-| `apiKey` | `$DEEPSEEK_API_KEY` | DeepSeek API key。为空／缺失时提供方不可用。同时作为 `x-api-key` 与 `Authorization: Bearer` 发送（官方与 Anthropic 兼容 proxy）。 |
+| `apiKey` | `$DEEPSEEK_API_KEY` | DeepSeek API 密钥。为空或缺失时提供方不可用。同时通过 `x-api-key` 和 `Authorization: Bearer` 发送（分别用于官方接口与 Anthropic 兼容代理）。 |
 | `baseURL` | `https://api.deepseek.com/anthropic/v1` | Anthropic 兼容端点基址；追加 `/messages`。覆盖时使用 `$DEEPSEEK_SEARCH_BASE_URL` 等独立环境变量；禁止复用属于 chat-completions LLM 适配器的 `$DEEPSEEK_BASE_URL`。无法解析时提供方不可用。 |
 | `model` | `deepseek-v4-flash` | Anthropic 格式模型名称。 |
 | `apiVersion` | `2023-06-01` | `anthropic-version` 标头值。 |
@@ -47,7 +47,7 @@ DeepSeek 不返回该提供方可作为 `content` 信任的提供方生成答案
 
 #### 模型看到的内容
 
-独立的 DeepSeek 模型会接收精确的 `Perform a web search for the query: <query>` 作为 user 文本，并收到一个原生 `web_search` 服务器工具定义。该请求不属于会话模型上下文。
+独立的 DeepSeek 模型会原样接收 `Perform a web search for the query: <query>` 作为用户文本，并收到一个原生 `web_search` 服务器工具定义。该请求不属于会话模型上下文。
 
 #### Token 影响
 
@@ -55,13 +55,13 @@ DeepSeek 不返回该提供方可作为 `content` 信任的提供方生成答案
 
 #### KV Cache 影响
 
-与会话请求 cache 相互独立。辅助指令与原生工具定义可以形成稳定前缀，但查询或模型路由的每次变化都会阻止从首个差异起的复用。
+与会话请求缓存相互独立。辅助指令与原生工具定义可以形成稳定前缀，但查询或模型路由的每次变化都会阻止从首个差异起的复用。
 
 ### 间接的会话工具结果
 
 #### 模型看到的内容
 
-通过 [`dsh-tool-web`](../tool-web/README.md)，会话模型会看到结构化搜索块中去重后的 URL、标题、日期与引用 snippet；提供方文本不会作为答案受到信任。该提供方的精确失败是 `DeepSeek search aborted`、`DeepSeek search request failed: <error>`、`DeepSeek returned no web_search_tool_result blocks; the request may not have triggered native web search` 和 `DeepSeek returned an unprocessable response body: <error>`；HTTP 失败保留提供方消息。错误包装属于消费方。
+通过 [`dsh-tool-web`](../tool-web/README.md)，会话模型会看到结构化搜索块中去重后的 URL、标题、日期与引用 snippet；提供方文本不会作为答案受到信任。该提供方的具体错误消息为 `DeepSeek search aborted`、`DeepSeek search request failed: <error>`、`DeepSeek returned no web_search_tool_result blocks; the request may not have triggered native web search` 和 `DeepSeek returned an unprocessable response body: <error>`；HTTP 失败保留提供方消息。错误包装属于消费方。
 
 #### Token 影响
 
@@ -69,11 +69,11 @@ DeepSeek 不返回该提供方可作为 `content` 信任的提供方生成答案
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
 - **一次搜索需要完整的 Messages 模型轮次**：会产生延迟与生成 token，并且最多执行 `maxUses` 次服务器侧搜索；DeepSeek 不公开专用检索端点。
 - **超量返回的源仍消耗 token**：协议没有结果数量旋钮，`maxResults` 只能由 seam 在事后截断。
 - **未引用的结果没有 `snippet`**：只有 `text` 块中的引用（`cited_text`）匹配其 URL 时，源才会获得 snippet。
-- **按错误形状分类中止**：只有 `DOMException` 且名为 `AbortError` 时才映射为 `WEB_ABORTED`；携带自定义原因的中止（例如 `dsh-timeout` 的 `TimeoutReason`）会呈现为 `WEB_PROVIDER_ERROR`。
+- **中止分类基于错误结构**：只有 `DOMException` 且名为 `AbortError` 时才映射为 `WEB_ABORTED`；携带自定义原因的中止（例如 `dsh-timeout` 的 `TimeoutReason`）会呈现为 `WEB_PROVIDER_ERROR`。
diff --git a/packages/web/web-search-exa/README.i18n.yaml b/packages/web/web-search-exa/README.i18n.yaml
index b8ffd244b1..e9d70dc2a8 100644
--- a/packages/web/web-search-exa/README.i18n.yaml
+++ b/packages/web/web-search-exa/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/web/web-search-exa/README.md
 README.md: c24f952eede90aae6fa23eec976255cb99625c19
-README.zh.md: 39b6fb9ffc0ce9ba3291c17aa88781e90ed18e08
+README.zh.md: 9b42c81d1fb08ac781dd8f4721bfc70fb19cfcea
diff --git a/packages/web/web-search-exa/README.zh.md b/packages/web/web-search-exa/README.zh.md
index 39b6fb9ffc..9b42c81d1f 100644
--- a/packages/web/web-search-exa/README.zh.md
+++ b/packages/web/web-search-exa/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-由 [Exa](https://exa.ai) 支持的 `WebSearchProvider`，用于 harness [web 能力 seam](../web/README.md)（`ctx.web`）。它调用 Exa 的 `POST /search` 端点并请求 highlight 内容，把扁平 `results[]` 映射为 seam 规范化的 `WebSearchResult`。
+由 [Exa](https://exa.ai) 支持的 `WebSearchProvider`，用于 harness [web 能力 seam](../web/README.md)（`ctx.web`）。它调用 Exa 的 `POST /search` 端点并请求高亮摘要内容，把扁平 `results[]` 映射为 seam 规范化的 `WebSearchResult`。
 
-这是一个**实现** 包：它向 `ctx.web` 注册提供方，不拥有 `ctx.web` key，也不注册面向模型的工具（后者属于 `@deepseek-ai/dsh-tool-web`）。与 `@deepseek-ai/dsh-llm-deepseek` 一样，它是函数／namespace 插件（`inject: ['web']`），负责注册后端，而非默认导出服务。
+这是一个**实现**包（package）：它向 `ctx.web` 注册提供方，不拥有 `ctx.web` 键，也不注册面向模型的工具（后者属于 `@deepseek-ai/dsh-tool-web`）。与 `@deepseek-ai/dsh-llm-deepseek` 一样，它是函数／命名空间插件（`inject: ['web']`），负责注册后端，而非默认导出服务。
 
 ## 配置
 
-| Key | 默认值 | 含义 |
+| 配置键 | 默认值 | 含义 |
 |---|---|---|
-| `apiKey` | `$EXA_API_KEY` | Exa API key。为空／缺失时提供方不可用。 |
+| `apiKey` | `$EXA_API_KEY` | Exa API 密钥。为空或缺失时提供方不可用。 |
 | `baseURL` | `https://api.exa.ai` | 端点基址；追加 `/search`。无法解析时提供方不可用。 |
 | `searchType` | `auto` | 以 Exa `type` 发送的检索模式：`auto`（由 Exa 决定）、`keyword` 或 `neural`。 |
 | `numResults` | （未设置） | 请求不含 `maxResults` 时使用的默认结果数。未设置时不发送默认值。必须是正整数。 |
@@ -25,18 +25,18 @@
 
 ## 映射
 
-Exa 返回扁平 `results[]`，不返回生成答案，因此省略 `content`。每项结果映射为 `WebSearchSource`：`url` ← `url`、`title` ← `title`、`snippet` ← 第一个非空 `highlights[]` 配置项（没有 highlight 的结果缺少可移植 snippet，会被丢弃）、`publishedAt` ← `publishedDate`。请求的 `maxResults` 优先于已配置的默认 `numResults`，并作为 Exa `numResults` 发送，以优化成本／延迟；最终边界由 seam 强制执行。提供方失败（HTTP 错误、网络失败、无法解析或形状错误的主体）以 `WebError` `WEB_PROVIDER_ERROR` 呈现；中止请求以 `WEB_ABORTED` 呈现。HTTP 重定向会在接触 `Location` 目标前被拒绝，并以 `WEB_PROVIDER_ERROR` 呈现。
+Exa 返回扁平 `results[]`，不返回生成答案，因此省略 `content`。每项结果映射为 `WebSearchSource`：`url` ← `url`、`title` ← `title`、`snippet` ← 第一个非空的 `highlights[]` 条目（没有高亮摘要的结果缺少可移植的 snippet，会被丢弃）、`publishedAt` ← `publishedDate`。请求的 `maxResults` 优先于已配置的默认 `numResults`，并作为 Exa `numResults` 发送，以优化成本和延迟；最终上限由 seam 强制执行。提供方失败（HTTP 错误、网络失败、响应体无法解析或结构不符）以 `WebError` `WEB_PROVIDER_ERROR` 呈现；中止请求以 `WEB_ABORTED` 呈现。HTTP 重定向会在访问 `Location` 指向的目标之前被拒绝，并以 `WEB_PROVIDER_ERROR` 呈现。
 
 ## 模型体验
 
-通过 [`dsh-tool-web`](../tool-web/README.md) 间接影响；该工具保留此提供方经 `maxResults` 限制的 URL、标题、首条 highlight 与发布日期，或将精确的 `Exa search aborted`、`Exa search request failed: <error>` 和 `Exa returned an unprocessable response body: <error>` 失败置于消费方错误包装内；生成答案与提供方私有字段不进入上下文。
+通过 [`dsh-tool-web`](../tool-web/README.md) 间接影响；该工具保留此提供方经 `maxResults` 限制的 URL、标题、首条 highlight 与发布日期，或将确切的错误消息 `Exa search aborted`、`Exa search request failed: <error>` 和 `Exa returned an unprocessable response body: <error>` 置于消费方的错误包装层内；生成答案与提供方私有字段不进入上下文。
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **没有非空 highlight 的结果会被整个丢弃**：没有可映射的可移植 snippet，因此返回源可能少于请求数量。
-- **只公开 `searchType`／`numResults`／`highlightsPerResult`**：Exa 的其他控制项（livecrawl、category、domain/date filter、全文内容）等待提供方无关 seam 字段（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。
+- **没有非空白高亮摘要的结果会被整个丢弃**：没有可映射的可移植 snippet，因此返回源可能少于请求数量。
+- **只公开 `searchType`／`numResults`／`highlightsPerResult`**：Exa 的其他控制项（livecrawl、category、域名／日期过滤条件、全文内容）等待提供方无关 seam 字段（见 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。
 - **按错误形状分类中止**：只有 `DOMException` 且名为 `AbortError` 时才映射为 `WEB_ABORTED`；携带自定义原因的中止（例如 `dsh-timeout` 的 `TimeoutReason`）会呈现为 `WEB_PROVIDER_ERROR`。
diff --git a/packages/web/web-search-perplexity/README.i18n.yaml b/packages/web/web-search-perplexity/README.i18n.yaml
index f69bb5d31e..8da0df01bc 100644
--- a/packages/web/web-search-perplexity/README.i18n.yaml
+++ b/packages/web/web-search-perplexity/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/web/web-search-perplexity/README.md
 README.md: 80f6d34d63ddf0cc7d1f0d4f6744d64c91f9a269
-README.zh.md: 2e631c11f99d9b24ba44fe1fe5f5238916ecc175
+README.zh.md: 7bb3721eedb2da27ee881f86f10368bec8be57a9
diff --git a/packages/web/web-search-perplexity/README.zh.md b/packages/web/web-search-perplexity/README.zh.md
index 2e631c11f9..7bb3721eed 100644
--- a/packages/web/web-search-perplexity/README.zh.md
+++ b/packages/web/web-search-perplexity/README.zh.md
@@ -4,17 +4,17 @@
 
 由 [Perplexity](https://perplexity.ai) 支持的 `WebSearchProvider`，用于 harness [web 能力 seam](../web/README.md)（`ctx.web`）。它调用 Perplexity 的 OpenAI 兼容 `POST /chat/completions` 端点，把生成答案与引用映射为 seam 规范化的 `WebSearchResult`。
 
-这是一个**实现** 包：它向 `ctx.web` 注册提供方，不拥有该 key，也不注册面向模型的工具。与 `@deepseek-ai/dsh-llm-deepseek` 一样，它是函数／namespace 插件（`inject: ['web']`）。OpenAI 兼容协议形状是提供方私有细节，并**不** 使该提供方依赖 `ctx.llm`。
+这是一个**实现**包（package）：它向 `ctx.web` 注册提供方，不拥有该键，也不注册面向模型的工具。与 `@deepseek-ai/dsh-llm-deepseek` 一样，它是函数／命名空间插件（`inject: ['web']`）。OpenAI 兼容协议格式（wire format）是提供方私有细节，并**不**使该提供方依赖 `ctx.llm`。
 
 ## 配置
 
-| Key | 默认值 | 含义 |
+| 配置键 | 默认值 | 含义 |
 |---|---|---|
-| `apiKey` | `$PERPLEXITY_API_KEY` | Perplexity API key。为空／缺失时提供方不可用。 |
+| `apiKey` | `$PERPLEXITY_API_KEY` | Perplexity API 密钥。为空或缺失时提供方不可用。 |
 | `baseURL` | `https://api.perplexity.ai` | 端点基址；追加 `/chat/completions`。无法解析时提供方不可用。 |
 | `model` | `sonar` | 搜索模型名称。 |
 | `maxTokens` | `1024` | 生成答案 token 上限（`max_tokens`）。必须是正整数。 |
-| `searchRecency` | （未设置） | 以 `search_recency_filter` 发送的新近程度窗口：`day`、`week`、`month` 或 `year`。未设置时不发送 filter。 |
+| `searchRecency` | （未设置） | 以 `search_recency_filter` 发送的新近程度窗口：`day`、`week`、`month` 或 `year`。未设置时不发送过滤条件。 |
 
 ```yaml
 - id: web-search-perplexity
@@ -25,7 +25,7 @@
 
 ## 映射
 
-`content` ← `choices[0].message.content`（生成答案）。`sources[]` 优先使用结构化 `search_results[]`（`url`、`title`、`snippet`、`publishedAt` ← `date`），否则回退到只含 URL 的 `citations[]` 数组；仅当不存在 `search_results` 时才采取这条回退路径。这些源只携带 `url`，因此 seam 上的 `title`／`snippet`／`publishedAt` 是可选字段。提供方失败以 `WebError` `WEB_PROVIDER_ERROR` 呈现；中止请求以 `WEB_ABORTED` 呈现。HTTP 重定向会在接触 `Location` 目标前被拒绝，并以 `WEB_PROVIDER_ERROR` 呈现。Perplexity 没有结果数量控制，因此 seam 会强制执行 `maxResults`（截断 `sources[]` 并设置 `truncated`）。
+`content` ← `choices[0].message.content`（生成答案）。`sources[]` 优先使用结构化 `search_results[]`（`url`、`title`、`snippet`、`publishedAt` ← `date`），否则回退到只含 URL 的 `citations[]` 数组；仅当不存在 `search_results` 时才采取这条回退路径。这些源只携带 `url`，因此 seam 上的 `title`／`snippet`／`publishedAt` 是可选字段。提供方失败以 `WebError` `WEB_PROVIDER_ERROR` 呈现；中止请求以 `WEB_ABORTED` 呈现。HTTP 重定向会在访问 `Location` 指向的目标之前被拒绝，并以 `WEB_PROVIDER_ERROR` 呈现。Perplexity 没有结果数量控制，因此 seam 会强制执行 `maxResults`（截断 `sources[]` 并设置 `truncated`）。
 
 ## 模型体验
 
@@ -33,7 +33,7 @@
 
 #### 模型看到的内容
 
-独立的 Perplexity 模型通过 chat-completions 端点接收逐字的 `<query>` 作为唯一 user 消息。该请求不属于会话模型上下文。
+独立的 Perplexity 模型通过 chat-completions 端点将 `<query>` 原样作为唯一用户消息接收。该请求不属于会话模型上下文。
 
 #### Token 影响
 
@@ -41,25 +41,25 @@
 
 #### KV Cache 影响
 
-与会话请求 cache 相互独立。同一模型路由下的相同查询可能复用提供方 cache；查询或路由改变会建立不同前缀。
+与会话请求缓存相互独立。同一模型路由下的相同查询可能复用提供方缓存；查询或路由改变会建立不同前缀。
 
 ### 间接的会话工具结果
 
 #### 模型看到的内容
 
-通过 [`dsh-tool-web`](../tool-web/README.md)，会话模型会看到生成答案及结构化结果元数据，或只含 URL 的引用。该提供方的精确失败是 `Perplexity search aborted`、`Perplexity search request failed: <error>` 和 `Perplexity returned an unprocessable response body: <error>`；HTTP 失败保留提供方消息。错误包装属于消费方。
+通过 [`dsh-tool-web`](../tool-web/README.md)，会话模型会看到生成答案及结构化结果元数据，或只含 URL 的引用。该提供方确切的错误消息为 `Perplexity search aborted`、`Perplexity search request failed: <error>` 和 `Perplexity returned an unprocessable response body: <error>`；HTTP 失败保留提供方消息。错误包装层属于消费方。
 
 #### Token 影响
 
-注册不会直接产生会话 token。答案与源 token 取决于数据，源数量受 seam 限制；保留的结果或错误会重复发送直到压缩。
+注册不会直接产生会话 token。答案与源 token 取决于数据，源数量受 seam 限制；保留的结果或错误会重复发送，直到发生压缩（compaction）。
 
 #### KV Cache 影响
 
-仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 配置项失效。
+仅追加；新可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
 ## 已知限制与暂缓事项
 
-- **引用回退源只含 URL**：Perplexity 省略结构化 `search_results[]` 时，源不含 `title`／`snippet`／`publishedAt`，因此工具只渲染裸 hostname label。
-- **超量返回的源仍消耗 token 与延迟**：协议没有结果数量控制，`maxResults` 只能由 seam 在事后截断。
-- **只公开 `model`／`maxTokens`／`searchRecency`**：Perplexity 的其他搜索控制项（domain filter、`web_search_options` 上下文大小、图片）等待提供方无关 seam 字段（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。
+- **引用回退源只含 URL**：Perplexity 省略结构化 `search_results[]` 时，源不含 `title`／`snippet`／`publishedAt`，因此工具只渲染纯主机名标签。
+- **超量返回的来源仍会增加 token 消耗和延迟**：协议没有结果数量控制，`maxResults` 只能由 seam 在事后截断。
+- **只公开 `model`／`maxTokens`／`searchRecency`**：Perplexity 的其他搜索控制项（域名过滤条件、`web_search_options` 上下文大小、图片）等待提供方无关 seam 字段（见 [seam Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。
 - **按错误形状分类中止**：只有 `DOMException` 且名为 `AbortError` 时才映射为 `WEB_ABORTED`；携带自定义原因的中止（例如 `dsh-timeout` 的 `TimeoutReason`）会呈现为 `WEB_PROVIDER_ERROR`。
diff --git a/packages/web/web/README.i18n.yaml b/packages/web/web/README.i18n.yaml
index 591b93b49a..7c2338f169 100644
--- a/packages/web/web/README.i18n.yaml
+++ b/packages/web/web/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/web/web/README.md
 README.md: 73765fe060cc0a2b0fa3d69703a670f488a29ac9
-README.zh.md: 0b89fb21769ca58e4a7421f69431a446614cf7b8
+README.zh.md: 49aef50bc678b523d5ee44d2bc4a015106ea7ae1
diff --git a/packages/web/web/README.zh.md b/packages/web/web/README.zh.md
index 0b89fb2176..49aef50bc6 100644
--- a/packages/web/web/README.zh.md
+++ b/packages/web/web/README.zh.md
@@ -4,7 +4,7 @@
 
 **web 访问 seam**：抽象 `WebService`（`ctx.web`）定义 harness 具备哪些 web 访问能力（搜索 web、抓取 URL），并通过多个提供方实现，不把模型契约绑定到某个厂商的 API 形状。
 
-该包是 web 能力中负责接口的三分之一。与 bash/fs 不同，它在一个 seam 上跨越搜索与抓取两种能力，每种能力都可能有多个提供方：
+该包（package）是 web 能力中负责接口的三分之一。与 bash/fs 不同，它在一个 seam 上跨越搜索与抓取两种能力，每种能力都可能有多个提供方：
 
 | 包 | 职责 |
 |---|---|
@@ -14,21 +14,21 @@
 | `@deepseek-ai/dsh-web-fetch-local` | 抓取实现：匿名公共 HTTP(S) |
 | `@deepseek-ai/dsh-tool-web` | 面向模型的 `web_search`／`web_fetch` 工具 schema，构建于 `ctx.web` 之上 |
 
-搜索与抓取没有共享请求 schema 或业务逻辑，但有意共用一个 seam：`ctx.web` 是单一 web 访问中间层，拥有一项提供方选择策略、一套中止／错误词汇和一个面向产品的「该 harness 如何访问 web」配置表层。代价是成对的并行 `Search`／`Fetch` 方法；这种并行是有意设计，不是遗漏提取。
+搜索与抓取没有共享请求 schema 或业务逻辑，但有意共用一个 seam：`ctx.web` 是单一 web 访问中间层，拥有一项提供方选择策略、一套中止／错误词汇和一个面向产品的「该 harness 如何访问 web」配置接口。代价是成对的并行 `Search`／`Fetch` 方法；这种并行是有意设计，并非遗漏了可抽取的共性。
 
 ## 服务 API（`ctx.web`）
 
 | 成员 | 语义 |
 |---|---|
-| `registerSearchProvider(provider)`／`registerFetchProvider(provider)` | 注册后端。同一能力 kind 下 id 重复时抛出 `WebError` `WEB_DUPLICATE_PROVIDER`。返回 disposer。随调用 fiber 释放。 |
+| `registerSearchProvider(provider)`／`registerFetchProvider(provider)` | 注册后端。同一能力类型下 id 重复时抛出 `WebError` `WEB_DUPLICATE_PROVIDER`。返回 disposer。随调用 fiber 一并 dispose（资源释放）。 |
 | `search(request, signal?)` | 解析搜索提供方并运行一次搜索。在结果上强制执行 `request.maxResults`（截断 `sources[]`，设置 `truncated`）。能力无法运行时抛出 `WebError`。 |
 | `fetch(request, signal?)` | 解析抓取提供方并获取一个 URL。非 2xx 响应是结果，不会抛出异常。无法安全获取或表示资源时抛出 `WebError`。 |
 
-提供方注册的是**能力** 而非工具。`dsh-tool-web` 是面向模型名称、描述、提示词指引、JSON schema 和呈现的唯一 owner。
+提供方注册的是**能力**而非工具。`dsh-tool-web` 是面向模型的名称、描述、提示词指引、JSON Schema 和呈现的唯一归属方。
 
 ## 选择
 
-选择绝不依赖注册、配置或 HMR 顺序。能力要么具有显式提供方 id（配置 `searchProvider`／`fetchProvider`，或由环境变量 `$DSH_WEB_SEARCH_PROVIDER`／`$DSH_WEB_FETCH_PROVIDER` 提供相同字段），要么在恰好只注册一个可用提供方时自动选择。`search()`／`fetch()` 会在执行时解析提供方：
+选择绝不依赖注册、配置或热模块替换（HMR）顺序。能力要么具有显式提供方 id（配置 `searchProvider`／`fetchProvider`，或由环境变量 `$DSH_WEB_SEARCH_PROVIDER`／`$DSH_WEB_FETCH_PROVIDER` 提供相同字段），要么在恰好只注册一个可用提供方时自动选择。`search()`／`fetch()` 会在执行时解析提供方：
 
 | 情况 | 执行 |
 |---|---|
@@ -39,23 +39,23 @@
 | 无 id，没有可用提供方 | `WEB_PROVIDER_UNAVAILABLE` |
 | 无 id，多个可用提供方 | `WEB_PROVIDER_AMBIGUOUS` |
 
-失败分支会抛出 `WebError`；调用方按其结构化 code（加消息细节：缺失 id、歧义候选集合）路由。提供方自身的 `available()` 是便宜的局部检查（credential 是否存在、配置是否可解析），供执行时选择使用，且**禁止发起网络调用**；`dsh-tool-web` 永远不会调用它。工具通过 `ctx.web.search()`／`fetch()` 执行，并按抛出的 code 路由，因此提供方选择只有一个 owner。
+失败分支会抛出 `WebError`；调用方按其结构化 code（加消息细节：缺失 id、歧义候选集合）路由。提供方自身的 `available()` 是便宜的局部检查（凭据是否存在、配置是否可解析），供执行时选择使用，且**禁止发起网络调用**；`dsh-tool-web` 永远不会调用它。工具通过 `ctx.web.search()`／`fetch()` 执行，并按抛出的 code 路由，因此提供方选择只有一个归属方。
 
 ## 词汇
 
-`WebSearchRequest`（`query`、`maxResults?`）→ `WebSearchResult`（`content?`、`sources[]`、`truncated`）；每个 `WebSearchSource` 都有必填 `url` 与可选 `title`／`snippet`／`publishedAt`（Perplexity 引用可能只含 URL）。`WebFetchRequest`（`url`）→ `WebFetchResult`（最终 `url`、`statusCode`、`body`、`truncated`）；取消作为可选的直接 `AbortSignal` 参数传给 `search()`／`fetch()`。`WebFetchBody` 是这里拥有的封闭判别联合（`html` | `text`）；消费方使用 `switch` 实现穷尽检查，因此新增 kind 会破坏编译，直到处理完毕。完整契约见 `src/types.ts`，其中也包含 `WebError` code 分类体系。
+`WebSearchRequest`（`query`、`maxResults?`）→ `WebSearchResult`（`content?`、`sources[]`、`truncated`）；每个 `WebSearchSource` 都有必填 `url` 与可选 `title`／`snippet`／`publishedAt`（Perplexity 引用可能只含 URL）。`WebFetchRequest`（`url`）→ `WebFetchResult`（最终 `url`、`statusCode`、`body`、`truncated`）；取消作为可选的直接 `AbortSignal` 参数传给 `search()`／`fetch()`。`WebFetchBody` 是这里拥有的封闭判别联合（`html` | `text`）；消费方使用 `switch` 实现穷尽检查，因此新增类型会导致编译失败，直到处理完毕。完整契约见 `src/types.ts`，其中也包含 `WebError` code 分类体系。
 
 ## 模型体验
 
-通过 `dsh-tool-web` 间接影响；该工具保留有界的规范化提供方数据，或精确的已配置提供方、提供方不可用、无提供方、多提供方及 `Error: <message>` 失败，本注册表自身不贡献提示词或 schema。
+通过 `dsh-tool-web` 间接影响；该工具会保留有界的规范化提供方数据，或者原样保留以下失败：已配置的提供方缺失、提供方不可用、无提供方、存在多个提供方以及 `Error: <message>`；本注册表自身不贡献提示词或 schema。
 
 #### KV Cache 影响
 
-不会直接失效；请求前缀变更由命名消费方负责。
+不会直接导致 KV Cache 失效；请求前缀变更由上述消费方负责。
 
 ## 已知限制与暂缓事项
 
-- **没有观测表层**：没有提供方变更事件或能力状态查询；可用性只能通过执行 `search()`／`fetch()` 并按抛出的 `WebError` code 路由来观测，无提供方失败是通用的 `WEB_PROVIDER_UNAVAILABLE`，不会枚举逐提供方原因（见 [Agent Note](../../../.agents/notes/archived/simplification/2026-07-04-drop-unconsumed-web-observation-surface.md)）。
-- **`WebSearchRequest` 只携带 `query` + `maxResults`**：提供方无关的控制项（新近程度、domain filter、区域提示、搜索深度）暂缓至 Exa 与 Perplexity 都能诚实支持时（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。
+- **没有观测接口**：没有提供方变更事件或能力状态查询；可用性只能通过执行 `search()`／`fetch()` 并按抛出的 `WebError` code 路由来观测，无提供方失败是通用的 `WEB_PROVIDER_UNAVAILABLE`，不会枚举逐提供方原因（见 [Agent Note（agent 决策记录）](../../../.agents/notes/archived/simplification/2026-07-04-drop-unconsumed-web-observation-surface.md)）。
+- **`WebSearchRequest` 只携带 `query` + `maxResults`**：提供方无关的控制项（新近程度、域名过滤条件、区域提示、搜索深度）暂缓至 Exa 与 Perplexity 都能诚实支持时（见 [seam Agent Note](../../../.agents/notes/implemented/architecture/2026-06-24-web-capability-seam.md)）。
 - **`WebFetchBody` 没有 `pdf` 分支**：可提取文本的 PDF 支持属于明确的暂缓工作；封闭联合会使新增该分支成为三个 web 包中由编译强制执行的变更。
-- **提供方支持的页面提取不属于 `fetch()` 范围**：Firecrawl/Tavily 风格的 `web_extract` 能力暂缓，而不会扩宽抓取 seam。
+- **提供方支持的页面提取不属于 `fetch()` 范围**：Firecrawl/Tavily 风格的 `web_extract` 能力暂缓，而不会扩展抓取 seam。
diff --git a/packages/workflow/README.i18n.yaml b/packages/workflow/README.i18n.yaml
index 5b143480d5..cb4619569f 100644
--- a/packages/workflow/README.i18n.yaml
+++ b/packages/workflow/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/workflow/README.md
 README.md: 17da3397e4fa64837b271d89e9062e3621434338
-README.zh.md: 056d005de69a493a24b360583651bf313548b3fb
+README.zh.md: 1e424778781ba4f41c321af7fd4cb8b5159a49cb
diff --git a/packages/workflow/README.zh.md b/packages/workflow/README.zh.md
index 056d005de6..1e42477878 100644
--- a/packages/workflow/README.zh.md
+++ b/packages/workflow/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-workflow seam：由模型编写 JavaScript 编排脚本，大规模扇出 subagent（分阶段、每个 agent（智能体）的结构化结果、并发上限），其设计参考 Claude Code 动态工作流。这是 bash 形态的能力 seam（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：每个上下文只有一个引擎实现注册为 `ctx.workflows`；面向模型的工具消费它。
+工作流 seam：由模型编写 JavaScript 编排脚本，大规模扇出 subagent（分阶段、每个 agent（智能体）的结构化结果、并发上限），其设计参考 Claude Code 动态工作流。这是 bash 形态的能力 seam（见[能力 seam](../../.agents/notes/implemented/architecture/2026-06-13-capability-seams.md)）：每个上下文只有一个引擎实现注册为 `ctx.workflows`；面向模型的工具使用它。
 
-| 包 | 角色 | ctx 键 |
+| 包（package） | 角色 | ctx 键 |
 |---|---|---|
-| `workflow/` | 抽象 workflow seam：服务基类、运行词汇和 `workflow/*` 事件 | `ctx.workflows` |
+| `workflow/` | 抽象工作流 seam：服务基类、运行词汇和 `workflow/*` 事件 | `ctx.workflows` |
 | `workflow-workerthread/` | `node:worker_threads` 引擎：每次运行使用一个 worker；脚本的 vm 上下文位于 worker 内，`agent()` 通过消息端口桥接到 `ctx.subagents` | （提供 `ctx.workflows`） |
 | `tool-workflow/` | 面向模型的 `workflow` 工具，基于 `ctx.workflows` | （注册到 `ctx.tools`） |
 | `tool-ralph/` | 基于 `ctx.workflows` 和全新结构化输出 subagent 提供方的固定全新 agent Ralph 策略 | （注册到 `ctx.tools`） |
 
-接口位于 `workflow/workflow/`。引擎的 `agent()` 钩子使用 [subagent seam](../subagent/README.md)（任何已注册提供方；随产品交付的示例使用 `spawn`），`agent({ schema })` 则使用进程内后端实现的结构化输出支持。worker thread 隔离的是脚本：宿主绝不会被它阻塞，已取消运行经过宽限时间后的终止也会实际生效；但它不是安全边界。如果将来确有需要，可以在同一接口后替换为 isolated-vm/独立进程引擎，以实现真正的沙箱隔离。
+接口位于 `workflow/workflow/`。引擎的 `agent()` 钩子使用 [subagent seam](../subagent/README.md)（任何已注册提供方；随产品交付的示例使用 `spawn`），`agent({ schema })` 则使用进程内后端实现的结构化输出支持。worker thread 隔离的是脚本：宿主绝不会被它阻塞，已取消运行经过宽限时间后的终止也会实际生效；但它不是安全边界。如果将来确有需要，可以在同一接口背后换用 isolated-vm／独立进程引擎，以实现真正的沙箱隔离。
 
-通用脚本引擎的决策和延期工作见[动态工作流 Agent Note](../../.agents/notes/implemented/feature/2026-07-05-dynamic-workflows.md)。独立的 [Ralph 消费方](../../.agents/notes/implemented/feature/2026-07-19-fresh-agent-ralph-workflow-tool.md)会固定脚本和全新提供方策略，而不是再添加一个引擎或 agent loop（智能体循环）模式。
+通用脚本引擎的决策和暂缓事项见[动态工作流 Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-05-dynamic-workflows.md)。独立的 [Ralph 消费方](../../.agents/notes/implemented/feature/2026-07-19-fresh-agent-ralph-workflow-tool.md)会固定脚本和全新提供方策略，而不是再添加一个引擎或 agent loop（智能体循环）模式。
diff --git a/packages/workflow/tool-ralph/README.i18n.yaml b/packages/workflow/tool-ralph/README.i18n.yaml
index 85df85d098..5bde7f1f02 100644
--- a/packages/workflow/tool-ralph/README.i18n.yaml
+++ b/packages/workflow/tool-ralph/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/workflow/tool-ralph/README.md
 README.md: daf242364d1093cff30cd1dc95823e1ecb89a9c7
-README.zh.md: cc71630c097157182823342a873d673f6420f44e
+README.zh.md: db9564da1d1b7d8eabc67ba14c9d873201ac2807
diff --git a/packages/workflow/tool-ralph/README.zh.md b/packages/workflow/tool-ralph/README.zh.md
index cc71630c09..db9564da1d 100644
--- a/packages/workflow/tool-ralph/README.zh.md
+++ b/packages/workflow/tool-ralph/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-面向模型的 `ralph` 工具运行固定的前台工作流，把一个不可变目标依次交给多个全新子 agent（智能体）。它展示如何把专用编排策略实现为基于 [`ctx.workflows`](../workflow/README.md) 和 [`ctx.subagents`](../../subagent/subagent/README.md) 的普通插件：不会向 `agent-loop` 添加 Ralph 模式或全新 agent 循环，同会话的[目标领域](../../goal/goal/README.md)也保持独立。政策和延期工作由 [Ralph Agent Note](../../../.agents/notes/implemented/feature/2026-07-19-fresh-agent-ralph-workflow-tool.md)负责。
+面向模型的 `ralph` 工具运行固定的前台工作流，把一个不可变目标依次交给多个全新子 agent（智能体）。它展示如何把专用编排策略实现为基于 [`ctx.workflows`](../workflow/README.md) 和 [`ctx.subagents`](../../subagent/subagent/README.md) 的普通插件：不会向 `agent-loop` 添加 Ralph 模式或全新 agent loop（智能体循环），同会话的[目标领域](../../goal/goal/README.md)也保持独立。策略和暂缓事项由 [Ralph Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-19-fresh-agent-ralph-workflow-tool.md)负责。
 
 ## 契约
 
@@ -16,7 +16,7 @@
 
 ## 生命周期与取消
 
-调用方 agent 是每个全新子 agent 的父级，因此会保留 cwd 和谱系，但不会复制其对话。`exec.signal` 进入工作流引擎，同时也桥接到 `run.cancel()`，确保实现相互独立。工具等待 `run.result` 并调用 `run.dispose()`，后一个调用位于 `finally` 中，因此取消的父级步骤会等到引擎完成有界终止且子 agent 完全停稳后才返回。
+调用方 agent 是每个全新子 agent 的父级，因此会保留 cwd 和谱系，但不会复制其对话。`exec.signal` 进入工作流引擎，同时也桥接到 `run.cancel()`，以便不依赖具体实现。工具等待 `run.result` 并调用 `run.dispose()`，后一个调用位于 `finally` 中，因此取消的父级步骤会等到引擎完成有界终止且子 agent 完全停稳后才返回。
 
 ## 渲染意图
 
@@ -31,7 +31,7 @@
 | `maxHandoffChars` | `16384` | 一份 Round 报告序列化后的最大字符数。 |
 | `maxResultChars` | `16384` | 返回给父级的完整成功结果最大字符数。 |
 
-插件应用时会规范化并校验所有配置值；直接应用、未经过 Loader schema 规范化的情况也包括在内。每次调用前都会立即解析提供方能力，因为提供方注册可能随插件生命周期和 HMR（热模块替换）变化。
+插件应用时会规范化并校验所有配置值，也包括绕过 Loader schema 规范化而直接应用的情况。每次调用前都会立即解析提供方能力，因为提供方注册可能随插件生命周期和热模块替换（HMR）变化。
 
 ## 模型体验
 
@@ -49,21 +49,21 @@ Use the ralph tool ONLY when the direct human explicitly asks for a Ralph loop o
 
 #### Token 影响
 
-插件启用期间，每个请求支付少量固定指导成本。
+插件启用期间，每个请求都会产生少量固定的指导 token 开销。
 
 #### KV Cache 影响
 
-只要插件作用域和指导文本不变，前缀就保持稳定。启用或 dispose（资源释放）可能从该提示词段开始使复用失效。
+只要插件作用域和指导文本不变，前缀就保持稳定。启用或 dispose（资源释放）可能会使从该提示词段起的缓存复用失效。
 
 ### 工具 schema
 
 #### 模型看到的内容
 
-已生成的 [`ralph` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-ralph)公开一个必填 `objective` 字符串和一个可选 `maxRounds` 数字。提供方选择、交接大小、报告 schema、工作流脚本和编排行为均由部署拥有，不在调用接口中。
+已生成的 [`ralph` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-ralph)公开一个必填 `objective` 字符串和一个可选 `maxRounds` 数字。提供方选择、交接大小、报告 schema、工作流脚本和编排行为均由部署侧控制，不在调用接口中。
 
 #### Token 影响
 
-工具可见的每个请求都会支付少量固定 schema 成本。
+工具可见时，每个请求都会产生少量固定的 schema token 开销。
 
 #### KV Cache 影响
 
@@ -83,11 +83,11 @@ Use the ralph tool ONLY when the direct human explicitly asks for a Ralph loop o
 
 每个全新子 agent 都有独立的请求缓存。父级结果追加在可复用请求前缀之后。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **完成由 worker 自行声明**：没有独立的评估器或验证器判断目标是否实际完成；评估器政策及评估器驱动的延续均延期处理。
-- **仅支持前台**：没有 task id、后台收集、进程恢复检查点、调度器或基于墙上时钟的启动政策。
+- **完成由 worker 自行声明**：没有独立的评估器或验证器判断目标是否实际完成；评估器策略及评估器驱动的延续均暂缓处理。
+- **仅支持前台**：没有 task id、后台收集、进程恢复检查点、调度器或基于挂钟时间的启动策略。
 - **工作区是唯一的跨 Round 长期记忆**：一份有界报告作为显式交接内容，每个子 agent 结束后，未提交的对话推理都会消失。
 - **一个 Round 对应一个全新子 agent**：Round 内没有扇出、模型/提供方切换、fork 上下文或由模型调用选择的提供方。
 - **普通子 agent 失败会终止运行**：固定脚本报告失败的 Round 和上一次成功交接，但不会重试；致命的工作流基础设施失败可能在该状态返回前结束。
-- **聚合工作量仅受 Round 数量限制**：token、价格和已用时间预算均延期处理。
+- **聚合工作量仅受 Round 数量限制**：token、价格和耗时预算均暂缓处理。
diff --git a/packages/workflow/tool-workflow/README.i18n.yaml b/packages/workflow/tool-workflow/README.i18n.yaml
index 33d3461568..902438506c 100644
--- a/packages/workflow/tool-workflow/README.i18n.yaml
+++ b/packages/workflow/tool-workflow/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/workflow/tool-workflow/README.md
 README.md: 5afa68764bfe339377954f8912b6f9b795435be1
-README.zh.md: d24442f2dcf12dae11980ab6ab08d1a63c2b1432
+README.zh.md: a322fa363803a10b06c66d1aae9b769fad8e5c22
diff --git a/packages/workflow/tool-workflow/README.zh.md b/packages/workflow/tool-workflow/README.zh.md
index d24442f2dc..a322fa3638 100644
--- a/packages/workflow/tool-workflow/README.zh.md
+++ b/packages/workflow/tool-workflow/README.zh.md
@@ -2,26 +2,26 @@
 
 [English](README.md) | 中文
 
-面向模型的 **`workflow` 工具**：运行一段扇出 subagent 的 JavaScript 编排脚本，并返回脚本的最终值。本包负责基于 [`ctx.workflows`](../workflow/README.md) 塑造 schema 和生命周期；脚本解析、执行、上限与取消位于 seam 之后，消费方继续拥有面向父级的 schema 和结果包络。
+面向模型的 **`workflow` 工具**：运行一段扇出 subagent 的 JavaScript 编排脚本，并返回脚本的最终值。本包（package）负责基于 [`ctx.workflows`](../workflow/README.md) 塑造 schema 和生命周期；脚本解析、执行、上限与取消位于 seam 之后，消费方仍负责面向父级的 schema 和结果包络。
 
 ## 模型看到的内容
 
-工具有三个参数：`meta`（必需的身份数据：`name`、`description` 和可选的进度注解）、`script`（必需的纯 JavaScript 函数体，不含 `export const meta` 语句；工具描述包含完整的编写契约）以及 `args`（可选 JSON 对象，作为全局变量 `args` 向脚本公开；裸列表应包装到字段中，使协议 schema 如实表达形态）。插件还会贡献一个 `tool:<toolName>` 系统提示词段，其中包含使用政策：只有用户明确要求工作流/大型编排时才使用该工具；一两项委派优先使用普通 subagent 调用。这遵循工具指导随工具插件交付、绝不放入部署 persona 的约定。
+工具有三个参数：`meta`（必需的身份数据：`name`、`description` 和可选的进度注解）、`script`（必需的纯 JavaScript 脚本体，不含 `export const meta` 语句；工具描述包含完整的编写契约）以及 `args`（可选 JSON 对象，作为全局变量 `args` 向脚本公开；裸列表应包装到字段中，使协议 schema 如实表达形态）。插件还会贡献一个 `tool:<toolName>` 系统提示词段，其中包含使用策略：只有用户明确要求工作流／大型编排时才使用该工具；一两项委派优先使用普通 subagent 调用。这遵循工具指导随工具插件交付、绝不放入部署 persona 的约定。
 
 ## 生命周期
 
-当前版本采用同步收集（类似 [`dsh-tool-subagent`](../../subagent/tool-subagent/README.md)）：`execute` 启动运行并等待 `run.result`；这些操作位于 `try/finally` 中，该结构总会 dispose 运行，使脚本及其子 agent（智能体）在每条路径上完全停稳。`exec.signal` 会桥接到 `run.cancel()`，包括启动前已经中止的情况。非 `completed` 结束原因会映射为报告原因的 `isError` 结果，绝不会把局部输出当作成功；`start()` 同步抛出的解析/meta 失败会变成模型可据以修正的 `isError`。完成时返回规范值 `{ runId, agentsStarted, result }`；Native 渲染器保留 meta 名称、agent 数量和 JSON 值，只会在 `maxResultChars` 处截断该投影。
+当前版本采用同步收集（类似 [`dsh-tool-subagent`](../../subagent/tool-subagent/README.md)）：`execute` 启动运行并等待 `run.result`；这些操作位于 `try/finally` 中，该结构总会 dispose（资源释放）运行，使脚本及其子 agent（智能体）在每条路径上完全停稳。`exec.signal` 会桥接到 `run.cancel()`，包括启动前已经中止的情况。非 `completed` 结束原因会映射为报告原因的 `isError` 结果，绝不会把局部输出当作成功；`start()` 同步抛出的解析/meta 失败会变成模型可据以修正的 `isError`。完成时返回规范值 `{ runId, agentsStarted, result }`；Native 渲染器保留 meta 名称、agent 数量和 JSON 值，只会在 `maxResultChars` 处截断该投影。
 
 ## 渲染意图
 
-渲染意图预先确定（见[渲染意图 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-02-tool-render-intent-union.md)）：使用一个 `generic` 卡片，标题为 `workflow: <meta.name>`，直接从 `args.meta.name` 读取（呈现是参数的纯函数，不要求引擎解析）；脚本文本作为 `rawInput` 携带。结果继续使用 generic 卡片。
+渲染意图预先确定（见[渲染意图 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/architecture/2026-07-02-tool-render-intent-union.md)）：使用一个 `generic` 卡片，标题为 `workflow: <meta.name>`，直接从 `args.meta.name` 读取（呈现是参数的纯函数，不要求引擎解析）；脚本文本作为 `rawInput` 携带。结果继续使用 generic 卡片。
 
 ## 配置
 
 | 键 | 默认值 | 含义 |
 |---|---|---|
 | `toolName` | `workflow` | 要注册的面向模型工具名称。 |
-| `maxResultChars` | `50000` | 渲染结果上限；更长的 JSON 会连同提示一起截断。 |
+| `maxResultChars` | `50000` | 渲染结果上限；更长的 JSON 会被截断并附上提示。 |
 
 ## 模型体验
 
@@ -29,9 +29,9 @@
 
 #### 模型看到的内容
 
-在该插件的注册作用域内，每个父级请求都会收到下方的 workflow 指导。作用域工具限制可以隐藏 schema，而不移除这段独立注册的指导。
+在该插件的注册作用域内，每个父级请求都会收到下方的工作流指导。作用域工具限制可以隐藏 schema，而不移除这段独立注册的指导。
 
-##### Workflow 指导
+##### 工作流指导
 
 ```markdown
 Use the <toolName> tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.
@@ -39,25 +39,25 @@ Use the <toolName> tool ONLY when the user explicitly asks for a workflow or for
 
 #### Token 影响
 
-插件启用期间，每个请求支付少量固定指导成本。
+插件启用期间，每个请求都会产生少量固定的指导 token 开销。
 
 #### KV Cache 影响
 
-只要插件作用域和指导文本不变，前缀就保持稳定。启用或 dispose（资源释放）可能从该提示词段开始使复用失效。
+只要插件作用域和指导文本不变，前缀就保持稳定。启用或 dispose 可能会使从该提示词段起的缓存复用失效。
 
 ### 工具 schema
 
 #### 模型看到的内容
 
-工具可见时，已生成的默认 [`workflow` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-workflow)包含完整的 JavaScript 钩子与元数据契约；`toolName` 可以重命名该定义，模型会提交脚本、元数据和可选 args。
+工具可见时，已生成的默认 [`workflow` schema](../../../docs/tool-catalog.md#deepseek-aidsh-tool-workflow) 包含完整的 JavaScript 钩子与元数据契约；`toolName` 可以重命名该定义，模型会提交脚本、元数据和可选 args。
 
 #### Token 影响
 
-工具可见的每个请求都会支付较大的固定 schema 成本。
+工具可见时，每个请求都会产生较大的固定 schema token 开销。
 
 #### KV Cache 影响
 
-只要 `toolName`、定义和可见性不变，前缀就保持稳定。重命名、插件生命周期或作用域限制可能从该 schema 开始使复用失效。
+只要 `toolName`、定义和可见性不变，前缀就保持稳定。重命名、插件生命周期或作用域限制可能会使从该 schema 起的缓存复用失效。
 
 ### 工具调用历史与结果
 
@@ -67,14 +67,14 @@ Use the <toolName> tool ONLY when the user explicitly asks for a workflow or for
 
 #### Token 影响
 
-调用 token 可能很多，并会保留到上下文压缩（compaction）为止。结果渲染受 `maxResultChars` 限制；子模型 token 与父级保留的上下文相互独立。
+调用 token 可能很多，并会保留到压缩（compaction）为止。结果渲染受 `maxResultChars` 限制；子模型 token 与父级保留的上下文相互独立。
 
 #### KV Cache 影响
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **父级轮次会阻塞到整个工作流结算**：没有后台启动/轮询接口，取消会把局部输出作为错误丢弃。
-- **`args` 必须是对象，Native 结果文本有界**：调用方把顶层数组/标量包装到字段中；规范工作流结果保持完整，超过 `maxResultChars` 的 JSON 会在面向模型的投影中截断，而不是存入检索句柄之后。
-- **每次工具注册的工作流政策固定**：提供方选择、上限和工具名称属于部署配置，不是模型调用参数。
+- **`args` 必须是对象，Native 结果文本有界**：调用方把顶层数组/标量包装到字段中；规范工作流结果保持完整，超过 `maxResultChars` 的 JSON 会在面向模型的投影中截断，而不是存储在检索句柄背后。
+- **每次工具注册的工作流策略固定**：提供方选择、上限和工具名称属于部署配置，不是模型调用参数。
diff --git a/packages/workflow/workflow-workerthread/README.i18n.yaml b/packages/workflow/workflow-workerthread/README.i18n.yaml
index 1c183d526a..f0aa44859c 100644
--- a/packages/workflow/workflow-workerthread/README.i18n.yaml
+++ b/packages/workflow/workflow-workerthread/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/workflow/workflow-workerthread/README.md
 README.md: 9da420bdd67ac5b0a4bfffa312c6fe7b2dabf8f5
-README.zh.md: e280562b95e9dbebb243aad0f765771f4737b312
+README.zh.md: cc2ed2043db7f6a51088930c07c5cf5a2223f480
diff --git a/packages/workflow/workflow-workerthread/README.zh.md b/packages/workflow/workflow-workerthread/README.zh.md
index e280562b95..cc2ed2043d 100644
--- a/packages/workflow/workflow-workerthread/README.zh.md
+++ b/packages/workflow/workflow-workerthread/README.zh.md
@@ -2,9 +2,9 @@
 
 [English](README.md) | 中文
 
-本包为 `WorkflowService` 提供实现，每次运行使用一个 Node worker thread。worker 执行编排脚本；子 agent（智能体）留在宿主上，通过带类型的宿主/worker 协议访问 `ctx.subagents`。
+本包（package）为 `WorkflowService` 提供实现，每次运行使用一个 Node worker thread。worker 执行编排脚本；子 agent（智能体）留在宿主上，脚本通过带类型的宿主／worker 协议经由 `ctx.subagents` 访问它们。
 
-包根目录默认导出引擎插件及其 `Config`；worker 协议、运行时和会话模块均为实现私有。操作入口 `./worker` 仍是引擎的派生目标。
+包根目录默认导出引擎插件及其 `Config`；worker 协议、运行时和会话模块均为实现私有。操作入口 `./worker` 仍是引擎的 spawn 目标。
 
 这种拆分只有一个主要目的：同步脚本循环不能阻塞 harness 事件循环，忽略取消的脚本可以连同其 worker 一起终止。它不是安全沙箱。
 
@@ -16,14 +16,14 @@ worker 仍提供实用的隔离：
 
 - 脚本 CPU 工作和同步自旋不会占用宿主事件循环；
 - `worker.terminate()` 为 dispose（资源释放）提供真实的最终停止手段；
-- 除未构建 loader 的管道变量外，worker 以空环境启动，因此环境凭据不会通过 `process.env` 跨越边界；
+- 除未构建 loader 所需的衔接配置外，worker 以空环境启动，因此环境凭据不会通过 `process.env` 跨越边界；
 - 宿主/worker 消息使用结构化克隆数据，并在脚本边界执行普通 JSON 校验。
 
-真正的不可信脚本沙箱需要在同一 workflow seam 后采用不同引擎。
+真正的不可信脚本沙箱需要在同一工作流 seam 背后采用不同引擎。
 
 ## 脚本契约
 
-工作流的 `meta` 是宿主提供的数据，而不是待求值的脚本文本。引擎会校验必需的 `name` 和 `description`、拒绝未知字段，并在返回运行前检查函数体能否解析。
+工作流的 `meta` 是宿主提供的数据，而不是待求值的脚本文本。引擎会校验必需的 `name` 和 `description`、拒绝未知字段，并在返回运行前检查脚本正文能否解析。
 
 在 worker 内，脚本会收到 `args` 以及以下钩子：
 
@@ -32,16 +32,16 @@ worker 仍提供实用的隔离：
 - `pipeline(items, ...stages)` 在没有跨阶段屏障的情况下传递 `(previous, item, index)`；
 - `phase(title)` 和 `log(message)` 发出观察器叙述。
 
-未知选项、格式错误的参数、不支持的 schema、触发的上限、提供方启动失败和基础设施结果失败都属于致命工作流错误。有意不注入 timer、文件系统 API 或 Node 全局变量，但上述信任注意事项仍然适用。
+未知选项、格式错误的参数、不支持的 schema、超出上限、提供方启动失败和基础设施结果失败都属于致命工作流错误。有意不注入 timer、文件系统 API 或 Node 全局变量，但上述信任注意事项仍然适用。
 
 ## 运行顺序
 
-`start()` 会校验 meta、解析函数体、解析一个已注册且规范化的提供方路由，并解析每次运行的子 agent 总数上限，然后才创建 worker 或发布 `workflow/start`。请求的 `maxTotalAgents` 必须是正安全整数，且不能超过引擎配置的部署上限。源代码模式通过 data URL bootstrap 安装 TypeScript 转换；构建模式把同级 `lib/worker.cjs` 作为文件系统路径传入，因为 pkg 的 VFS 钩子要求 CommonJS。两者都能在普通 Node 下运行。ready/go 握手可以避免启动信号取消与 worker 启动发生竞态，导致脚本最初的同步片段被执行。
+`start()` 会校验 meta、解析脚本正文、解析一个已注册且规范化的提供方路由，并解析每次运行的子 agent 总数上限，然后才创建 worker 或发布 `workflow/start`。请求的 `maxTotalAgents` 必须是正安全整数，且不能超过引擎配置的部署上限。源代码模式通过 data URL bootstrap 安装 TypeScript 转换；构建模式把同级 `lib/worker.cjs` 作为文件系统路径传入，因为 pkg 的虚拟文件系统（VFS）钩子要求 CommonJS。两者都能在普通 Node 下运行。ready/go 握手可以避免启动信号取消与 worker 启动发生竞态，导致脚本最初的同步片段被执行。
 
 对于每次 `agent()` 调用：
 
 1. worker 发送 `child-start`，其中包含普通数据提示词和选项。
-2. 宿主通过异步 `SubagentService.start` 调用启动请求中的提供方覆盖值，否则调用已配置提供方；调用会传入工作流父级和每次运行唯一的规范中止信号。提供方选择应用于该次运行的每个子 agent，对脚本不可见。
+2. 宿主通过异步 `SubagentService.start` 调用启动请求中指定的提供方，否则调用已配置的提供方；调用会传入工作流父级和该次运行共用的唯一中止信号。提供方选择应用于该次运行的每个子 agent，对脚本不可见。
 3. 如果启动被拒绝，宿主会发送 `child-start-error`；提供方启动已经完全停稳，不会发出子 agent 生命周期事件。
 4. 如果启动兑现时工作流仍接纳工作，宿主会记录该运行、观察 `result`，然后发送 `child-started`。即使结果已经结算，也只会随后转发，以保持先启动、后结果的顺序。
 5. worker 发出成对的 `workflow/agent-start` 和 `workflow/agent-end` 叙述，并在收集后请求 dispose 子 agent。
@@ -56,9 +56,9 @@ worker 仍提供实用的隔离：
 
 ## 取消与 dispose
 
-`WorkflowRun.cancel()` 会记录第一个原因、通知 worker 取消、中止每个待处理及已发布子 agent 共享的唯一信号，并启动 `disposeGraceMs` timer。worker 钩子会在下次 await 时抛出 `CANCELLED`。如果运行到期限仍未结算，宿主会将其以已取消状态兑现、为悬空的子 agent 生命周期事件配对，并终止 worker。
+`WorkflowRun.cancel()` 会记录第一个原因、通知 worker 取消、中止每个待处理及已发布子 agent 共享的唯一信号，并启动 `disposeGraceMs` 定时器。worker 钩子会在下次 await 时抛出 `CANCELLED`。如果运行到期限仍未结算，宿主会将其以已取消状态兑现、为悬空的子 agent 生命周期事件配对，并终止 worker。
 
-subagent seam 只有一个取消通道：请求信号。不存在单独的子 agent 取消 RPC。已发布子 agent 使用 `run.dispose()` 清理；待处理提供方启动在其 promise 拒绝或兑现前仍由提供方拥有。
+subagent seam 只有一个取消通道：请求信号。不存在单独的子 agent 取消 RPC。已发布子 agent 使用 `run.dispose()` 清理；待处理的提供方启动在其 promise 拒绝或兑现前仍由提供方负责。
 
 正常结算也会中止待处理启动，并在结果对外结算前开始 dispose 所有已发布但无需等待的子 agent。宿主的完全停稳条件同时包括待处理启动和已发布子 agent 的 dispose，因此清理不会遗漏异步启动事务。
 
@@ -66,9 +66,9 @@ subagent seam 只有一个取消通道：请求信号。不存在单独的子 ag
 
 ## 结果与事件保证
 
-在宿主主张点，终态结果遵循先到者胜。已接受的外部取消会覆盖后到的非取消 worker 结果；先完成主张的结果或 worker 死亡不能被可重入清理回调改写。
+在宿主的结果确认点，终态结果遵循先到者胜。已接受的外部取消会覆盖后到的非取消 worker 结果；先完成确认的结果或 worker 死亡不能被可重入清理回调改写。
 
-worker 错误、消息失败或提前退出会在清理前关闭消息接纳，然后以 `error` 兑现；如果取消已经拥有该运行，则不覆盖取消。后到的排队消息无法在该逻辑边界后创建子 agent 或发出叙述。
+worker 错误、消息失败或提前退出会在清理前关闭消息接纳，然后以 `error` 兑现；如果取消已经接管该运行，则不覆盖取消。后到的排队消息无法在该逻辑边界后创建子 agent 或发出叙述。
 
 宿主会维护已转发子 agent 启动的台账。优雅退出的 worker 会提供对应的结束事件；死亡或强制终止会把缺失的结束事件合成为已取消。因此，每个已转发的 `workflow/agent-start` 都会且只会配对一次，不过已经到达的工作流结果之后的清理可能稍后才完成。
 
@@ -83,7 +83,7 @@ worker 错误、消息失败或提前退出会在清理前关闭消息接纳，
 | `syncTimeoutMs` | `5000` | 脚本最初同步片段的 VM 超时时间。 |
 | `disposeGraceMs` | `5000` | 强制结算/终止之前的期限，也是公开 dispose 的期限。 |
 
-所属消费方可以为一次运行设置 `WorkflowStartRequest.subagentProvider` 和 `WorkflowStartRequest.maxTotalAgents`。它们属于引擎级政策，不是脚本钩子或面向模型的选项；普通 `workflow` 工具不会设置两者。每次运行的子 agent 总数上限可以降低、但绝不能提高已配置的 `maxTotalAgents` 上限。
+负责该引擎的消费方可以为一次运行设置 `WorkflowStartRequest.subagentProvider` 和 `WorkflowStartRequest.maxTotalAgents`。它们属于引擎级策略，不是脚本钩子或面向模型的选项；普通 `workflow` 工具不会设置两者。每次运行的子 agent 总数上限可以降低、但绝不能提高已配置的 `maxTotalAgents` 上限。
 
 ## 模型体验
 
@@ -91,7 +91,7 @@ worker 错误、消息失败或提前退出会在清理前关闭消息接纳，
 
 #### 模型看到的内容
 
-脚本每次调用 `agent()`，都会把提示词逐字发送给 subagent 提供方，并附带可选模型或结构化输出 schema。每个子 agent 看到该提供方自己的上下文；phase 和 log 叙述只留在观察器事件中。
+脚本每次调用 `agent()`，都会把提示词原样发送给 subagent 提供方，并附带可选模型或结构化输出 schema。每个子 agent 看到该提供方自己的上下文；phase 和 log 叙述只留在观察器事件中。
 
 #### Token 影响
 
@@ -109,16 +109,16 @@ worker 错误、消息失败或提前退出会在清理前关闭消息接纳，
 
 #### Token 影响
 
-本引擎不会直接向父级添加 token。最终结果大小由工具消费方限制，并保留到上下文压缩（compaction）为止。
+本引擎不会直接向父级添加 token。最终结果大小由工具消费方限制，并保留到压缩（compaction）为止。
 
 #### KV Cache 影响
 
 仅追加；新增可见内容位于可复用请求前缀之后，不会使现有 KV-cache 条目失效。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
 - **worker/vm 不是安全边界**：模型编写的代码可以逃逸 `node:vm` 并取得 worker 的进程权限；不可信代码部署需要独立进程或容器引擎。
 - **每次运行都要支付一个 worker thread 的成本**：没有池、预热运行时或跨运行脚本缓存。
-- **不注入环境 timer、文件系统或网络，但逃逸代码仍可访问 Node**：缺失的全局变量用于保证 API 可移植性，而非隔离。
-- **终止只能报告宿主观察到的启动**：`agentsStarted` 不包括仍在 worker 侧排队等待并发、且在强制终止后无法得知的调用。
+- **不注入默认可用的定时器、文件系统或网络，但逃逸代码仍可访问 Node**：这些缺失的全局变量属于可移植性 API 设计，而非隔离措施。
+- **终止只能报告宿主观察到的启动**：`agentsStarted` 不包括因并发限制仍在 worker 侧排队、且在强制终止后无法得知的调用。
 - **跨 realm 错误在脚本内无法通过 `instanceof Error`**：工作流作者必须根据 `name` 和 `code` 等稳定字段分支。
diff --git a/packages/workflow/workflow/README.i18n.yaml b/packages/workflow/workflow/README.i18n.yaml
index 33cc6468d9..08ebb11745 100644
--- a/packages/workflow/workflow/README.i18n.yaml
+++ b/packages/workflow/workflow/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write packages/workflow/workflow/README.md
 README.md: 9d01cfd3d2504d6b5af5a1af6cba735a4abb5795
-README.zh.md: 85dc6bb1897678f72eb715514ab85cad7f7b9589
+README.zh.md: 20dde758d82b38658c39157f9d72d1f3fba046a9
diff --git a/packages/workflow/workflow/README.zh.md b/packages/workflow/workflow/README.zh.md
index 85dc6bb189..20dde758d8 100644
--- a/packages/workflow/workflow/README.zh.md
+++ b/packages/workflow/workflow/README.zh.md
@@ -2,15 +2,15 @@
 
 [English](README.md) | 中文
 
-workflow seam（`ctx.workflows`）执行由模型编写、可扇出 subagent 的编排脚本。该 seam 定义脚本、运行、结果、错误和事件契约；引擎负责决定如何隔离并执行脚本。
+工作流 seam（`ctx.workflows`）执行由模型编写、可扇出 subagent 的编排脚本。该 seam 定义脚本、运行、结果、错误和事件契约；引擎负责决定如何隔离并执行脚本。
 
 `@deepseek-ai/dsh-workflow-workerthread` 是当前引擎，`@deepseek-ai/dsh-tool-workflow` 是面向模型的消费方。未来的进程或沙箱引擎可以替换实现，而无需更改工具。
 
 ## 服务与运行契约
 
-`WorkflowService.start(request): WorkflowRun` 会同步完成足够多的校验，在运行存在前拒绝格式错误的 meta 块、无法解析的脚本、不可用的提供方路由或不受支持的单次运行限制。返回后，`WorkflowRun.result` 绝不拒绝：执行失败以 `stopReason: 'error'` 兑现，取消则在引擎有限的宽限时间内以 `cancelled` 兑现。
+`WorkflowService.start(request): WorkflowRun` 会同步完成足够多的校验，在运行创建前拒绝格式错误的 meta 块、无法解析的脚本、不可用的提供方路由或不受支持的单次运行限制。返回后，`WorkflowRun.result` 绝不拒绝：执行失败以 `stopReason: 'error'` 兑现，取消则在引擎有限的宽限时间内以 `cancelled` 兑现。
 
-运行由持有方拥有。引擎插件卸载会阻止新的启动，但不会撤销已接受的运行。持有方必须在每条路径上调用 `dispose()`；dispose 会取消剩余工作，并在文档规定的期限内达到或放弃完全停稳。
+运行由持有方负责。引擎插件卸载会阻止新的启动，但不会撤销已接受的运行。持有方必须在每条路径上调用 `dispose()`；dispose（资源释放）会取消剩余工作，并在文档规定的期限内达到或放弃完全停稳。
 
 `WorkflowStartRequest` 包含 `{ meta, script, args?, subagentProvider?, maxTotalAgents?, parent, signal? }`。`parent` 把每个子 agent（智能体）归属于调用 agent。`subagentProvider` 可以为该次运行的所有子 agent 指定路由，同时不向脚本公开提供方选择；省略时使用引擎配置的提供方。`maxTotalAgents` 可以为一次运行降低引擎的部署上限，同样对脚本不可见。实现会同步拒绝无效路由和限制。`meta` 与 `args` 是普通数据，不是脚本片段。
 
@@ -18,13 +18,13 @@ workflow seam（`ctx.workflows`）执行由模型编写、可扇出 subagent 的
 
 ## 事件
 
-工作流事件只供观察。它们携带 `WorkflowRunInfo`（`id` 加 `meta`），而不是实时运行，因此监听器无法取得取消或 dispose（资源释放）权限。
+工作流事件只供观察。它们携带 `WorkflowRunInfo`（`id` 加 `meta`），而不是活动运行，因此监听器无法取得取消或 dispose 权限。
 
 - `workflow/start` / `workflow/end` 为运行配对；
 - `workflow/phase` 和 `workflow/log` 公开脚本叙述；
-- `workflow/agent-start` / `workflow/agent-end` 按 `seq` 为每次子 agent 调用配对；异步提供方启动被拒绝的子 agent 不会发出其中任何一个事件。
+- `workflow/agent-start` / `workflow/agent-end` 按 `seq` 为每次子 agent 调用配对；提供方的异步启动调用被拒绝时，该子 agent 不会发出其中任何一个事件。
 
-同进程事件 payload 是以不可变方式借用的值。每个监听器都独立隔离：同步抛出或返回的 promise 被拒绝时，只会记录日志，不会阻塞同级监听器或改变执行。
+同进程事件 payload 是以不可变方式借用的值。每个监听器都独立隔离：同步抛出异常或返回的 promise 被拒绝时，只会记录日志，不会阻塞同级监听器或改变执行。
 
 ## 失败纪律
 
@@ -33,10 +33,10 @@ workflow seam（`ctx.workflows`）执行由模型编写、可扇出 subagent 的
 - `SCRIPT_PARSE` / `META_INVALID`：工作流无法启动；
 - `INVALID_ARGUMENT` / `UNSUPPORTED_OPTION` / `UNSUPPORTED_SCHEMA`：钩子调用违反引擎契约；
 - `AGENT_CAP` / `ITEM_CAP`：超过已配置的安全上限；
-- `AGENT_START`：提供方异步启动被拒绝；
+- `AGENT_START`：提供方的异步启动调用被拒绝；
 - `AGENT_RESULT`：已就绪子 agent 的结果因基础设施故障而拒绝；
 - `RESULT_UNSERIALIZABLE`：脚本/worker 值不是普通 JSON 数据；
-- `CANCELLED`：取消拥有该运行，待处理和未来的钩子都会拒绝。
+- `CANCELLED`：取消会接管该运行，待处理和未来的钩子都会拒绝。
 
 子 agent 若以非完成的结束原因正常兑现，并不属于基础设施异常：`agent()` 返回 `null`，使脚本可以处理普通的子 agent 失败。
 
@@ -46,14 +46,14 @@ workflow seam（`ctx.workflows`）执行由模型编写、可扇出 subagent 的
 
 #### KV Cache 影响
 
-不会直接使缓存失效；具名消费方负责请求前缀的任何变化。
+不会直接导致 KV Cache 失效；请求前缀的任何变化均由上述消费方负责。
 
-## 已知限制与延期工作
+## 已知限制与暂缓事项
 
-- **仅支持前台收集**：调用方拥有一个实时运行并等待它；后台启动/轮询、spill 句柄和分离收集均延期处理。
-- **没有日志记录或恢复**：脚本、子 agent 进度和中间值均不设检查点，因此进程重启后无法继续运行。
+- **仅支持前台收集**：调用方负责一个活动运行并等待它；后台启动／轮询、spill 句柄和分离收集均暂缓处理。
+- **没有日志化或恢复**：脚本、子 agent 进度和中间值均不设检查点，因此进程重启后无法继续运行。
 - **没有已保存或嵌套工作流**：该 seam 只启动调用方提供的脚本，工作流脚本不会收到用于递归编排的 `workflow()` 钩子。
 - **没有 token 预算词汇**：引擎会限制并发、条目和子 agent，但请求与结果都不会统计跨子 agent 的模型 token。
-- **运行由持有方拥有，不由服务跟踪**：卸载引擎不会发现独立的实时句柄；每个消费方都必须 dispose 自己启动的运行。
+- **运行由持有方负责，不由服务跟踪**：卸载引擎不会发现独立的活动句柄；每个消费方都必须 dispose 自己启动的运行。
 
-延期的工作流接口见[动态工作流 Agent Note](../../../.agents/notes/implemented/feature/2026-07-05-dynamic-workflows.md)。
+暂缓实现的工作流接口见[动态工作流 Agent Note（agent 决策记录）](../../../.agents/notes/implemented/feature/2026-07-05-dynamic-workflows.md)。
diff --git a/packages/workspace/README.i18n.yaml b/packages/workspace/README.i18n.yaml
index 6d62c08c0b..b86bee367f 100644
--- a/packages/workspace/README.i18n.yaml
+++ b/packages/workspace/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/workspace/README.md
 README.md: ba92e95d3cde0a95eaaeae5a9b4384c3b8c9c4b8
-README.zh.md: 8c8146bba5fa6d81c0ce5d2ed29add77b4083270
+README.zh.md: 2646a5fbcc9921362978fe129fe59d022efdd1c4
diff --git a/packages/workspace/README.zh.md b/packages/workspace/README.zh.md
index 8c8146bba5..2646a5fbcc 100644
--- a/packages/workspace/README.zh.md
+++ b/packages/workspace/README.zh.md
@@ -2,10 +2,10 @@
 
 [English](README.md) | 中文
 
-Workspace 系列拥有持久 workspace 概念：用户工作所在的目录，包含标题以及属于它的有序会话列表。设计记录：[领域 KV 存储 Agent Note](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md)。
+Workspace 系列负责持久 workspace 概念：用户工作所在的目录，包含标题以及属于它的有序会话列表。设计记录：[领域 KV 存储 Agent Note（agent 决策记录）](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md)。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
-| `workspace/` | 位于存储领域形式之上的 `WorkspaceRegistry` 服务：按 realpath 唯一的路径、会话所有权计数、实体缓存 | `ctx.workspace` |
+| `workspace/` | 基于存储领域数据形式的 `WorkspaceRegistry` 服务：按 realpath 去重的路径、会话归属记账、实体缓存 | `ctx.workspace` |
 
-所有权真相存在 workspace 记录的 `sessionIds`（有序）中，绝不从会话 cwd 派生；`attachSession` 会验证会话头的 cwd 解析到 workspace 路径，因此一个会话在结构上最多属于一个 workspace。删除 Workspace 只会移除该注册表记录及账本：目录、用户文件和会话日志都会保留，相关会话则进入 Ungrouped（参见[决策记录](../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)）。
+所有权信息以 workspace 记录中的 `sessionIds`（有序）为准，绝不从会话 cwd 派生；`attachSession` 会验证会话头的 cwd 解析到 workspace 路径，因此一个会话在结构上最多属于一个 workspace。删除 Workspace 只会移除该注册表记录及会话归属记录：目录、用户文件和会话日志都会保留，相关会话则进入 Ungrouped（参见[决策记录](../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)）。
diff --git a/packages/workspace/workspace/README.i18n.yaml b/packages/workspace/workspace/README.i18n.yaml
index b5eaefa98c..a24e81d977 100644
--- a/packages/workspace/workspace/README.i18n.yaml
+++ b/packages/workspace/workspace/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/workspace/workspace/README.md
 README.md: bee3e4fcb5dded273f30942ee2e42ee93b839e62
-README.zh.md: 7960a2d13df4f881687fd88cdb07e237b3abb7c8
+README.zh.md: 9a052f796fb7cc8756999bdc9b2ce905805730ab
diff --git a/packages/workspace/workspace/README.zh.md b/packages/workspace/workspace/README.zh.md
index 7960a2d13d..9a052f796f 100644
--- a/packages/workspace/workspace/README.zh.md
+++ b/packages/workspace/workspace/README.zh.md
@@ -4,27 +4,27 @@
 
 DeepSeek Harness 的 Workspace 实体注册表（`ctx.workspace`）：通过领域数据形式存储持久 workspace 记录、稳定 workspace 顺序和按新到旧排列的候选会话索引。消费方看到 `Workspace` 接口；实体实现保持包（package）私有。
 
-实体／存储理由见[领域 Agent Note](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md)；仅头部启动和 GUI 排序见 [Workspace UI 产品流 Agent Note](../../../.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md)。
+实体／存储理由见[领域 Agent Note（agent 决策记录）](../../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md)；仅使用头部的引导初始化和 GUI 排序见 [Workspace UI 产品流 Agent Note](../../../.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md)。
 
-## 形状
+## 结构
 
 - `ctx.workspace.create(path, title?)`：规范化 `path` 时使用 `fs.realpath`，拒绝不存在或非目录的路径，每个规范路径最多创建一条记录，并将新记录前置到持久 workspace 顺序。对同一路径重复调用会返回现有 workspace，且不改变其标题；不同路径不能创建重复标题。
-- `ctx.workspace.get(id)`/`list()`/`resolveByPath(path)`：由缓存提供的查找。`list()` 为同步操作，并遵循持久注册表顺序；`resolveByPath` 为异步操作，因为它应用同一 `realpath` 规范，并会拒绝缺失路径，而不是创建路径。
-- `ctx.workspace.delete(id)`：只移除 Workspace 注册记录、对应的持久顺序条目及会话账本。未知 id 返回 `false`，成功移除记录则返回 `true`。目录、用户文件、实时会话和持久化会话日志绝不受影响，因此相关会话会进入 Ungrouped。表写入失败时会恢复原顺序和此前发布的实体。
+- `ctx.workspace.get(id)`/`list()`/`resolveByPath(path)`：由缓存提供的查找。`list()` 为同步操作，并遵循持久注册表顺序；`resolveByPath` 为异步操作，因为它采用相同的 `realpath` 规范化方式，并会拒绝缺失路径，而不是创建路径。
+- `ctx.workspace.delete(id)`：只移除 Workspace 注册记录、对应的持久顺序条目及会话归属记录。未知 id 返回 `false`，成功移除记录则返回 `true`。目录、用户文件、活跃会话和持久化会话日志绝不受影响，因此相关会话会进入 Ungrouped。表写入失败时会恢复原顺序和此前发布的实体。
 - `Workspace.attachSession(id)`：对照 workspace 路径验证实时或已持久化的会话头 cwd，并将新 id 前置。未知会话、缺失／无法解析／非目录的 cwd 值和不匹配情况都会在不写入的前提下被拒绝。`detachSession` 只移除候选索引条目。
-- `ctx.workspace.touchSession(id)`：仅将已验证、已记账的会话移到最前。未分组或被过滤的会话为空操作，workspace 顺序绝不改变。
-- `Workspace.sessionIds`：按持久候选顺序提供同步 id 加规范 cwd 成员投影。缺失头部、无效 cwd 值和不匹配情况都被过滤；下一次 workspace 变更会剪除它们。如果同一存储介质将一个会话索引到两个 workspace 下、从两条记录声明同一路径，或偏离持久 workspace 顺序，启动会被拒绝。
+- `ctx.workspace.touchSession(id)`：仅将已验证、已记账的会话移到最前。未分组或被过滤的会话不会触发任何操作，workspace 顺序绝不改变。
+- `Workspace.sessionIds`：按持久候选顺序提供同步 id 加规范 cwd 成员投影。缺失头部、无效 cwd 值和不匹配情况都被过滤；下一次 workspace 变更会剪除它们。如果同一存储介质将一个会话索引到两个 workspace 下、用两条记录声明同一路径，或偏离持久 workspace 顺序，启动会被拒绝。
 - `Workspace.status()`：未缓存的目录检查，返回 `'ok' | 'missing-dir'`；目录缺失绝不会改动记录。
 
-`storageDomain` 和 `sessionPersistence` 是启动必需依赖。对等服务不可用时，插件保持待处理，且不能提交空的已初始化标记。首次成功启动时，注册表调用 `SessionPersistence.list()`，仅使用头部 `id`、`cwd` 和 `createdAt` 对有效历史目录分组并持久化初始顺序；它绝不读取事件正文。已初始化标记最后写入，因此重启后可安全复用部分启动写入。后续仅有 cwd 的会话仍属于 Ungrouped。
+`storageDomain` 和 `sessionPersistence` 是启动必需依赖。任一依赖服务不可用时，插件保持待处理，且不能提交空的已初始化标记。首次成功启动时，注册表调用 `SessionPersistence.list()`，仅使用头部 `id`、`cwd` 和 `createdAt` 对有效历史目录分组并持久化初始顺序；它绝不读取事件正文。已初始化标记最后写入，因此重启后可安全复用引导初始化期间的部分写入。后续仅能通过 cwd 识别的会话仍属于 Ungrouped。
 
-Create 与 delete 会在记录和顺序可能分叉之前，先持久化明确的待处理变更标记。启动时只补全被该标记证明的变更，随后清除标记；没有标记的顺序／表不一致仍属于来源不明的损坏，并会直接失败。删除后重新注册同一路径会生成新的 Workspace id，且不会自动重新接纳保留下来的 Session。
+创建和删除操作会在记录和顺序可能分叉之前，先持久化明确的待处理变更标记。启动时只补全该标记所指明的变更，随后清除标记；没有标记的顺序／表不一致仍属于来源不明的损坏，并会明确报错。删除后重新注册同一路径会生成新的 Workspace id，且不会自动重新接纳保留下来的会话。
 
 ## 模型体验
 
 ### Workspace 记录与会话记账
 
-#### 模型所见内容
+#### 模型看到的内容
 
 没有。`ctx.workspace` 只向宿主侧消费方提供 workspace 记录：此包不注册工具、不注入提示词、不写入会话事件，因此没有请求字段会携带此包数据。
 
@@ -34,9 +34,9 @@ Create 与 delete 会在记录和顺序可能分叉之前，先持久化明确
 
 #### KV Cache 影响
 
-与实时请求无关：此包绝不触及请求前缀，因此无法使提供方缓存复用失效。
+与实时请求无关：此包绝不触及请求前缀，因此不会使提供方缓存复用失效。
 
-## 已知限制与延后工作
+## 已知限制与暂缓事项
 
 - 会话删除与破坏性的文件夹移除是彼此独立且尚未提供的功能；删除 Workspace 注册记录绝不能替代二者（参见[决策记录](../../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)）。
-- 头部索引会在启动时刷新，也会在 attach 必须解析未缓存持久 id 时刷新；另一进程执行的删除或 cwd 破坏会在下次刷新或重启后被观测。
+- 头部索引会在启动时刷新，也会在 attach 必须解析未缓存持久 id 时刷新；另一进程执行的删除或造成的 cwd 损坏会在下次刷新或重启后被发现。
diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml
index 66ea5bea26..99ac494365 100644
--- a/pnpm-lock.yaml
+++ b/pnpm-lock.yaml
@@ -1049,6 +1049,9 @@ importers:
         specifier: ^2.0.0
         version: 2.1.1
     devDependencies:
+      '@deepseek-ai/dsh-client-locale':
+        specifier: workspace:^
+        version: link:../locale
       '@deepseek-ai/dsh-client-runtime':
         specifier: workspace:^
         version: link:../runtime
@@ -1064,6 +1067,9 @@ importers:
       '@deepseek-ai/dsh-client-ui-slots':
         specifier: workspace:^
         version: link:../ui-slots
+      '@deepseek-ai/dsh-goal':
+        specifier: workspace:^
+        version: link:../../goal/goal
       '@deepseek-ai/dsh-invariants':
         specifier: workspace:^
         version: link:../../support/invariants
@@ -1280,9 +1286,21 @@ importers:
       '@shikijs/langs':
         specifier: ^4.3.1
         version: 4.3.1
+      anser:
+        specifier: ^2.3.5
+        version: 2.3.5
       clsx:
         specifier: ^2.0.0
         version: 2.1.1
+      mdast-util-from-markdown:
+        specifier: ^2.0.3
+        version: 2.0.3
+      mdast-util-gfm:
+        specifier: ^3.1.0
+        version: 3.1.0
+      micromark-extension-gfm:
+        specifier: ^3.0.0
+        version: 3.0.0
       react:
         specifier: ^18.2.0
         version: 18.3.1
@@ -1483,9 +1501,15 @@ importers:
         specifier: ^2.0.0
         version: 2.1.1
     devDependencies:
+      '@deepseek-ai/dsh-client-locale':
+        specifier: workspace:^
+        version: link:../locale
       '@deepseek-ai/dsh-client-runtime':
         specifier: workspace:^
         version: link:../runtime
+      '@deepseek-ai/dsh-client-ui-primitives':
+        specifier: workspace:^
+        version: link:../ui-primitives
       '@deepseek-ai/dsh-client-ui-slots':
         specifier: workspace:^
         version: link:../ui-slots
@@ -1564,6 +1588,10 @@ importers:
         version: 18.3.1
 
   packages/client/ui-trajectory:
+    dependencies:
+      diff:
+        specifier: ^9.0.0
+        version: 9.0.0
     devDependencies:
       '@deepseek-ai/dsh-client-runtime':
         specifier: workspace:^
@@ -1571,6 +1599,9 @@ importers:
       '@deepseek-ai/dsh-client-ui-conversation':
         specifier: workspace:^
         version: link:../ui-conversation
+      '@deepseek-ai/dsh-client-ui-primitives':
+        specifier: workspace:^
+        version: link:../ui-primitives
       '@deepseek-ai/dsh-client-ui-slots':
         specifier: workspace:^
         version: link:../ui-slots
@@ -1880,6 +1911,34 @@ importers:
         specifier: ^4.0.0-rc.7
         version: 4.0.0-rc.7(@cordisjs/plugin-include@1.0.4)(@cordisjs/plugin-loader@1.0.0-rc.5)
 
+  packages/context/tmux-context:
+    dependencies:
+      schemastery:
+        specifier: ^3.18.0
+        version: 3.18.0
+    devDependencies:
+      '@deepseek-ai/dsh-agent':
+        specifier: workspace:^
+        version: link:../../core/agent
+      '@deepseek-ai/dsh-bash':
+        specifier: workspace:^
+        version: link:../../bash/bash
+      '@deepseek-ai/dsh-invariants':
+        specifier: workspace:^
+        version: link:../../support/invariants
+      '@deepseek-ai/dsh-llm':
+        specifier: workspace:^
+        version: link:../../llm/llm
+      '@deepseek-ai/dsh-session':
+        specifier: workspace:^
+        version: link:../../core/session
+      '@deepseek-ai/dsh-system-prompt':
+        specifier: workspace:^
+        version: link:../../core/system-prompt
+      cordis:
+        specifier: ^4.0.0-rc.7
+        version: 4.0.0-rc.7(@cordisjs/plugin-include@1.0.4)(@cordisjs/plugin-loader@1.0.0-rc.5)
+
   packages/context/workspace-context:
     dependencies:
       schemastery:
@@ -2890,6 +2949,9 @@ importers:
       '@deepseek-ai/dsh-session-projection-cache':
         specifier: workspace:^
         version: link:../../session-projection/session-projection-cache
+      '@deepseek-ai/dsh-session-title':
+        specifier: workspace:^
+        version: link:../../session-title/session-title
       '@deepseek-ai/dsh-skill':
         specifier: workspace:^
         version: link:../../skill/skill
@@ -5017,6 +5079,9 @@ importers:
       '@deepseek-ai/dsh-skill':
         specifier: workspace:^
         version: link:../../skill/skill
+      '@deepseek-ai/dsh-subprocess':
+        specifier: workspace:^
+        version: link:../../subprocess/subprocess
       '@deepseek-ai/dsh-system-prompt':
         specifier: workspace:^
         version: link:../../core/system-prompt
@@ -8274,6 +8339,9 @@ packages:
     resolution: {integrity: sha512-OyacJsaeuLUvGWOynNqYc6sx88XvyoG39wMT8SYqL3l9wwaorDW/LPRbUPfhzw0bWsUWzNCZTnFYOrWFBKsUaw==}
     engines: {node: '>= 14.0.0'}
 
+  anser@2.3.5:
+    resolution: {integrity: sha512-vcZjxvvVoxTeR5XBNJB38oTu/7eDCZlwdz32N1eNgpyPF7j/Z7Idf+CUwQOkKKpJ7RJyjxgLHCM7vdIK0iCNMQ==}
+
   ansi-regex@5.0.1:
     resolution: {integrity: sha512-quJQXlTSUGL2LH9SUXo8VwsY4soanhgo6LNSm84E1LBcE8s3O0wpdiRzyR9z/ZZJMlMWv37qOOb9pdJlMUEKFQ==}
     engines: {node: '>=8'}
@@ -13258,6 +13326,8 @@ snapshots:
       '@algolia/requester-fetch': 5.55.2
       '@algolia/requester-node-http': 5.55.2
 
+  anser@2.3.5: {}
+
   ansi-regex@5.0.1: {}
 
   ansi-regex@6.2.2: {}
diff --git a/python/README.i18n.yaml b/python/README.i18n.yaml
index 5df1ca291d..a679815f4e 100644
--- a/python/README.i18n.yaml
+++ b/python/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write python/README.md
 README.md: d2b6a1cfe9897026d567b2def301799069c350fb
-README.zh.md: 2ffccef922d52923f7c6e373a01ed8b19d9a55c1
+README.zh.md: f519fbc0a85539e4dc053785e1a8d6759e6fc894
diff --git a/python/README.zh.md b/python/README.zh.md
index 2ffccef922..f519fbc0a8 100644
--- a/python/README.zh.md
+++ b/python/README.zh.md
@@ -2,13 +2,13 @@
 
 [English](README.md) | 中文
 
-以子进程方式驱动 DeepSeek Harness 的 Python 包：客户端 SDK spawn `dsh-jsonrpc-agent` 二进制，并通过 stdio 上按行分隔的 JSON-RPC 与之通信。运行时载体是本仓库产出的单文件可执行文件；设计、构建与验收细节见 [.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md](../.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md)。
+以子进程方式驱动 DeepSeek Harness 的 Python 包：客户端 SDK 启动 `dsh-jsonrpc-agent` 二进制，并通过 stdio 上按行分隔的 JSON-RPC 与之通信。运行时载体是本仓库产出的单文件可执行文件；设计、构建与验收细节见 [.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md](../.agents/notes/implemented/architecture/2026-07-10-single-file-executable-sdk-runtime-distribution.md)。
 
 ## 包
 
 | 目录 | 分发名 / 模块 | 职责 |
 |---|---|---|
-| [sdk](sdk/) | `deepseek-harness` / `deepseek_harness` | 客户端 SDK：高层回合 API `DeepSeekHarness` 与低层 JSON-RPC 客户端 `HarnessClient` |
+| [sdk](sdk/) | `deepseek-harness` / `deepseek_harness` | 客户端 SDK：高层轮次 API `DeepSeekHarness` 与低层 JSON-RPC 客户端 `HarnessClient` |
 | [sdk-runtime](sdk-runtime/) | `deepseek-harness-runtime-bin` / `deepseek_harness_runtime` | 运行时载体：定位内置的运行时二进制，并携带默认的 agent（智能体）配置 |
 
 ## 构建运行时可执行文件
@@ -63,7 +63,7 @@ pip install --find-links dist-python deepseek-harness=="$version"
 
 ## 零配置语义
 
-运行时二进制本身始终要求显式配置（`$DSH_CORDIS_CONFIG`，或作为首个 argv 参数的配置路径），没有内置兜底，也只启动配置里列出的内容。零配置是 SDK 包装层的行为：调用方没有使用任何显式通道时，客户端把运行时包检入的默认配置（[runtime/cordis.yml](sdk-runtime/src/deepseek_harness_runtime/runtime/cordis.yml)）注入 `DSH_CORDIS_CONFIG`；任一显式通道存在即优先采用，并禁用注入。注入条件的完整定义见 [sdk README](sdk/README.md)，默认配置的内容与硬语义见 [sdk-runtime README](sdk-runtime/README.md)。
+运行时二进制本身始终要求显式配置（`$DSH_CORDIS_CONFIG`，或作为首个 argv 参数的配置路径），没有内置兜底，也只启动配置里列出的内容。零配置是 SDK 包装层的行为：调用方没有使用任何显式通道时，客户端把运行时包中检入的默认配置（[runtime/cordis.yml](sdk-runtime/src/deepseek_harness_runtime/runtime/cordis.yml)）注入 `DSH_CORDIS_CONFIG`；任一显式通道存在即优先采用，并禁用注入。注入条件的完整定义见 [sdk README](sdk/README.md)，默认配置的内容与强制语义见 [sdk-runtime README](sdk-runtime/README.md)。
 
 可执行文件也支持直接调用；在 NDJSON JSON-RPC 交互期间保持 stdin 打开，并显式提供配置：
 
@@ -73,4 +73,4 @@ DSH_CORDIS_CONFIG=/absolute/path/cordis.yml ./dsh-jsonrpc-agent-pkg-macos-arm64
 
 ## 测试布局
 
-`test_client.py` 完全无需密钥（对端是 Python 假运行时）。`test_bundled_runtime.py` 逐个启动内置载体，某个载体产物缺失时跳过对应用例。`test_runtime_resolution.py` 覆盖载体解析规则，不 spawn 任何进程。
+`test_client.py` 完全无需密钥（对端是 Python 假运行时）。`test_bundled_runtime.py` 逐个启动内置载体，某个载体产物缺失时跳过对应用例。`test_runtime_resolution.py` 覆盖载体解析规则，不启动任何进程。
diff --git a/python/sdk-runtime/README.i18n.yaml b/python/sdk-runtime/README.i18n.yaml
index 90433c4f5f..78c6fcc783 100644
--- a/python/sdk-runtime/README.i18n.yaml
+++ b/python/sdk-runtime/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
+#   pnpm run verify-translation-pairing --write python/sdk-runtime/README.md
 README.md: f2ccd8939e497d10359aafe8b1bd8b364875ed98
-README.zh.md: 30bdf46fee03c38a1f4b6e8b2b39d87e8174a3e0
+README.zh.md: 5e11ee27568a7b1d7321ffb8486aba4adb1603e8
diff --git a/python/sdk-runtime/README.zh.md b/python/sdk-runtime/README.zh.md
index 30bdf46fee..5e11ee2756 100644
--- a/python/sdk-runtime/README.zh.md
+++ b/python/sdk-runtime/README.zh.md
@@ -9,21 +9,21 @@ Python SDK 的运行时载体包（分发名 `deepseek-harness-runtime-bin`，
 两种载体并存于 `src/deepseek_harness_runtime/runtime/` 之下，均由仓库的 `scripts/build-exe-for-python-sdk.ts` 构建注入，且均被 git 忽略：
 
 - **exe（生产）**——单文件可执行程序 `dsh-jsonrpc-agent-pkg-<platform>-<arch>`（platform：`linux`/`macos`；arch：`x64`/`arm64`）。目标机器无需安装 Node。这是唯一随 wheel 包分发的载体；本包不发布 sdist。
-- **`node`（仅限开发）**——`runtime/node/` 下的完整部署闭包（`package.json` + `node_modules/`），在系统 Node >= 22.19 上以 `node runtime/node/node_modules/@deepseek-ai/dsh-jsonrpc-demo/lib/bin.js` 执行。它是当前检出的源码构建，仅用于仓库本地的开发与验证；不会被自动选中，也不进入分发物。
+- **node（仅限开发）**——`runtime/node/` 下的完整部署闭包（`package.json` + `node_modules/`），在系统 Node >= 22.19 上以 `node runtime/node/node_modules/@deepseek-ai/dsh-jsonrpc-demo/lib/bin.js` 执行。它是当前检出的源码构建，仅用于仓库本地的开发与验证；不会被自动选中，也不进入分发物。
 
 两种载体承载相同的内容，且只定义一次：本包根目录的 [package.json](package.json) 是 single-exe 流水线的部署根目录——一份零代码的纯依赖 manifest，其依赖闭包既是编译进 exe 的插件集，也是物化到 `runtime/node/` 的文件树。往分发物里加插件，就是在那里加一行依赖再重新构建。
 
-exe 缺失时抛出 `FileNotFoundError`，并写明两种获取途径：在 deepseek-harness 检出中经 `scripts/build-exe-for-python-sdk.ts` 构建，或安装 `build-exe-for-python-sdk` CI 工作流生成的对应平台运行时 wheel 包。仅限开发的 `node` 载体缺失时只提示构建脚本这一条途径。该工作流只保留 wheel 包，不保留独立 exe 归档。获取策略与查找接口刻意分离，之后可以换成按需下载而不改动任何调用方。
+exe 缺失时抛出 `FileNotFoundError`，并写明两种获取途径：在 deepseek-harness 检出中经 `scripts/build-exe-for-python-sdk.ts` 构建，或安装 `build-exe-for-python-sdk` CI 工作流生成的对应平台运行时 wheel 包。仅限开发的 node 载体缺失时只提示构建脚本这一条途径。该工作流只保留 wheel 包，不保留独立 exe 归档。获取策略与查找接口刻意分离，之后可以换成按需下载而不改动任何调用方。
 
-每个 wheel 包只包含一个可执行文件。固定标签为 `py3-none-manylinux_2_28_x86_64`、`py3-none-manylinux_2_28_aarch64` 与 `py3-none-macosx_11_0_arm64`；构建钩子会拒绝 `py3-none-any`、可执行文件缺失或重复以及不支持的平台标签。仓库根目录的 `package.json` 为本包和 SDK 提供共同版本，`python-vX.Y.Z` 发布标签必须与其匹配。
+每个 wheel 包只包含一个可执行文件。固定标签为 `py3-none-manylinux_2_28_x86_64`、`py3-none-manylinux_2_28_aarch64` 与 `py3-none-macosx_11_0_arm64`；构建钩子会拒绝 `py3-none-any`、不存在可执行文件、存在多个可执行文件以及不支持的平台标签。仓库根目录的 `package.json` 为本包和 SDK 提供共同版本，`python-vX.Y.Z` 发布标签必须与其匹配。
 
 ## 解析 API
 
-- `resolve_bundled_launch_args(mode=None) -> tuple[str, ...]`——启动内置运行时的 argv 元组：exe 模式下为 `(exe_path,)`，`node` 模式下为 `(node_path, bin_js_path)`。模式选择：显式参数 > `DSH_RUNTIME_MODE` 环境变量（`exe` | `node`）> 自动。自动解析只找生产 exe——仅限开发的 `node` 载体必须显式选用，从而生产部署绝不会悄悄跑在源码构建上。
-- `bundled_runtime_path() -> Path`——平台 exe 路径（仅 exe 载体；`node` 载体没有单一路径的等价物，经由上面的 argv 元组启动）。
+- `resolve_bundled_launch_args(mode=None) -> tuple[str, ...]`——启动内置运行时的 argv 元组：exe 模式下为 `(exe_path,)`，node 模式下为 `(node_path, bin_js_path)`。模式选择：显式参数 > `DSH_RUNTIME_MODE` 环境变量（`exe` | `node`）> 自动。自动解析只找生产 exe——仅限开发的 node 载体必须显式选用，从而生产部署绝不会悄悄跑在源码构建上。
+- `bundled_runtime_path() -> Path`——平台 exe 路径（仅 exe 载体；node 载体没有单一路径的等价物，经由上面的 argv 元组启动）。
 - `bundled_default_config_path() -> Path`——检入的默认配置（见下文）。
 - `bundled_package_dir() -> Path`——已安装包的数据根目录。
 
 ## 零配置设计
 
-运行时二进制始终要求显式配置（`$DSH_CORDIS_CONFIG`，或作为 argv 位置参数的配置路径），缺了就报错退出——这一硬语义是运行时设计的一部分，本包不软化它。`bin`（`dsh-jsonrpc-agent`）只启动配置里列出的插件；对外服务接口（stdio JSON-RPC 服务器）也是其中一个条目（`@deepseek-ai/dsh-jsonrpc`），缺了它，启动出的 agent 就没有对外通道。本包检入的 `runtime/cordis.yml` 包含 JSON-RPC 服务条目、`agent-core`、预载的 DeepSeek 适配器、JSONL 持久化、显式组合的语义检查点策略、本地 bash，以及用于有界加载工作区指令的本地文件系统 provider。持久化后端负责持久存储，独立的策略则选择请求、工具分发和已完成步骤的检查点。DeepSeek 适配器读取 `DEEPSEEK_API_KEY` 与 `DEEPSEEK_BASE_URL`，持久化、bash 和文件系统 provider 则使用 `DSH_SESSION_ROOT` 和 `DSH_CWD`，并为手动运行提供回退值。调用方未使用任何显式配置通道时，`deepseek_harness` 客户端把该文件路径注入 `DSH_CORDIS_CONFIG`（注入条件见 [sdk README](../sdk/README.md)）。因此，零配置是包装层中一次显式、可见的参数传递，而不是运行时中的隐藏回退。
+运行时二进制始终要求显式配置（`$DSH_CORDIS_CONFIG`，或作为 argv 位置参数的配置路径），缺了就报错退出——这一强制语义是运行时设计的一部分，本包不会弱化它。bin（`dsh-jsonrpc-agent`）只启动配置里列出的插件；对外服务接口（stdio JSON-RPC 服务器）也是其中一个条目（`@deepseek-ai/dsh-jsonrpc`），缺了它，启动出的 agent（智能体）就没有对外通道。本包检入的 `runtime/cordis.yml` 包含 JSON-RPC 服务条目、agent 核心、预载的 DeepSeek 适配器、JSONL 持久化、显式组合的语义检查点策略、本地 bash，以及用于有界加载工作区指令的本地文件系统提供方。持久化后端负责持久存储，独立的策略则选择请求、工具分发和已完成步骤的检查点。DeepSeek 适配器读取 `DEEPSEEK_API_KEY` 与 `DEEPSEEK_BASE_URL`，持久化、bash 和文件系统提供方则使用 `DSH_SESSION_ROOT` 和 `DSH_CWD`，并为手动运行提供回退值。调用方未使用任何显式配置通道时，`deepseek_harness` 客户端把该文件路径注入 `DSH_CORDIS_CONFIG`（注入条件见 [sdk README](../sdk/README.md)）。因此，零配置是包装层中一次显式、可见的参数传递，而不是运行时中的隐藏回退。
diff --git a/python/sdk/README.i18n.yaml b/python/sdk/README.i18n.yaml
index 048fecb6f9..0115e41ce9 100644
--- a/python/sdk/README.i18n.yaml
+++ b/python/sdk/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write python/sdk/README.md
 README.md: f1e16e724efd6f71f63e475e47d7e4d704b8ceac
-README.zh.md: e56ae31020d1068e009056c20d8f11bab145dc8a
+README.zh.md: 11ff235a0eea23acab5f1737537c799021c35bea
diff --git a/python/sdk/README.zh.md b/python/sdk/README.zh.md
index e56ae31020..11ff235a0e 100644
--- a/python/sdk/README.zh.md
+++ b/python/sdk/README.zh.md
@@ -31,7 +31,7 @@ with DeepSeekHarness(
 
 `provider` 用于选择当前 Cordis 组合已注册的提供方路由；`model` 是该适配器解析的模型 ID。`max_tokens` 是可选的正整数，用于限制根 agent 及其进程内后代每次请求的输出 token；省略时由提供方默认值控制。压缩摘要继续使用压缩插件单独配置的上限。内置默认组合注册 `deepseek`。自定义组合可以挂载 `llm-pi-ai`，在其中配置各提供方的凭据与端点，再选择 pi-ai 已安装目录中的任意提供方/模型组合。
 
-`HarnessClient` 会在运行时进程的生命周期内保留已发现的 subagent（子 agent）祖先关系。每次执行 `Session.run()` 时，`TurnResult.notifications` 与 `on_notification` 会按线上的原始顺序收到根会话及所有已知后代的通知，其中包括嵌套 subagent 的生命周期与会话事件。`TurnResult.events` 仍只保存根会话的完整事件流，`TurnResult.final_response` 则取该会话最后一个 `assistant/message` 的文本内容，因此后代消息不会覆盖根会话回复。
+`HarnessClient` 会在运行时进程的生命周期内保留已发现的 subagent（子 agent）祖先关系。每次执行 `Session.run()` 时，`TurnResult.notifications` 与 `on_notification` 会按协议传输顺序收到根会话及所有已知后代的通知，其中包括嵌套 subagent 的生命周期事件与会话事件。`TurnResult.events` 仍只保存根会话的完整事件流，`TurnResult.final_response` 则取该会话最后一个 `assistant/message` 的文本内容，因此后代消息不会覆盖根会话回复。
 
 同样的行为也可以通过 `DSH_CORDIS_CONFIG` 为运行时子进程选定。注入逻辑位于 `HarnessClient.start()`，因此底层客户端的默认启动也具有此行为：当启动解析到内置运行时，且 `cordis` 与非空的 `DSH_CORDIS_CONFIG` 均未设置时（运行时把空值视为缺省，注入检查与之一致），使用内置的默认配置；显式给出 `runtime_bin`、`bridge_bin` 或 `launch_args_override` 则完全禁用注入。运行时载体（生产用 exe 与仅限开发的 `node` 闭包）及其获取方式见 [sdk-runtime README](../sdk-runtime/README.md)。
 
diff --git a/scripts/client-bundle-css.spec.ts b/scripts/client-bundle-css.spec.ts
new file mode 100644
index 0000000000..30350241fc
--- /dev/null
+++ b/scripts/client-bundle-css.spec.ts
@@ -0,0 +1,47 @@
+/**
+ * CSS Modules enter client bundles through virtual modules, so the loader must
+ * explicitly register the underlying stylesheet as a watch dependency.
+ */
+import { mkdtemp, rm, writeFile } from 'node:fs/promises'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
+import { describe, expect, it } from 'vitest'
+import { clientBundle } from '../packages/client/tsdown.client.ts'
+
+interface CssPlugin {
+  name: string
+  resolveId?: (source: string, importer?: string) => string | null
+  load?: (this: { addWatchFile(id: string): void }, id: string) => Promise<string | null>
+}
+
+function cssPlugin(): CssPlugin {
+  const configs = clientBundle('@deepseek-ai/dsh-client-test', ['lib/types/index.js', 'lib/types/invariant.js'])
+  const plugins = (configs[1] as { plugins: CssPlugin[] }).plugins
+  const plugin = plugins.find(candidate => candidate.name === 'dsh-css-modules-inline')
+  if (plugin === undefined) throw new Error('CSS Modules plugin missing from client config')
+  return plugin
+}
+
+describe('client bundle CSS Modules', () => {
+  it('registers the source stylesheet as a watch dependency', async () => {
+    const root = await mkdtemp(join(tmpdir(), 'dsh-client-css-watch-'))
+    try {
+      const stylesheet = join(root, 'Fixture.module.css')
+      const importer = join(root, 'index.ts')
+      await writeFile(stylesheet, '.root { color: red; }\n')
+      const plugin = cssPlugin()
+      const virtualId = plugin.resolveId?.('./Fixture.module.css', importer)
+      if (typeof virtualId !== 'string' || plugin.load === undefined) {
+        throw new Error('CSS Modules plugin hooks are incomplete')
+      }
+      const watched: string[] = []
+
+      const output = await plugin.load.call({ addWatchFile: id => watched.push(id) }, virtualId)
+
+      expect(watched).toEqual([stylesheet])
+      expect(output).toContain('data-plugin-css')
+    } finally {
+      await rm(root, { recursive: true, force: true })
+    }
+  })
+})
diff --git a/scripts/demo-code-mode.mjs b/scripts/demo-code-mode.mjs
index d790fe15d5..994bea32d0 100644
--- a/scripts/demo-code-mode.mjs
+++ b/scripts/demo-code-mode.mjs
@@ -8,9 +8,8 @@ import { spawn } from 'node:child_process'
 // Each UI's node invocation matches its base demo script plus the overlay config.
 const UIS = new Map([
   ['tui', [
-    '--experimental-transform-types',
     '--import',
-    './scripts/tspath-loader.ts',
+    'tsx/esm',
     'apps/cli/src/bin.ts',
     '--config',
     'examples/tui-agent/code-mode.cordis.yml',
diff --git a/scripts/install-lefthook.mjs b/scripts/install-lefthook.mjs
index 49c246df69..f94fb3bf11 100644
--- a/scripts/install-lefthook.mjs
+++ b/scripts/install-lefthook.mjs
@@ -2,7 +2,7 @@
 import { randomUUID } from 'node:crypto'
 import { existsSync, lstatSync, mkdirSync, readdirSync, readFileSync, unlinkSync, writeFileSync } from 'node:fs'
 import { spawnSync } from 'node:child_process'
-import { isAbsolute, join, resolve } from 'node:path'
+import { dirname, isAbsolute, join, resolve } from 'node:path'
 
 const MINIMUM_GIT = [2, 26, 0]
 const HOOKS_DIRECTORY = 'dsh-hooks'
@@ -437,6 +437,15 @@ function inspectOwnedHooksDirectory(hooksPath) {
   return { markerPath, ...marker }
 }
 
+function isRegisteredOwnedHooksPath(commonDirectory, hooksPath) {
+  const normalizedHooksPath = normalizedPath(hooksPath)
+  const isRegistered = registeredWorktreeConfigPaths(commonDirectory).some(
+    configPath => normalizedPath(join(dirname(configPath), HOOKS_DIRECTORY)) === normalizedHooksPath,
+  )
+  if (!isRegistered) return false
+  return inspectOwnedHooksDirectory(hooksPath)?.hooksPath === hooksPath
+}
+
 function ensureOwnedHooksDirectory(hooksPath) {
   const inspected = inspectOwnedHooksDirectory(hooksPath)
   if (inspected !== undefined) return inspected
@@ -561,14 +570,22 @@ async function main() {
       'worktree core.hooksPath',
     )
     let ownedHooksDirectory
+    let copiedWorktreePathIsOwned = false
     if (worktreePath !== undefined && worktreePath !== hooksPath) {
       ownedHooksDirectory = inspectOwnedHooksDirectory(hooksPath)
-      if (ownedHooksDirectory === undefined || ownedHooksDirectory.hooksPath !== worktreePath) {
+      const worktreePathIsRelocated = ownedHooksDirectory?.hooksPath === worktreePath
+      copiedWorktreePathIsOwned = !worktreePathIsRelocated
+        && isRegisteredOwnedHooksPath(commonDirectory, worktreePath)
+      if (!worktreePathIsRelocated && !copiedWorktreePathIsOwned) {
         refuseScopedHooksPath({ origin: `file:${worktreeConfigPath}`, scope: 'worktree', value: worktreePath })
       }
     }
     const directWorktreePathIsOwned = worktreePath !== undefined
-      && (worktreePath === hooksPath || ownedHooksDirectory?.hooksPath === worktreePath)
+      && (
+        worktreePath === hooksPath
+        || ownedHooksDirectory?.hooksPath === worktreePath
+        || copiedWorktreePathIsOwned
+      )
     const effectiveEntry = effectiveConfigEntry(root, 'core.hooksPath')
     if (effectiveEntry !== undefined) {
       const effectivePathIsOwned = effectiveEntry.scope === 'worktree'
@@ -593,6 +610,7 @@ async function main() {
       worktreePath !== undefined
       && worktreePath !== hooksPath
       && ownedHooksDirectory.hooksPath !== worktreePath
+      && !copiedWorktreePathIsOwned
     ) {
       throw new Error(`hooks directory ownership changed while relocating ${JSON.stringify(worktreePath)}`)
     }
diff --git a/scripts/install-lefthook.spec.ts b/scripts/install-lefthook.spec.ts
index 7e30c887ec..a33c245e3d 100644
--- a/scripts/install-lefthook.spec.ts
+++ b/scripts/install-lefthook.spec.ts
@@ -262,6 +262,30 @@ describe('worktree-local Lefthook installer', () => {
     expect(readFileSync(legacyHook, 'utf8')).toBe('#!/bin/sh\n# legacy hook\n')
   })
 
+  it('replaces the owned hook path Git copies into a newly added worktree', async () => {
+    const fixture = createFixture()
+    const mainInstall = await runInstaller(fixture, fixture.main)
+    expect(mainInstall.status, mainInstall.stderr).toBe(0)
+    const mainHooks = hooksPath(fixture, fixture.main)
+    const mainHookBefore = readFileSync(join(mainHooks, 'pre-commit'), 'utf8')
+    const lateLinked = join(fixture.container, 'late-linked')
+    git(fixture, fixture.main, ['worktree', 'add', '-b', 'late-linked', lateLinked])
+    write(join(lateLinked, 'lefthook.yml'), 'late-linked-worktree-config\n')
+    installFakeLefthook(lateLinked)
+    expect(git(fixture, lateLinked, ['config', '--worktree', '--get', 'core.hooksPath'])).toBe(mainHooks)
+
+    const linkedInstall = await runInstaller(fixture, lateLinked)
+
+    expect(linkedInstall.status, linkedInstall.stderr).toBe(0)
+    const linkedHooks = hooksPath(fixture, lateLinked)
+    expect(linkedHooks).not.toBe(mainHooks)
+    expect(git(fixture, lateLinked, ['config', '--worktree', '--get', 'core.hooksPath'])).toBe(linkedHooks)
+    expect(readFileSync(join(linkedHooks, 'pre-commit'), 'utf8')).toContain(
+      '# config=late-linked-worktree-config',
+    )
+    expect(readFileSync(join(mainHooks, 'pre-commit'), 'utf8')).toBe(mainHookBefore)
+  })
+
   it('serializes concurrent installs and keeps repeated output stable', async () => {
     const fixture = createFixture()
     const delayed = { DSH_TEST_LEFTHOOK_DELAY_MS: '150' }
@@ -509,6 +533,30 @@ describe('worktree-local Lefthook installer', () => {
     expect(git(fixture, fixture.linked, ['config', '--worktree', '--get', 'core.hooksPath'])).toBe('linked-custom-hooks')
   })
 
+  it('does not trust an ownership marker outside a registered worktree hook path', async () => {
+    const fixture = createFixture()
+    const mainInstall = await runInstaller(fixture, fixture.main)
+    expect(mainInstall.status, mainInstall.stderr).toBe(0)
+    const externalHooks = join(fixture.container, 'external-owned-hooks')
+    write(
+      join(externalHooks, '.dsh-lefthook-owned'),
+      `${JSON.stringify({
+        version: 1,
+        owner: 'deepseek-harness worktree-local lefthook hooks',
+        hooksPath: externalHooks,
+      })}\n`,
+      0o600,
+    )
+    git(fixture, fixture.linked, ['config', '--worktree', 'core.hooksPath', externalHooks])
+
+    const result = await runInstaller(fixture, fixture.linked)
+
+    expect(result.status).toBe(1)
+    expect(result.stderr).toContain('worktree-scoped core.hooksPath')
+    expect(git(fixture, fixture.linked, ['config', '--worktree', '--get', 'core.hooksPath'])).toBe(externalHooks)
+    expect(existsSync(hooksPath(fixture, fixture.linked))).toBe(false)
+  })
+
   it('refuses to activate a sibling worktree dormant hook path', async () => {
     const fixture = createFixture()
     const linkedConfig = join(gitDirectory(fixture, fixture.linked), 'config.worktree')
diff --git a/scripts/run-gates.ts b/scripts/run-gates.ts
index aa2378b24f..d10a9ccba7 100644
--- a/scripts/run-gates.ts
+++ b/scripts/run-gates.ts
@@ -287,6 +287,11 @@ function nodeCompatSmokeGates(): Gate[] {
       'run',
       'packages/session-persistence/session-persistence-jsonl/tests/zstd.compat.spec.ts',
     ], { label: 'JSONL Zstandard smoke' }),
+    pnpmExec('dsh-source-launch-smoke', [
+      'vitest',
+      'run',
+      'apps/cli/tests/source-launch.compat.spec.ts',
+    ], { label: 'dsh source-launch smoke' }),
   ]
 }
 
diff --git a/scripts/snapshots/translation-prompt-v4/request-response.expected.json b/scripts/snapshots/translation-prompt-v4/request-response.expected.json
index 96eaad8c33..d26d2e8d22 100644
--- a/scripts/snapshots/translation-prompt-v4/request-response.expected.json
+++ b/scripts/snapshots/translation-prompt-v4/request-response.expected.json
@@ -4,7 +4,7 @@
     "messages": [
       {
         "role": "system",
-        "content": "# Translation Prompt\n\nYou are a senior technical translator specializing in LLM and agent development documentation. Your task is to translate the given source document from English to Chinese, producing natural, professional technical prose.\n\n## Quality Requirements\n\n### Structure and Format Preservation\n- Output a complete translated document that maintains exactly the same structure as the source: heading hierarchy, list shape, table columns, link targets, and code blocks.\n- Fenced code blocks must be byte-identical to the source, including ALL comments inside them. Do NOT translate comments inside code blocks. This is a hard rule with no exceptions.\n- Inline code spans (commands, flags, paths, API names, version numbers) must be kept verbatim. Never translate or reformat them.\n- Every relative link must point to the same target as in the source. Link text is translated; link targets are not.\n- Language switcher line: when translating into Chinese, write `[English](source-filename.md) | 中文`. When translating into English, write `English | [中文](source-filename.zh.md)`. Do NOT copy the switcher line from the source file unchanged — you must flip the link direction.\n- After a closing bold marker `**`, insert a space before the next character when that character is a Latin letter, digit, or CJK ideograph. Never insert a space before any punctuation (full-width or half-width).\n\n### Tone and Style\n- The translation must read as if originally written in the target language by a native speaker. If an expression sounds like a word-for-word rendering from the source language, rephrase it.\n- Write in a professional, formal tone appropriate for developer documentation. Never use colloquial or casual expressions.\n- Use polite imperative forms where the text instructs the reader to do something.\n- Keep the author's register: concise stays concise, detailed stays detailed.\n\n### Sentence Structure\n- Break long sentences with commas or semicolons. Avoid run-on sentences.\n- Prefer active voice. Convert passive constructions to active if it reads more naturally.\n- Translate meaning, not words. Restructure sentences where the target language grammar requires it.\n- Do not invent words or expressions that do not exist in natural technical writing of the target language.\n\n### Word Choice\n- Prefer precise, formal vocabulary over casual or colloquial alternatives.\n- When multiple synonyms exist, choose the one most commonly used in professional technical documentation of the target language.\n- Avoid slang, internal jargon, or overly literal translations that would not be recognized by the general developer audience.\n- Do not use the same word to translate two different source-language terms that carry distinct meanings.\n- Avoid repeating the same verb in close proximity; vary word choice for readability.\n\n#### When translating into Chinese\n- When a number modifies a noun, always include a Chinese classifier or measure word (量词). For example: \"three-package seam\" → \"由三个包构成的 seam\", not \"三包 seam\".\n\n### Punctuation\n\n#### When translating into Chinese\n- Use full-width Chinese punctuation in prose: `，。：；？！（）「」`.\n- Strongly prefer replacing all em-dashes (——) with colons, periods, commas, or parentheses. Keep an em-dash only if no other punctuation works at all.\n- Use enumeration commas (、) between parallel items, not regular commas.\n- List item endings: use semicolons or no punctuation. Do not end list items with commas.\n- Put one half-width space between Chinese text and Latin words/numbers.\n- For RFC 2119 keywords (MUST, MUST NOT, SHOULD, MAY), translate to the corresponding Chinese term (必须、禁止、应当、可以) and keep the SOURCE emphasis marker: plain source stays plain (必须), italic source stays italic (*必须*), and bold source stays bold (**必须**).\n\n#### When translating into English\n(To be added.)\n\n## Terminology\n\nA terminology table is provided below. Follow it strictly:\n- Render every listed term exactly as specified.\n- When the target language is Chinese, use the \"中文\" column. On first occurrence, write the \"首次出现\" value with its parenthetical gloss; on subsequent occurrences, write only the part before the parentheses.\n- When the target language is English, use the \"English\" column without a Chinese gloss; do not copy the \"中文\" or \"首次出现\" value into English prose.\n- If a term has already been glossed as part of a compound term, do not gloss it again when it appears alone later.\n- NEVER use translations listed in the \"不要译作\" column.\n- For technical terms not in the table, follow the target language: for a Chinese target, use an established Chinese rendering from a major Chinese-language OSS or vendor source, or keep the source term and flag it as pending when no such precedent exists; for an English target, use the established English technical term, or preserve an ambiguous source term with a short English gloss and flag it as pending. Do not invent a translation. This rule applies to terminology only; for general prose, freely restructure and paraphrase for natural expression.\n\n# Terminology\n\n本表约定本仓库的中英术语统一译法。\n\n**通用规则：**\n- \"中文\"列为中文译文的正文默认用词。若该列为英文，则中文译文的正文中保留英文不翻译。\n- 首次出现按\"首次出现\"列书写（带括号注释）；后续出现只写括号前的部分（可能为中文，也可能为英文），不出现括号内的注释。\n- \"不要译作\"列为严格禁止的译法。\n- 如果某术语已经作为另一个术语的组成部分被括注过（如 `agent loop（智能体循环）` 中已包含 `agent` 的括注），则该术语后续单独出现时无需再次括注。\n\n## 缩写类（中英文文本中均使用缩写）\n\n| English | 中文 | 首次出现 | 不要译作 | 备注 |\n|---|---|---|---|---|\n| ACP | ACP | ACP（Agent Client Protocol） | | |\n| AI | AI | AI（人工智能） | | |\n| API | API | | | |\n| CI | CI | | | |\n| CLI | CLI | CLI（命令行界面） | | |\n| e2e | e2e | | | |\n| HMR | HMR | HMR（热模块替换） | | |\n| JSON Schema | JSON Schema | | | |\n| JSONL | JSONL | | | |\n| LLM | LLM | LLM（大语言模型） | | |\n| MCP | MCP | | | |\n| PR | PR | PR（Pull Request） | | |\n| RAG | RAG | RAG（检索增强生成） | | |\n| SDK | SDK | | | |\n| SSE | SSE | SSE（Server-Sent Events） | | |\n\n## 英文类（中英文文本中均使用英文）\n\n| English | 中文 | 首次出现 | 不要译作 | 备注 |\n|---|---|---|---|---|\n| agent | agent | agent（智能体） | | |\n| Agent Note | Agent Note | Agent Note（agent 决策记录） | 智能体注记、智能体笔记 | 本仓库中由 agent 撰写的提案与决策记录 |\n| agent harness | agent harness | agent harness（智能体框架） | | agent 组合词（agent harness/workflow/loop/skill 等）整体保留英文；未括注过 agent 时首现按对应组合词或 agent 行处理 |\n| agent loop | agent loop | agent loop（智能体循环） | | |\n| blob hash | blob hash | | | `git hash-object` 的结果 |\n| Cordis | Cordis | | | |\n| dispose | dispose | dispose（资源释放） | | |\n| doc-sync | doc-sync | doc-sync（文档同步门禁） | | |\n| fiber | fiber | | | |\n| fixture | fixture | fixture（测试前置数据） | | |\n| fork | fork | | | |\n| Function Calling | Function Calling | Function Calling（函数调用） | | |\n| harness | harness | | | |\n| harness engineering | harness engineering | | | |\n| lint | lint | | | |\n| mock | mock | | | 保留英文；指测试替身 |\n| loader | loader | | | |\n| manifest | manifest | manifest（元数据清单） | | |\n| monorepo | monorepo | | | |\n| Round | Round | | 回合、目标回合、Ralph 回合 | 外层策略使用 Round 时，领域层级为 Session > Round > Turn（轮次） > Step（步骤）；Round 是可选的外层策略迭代，并非每个会话轮次都具有的通用层级。Goal Round 与 Ralph Round 均保留英文。一个 Round 承载一个轮次，步骤隶属于该轮次；明确的零步骤轮次仍保持原义。 |\n| schema | schema | | | |\n| schema DSL | schema DSL | | | |\n| seam | seam | | | 与 `extension point` 是不同概念；根据具体语境，可译为`服务边界`或`可替换点` |\n| skill | skill | skill（技能） | | |\n| spawn | spawn | | | |\n| steering | steering | steering（中途引导） | | |\n| task id | task id | | 任务 id | 保留英文 |\n| subagent | subagent | | | |\n| thinking | thinking | | | API 字段保留英文；描述模型模式时译为`思考` |\n| transcript | transcript | transcript（文本记录） | | 指会话渲染给用户或编辑器的完整文本，区别于事件日志 |\n| waterfall | waterfall | waterfall（瀑布式事件） | | |\n| wheel | wheel 包 | | | Python 打包格式 |\n| worktree | worktree | | | git 工作区概念 |\n| Zstandard | Zstandard | | | RFC 8878 compression format; `zstd` remains a code value. |\n\n## 双语类（中英文文本各自使用中英文）\n\n| English | 中文 | 首次出现 | 不要译作 | 备注 |\n|---|---|---|---|---|\n| adapter | 适配器 | | | |\n| adapter contract | 适配器契约 | 适配器契约（adapter contract） | | |\n| append-only | 仅追加 | | | |\n| artifact | 产物 | | | |\n| backend | 后端 | | | |\n| background task | 后台任务 | | | |\n| block | 块 | | | |\n| build target | 构建目标 | | | |\n| cancel | 取消 | | | |\n| feature | 功能 | | 能力 | SDK 产品与工程模型中的可管理产品单元 |\n| feature option | 功能选项 | | variant | 一项 SDK 功能内有限、可选择的实现或配置 |\n| checkpoint | 检查点 | | | |\n| chunk | 分片 | | | |\n| compaction | 压缩 | 压缩（compaction） | | |\n| companion tool | 配套工具 | | | |\n| Cordis plugin config | Cordis 插件配置 | | | Cordis 插件公开的 `Config` 对象或配置结构 |\n| config key | 配置键 | | | Cordis 插件配置中的单个字段 |\n| consumer | 消费方 | | | |\n| content block | 内容块 | | | |\n| Cookbook | 实操手册 | | | 文档标题用语 |\n| context | 上下文 | | | |\n| counterpart | 对侧文件 | | 对应物、配对物 | 双语配对语境；泛指\"另一侧\"时可写「另一侧」 |\n| context compaction | 上下文压缩 | 上下文压缩（context compaction） | | |\n| contract | 契约 | | | 如：`pairing contract` →`配对契约` |\n| Cordis config entry | Cordis 配置项 | | | 指 `cordis.yml` 插件列表中的一项；插件实现本身写`Cordis 插件` |\n| Cordis plugin | Cordis 插件 | | | Cordis 加载的插件实现，不指 `cordis.yml` 中的一项配置 |\n| coverage | 覆盖率 | | | |\n| crash recovery | 崩溃恢复 | | | |\n| deploy root | 部署根目录 | | | |\n| durability | 持久性 | | | |\n| feature requirement | 功能依赖 | | | 功能或功能选项通过 `requires` 声明的关系 |\n| ergonomics | 易用性 / 开发体验 | | 人体工学 | API 或面向模型的接口用「易用性」；工具链或开发者工作流用「开发体验」 |\n| event | 事件 | | | |\n| event log | 事件日志 | | | |\n| event stream | 事件流 | | | |\n| event-sourced | 事件溯源 | | | 沿用 DDD 社区通行译法 |\n| Executive summary | 摘要 | | | 事故复盘标题用语 |\n| executor | 执行器 | | | |\n| expected output | 预期输出 | | 金标 | 指 snapshot 比较产物；翻译语料的人工校准样例不在此列 |\n| extension | 扩展 | | | |\n| extension point | 扩展点 | | | 注意与 `seam` 区分 |\n| fail-fast | 快速失败 | | | |\n| fenced code block | 围栏代码块 | | | 沿用 MDN 中文翻译 |\n| fingerprint | 指纹 | | | 通用内容指纹；双语配对机制使用 sidecar record 记录两侧 blob hash |\n| finish reason | 结束原因 | | | |\n| foreground run | 前台运行 | | | |\n| freshness | 新鲜度 | | | 沿用 MDN 中文翻译；在本项目中指译文相对源文的同步状态 |\n| hook | 钩子 | | | |\n| implementation | 实现 | | | |\n| inference | 推理 | 推理（inference） | | 需要和 `reasoning` 区分时保留英文括注 |\n| info string | 信息字符串 | | | 沿用 CommonMark 中文翻译；指代码围栏 ``` 之后的语言标注 |\n| injection | 注入 | | | |\n| integration | 集成 | | | |\n| interface | 接口 | | | |\n| language switcher | 语言切换行 | | | i18n 配对机制用语：双语配对文件顶部的互链行 |\n| memory | 记忆 / 内存 | | | 与 `agent` 搭配时译为`记忆`（如 `agent memory` →`智能体记忆`）；指系统资源时译为`内存` |\n| merge | 合并 | | | |\n| message | 消息 | | | |\n| mod | 模组 | | | |\n| model provider | 模型提供方 | | | |\n| module | 模块 | | | |\n| non-escalation | 非升权 | | 非升级、不可升级 | 仅用于安全与权限语境，指主体不得获得超出既有授权的权限；普通升级不适用此行 |\n| npm dependency | NPM 依赖 | | | `package.json` 中的包关系；`dependencies`、`devDependencies` 等字段保持原样 |\n| opt-out ratio | opt-out 比例 | | 退出检查比例 | |\n| orphan | 遗留 | | 孤儿、孤立 | 指英文源已不存在的 `.zh.md`（如「遗留译文」）；进程语境按 OS 惯用语译「孤儿进程」 |\n| orphan branch | 孤立分支 | | 孤儿分支 | 沿用 git 官方中文翻译 |\n| package | 包 | 包（package） | | 指 npm 包（`@deepseek-ai/dsh-*`）；`package.json` 等代码标识保持原样 |\n| pairing | 配对 | | | |\n| parent-subset grants | 父级子集授权 | | 父集合授权 | 指授权范围仅限于父级所持授权的子集 |\n| peer dependency | 对等依赖 | 对等依赖（peer dependency） | | |\n| permission | 权限 | | | |\n| persistence | 持久化 | | | |\n| pipeline | 流水线 | | | |\n| plugin | 插件 | | | |\n| prompt | 提示词 | | | |\n| provider | 提供方 | | | |\n| provider-neutral | 提供方无关 | | | |\n| quality gate | 质量门禁 | | | |\n| quiescence | 完全停稳 | | 静默、静止状态 | 指生命周期工作全部结算后的状态 |\n| reasoning | 推理 | 推理（reasoning） | | 需要和 `inference` 区分时保留英文括注 |\n| reasoning_content | 思考内容 | | | |\n| registry | 注册表 | | | |\n| replay | 回放 | | | |\n| resume | 恢复 | | | |\n| runtime | 运行时 | | | |\n| same-world subprocess | 与宿主共享文件系统和内核的子进程 | | 同世界子进程 | |\n| sandbox | 沙箱 | | | |\n| service | 服务 | | | |\n| serving surface | 对外服务接口 | | | |\n| session | 会话 | | | |\n| session event | 会话事件 | | | |\n| sidecar record | 伴随记录 | | 旁挂记录 | 指与文档同目录的伴随记录文件 |\n| smoke test | 冒烟测试 | | | |\n| snapshot | 快照 | | | |\n| source of truth | 真源 | | | |\n| spine | 主干 | | | |\n| staged | 暂存 | | | 沿用 git 官方中文翻译 |\n| stale | 陈旧 | | 过期 | 与 `fresh`（`新鲜`）成对；门禁输出中保留英文 `stale` 不翻译；`expired` 才译为`过期` |\n| step | 步骤 | | | |\n| stream | 流 | | | |\n| streaming | 流式输出 | | | |\n| structural signature | 结构签名 | | | i18n 配对机制用语：门禁比对两侧文件时提取的有序结构序列（标题层级、代码块、列表等） |\n| Summary | 概述 | | | 事故复盘标题用语 |\n| system prompt | 系统提示词 | | | |\n| taxonomy | 分类体系 | | | |\n| token usage | token 用量 | | | |\n| tool | 工具 | | | |\n| tool call | 工具调用 | | | |\n| tool result | 工具结果 | | | |\n| tool schema | 工具 schema | | | |\n| toolkit | 工具包 | | | |\n| turn | 轮次 | | | |\n| VFS | VFS | 虚拟文件系统（VFS） | | |\n| typecheck | 类型检查 | | | |\n| vocabulary | 词汇 | | | |\n| wire format | 协议格式 | 协议格式（wire format） | | |\n| workflow | 工作流 | | | |\n| wrapper | 包装层 | | | 软件层或 SDK 包装层 |\n| wrapper script | 包装脚本 | | | 可执行脚本包装层 |\n\n\n## Output Format\n\nProduce your output in three XML sections:\n\nThe outer section tags are framing. If Markdown inside any section body contains a line consisting only of `<translation>`, `</translation>`, `<review>`, `</review>`, `<final>`, or `</final>`, prefix that line with `\\`. If the original line already has one or more backslashes immediately before the tag, add one more. The parser removes exactly one framing escape; tags mentioned inline need no escaping.\n\n```xml\n<translation>\n(Complete translation of the source document)\n</translation>\n\n<review>\n(Self-review notes, one correction per line with category tag, e.g.)\n- [Tone] \"旁挂记录\" → \"伴随记录\"（生造词）\n- [Sentence] 第 3 段补充逗号断句\n- [Punctuation] 两处破折号替换为冒号\n- 无修正\n</review>\n\n<final>\n(Final translation after corrections)\n</final>\n```\n\n## Self-Review Instructions\n\nAfter writing `<translation>`, re-read it in the target language only, without looking at the source. Check by category:\n\n**Structure**\n- Is the heading hierarchy, list shape, and code block content identical to the source?\n- Are ALL comments inside code blocks left untranslated (byte-identical to source)?\n- Is the language switcher line correctly flipped (not copied from source)?\n- Are link targets preserved, and are spaces after bold markers present only before Latin letters, digits, or CJK ideographs?\n- Are wrapper-tag lines inside section bodies escaped with one additional backslash?\n\n**Tone & Style**\n- Does every sentence read as if originally written by a native speaker?\n- Is there any colloquial, casual, or overly informal phrasing?\n\n**Sentence Structure**\n- Are there run-on sentences that need breaking?\n- Are there stiff passive constructions that should be converted to active voice?\n\n**Word Choice**\n- Are there overly literal translations that sound unnatural?\n- Is the same target-language word used to translate two distinct source concepts?\n- Is any slang or internal jargon present?\n\n**Terminology**\n- For a Chinese target, are first-occurrence glosses correctly applied (not missing, not repeated)? For an English target, are Chinese glosses absent?\n- Are any \"不要译作\" forbidden translations present?\n- For unlisted terms, does a Chinese target use established Chinese precedent or retain the source term as pending, and does an English target use established English terminology or preserve only an ambiguous source term with a short English gloss?\n\n**Punctuation** (when target is Chinese)\n- Are there em-dashes that should be replaced with colons, periods, or commas?\n- Are list items ending with commas instead of semicolons?\n- Do RFC 2119 keywords preserve the source emphasis exactly?\n\nRecord corrections in `<review>` with category tags. Then output the corrected version in `<final>`. If no corrections are needed, write \"无修正\" in `<review>` and copy the translation unchanged into `<final>`.\n\n## Examples\n\nBelow are representative examples of common problems and their corrections. Follow the \"Good\" versions.\n\n### Colloquial verb → Professional verb\n- Source: `The repo pins pnpm@11.7.0 in package.json`\n- Bad: `仓库在 package.json 中钉住 pnpm@11.7.0`\n- Good: `该仓库在 package.json 中固定使用 pnpm@11.7.0`\n\n### Run-on sentence → Natural phrasing with pause\n- Source: `Read docs/architecture.md before changing anything under packages/.`\n- Bad: `改动 packages/ 下的任何东西之前先读 docs/architecture.md。`\n- Good: `在修改 packages/ 目录下的任何内容之前，请先阅读 docs/architecture.md。`\n\n### Stiff passive voice → Active and natural\n- Source: `a green gate means the pair was confirmed consistent at these exact contents, not that the confirmation was sound.`\n- Bad: `门禁绿意味着这对文档曾在当前内容上被确认一致，不意味着这次确认本身是对的。`\n- Good: `门禁通过意味着这组文档在当前内容上的一致性得到了确认，不代表确认本身正确可靠。`\n\n### Invented word → Natural expression\n- Source: `A sidecar record of both blob hashes makes consistency checkable`\n- Bad: `旁挂记录两侧 blob hash，使一致性可检查`\n- Good: `伴随记录保存两侧 blob hash，使一致性可检查`\n\n### Em-dash → Colon/period\n- Source: `FIXME — an issue that should block a new release. A release should not ship with an open FIXME unless reviewers explicitly agree the change can be merged anyway.`\n- Bad: `FIXME——应当阻塞新版本发布的问题。除非评审者明确同意可以照常合入，发布不应带着未解决的 FIXME 出门。`\n- Good: `FIXME：应当阻塞新版本发布的问题。除非评审者明确同意该更改可以合并，否则发布版本不应包含未解决的 FIXME。`\n\n### Overly literal → Meaningful rendering\n- Source: `awkward phrasing is easier to hear without the source anchoring you`\n- Bad: `没有源文锚着，别扭的表述更容易被听出来`\n- Good: `不对照原文时，更容易察觉别扭的表达`\n\n### Terminology — do not translate what should be kept in English\n- Source: `typed service seams, and explicit extension points`\n- Bad: `类型化的服务 seam（扩展点）与显式扩展点`\n- Good: `类型化的服务 seam 与显式扩展点`\n\n### Slang/jargon → Professional phrasing\n- Source: `The committed agent workflow lives in .agents/skills/dsh-translate-docs`\n- Bad: `进仓的 agent 工作流见 .agents/skills/dsh-translate-docs`\n- Good: `仓库内置的 agent 工作流见 .agents/skills/dsh-translate-docs`\n\n### \"For humans\" — translate the intent, not the word\n- Source: `For humans, start with the development guide`\n- Bad: `对于人工读者，请先从开发指南开始`（\"人工读者\"生硬）\n- Good: `面向开发者：请先阅读开发指南`（\"开发者\"自然，且中文里冒号在此处更自然）\n\n### Code block comments — NEVER translate\n- Source code block contains: `# full-screen TUI coding agent (needs DEEPSEEK_API_KEY)`\n- Bad: `# 全屏 TUI coding agent（需要 DEEPSEEK_API_KEY）`\n- Good: `# full-screen TUI coding agent (needs DEEPSEEK_API_KEY)` (keep exactly as-is, byte-for-byte)\n\n### Language switcher — flip direction\n- Source file (English) has: `English | [中文](README.zh.md)`\n- Bad (copying source unchanged): `English | [中文](README.zh.md)`\n- Good (flipped for Chinese file): `[English](README.md) | 中文`\n\n---\n\nNow translate the following document:"
+        "content": "# Translation Prompt\n\nYou are a senior technical translator specializing in LLM and agent development documentation. Your task is to translate the given source document from English to Chinese, producing natural, professional technical prose.\n\n## Quality Requirements\n\n### Structure and Format Preservation\n- Output a complete translated document that maintains exactly the same structure as the source: heading hierarchy, list shape, table columns, link targets, and code blocks.\n- Fenced code blocks must be byte-identical to the source, including ALL comments inside them. Do NOT translate comments inside code blocks. This is a hard rule with no exceptions.\n- Inline code spans (commands, flags, paths, API names, version numbers) must be kept verbatim. Never translate or reformat them.\n- Every relative link must point to the same target as in the source. Link text is translated; link targets are not.\n- Language switcher line: when translating into Chinese, write `[English](source-filename.md) | 中文`. When translating into English, write `English | [中文](source-filename.zh.md)`. Do NOT copy the switcher line from the source file unchanged — you must flip the link direction.\n- After a closing bold marker `**`, insert a space before the next character when that character is a Latin letter, digit, or CJK ideograph. Never insert a space before any punctuation (full-width or half-width).\n\n### Tone and Style\n- The translation must read as if originally written in the target language by a native speaker. If an expression sounds like a word-for-word rendering from the source language, rephrase it.\n- Write in a professional, formal tone appropriate for developer documentation. Never use colloquial or casual expressions.\n- Use polite imperative forms where the text instructs the reader to do something.\n- Keep the author's register: concise stays concise, detailed stays detailed.\n\n### Sentence Structure\n- Break long sentences with commas or semicolons. Avoid run-on sentences.\n- Prefer active voice. Convert passive constructions to active if it reads more naturally.\n- Translate meaning, not words. Restructure sentences where the target language grammar requires it.\n- Do not invent words or expressions that do not exist in natural technical writing of the target language.\n\n### Word Choice\n- Prefer precise, formal vocabulary over casual or colloquial alternatives.\n- When multiple synonyms exist, choose the one most commonly used in professional technical documentation of the target language.\n- Avoid slang, internal jargon, or overly literal translations that would not be recognized by the general developer audience.\n- Do not use the same word to translate two different source-language terms that carry distinct meanings.\n- Avoid repeating the same verb in close proximity; vary word choice for readability.\n\n#### When translating into Chinese\n- When a number modifies a noun, always include a Chinese classifier or measure word (量词). For example: \"three-package seam\" → \"由三个包构成的 seam\", not \"三包 seam\".\n\n### Punctuation\n\n#### When translating into Chinese\n- Use full-width Chinese punctuation in prose: `，。：；？！（）「」`.\n- Strongly prefer replacing all em-dashes (——) with colons, periods, commas, or parentheses. Keep an em-dash only if no other punctuation works at all.\n- Use enumeration commas (、) between parallel items, not regular commas.\n- List item endings: use semicolons or no punctuation. Do not end list items with commas.\n- Put one half-width space between Chinese text and Latin words/numbers.\n- For RFC 2119 keywords (MUST, MUST NOT, SHOULD, MAY), translate to the corresponding Chinese term (必须、禁止、应当、可以) and keep the SOURCE emphasis marker: plain source stays plain (必须), italic source stays italic (*必须*), and bold source stays bold (**必须**).\n\n#### When translating into English\n(To be added.)\n\n## Terminology\n\nA terminology table is provided below. Follow it strictly:\n- Render every listed term exactly as specified.\n- When the target language is Chinese, use the \"中文\" column. On first occurrence, write the \"首次出现\" value with its parenthetical gloss; on subsequent occurrences, write only the part before the parentheses.\n- When the target language is English, use the \"English\" column without a Chinese gloss; do not copy the \"中文\" or \"首次出现\" value into English prose.\n- If a term has already been glossed as part of a compound term, do not gloss it again when it appears alone later.\n- NEVER use translations listed in the \"不要译作\" column.\n- For technical terms not in the table, follow the target language: for a Chinese target, use an established Chinese rendering from a major Chinese-language OSS or vendor source, or keep the source term and flag it as pending when no such precedent exists; for an English target, use the established English technical term, or preserve an ambiguous source term with a short English gloss and flag it as pending. Do not invent a translation. This rule applies to terminology only; for general prose, freely restructure and paraphrase for natural expression.\n\n# Terminology\n\n本表约定本仓库的中英术语统一译法。\n\n**通用规则：**\n- \"中文\"列为中文译文的正文默认用词。若该列为英文，则中文译文的正文中保留英文不翻译。\n- 首次出现按\"首次出现\"列书写（带括号注释）；后续出现只写括号前的部分（可能为中文，也可能为英文），不出现括号内的注释。\n- \"不要译作\"列为严格禁止的译法。\n- 如果某术语已经作为另一个术语的组成部分被括注过（如 `agent loop（智能体循环）` 中已包含 `agent` 的括注），则该术语后续单独出现时无需再次括注。\n\n## 缩写类（中英文文本中均使用缩写）\n\n| English | 中文 | 首次出现 | 不要译作 | 备注 |\n|---|---|---|---|---|\n| ACP | ACP | ACP（Agent Client Protocol） | | |\n| AI | AI | AI（人工智能） | | |\n| API | API | | | |\n| CI | CI | | | |\n| CLI | CLI | CLI（命令行界面） | | |\n| e2e | e2e | | | |\n| HMR | HMR | HMR（热模块替换） | | |\n| JSON Schema | JSON Schema | | | |\n| JSONL | JSONL | | | |\n| LLM | LLM | LLM（大语言模型） | | |\n| MCP | MCP | | | |\n| PR | PR | PR（Pull Request） | | |\n| RAG | RAG | RAG（检索增强生成） | | |\n| SDK | SDK | | | |\n| SSE | SSE | SSE（Server-Sent Events） | | |\n\n## 英文类（中英文文本中均使用英文）\n\n| English | 中文 | 首次出现 | 不要译作 | 备注 |\n|---|---|---|---|---|\n| agent | agent | agent（智能体） | | |\n| Agent Note | Agent Note | Agent Note（agent 决策记录） | 智能体注记、智能体笔记 | 本仓库中由 agent 撰写的提案与决策记录 |\n| agent harness | agent harness | agent harness（智能体框架） | | agent 组合词（agent harness/workflow/loop/skill 等）整体保留英文；未括注过 agent 时首现按对应组合词或 agent 行处理 |\n| agent loop | agent loop | agent loop（智能体循环） | | |\n| blob hash | blob hash | | | `git hash-object` 的结果 |\n| Cordis | Cordis | | | |\n| dispose | dispose | dispose（资源释放） | | |\n| doc-sync | doc-sync | doc-sync（文档同步门禁） | | |\n| fiber | fiber | | | |\n| fixture | fixture | fixture（测试前置数据） | | |\n| fork | fork | | | |\n| Function Calling | Function Calling | Function Calling（函数调用） | | |\n| harness | harness | | | |\n| harness engineering | harness engineering | | | |\n| lint | lint | | | |\n| mock | mock | | | 保留英文；指测试替身 |\n| loader | loader | | | |\n| manifest | manifest | manifest（元数据清单） | | |\n| monorepo | monorepo | | | |\n| Round | Round | | 回合、目标回合、Ralph 回合 | 外层策略使用 Round 时，领域层级为 Session > Round > Turn（轮次） > Step（步骤）；Round 是可选的外层策略迭代，并非每个会话轮次都具有的通用层级。Goal Round 与 Ralph Round 均保留英文。一个 Round 承载一个轮次，步骤隶属于该轮次；明确的零步骤轮次仍保持原义。 |\n| schema | schema | | | |\n| schema DSL | schema DSL | | | |\n| seam | seam | | 接缝 | 与 `extension point` 是不同概念；根据具体语境，可译为`服务边界`或`可替换点` |\n| skill | skill | skill（技能） | | |\n| spawn | spawn | | | |\n| steering | steering | steering（中途引导） | | |\n| task id | task id | | 任务 id | 保留英文 |\n| subagent | subagent | | | |\n| thinking | thinking | | | API 字段保留英文；描述模型模式时译为`思考` |\n| transcript | transcript | transcript（文本记录） | | 指会话渲染给用户或编辑器的完整文本，区别于事件日志 |\n| waterfall | waterfall | waterfall（瀑布式事件） | | |\n| wheel | wheel 包 | | | Python 打包格式 |\n| worktree | worktree | | | git 工作区概念 |\n| Zstandard | Zstandard | | | RFC 8878 compression format; `zstd` remains a code value. |\n\n## 双语类（中英文文本各自使用中英文）\n\n| English | 中文 | 首次出现 | 不要译作 | 备注 |\n|---|---|---|---|---|\n| adapter | 适配器 | | | |\n| adapter contract | 适配器契约 | 适配器契约（adapter contract） | | |\n| append-only | 仅追加 | | | |\n| artifact | 产物 | | 制品 | |\n| backend | 后端 | | | |\n| background task | 后台任务 | | | |\n| block | 块 | | | |\n| build target | 构建目标 | | | |\n| cancel | 取消 | | | |\n| canary test | canary 测试 | | 金丝雀测试 | 本仓库保留 `canary` |\n| capability seam | 能力 seam | | 功能 seam、能力接缝 | 本仓库接口、实现与消费方分离的命名架构概念；普通 `seam` 仍按其词条处理 |\n| feature | 功能 | | 能力 | SDK 产品与工程模型中的可管理产品单元 |\n| feature option | 功能选项 | | variant | 一项 SDK 功能内有限、可选择的实现或配置 |\n| checkpoint | 检查点 | | | |\n| chunk | 分片 | | | |\n| compaction | 压缩 | 压缩（compaction） | | |\n| companion tool | 配套工具 | | | |\n| composition bundle | 组合包 | | | 只约束应用或插件的组合语境，不约束所有 `bundle` |\n| Cordis plugin config | Cordis 插件配置 | | | Cordis 插件公开的 `Config` 对象或配置结构 |\n| config key | 配置键 | | | Cordis 插件配置中的单个字段 |\n| consumer | 消费方 | | 消费者 | |\n| content block | 内容块 | | | |\n| Cookbook | 实操手册 | | | 文档标题用语 |\n| context | 上下文 | | | |\n| counterpart | 对侧文件 | | 对应物、配对物 | 双语配对语境；泛指\"另一侧\"时可写「另一侧」 |\n| context compaction | 上下文压缩 | 上下文压缩（context compaction） | | |\n| contract | 契约 | | | 如：`pairing contract` →`配对契约` |\n| Cordis config entry | Cordis 配置项 | | | 指 `cordis.yml` 插件列表中的一项；插件实现本身写`Cordis 插件` |\n| Cordis plugin | Cordis 插件 | | | Cordis 加载的插件实现，不指 `cordis.yml` 中的一项配置 |\n| coverage | 覆盖率 | | | |\n| crash recovery | 崩溃恢复 | | | |\n| deploy root | 部署根目录 | | | |\n| durability | 持久性 | | | |\n| feature requirement | 功能依赖 | | | 功能或功能选项通过 `requires` 声明的关系 |\n| ergonomics | 易用性 / 开发体验 | | 人体工学 | API 或面向模型的接口用「易用性」；工具链或开发者工作流用「开发体验」 |\n| event | 事件 | | | |\n| event log | 事件日志 | | | |\n| event stream | 事件流 | | | |\n| event-sourced | 事件溯源 | | | 沿用 DDD 社区通行译法 |\n| Executive summary | 摘要 | | | 事故复盘标题用语 |\n| executor | 执行器 | | | |\n| expected output | 预期输出 | | 金标 | 指 snapshot 比较产物；翻译语料的人工校准样例不在此列 |\n| extension | 扩展 | | | |\n| extension point | 扩展点 | | | 注意与 `seam` 区分 |\n| fail-fast | 快速失败 | | | |\n| fenced code block | 围栏代码块 | | | 沿用 MDN 中文翻译 |\n| fingerprint | 指纹 | | | 通用内容指纹；双语配对机制使用 sidecar record 记录两侧 blob hash |\n| finish reason | 结束原因 | | | |\n| foreground run | 前台运行 | | | |\n| freshness | 新鲜度 | | | 沿用 MDN 中文翻译；在本项目中指译文相对源文的同步状态 |\n| hook | 钩子 | | | |\n| implementation | 实现 | | | |\n| inference | 推理 | 推理（inference） | | 需要和 `reasoning` 区分时保留英文括注 |\n| info string | 信息字符串 | | | 沿用 CommonMark 中文翻译；指代码围栏 ``` 之后的语言标注 |\n| injection | 注入 | | | |\n| integration | 集成 | | | |\n| interface | 接口 | | | |\n| language switcher | 语言切换行 | | | i18n 配对机制用语：双语配对文件顶部的互链行 |\n| memory | 记忆 / 内存 | | | 与 `agent` 搭配时译为`记忆`（如 `agent memory` →`智能体记忆`）；指系统资源时译为`内存` |\n| merge | 合并 | | | |\n| message | 消息 | | | |\n| mod | 模组 | | | |\n| model provider | 模型提供方 | | | |\n| module | 模块 | | | |\n| non-escalation | 非升权 | | 非升级、不可升级 | 仅用于安全与权限语境，指主体不得获得超出既有授权的权限；普通升级不适用此行 |\n| npm dependency | NPM 依赖 | | | `package.json` 中的包关系；`dependencies`、`devDependencies` 等字段保持原样 |\n| opt-out ratio | opt-out 比例 | | 退出检查比例 | |\n| orphan | 遗留 | | 孤儿、孤立 | 指英文源已不存在的 `.zh.md`（如「遗留译文」）；进程语境按 OS 惯用语译「孤儿进程」 |\n| orphan branch | 孤立分支 | | 孤儿分支 | 沿用 git 官方中文翻译 |\n| package | 包 | 包（package） | | 指 npm 包（`@deepseek-ai/dsh-*`）；`package.json` 等代码标识保持原样 |\n| pairing | 配对 | | | |\n| parent-subset grants | 父级子集授权 | | 父集合授权 | 指授权范围仅限于父级所持授权的子集 |\n| peer dependency | 对等依赖 | 对等依赖（peer dependency） | | |\n| permission | 权限 | | | |\n| persistence | 持久化 | | | |\n| pipeline | 流水线 | | | |\n| plugin | 插件 | | | |\n| postmortem | 事故复盘 | 事故复盘（postmortem） | 事后分析、事故记录 | 事故记录与分析文档；目录或路径中的 `postmortem` 保持代码形式 |\n| prompt | 提示词 | | | |\n| provider | 提供方 | | | |\n| provider-neutral | 提供方无关 | | 提供方中立 | |\n| quality gate | 质量门禁 | | | |\n| quiescence | 完全停稳 | | 静默、静止状态 | 指生命周期工作全部结算后的状态 |\n| reasoning | 推理 | 推理（reasoning） | | 需要和 `inference` 区分时保留英文括注 |\n| reasoning_content | 思考内容 | | | |\n| registry | 注册表 | | | |\n| replay | 回放 | | | |\n| resume | 恢复 | | | |\n| runtime | 运行时 | | | |\n| same-world subprocess | 与宿主共享文件系统和内核的子进程 | | 同世界子进程 | |\n| sandbox | 沙箱 | | | |\n| service | 服务 | | | |\n| serving surface | 对外服务接口 | | | |\n| session | 会话 | | | |\n| session event | 会话事件 | | | |\n| sidecar record | 伴随记录 | | 旁挂记录 | 指与文档同目录的伴随记录文件 |\n| smoke test | 冒烟测试 | | | |\n| snapshot | 快照 | | | |\n| source of truth | 真源 | | 事实来源、唯一来源 | |\n| spine | 主干 | | | |\n| staged | 暂存 | | | 沿用 git 官方中文翻译 |\n| stale | 陈旧 | | 过期 | 与 `fresh`（`新鲜`）成对；门禁输出中保留英文 `stale` 不翻译；`expired` 才译为`过期` |\n| step | 步骤 | | | |\n| stream | 流 | | | |\n| streaming | 流式输出 | | | |\n| structural signature | 结构签名 | | | i18n 配对机制用语：门禁比对两侧文件时提取的有序结构序列（标题层级、代码块、列表等） |\n| Summary | 概述 | | | 事故复盘标题用语 |\n| system prompt | 系统提示词 | | | |\n| taxonomy | 分类体系 | | | |\n| token usage | token 用量 | | | |\n| tool | 工具 | | | |\n| tool call | 工具调用 | | | |\n| tool result | 工具结果 | | | |\n| tool schema | 工具 schema | | | |\n| toolkit | 工具包 | | | |\n| turn | 轮次 | | | |\n| VFS | VFS | 虚拟文件系统（VFS） | | |\n| typecheck | 类型检查 | | | |\n| vocabulary | 词汇 | | | |\n| wire format | 协议格式 | 协议格式（wire format） | | |\n| workflow | 工作流 | | | |\n| wrapper | 包装层 | | | 软件层或 SDK 包装层 |\n| wrapper script | 包装脚本 | | | 可执行脚本包装层 |\n\n\n## Output Format\n\nProduce your output in three XML sections:\n\nThe outer section tags are framing. If Markdown inside any section body contains a line consisting only of `<translation>`, `</translation>`, `<review>`, `</review>`, `<final>`, or `</final>`, prefix that line with `\\`. If the original line already has one or more backslashes immediately before the tag, add one more. The parser removes exactly one framing escape; tags mentioned inline need no escaping.\n\n```xml\n<translation>\n(Complete translation of the source document)\n</translation>\n\n<review>\n(Self-review notes, one correction per line with category tag, e.g.)\n- [Tone] \"旁挂记录\" → \"伴随记录\"（生造词）\n- [Sentence] 第 3 段补充逗号断句\n- [Punctuation] 两处破折号替换为冒号\n- 无修正\n</review>\n\n<final>\n(Final translation after corrections)\n</final>\n```\n\n## Self-Review Instructions\n\nAfter writing `<translation>`, re-read it in the target language only, without looking at the source. Check by category:\n\n**Structure**\n- Is the heading hierarchy, list shape, and code block content identical to the source?\n- Are ALL comments inside code blocks left untranslated (byte-identical to source)?\n- Is the language switcher line correctly flipped (not copied from source)?\n- Are link targets preserved, and are spaces after bold markers present only before Latin letters, digits, or CJK ideographs?\n- Are wrapper-tag lines inside section bodies escaped with one additional backslash?\n\n**Tone & Style**\n- Does every sentence read as if originally written by a native speaker?\n- Is there any colloquial, casual, or overly informal phrasing?\n\n**Sentence Structure**\n- Are there run-on sentences that need breaking?\n- Are there stiff passive constructions that should be converted to active voice?\n\n**Word Choice**\n- Are there overly literal translations that sound unnatural?\n- Is the same target-language word used to translate two distinct source concepts?\n- Is any slang or internal jargon present?\n\n**Terminology**\n- For a Chinese target, are first-occurrence glosses correctly applied (not missing, not repeated)? For an English target, are Chinese glosses absent?\n- Are any \"不要译作\" forbidden translations present?\n- For unlisted terms, does a Chinese target use established Chinese precedent or retain the source term as pending, and does an English target use established English terminology or preserve only an ambiguous source term with a short English gloss?\n\n**Punctuation** (when target is Chinese)\n- Are there em-dashes that should be replaced with colons, periods, or commas?\n- Are list items ending with commas instead of semicolons?\n- Do RFC 2119 keywords preserve the source emphasis exactly?\n\nRecord corrections in `<review>` with category tags. Then output the corrected version in `<final>`. If no corrections are needed, write \"无修正\" in `<review>` and copy the translation unchanged into `<final>`.\n\n## Examples\n\nBelow are representative examples of common problems and their corrections. Follow the \"Good\" versions.\n\n### Colloquial verb → Professional verb\n- Source: `The repo pins pnpm@11.7.0 in package.json`\n- Bad: `仓库在 package.json 中钉住 pnpm@11.7.0`\n- Good: `该仓库在 package.json 中固定使用 pnpm@11.7.0`\n\n### Run-on sentence → Natural phrasing with pause\n- Source: `Read docs/architecture.md before changing anything under packages/.`\n- Bad: `改动 packages/ 下的任何东西之前先读 docs/architecture.md。`\n- Good: `在修改 packages/ 目录下的任何内容之前，请先阅读 docs/architecture.md。`\n\n### Stiff passive voice → Active and natural\n- Source: `a green gate means the pair was confirmed consistent at these exact contents, not that the confirmation was sound.`\n- Bad: `门禁绿意味着这对文档曾在当前内容上被确认一致，不意味着这次确认本身是对的。`\n- Good: `门禁通过意味着这组文档在当前内容上的一致性得到了确认，不代表确认本身正确可靠。`\n\n### Invented word → Natural expression\n- Source: `A sidecar record of both blob hashes makes consistency checkable`\n- Bad: `旁挂记录两侧 blob hash，使一致性可检查`\n- Good: `伴随记录保存两侧 blob hash，使一致性可检查`\n\n### Em-dash → Colon/period\n- Source: `FIXME — an issue that should block a new release. A release should not ship with an open FIXME unless reviewers explicitly agree the change can be merged anyway.`\n- Bad: `FIXME——应当阻塞新版本发布的问题。除非评审者明确同意可以照常合入，发布不应带着未解决的 FIXME 出门。`\n- Good: `FIXME：应当阻塞新版本发布的问题。除非评审者明确同意该更改可以合并，否则发布版本不应包含未解决的 FIXME。`\n\n### Overly literal → Meaningful rendering\n- Source: `awkward phrasing is easier to hear without the source anchoring you`\n- Bad: `没有源文锚着，别扭的表述更容易被听出来`\n- Good: `不对照原文时，更容易察觉别扭的表达`\n\n### Terminology — do not translate what should be kept in English\n- Source: `typed service seams, and explicit extension points`\n- Bad: `类型化的服务 seam（扩展点）与显式扩展点`\n- Good: `类型化的服务 seam 与显式扩展点`\n\n### Slang/jargon → Professional phrasing\n- Source: `The committed agent workflow lives in .agents/skills/dsh-translate-docs`\n- Bad: `进仓的 agent 工作流见 .agents/skills/dsh-translate-docs`\n- Good: `仓库内置的 agent 工作流见 .agents/skills/dsh-translate-docs`\n\n### \"For humans\" — translate the intent, not the word\n- Source: `For humans, start with the development guide`\n- Bad: `对于人工读者，请先从开发指南开始`（\"人工读者\"生硬）\n- Good: `面向开发者：请先阅读开发指南`（\"开发者\"自然，且中文里冒号在此处更自然）\n\n### Code block comments — NEVER translate\n- Source code block contains: `# full-screen TUI coding agent (needs DEEPSEEK_API_KEY)`\n- Bad: `# 全屏 TUI coding agent（需要 DEEPSEEK_API_KEY）`\n- Good: `# full-screen TUI coding agent (needs DEEPSEEK_API_KEY)` (keep exactly as-is, byte-for-byte)\n\n### Language switcher — flip direction\n- Source file (English) has: `English | [中文](README.zh.md)`\n- Bad (copying source unchanged): `English | [中文](README.zh.md)`\n- Good (flipped for Chinese file): `[English](README.md) | 中文`\n\n---\n\nNow translate the following document:"
       },
       {
         "role": "user",
@@ -16,19 +16,19 @@
       },
       {
         "role": "user",
-        "content": "# Development guide\n\nEnglish | [中文](development.zh.md)\n\nThis onboarding guide helps project contributors get started with the local environment, daily workflow, and CI flow; see the Agent Notes for design rationale and technical trade-offs.\n\n## Prerequisites\n\n- Node.js supports 22.19+ and 24+. CI covers 22.19, 24, and 26; see the [Node engine floor Agent Note](../.agents/notes/implemented/process/2026-07-06-node-engine-floor.md).\n- Corepack-enabled pnpm. The repo pins `pnpm@11.7.0` in `package.json`; run `corepack enable` if `pnpm --version` does not resolve through Corepack.\n- Git 2.26 or newer; hook setup enables Git's worktree-specific configuration extension.\n- Optional: a DeepSeek API key for the TUI, headless, and ACP automation demos and real-API e2e tests.\n\n## First-time setup\n\nInstall dependencies from the repo root:\n\n```sh\npnpm install\n```\n\nThe install also runs the root `postinstall` script, which installs lefthook from the repo dev dependency through `scripts/install-lefthook.mjs`. With `CI=true` or `GITHUB_ACTIONS=true`, the wrapper returns before Git discovery because automated jobs do not consume contributor hooks. Otherwise, it requires Git 2.26 or newer and gives the current worktree an explicit hook directory under its own Git directory; linked worktrees therefore use their own lefthook binary and configuration instead of rewriting common hooks. The first install enables Git's worktree-specific configuration extension and repository format 1; see the [worktree-local hooks Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md).\n\nIf hooks are missing because dependencies were restored from cache or `postinstall` was skipped, install them manually:\n\n```sh\nnode scripts/install-lefthook.mjs\n```\n\nThe wrapper refuses user-owned `core.hooksPath` values. An inherited system, global, or common-repository path requires `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`; command-scoped and worktree-scoped custom paths must be integrated or removed explicitly.\n\nBefore enabling worktree config, migrate direct `extensions.*` in a format-0 common config, direct `core.worktree` or `core.bare=true`, and any non-empty dormant `config.worktree`. The common config and every worktree config must be regular files, while the owned hook directory may contain only unaliased regular files.\n\nAfter moving a checkout, rerun the wrapper to relocate its owned path and regenerate hooks. For a stale or invalid installer lock, first confirm no installer is running, then remove the reported lock and retry. If installation and hook-path rollback both fail, inspect the reported worktree config before retrying. The [worktree-local hooks Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md) owns the full safety contract.\n\nRun typecheck once after a fresh clone:\n\n```sh\npnpm run typecheck\n```\n\nThat first typecheck runs the whole-repo `tsc -b` graph: it emits every package/vendor `lib/types` and checks examples, tests, and scripts through the two no-emit aggregates described below.\n\n## TypeScript project layout\n\nThe repository's TypeScript configuration has exactly three roles; every tsconfig file plays one of them.\n\n| File | Role | Forms a program? |\n|---|---|---|\n| `tsconfig.json` | Solution root: `extends` base, `files: []`, references to the two aggregates. The whole-repo `tsc -b tsconfig.json` graph, the tsserver discovery entry, and — through the inherited `paths` — the resolution config for tsx running `examples/` and `scripts/` (their nearest tsconfig is this file). | No |\n| `tsconfig.host.json` | Host aggregate: host-side packages (via references), examples, tests, scripts, website. Excludes `packages/client`. | Yes |\n| `tsconfig.client.json` | Client aggregate: `packages/client/*` packages and their tests, `apps/web`. | Yes |\n| `tsconfig.base.json` | Shared compilerOptions and the source `paths` map. Also the resolution facade the vitest configs point vite-tsconfig-paths at: it has no `include`, so its `paths` apply to every importer. | No |\n| `tsconfig.base.client.json` | Browser compiler shape (`jsx`, DOM libs, `types: []`) extended by the client aggregate and every `packages/client/*` package. | No |\n\nHost and client stay two aggregate programs because both sides declaration-merge the cordis `Context` interface under the same keys with different services; one program seeing both merges reports a collision. The collision exists only inside a `ts.Program` — module resolution never triggers it — which is why the solution may reference both aggregates and one paths facade may span both sides. Two disciplines follow:\n\n- `tsconfig.base.json` never gains `include` or `files`: they would leak into every extending package project and narrow the facade's match-all scope.\n- A script that builds a repo-wide `ts.Program` seeds `tsconfig.host.json` or `tsconfig.client.json` explicitly — never the root solution, because flattening both aggregates into one program collides the `Context` merges. Program-backed generators and gates (`scripts/ts-project.ts` consumers, doc-typecheck standalone mode) are host-only by decision; the client side gains program-backed tooling only with a concrete need.\n\nStatic analysis and tests resolve workspace imports through the base `paths` map to `src` and must pass on a clean tree; gates that consume built `lib/` output declare that dependency explicitly. Decision record: [solution-root note](../.agents/notes/implemented/process/2026-07-22-tsconfig-solution-root-two-aggregates.md); the tsc-first emit pipeline is the [ts-build-config note](../.agents/notes/implemented/process/2026-06-17-ts-build-config.md).\n\nIf a relevant local check consumes built package output, build once first:\n\n```sh\npnpm run build\n```\n\n`pnpm run hygiene` includes `publint`, which validates package entrypoints against the built `lib/*.js` files, and `verify-node-next-types`, which validates built declarations against a temporary NodeNext consumer. A fresh worktree has no bundled JS or declarations until `pnpm run build` runs; ordinary commits and pushes do not require that build unless their selected checks consume it.\n\n## Environment variables\n\nThe real DeepSeek adapter and key-backed agent demos read credentials from the environment or from a gitignored `.env` at the repo root:\n\n```sh\nDEEPSEEK_API_KEY=sk-...\nDEEPSEEK_BASE_URL=https://... # optional\n```\n\n`DEEPSEEK_BASE_URL` is optional and defaults to the public API. Never commit real credentials. The real-API e2e suites self-skip when `DEEPSEEK_API_KEY` is not set.\n\n## Git hooks\n\nlefthook is configured in `lefthook.yml` as a fast local checkpoint:\n\n- `pre-commit` runs staged-file ESLint fixes, checks the staged diff for whitespace errors, and runs the vendor manifest guard.\n- `pre-push` runs only the incremental repository typecheck (`tsc -b` over the root solution, covering both the host and client aggregates).\n\nThe vendor manifest guard checks that changes under `vendor/*/src` are staged with the matching `vendor/README.md` manifest update. See `vendor/README.md` before editing vendored code.\n\nThe hooks intentionally do not run tests, snapshots, documentation checks, builds, or hygiene. Contributors run the [checks relevant to the changed behavior](../AGENTS.md#run-relevant-checks-locally) once; CI owns exhaustive coverage, built-artifact smokes, and the Node 22.19, 24, and 26 compatibility matrix.\n\nContributors can opt into the comprehensive local gate set with `pnpm run check:all`. The command is independent of both Git hooks and is not an agent instruction.\n\n## CI gates\n\nThe keyless [CI workflow](../.github/workflows/ci.yml) groups independent gates into broad lanes and runs a smaller compatibility signal across supported Node versions. Artifact consumers wait for one build within their lane. The separate real-API workflow runs `pnpm run test:e2e` with its configured worker bound. See [scripts/run-gates.ts](../scripts/run-gates.ts) and the workflow files for the current gate and job inventory.\n\n## Daily commands\n\nUse these from the repo root:\n\n```sh\npnpm run test           # unit tests\npnpm run test:coverage  # unit tests with per-file coverage gates\npnpm run test:e2e       # real-API tests; self-skips without DEEPSEEK_API_KEY\npnpm run check:all      # comprehensive opt-in gate set; not wired to Git hooks\npnpm run typecheck      # tsc -b over the root solution: emits package/vendor lib/types, checks both aggregates\npnpm run lint           # eslint .\npnpm run lint:fix       # eslint . --fix\npnpm run doc-typecheck  # compile checked TypeScript snippets in Markdown docs\npnpm run gen-cordis-catalog     # regenerate docs/cordis-catalog/events.md + services.md from source\npnpm run verify-cordis-catalog  # fail if either cordis catalog is stale\npnpm run verify-export-jsdoc    # fail if a module-level package export lacks complete JSDoc\npnpm run gen-doc-graphs     # regenerate generated relationship docs from source and curated graph definitions\npnpm run verify-doc-graphs  # fail if generated relationship docs are stale\npnpm run verify-md-wrap  # fail on hard-wrapped prose paragraphs in docs/README markdown\npnpm run verify-mermaid  # fail if a ```mermaid diagram has invalid Mermaid syntax\npnpm run verify-type-equiv  # fail if a ```ts type-equiv doc block drifts from its source type\npnpm run verify-doc-budgets  # fail if a budgeted standing doc exceeds its word ceiling\npnpm run gen-translation-brief   # print the minimal-update briefing for out-of-sync translation pairs (--apply splices code-only edits)\npnpm run doc-sync       # all Markdown/doc gates, scheduled concurrently; the doc-sync leaf list in scripts/run-gates.ts is the full list\npnpm run gen-module-graph     # regenerate docs/module-graph.md from package peerDeps\npnpm run verify-module-graph  # fail if docs/module-graph.md is stale\npnpm run build          # emit lib/types intermediates, then bundle lib/index.* runtime files\npnpm run verify-node-next-types  # fail if built declarations are not NodeNext-consumable\npnpm run hygiene        # knip, publint, workspace constraints, and NodeNext declaration check\n```\n\nWhen changing package public behavior, update the relevant README or JSDoc in the same change. `pnpm run doc-sync` catches checked TypeScript snippets, generated doc freshness, markdown wrap/link drift, type equivalence, translation pairing, Mermaid syntax, and doc budgets, but broader prose/API sync still needs review.\n\n## Demos\n\nThe one-shot Headless coding agent needs `DEEPSEEK_API_KEY` in the environment or repo-root `.env`:\n\n```sh\npnpm run demo:headless \"summarize this workspace\"\n```\n\nThe full-screen interactive coding agent needs `DEEPSEEK_API_KEY` in the environment or repo-root `.env`:\n\n```sh\npnpm run demo:tui\n```\n\nThe self-referential cordis-agent demo can inspect and modify its live plugin runtime and needs the same credentials:\n\n```sh\npnpm run demo:cordis\n```\n\nThe ACP automation server exposes fresh agent sessions over JSON-RPC stdio and also needs `DEEPSEEK_API_KEY`:\n\n```sh\npnpm run demo:acp\n```\n\n## TODO markers\n\nUse one of three comment tags to flag known issues in the code, ordered by urgency:\n\n- `FIXME` — an issue that should block a new release. A release should not ship with an open `FIXME` unless reviewers explicitly agree the change can be merged anyway.\n- `TODO` — an issue that should be fixed soon, once we have the resources.\n- `XXX` — an issue that we may fix someday; lowest priority, no commitment.\n\nPick the tag that matches the urgency so anyone scanning the code can tell a release blocker from a someday-maybe.\n\n## Documenting types verbatim (`ts type-equiv`)\n\nThe [core data structures](core-data-structures/core.md) docs paste source-equivalent declarations together with their original JSDoc so a reader sees the exact shape and source contract. To keep a paste from drifting when source changes, fence it as ` ```ts type-equiv ` (instead of ` ```ts `) and register it in `scripts/type-equiv.manifest.json` with the source file and symbol it mirrors:\n\n```json\n{ \"doc\": \"docs/core-data-structures/session.md\", \"symbol\": \"SessionEvent\", \"source\": \"packages/core/session/src/types.ts\" }\n```\n\n`pnpm run verify-type-equiv` (part of `doc-sync`) then extracts that symbol's declaration and attached JSDoc from source via the TypeScript parser and asserts the block matches both. For a class whose implementation bodies do not belong in the catalog, use ` ```ts public-api ` and set `\"projection\": \"public-api\"`; the checked projection retains the public fields, constructor, accessors, methods, and original class/member JSDoc while omitting bodies and private or protected members. Comparison ignores whitespace and non-JSDoc comments but requires every original JSDoc comment, including member documentation, so readers see the source contract beside the exact shape. The gate enforces a 1:1 correspondence by document, symbol, and projection between primary blocks and manifest entries; a paired `.zh.md` block reuses its unsuffixed sibling's entry only when the whole tracked fence sequence is byte-identical and ordered identically. `doc-typecheck` applies the same derivative rule to compilable fences, while skipping both source-equivalence fence kinds from compilation and its opt-out ratio. When you change a documented declaration or its JSDoc, the gate fails until you update the paste; when you add or remove a primary block, update the manifest in the same change.\n\n## Architecture context\n\nRead `docs/architecture.md` before changing anything under `packages/`. The codebase is built around Cordis plugins, event-sourced sessions, typed service seams, and explicit extension points.\n"
+        "content": "# Development guide\n\nEnglish | [中文](development.zh.md)\n\nThis onboarding guide helps project contributors get started with the local environment, daily workflow, and CI flow; see the Agent Notes for design rationale and technical trade-offs.\n\n## Prerequisites\n\n- Node.js supports 22.19+ and 24+. CI covers 22.19, 24, and 26; see the [Node engine floor Agent Note](../.agents/notes/implemented/process/2026-07-06-node-engine-floor.md).\n- Corepack-enabled pnpm. The repo pins `pnpm@11.7.0` in `package.json`; run `corepack enable` if `pnpm --version` does not resolve through Corepack.\n- Git 2.26 or newer; hook setup enables Git's worktree-specific configuration extension.\n- Optional: a DeepSeek API key for the TUI, headless, and ACP automation demos and real-API e2e tests.\n\n## First-time setup\n\nInstall dependencies from the repo root:\n\n```sh\npnpm install\n```\n\nThe install also runs the root `postinstall` script, which installs lefthook from the repo dev dependency through `scripts/install-lefthook.mjs`. With `CI=true` or `GITHUB_ACTIONS=true`, the wrapper returns before Git discovery because automated jobs do not consume contributor hooks. Otherwise, it requires Git 2.26 or newer and gives the current worktree an explicit hook directory under its own Git directory; linked worktrees therefore use their own lefthook binary and configuration instead of rewriting common hooks. The first install enables Git's worktree-specific configuration extension and repository format 1; see the [worktree-local hooks Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md).\n\nIf hooks are missing because dependencies were restored from cache or `postinstall` was skipped, install them manually:\n\n```sh\nnode scripts/install-lefthook.mjs\n```\n\nThe wrapper refuses user-owned `core.hooksPath` values. An inherited system, global, or common-repository path requires `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`. When Git seeds a new worktree with another registered worktree's marker-backed hook path, the wrapper replaces that copied value with the new worktree's own path; command-scoped and other worktree-scoped paths must be integrated or removed explicitly.\n\nBefore enabling worktree config, migrate direct `extensions.*` in a format-0 common config, direct `core.worktree` or `core.bare=true`, and any non-empty dormant `config.worktree`. The common config and every worktree config must be regular files, while the owned hook directory may contain only unaliased regular files.\n\nAfter moving a checkout, rerun the wrapper to relocate its owned path and regenerate hooks. For a stale or invalid installer lock, first confirm no installer is running, then remove the reported lock and retry. If installation and hook-path rollback both fail, inspect the reported worktree config before retrying. The [worktree-local hooks Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md) owns the full safety contract.\n\nRun typecheck once after a fresh clone:\n\n```sh\npnpm run typecheck\n```\n\nThat first typecheck runs the whole-repo `tsc -b` graph: it emits every package/vendor `lib/types` and checks examples, tests, and scripts through the two no-emit aggregates described below.\n\n## TypeScript project layout\n\nThe repository's TypeScript configuration has exactly three roles; every tsconfig file plays one of them.\n\n| File | Role | Forms a program? |\n|---|---|---|\n| `tsconfig.json` | Solution root: `extends` base, `files: []`, references to the two aggregates. The whole-repo `tsc -b tsconfig.json` graph, the tsserver discovery entry, and — through the inherited `paths` — the resolution config for tsx running `examples/` and `scripts/` (their nearest tsconfig is this file). | No |\n| `tsconfig.host.json` | Host aggregate: host-side packages (via references), examples, tests, scripts, website. Excludes `packages/client`. | Yes |\n| `tsconfig.client.json` | Client aggregate: `packages/client/*` packages and their tests, `apps/web`. | Yes |\n| `tsconfig.base.json` | Shared compilerOptions and the source `paths` map. Also the resolution facade the vitest configs point vite-tsconfig-paths at: it has no `include`, so its `paths` apply to every importer. | No |\n| `tsconfig.base.client.json` | Browser compiler shape (`jsx`, DOM libs, `types: []`) extended by the client aggregate and every `packages/client/*` package. | No |\n\nHost and client stay two aggregate programs because both sides declaration-merge the cordis `Context` interface under the same keys with different services; one program seeing both merges reports a collision. The collision exists only inside a `ts.Program` — module resolution never triggers it — which is why the solution may reference both aggregates and one paths facade may span both sides. Two disciplines follow:\n\n- `tsconfig.base.json` never gains `include` or `files`: they would leak into every extending package project and narrow the facade's match-all scope.\n- A script that builds a repo-wide `ts.Program` seeds `tsconfig.host.json` or `tsconfig.client.json` explicitly — never the root solution, because flattening both aggregates into one program collides the `Context` merges. Program-backed generators and gates (`scripts/ts-project.ts` consumers, doc-typecheck standalone mode) are host-only by decision; the client side gains program-backed tooling only with a concrete need.\n\nStatic analysis and tests resolve workspace imports through the base `paths` map to `src` and must pass on a clean tree; gates that consume built `lib/` output declare that dependency explicitly. Decision record: [solution-root note](../.agents/notes/implemented/process/2026-07-22-tsconfig-solution-root-two-aggregates.md); the tsc-first emit pipeline is the [ts-build-config note](../.agents/notes/implemented/process/2026-06-17-ts-build-config.md).\n\nIf a relevant local check consumes built package output, build once first:\n\n```sh\npnpm run build\n```\n\n`pnpm run hygiene` includes `publint`, which validates package entrypoints against the built `lib/*.js` files, and `verify-node-next-types`, which validates built declarations against a temporary NodeNext consumer. A fresh worktree has no bundled JS or declarations until `pnpm run build` runs; ordinary commits and pushes do not require that build unless their selected checks consume it.\n\n## Environment variables\n\nThe real DeepSeek adapter and key-backed agent demos read credentials from the environment or from a gitignored `.env` at the repo root:\n\n```sh\nDEEPSEEK_API_KEY=sk-...\nDEEPSEEK_BASE_URL=https://... # optional\n```\n\n`DEEPSEEK_BASE_URL` is optional and defaults to the public API. Never commit real credentials. The real-API e2e suites self-skip when `DEEPSEEK_API_KEY` is not set.\n\n## Git hooks\n\nlefthook is configured in `lefthook.yml` as a fast local checkpoint:\n\n- `pre-commit` runs staged-file ESLint fixes, checks the staged diff for whitespace errors, and runs the vendor manifest guard.\n- `pre-push` runs only the incremental repository typecheck (`tsc -b` over the root solution, covering both the host and client aggregates).\n\nThe vendor manifest guard checks that changes under `vendor/*/src` are staged with the matching `vendor/README.md` manifest update. See `vendor/README.md` before editing vendored code.\n\nThe hooks intentionally do not run tests, snapshots, documentation checks, builds, or hygiene. Contributors run the [checks relevant to the changed behavior](../AGENTS.md#run-relevant-checks-locally) once; CI owns exhaustive coverage, built-artifact smokes, and the Node 22.19, 24, and 26 compatibility matrix.\n\nContributors can opt into the comprehensive local gate set with `pnpm run check:all`. The command is independent of both Git hooks and is not an agent instruction.\n\n## CI gates\n\nThe keyless [CI workflow](../.github/workflows/ci.yml) groups independent gates into broad lanes and runs a smaller compatibility signal across supported Node versions. Artifact consumers wait for one build within their lane. The separate real-API workflow runs `pnpm run test:e2e` with its configured worker bound. See [scripts/run-gates.ts](../scripts/run-gates.ts) and the workflow files for the current gate and job inventory.\n\n## Daily commands\n\nUse these from the repo root:\n\n```sh\npnpm run test           # unit tests\npnpm run test:coverage  # unit tests with per-file coverage gates\npnpm run test:e2e       # real-API tests; self-skips without DEEPSEEK_API_KEY\npnpm run check:all      # comprehensive opt-in gate set; not wired to Git hooks\npnpm run typecheck      # tsc -b over the root solution: emits package/vendor lib/types, checks both aggregates\npnpm run lint           # eslint .\npnpm run lint:fix       # eslint . --fix\npnpm run doc-typecheck  # compile checked TypeScript snippets in Markdown docs\npnpm run gen-cordis-catalog     # regenerate docs/cordis-catalog/events.md + services.md from source\npnpm run verify-cordis-catalog  # fail if either cordis catalog is stale\npnpm run verify-export-jsdoc    # fail if a module-level package export lacks complete JSDoc\npnpm run gen-doc-graphs     # regenerate generated relationship docs from source and curated graph definitions\npnpm run verify-doc-graphs  # fail if generated relationship docs are stale\npnpm run verify-md-wrap  # fail on hard-wrapped prose paragraphs in docs/README markdown\npnpm run verify-mermaid  # fail if a ```mermaid diagram has invalid Mermaid syntax\npnpm run verify-type-equiv  # fail if a ```ts type-equiv doc block drifts from its source type\npnpm run verify-doc-budgets  # fail if a budgeted standing doc exceeds its word ceiling\npnpm run gen-translation-brief   # print the minimal-update briefing for out-of-sync translation pairs (--apply splices code-only edits)\npnpm run doc-sync       # all Markdown/doc gates, scheduled concurrently; the doc-sync leaf list in scripts/run-gates.ts is the full list\npnpm run gen-module-graph     # regenerate docs/module-graph.md from package peerDeps\npnpm run verify-module-graph  # fail if docs/module-graph.md is stale\npnpm run build          # emit lib/types intermediates, then bundle lib/index.* runtime files\npnpm run verify-node-next-types  # fail if built declarations are not NodeNext-consumable\npnpm run hygiene        # knip, publint, workspace constraints, and NodeNext declaration check\n```\n\nWhen changing package public behavior, update the relevant README or JSDoc in the same change. `pnpm run doc-sync` catches checked TypeScript snippets, generated doc freshness, markdown wrap/link drift, type equivalence, translation pairing, Mermaid syntax, and doc budgets, but broader prose/API sync still needs review.\n\n## Demos\n\nThe one-shot Headless coding agent needs `DEEPSEEK_API_KEY` in the environment or repo-root `.env`:\n\n```sh\npnpm run demo:headless \"summarize this workspace\"\n```\n\nThe full-screen interactive coding agent needs `DEEPSEEK_API_KEY` in the environment or repo-root `.env`:\n\n```sh\npnpm run demo:tui\n```\n\nThe self-referential cordis-agent demo can inspect and modify its live plugin runtime and needs the same credentials:\n\n```sh\npnpm run demo:cordis\n```\n\nThe ACP automation server exposes fresh agent sessions over JSON-RPC stdio and also needs `DEEPSEEK_API_KEY`:\n\n```sh\npnpm run demo:acp\n```\n\n## TODO markers\n\nUse one of three comment tags to flag known issues in the code, ordered by urgency:\n\n- `FIXME` — an issue that should block a new release. A release should not ship with an open `FIXME` unless reviewers explicitly agree the change can be merged anyway.\n- `TODO` — an issue that should be fixed soon, once we have the resources.\n- `XXX` — an issue that we may fix someday; lowest priority, no commitment.\n\nPick the tag that matches the urgency so anyone scanning the code can tell a release blocker from a someday-maybe.\n\n## Documenting types verbatim (`ts type-equiv`)\n\nThe [core data structures](core-data-structures/core.md) docs paste source-equivalent declarations together with their original JSDoc so a reader sees the exact shape and source contract. To keep a paste from drifting when source changes, fence it as ` ```ts type-equiv ` (instead of ` ```ts `) and register it in `scripts/type-equiv.manifest.json` with the source file and symbol it mirrors:\n\n```json\n{ \"doc\": \"docs/core-data-structures/session.md\", \"symbol\": \"SessionEvent\", \"source\": \"packages/core/session/src/types.ts\" }\n```\n\n`pnpm run verify-type-equiv` (part of `doc-sync`) then extracts that symbol's declaration and attached JSDoc from source via the TypeScript parser and asserts the block matches both. For a class whose implementation bodies do not belong in the catalog, use ` ```ts public-api ` and set `\"projection\": \"public-api\"`; the checked projection retains the public fields, constructor, accessors, methods, and original class/member JSDoc while omitting bodies and private or protected members. Comparison ignores whitespace and non-JSDoc comments but requires every original JSDoc comment, including member documentation, so readers see the source contract beside the exact shape. The gate enforces a 1:1 correspondence by document, symbol, and projection between primary blocks and manifest entries; a paired `.zh.md` block reuses its unsuffixed sibling's entry only when the whole tracked fence sequence is byte-identical and ordered identically. `doc-typecheck` applies the same derivative rule to compilable fences, while skipping both source-equivalence fence kinds from compilation and its opt-out ratio. When you change a documented declaration or its JSDoc, the gate fails until you update the paste; when you add or remove a primary block, update the manifest in the same change.\n\n## Architecture context\n\nRead `docs/architecture.md` before changing anything under `packages/`. The codebase is built around Cordis plugins, event-sourced sessions, typed service seams, and explicit extension points.\n"
       },
       {
         "role": "assistant",
-        "content": "# 开发指南\n\n[English](development.md) | 中文\n\n本指南覆盖参与 DeepSeek Harness 开发所需的本地环境搭建、日常工作流与 CI 流程；设计动机与技术权衡请查阅相应 Agent Note。\n\n## 前置条件\n\n- Node.js 支持 22.19+ 与 24+。CI 覆盖 22.19、24 和 26；见 [Node 引擎下限 Agent Note](../.agents/notes/implemented/process/2026-07-06-node-engine-floor.md)。\n- 启用了 Corepack 的 pnpm。仓库在 `package.json` 中固定使用 `pnpm@11.7.0`；如果 `pnpm --version` 无法通过 Corepack 解析，请先运行 `corepack enable`。\n- Git 2.26 或更高版本；钩子设置会启用 Git 的 worktree 专属配置扩展。\n- 可选：一个 DeepSeek API key，用于 TUI、headless 和 ACP（Agent Client Protocol）自动化 agent（智能体）演示以及真实 API 的 e2e 测试。\n\n## 首次搭建\n\n在仓库根目录安装依赖：\n\n```sh\npnpm install\n```\n\n安装过程同时会运行根目录的 `postinstall` 脚本，该脚本通过 `scripts/install-lefthook.mjs` 从仓库 dev 依赖安装 lefthook。当 `CI=true` 或 `GITHUB_ACTIONS=true` 时，该脚本会在探测 Git 前返回，因为自动化任务不会使用贡献者钩子。否则，包装脚本要求使用 Git 2.26 或更高版本，并会为当前 worktree 在其自身的 Git 目录下设置显式钩子目录；因此，关联 worktree 会使用各自的 lefthook 二进制文件和配置，而不会改写共用钩子。首次安装会启用 Git 的 worktree 专属配置扩展和仓库格式 1；见 [worktree 本地钩子 Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md)。\n\n如果依赖是从缓存恢复或 `postinstall` 被跳过而导致缺少钩子，请手动安装：\n\n```sh\nnode scripts/install-lefthook.mjs\n```\n\n包装层会拒绝用户自有的 `core.hooksPath` 值。继承自系统、全局或共用仓库配置的路径必须设置 `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`；命令作用域和 worktree 作用域的自定义路径必须显式集成或移除。\n\n启用 worktree 配置之前，请迁移格式 0 共用配置中直接设置的 `extensions.*`，并迁移直接设置的 `core.worktree` 或 `core.bare=true`，以及任何非空且尚未生效的 `config.worktree`。共用配置和每个 worktree 配置都必须是常规文件，而自有钩子目录只能包含不带别名的常规文件。\n\n检出目录移动后，请重新运行包装层，使其重新定位自有路径并重新生成钩子。对于陈旧或无效的安装程序锁，请先确认没有安装程序正在运行，再移除报告的锁并重试。若安装和钩子路径回滚都失败，请在重试前检查报告的 worktree 配置。完整安全契约由 [worktree 本地钩子 Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md) 统一定义。\n\n新克隆后请先运行一次类型检查：\n\n```sh\npnpm run typecheck\n```\n\n首次类型检查会执行全仓 `tsc -b tsconfig.json` 图：发射每个 package/vendor 的 `lib/types`，并通过下述两个 no-emit 聚合检查示例、测试和脚本。\n\n## TypeScript 项目布局\n\n仓库的 TypeScript 配置只有三种角色；每个 tsconfig 文件恰好扮演其中一种。\n\n| 文件 | 角色 | 是否构成 program？ |\n|---|---|---|\n| `tsconfig.json` | solution 根：`extends` base、`files: []`、引用两个聚合。全仓 `tsc -b tsconfig.json` 图、tsserver 发现入口，并经继承的 `paths` 充当 tsx 运行 `examples/` 与 `scripts/` 时的解析配置（它们最近的 tsconfig 就是此文件）。 | 否 |\n| `tsconfig.host.json` | host 聚合：host 侧各包（经 references）、示例、测试、脚本、website。排除 `packages/client`。 | 是 |\n| `tsconfig.client.json` | client 聚合：`packages/client/*` 各包及其测试、`apps/web`。 | 是 |\n| `tsconfig.base.json` | 共享 compilerOptions 与源码 `paths` 映射。同时是各 vitest 配置让 vite-tsconfig-paths 指向的解析门面：它没有 `include`，因此其 `paths` 适用于任何 importer。 | 否 |\n| `tsconfig.base.client.json` | 浏览器编译形状（`jsx`、DOM lib、`types: []`），由 client 聚合和每个 `packages/client/*` 包 extends。 | 否 |\n\nhost 与 client 保持两个聚合 program，是因为两侧在相同键下以不同服务对 cordis `Context` 接口做声明合并；单一 program 同时看到两份合并会报冲突。这种冲突只存在于 `ts.Program` 内部——模块解析永远不会触发它——所以 solution 可以同时引用两个聚合，一个 paths 门面也可以横跨两侧。由此推出两条纪律：\n\n- `tsconfig.base.json` 永不添加 `include` 或 `files`：它们会泄漏进每个 extends 它的包项目，并收窄门面的全匹配范围。\n- 构造全仓 `ts.Program` 的脚本显式种子 `tsconfig.host.json` 或 `tsconfig.client.json`——永不种子根 solution，因为把两个聚合展平进一个 program 会撞上 `Context` 合并冲突。基于 program 的生成器与门禁（`scripts/ts-project.ts` 的消费者、doc-typecheck standalone 模式）按决策仅覆盖 host 侧；client 侧只在出现真实需求时再获得基于 program 的工具。\n\n静态分析和测试通过 base 的 `paths` 映射把工作区 import 解析到 `src`，且必须在干净树上通过；消费构建产物 `lib/` 的门禁显式声明该依赖。决策记录：[solution-root note](../.agents/notes/implemented/process/2026-07-22-tsconfig-solution-root-two-aggregates.md)；tsc-first 发射管线见 [ts-build-config note](../.agents/notes/implemented/process/2026-06-17-ts-build-config.md)。\n\n如果相关的本地检查需要使用构建后的包产物，请先构建一次：\n\n```sh\npnpm run build\n```\n\n`pnpm run hygiene` 包含 `publint`（用构建出的 `lib/*.js` 文件校验 package 入口点）和 `verify-node-next-types`（用一个临时的 NodeNext 消费方校验构建出的声明文件）。新 worktree 在 `pnpm run build` 运行之前没有打包的 JS 和声明文件；普通提交和推送无需构建，除非所选检查会使用这些产物。\n\n## 环境变量\n\n真实的 DeepSeek 适配器和需要密钥的 agent 演示从环境变量或仓库根目录一个被 gitignore 的 `.env` 文件读取凭证：\n\n```sh\nDEEPSEEK_API_KEY=sk-...\nDEEPSEEK_BASE_URL=https://... # optional\n```\n\n`DEEPSEEK_BASE_URL` 可选，默认为公开 API。请勿提交真实凭证。未设置 `DEEPSEEK_API_KEY` 时，真实 API 的 e2e 套件会自动跳过。\n\n## Git 钩子\n\nlefthook 在 `lefthook.yml` 中配置，作为快速的本地检查点：\n\n- `pre-commit` 运行对暂存文件的 ESLint 修复，检查暂存 diff 中的空白错误，并运行 vendor manifest（元数据清单）守卫；\n- `pre-push` 只运行仓库增量类型检查（对根 solution 执行 `tsc -b`，覆盖 host 与 client 两个聚合）。\n\nvendor manifest 守卫检查 `vendor/*/src` 下的改动是否连同对应的 `vendor/README.md` manifest 更新一起暂存。请在编辑 vendor 代码前先阅读 `vendor/README.md`。\n\n这些钩子有意不运行测试、快照、文档检查、构建或 `hygiene`。贡献者只运行一次[与改动行为相关的检查](../AGENTS.md#run-relevant-checks-locally)；CI 负责全量覆盖率门禁、构建产物冒烟测试，以及 Node 22.19、24 和 26 兼容性矩阵。\n\n贡献者可以选择运行 `pnpm run check:all`，执行全面的本地门禁集。该命令独立于两个 Git 钩子，也不是对 agent 的指令。\n\n## CI 门禁\n\nkeyless [CI 工作流](../.github/workflows/ci.yml) 将独立门禁分组到若干宽粒度 lane，并在受支持的 Node 版本上运行一组较小的兼容性检查。产物消费方在各自 lane 内等待一次 build。单独的真实 API 工作流按其配置的 worker 上限运行 `pnpm run test:e2e`。当前门禁和 job 清单以 [scripts/run-gates.ts](../scripts/run-gates.ts) 和工作流文件为准。\n\n## 日常命令\n\n在仓库根目录使用：\n\n```sh\npnpm run test           # unit tests\npnpm run test:coverage  # unit tests with per-file coverage gates\npnpm run test:e2e       # real-API tests; self-skips without DEEPSEEK_API_KEY\npnpm run check:all      # comprehensive opt-in gate set; not wired to Git hooks\npnpm run typecheck      # tsc -b over the root solution: emits package/vendor lib/types, checks both aggregates\npnpm run lint           # eslint .\npnpm run lint:fix       # eslint . --fix\npnpm run doc-typecheck  # compile checked TypeScript snippets in Markdown docs\npnpm run gen-cordis-catalog     # regenerate docs/cordis-catalog/events.md + services.md from source\npnpm run verify-cordis-catalog  # fail if either cordis catalog is stale\npnpm run verify-export-jsdoc    # fail if a module-level package export lacks complete JSDoc\npnpm run gen-doc-graphs     # regenerate generated relationship docs from source and curated graph definitions\npnpm run verify-doc-graphs  # fail if generated relationship docs are stale\npnpm run verify-md-wrap  # fail on hard-wrapped prose paragraphs in docs/README markdown\npnpm run verify-mermaid  # fail if a ```mermaid diagram has invalid Mermaid syntax\npnpm run verify-type-equiv  # fail if a ```ts type-equiv doc block drifts from its source type\npnpm run verify-doc-budgets  # fail if a budgeted standing doc exceeds its word ceiling\npnpm run gen-translation-brief   # print the minimal-update briefing for out-of-sync translation pairs (--apply splices code-only edits)\npnpm run doc-sync       # all Markdown/doc gates, scheduled concurrently; the doc-sync leaf list in scripts/run-gates.ts is the full list\npnpm run gen-module-graph     # regenerate docs/module-graph.md from package peerDeps\npnpm run verify-module-graph  # fail if docs/module-graph.md is stale\npnpm run build          # emit lib/types intermediates, then bundle lib/index.* runtime files\npnpm run verify-node-next-types  # fail if built declarations are not NodeNext-consumable\npnpm run hygiene        # knip, publint, workspace constraints, and NodeNext declaration check\n```\n\n修改 package 的公开行为时，请在同一个变更中更新相关 README 或 JSDoc。`pnpm run doc-sync` 能检测到被检查的 TypeScript 片段、生成文档的新鲜度、Markdown 换行/链接漂移、type-equiv、翻译配对、Mermaid 语法和文档预算，但更广泛的行文/API 同步仍需评审把关。\n\n## 演示\n\n单次运行的 Headless coding agent 需要环境变量或仓库根目录 `.env` 中的 `DEEPSEEK_API_KEY`：\n\n```sh\npnpm run demo:headless \"summarize this workspace\"\n```\n\n全屏交互式 coding agent 需要环境变量或仓库根目录 `.env` 中的 `DEEPSEEK_API_KEY`：\n\n```sh\npnpm run demo:tui\n```\n\n自指的 cordis-agent 演示可以检查并修改其实时插件运行时，并需要相同的凭证：\n\n```sh\npnpm run demo:cordis\n```\n\nACP 自动化服务器通过 JSON-RPC stdio 提供全新 agent 会话，同样需要 `DEEPSEEK_API_KEY`：\n\n```sh\npnpm run demo:acp\n```\n\n## TODO 标记\n\n请使用以下三种注释标签之一标记代码中的已知问题，按紧急程度排序：\n\n- `FIXME`：应当阻塞新版本发布的问题。除非评审者明确同意该更改可以合并，否则发布版本不应包含未解决的 `FIXME`；\n- `TODO`：应当尽快修复的问题，等资源到位即可处理；\n- `XXX`：也许某天会修复的问题，优先级最低，不作承诺。\n\n请选择与紧急程度匹配的标签，让浏览代码的人一眼分清「发布阻塞」和「有空再说」。\n\n## 逐字记录类型（`ts type-equiv`）\n\n[核心数据结构](core-data-structures/core.md)文档会把与源码等价的声明及其原始 JSDoc 一并粘贴，让读者看到确切形状和源码契约。为防止粘贴内容在源码变化时漂移，请将其围栏为 ` ```ts type-equiv `（而不是 ` ```ts `），并在 `scripts/type-equiv.manifest.json` 中登记它镜像的源文件和符号：\n\n```json\n{ \"doc\": \"docs/core-data-structures/session.md\", \"symbol\": \"SessionEvent\", \"source\": \"packages/core/session/src/types.ts\" }\n```\n\n`pnpm run verify-type-equiv`（`doc-sync` 的一环）随后通过 TypeScript 解析器从源码提取该符号的声明及其附带的 JSDoc，并断言代码块同时匹配两者。对于不应把实现体写进目录的类，请使用 ` ```ts public-api ` 并设置 `\"projection\": \"public-api\"`；门禁检查的投影会保留公共字段、构造函数、访问器、方法以及类和成员的原始 JSDoc，同时省略实现体和私有或受保护成员。比对会忽略空白和非 JSDoc 注释，但要求保留每条原始 JSDoc（包括成员文档），让读者同时看到源码契约和确切形状。该门禁按文档、符号和投影，在主块与 manifest 条目之间强制 1:1 对应；只有当配对 `.zh.md` 块的完整受跟踪围栏序列与其无后缀兄弟文件按字节一致且顺序相同时，才会复用后者的条目。`doc-typecheck` 对可编译围栏应用同一派生规则，同时跳过两种源码等价围栏的编译，并将其排除在 opt-out 比例之外。当你改动一个已记录的类型声明或其 JSDoc 时，门禁会失败直到你更新粘贴内容；当你增删一个主块时，请在同一个变更里更新 manifest。\n\n## 架构上下文\n\n在修改 `packages/` 目录下的任何内容之前，请先阅读 `docs/architecture.md`。这套代码围绕 Cordis 插件、事件溯源的会话、类型化的服务 seam 与显式扩展点构建。\n"
+        "content": "# 开发指南\n\n[English](development.md) | 中文\n\n本指南覆盖参与 DeepSeek Harness 开发所需的本地环境搭建、日常工作流与 CI 流程；设计动机与技术权衡请查阅相应 Agent Note。\n\n## 前置条件\n\n- Node.js 支持 22.19+ 与 24+。CI 覆盖 22.19、24 和 26；见 [Node 引擎下限 Agent Note](../.agents/notes/implemented/process/2026-07-06-node-engine-floor.md)。\n- 启用了 Corepack 的 pnpm。仓库在 `package.json` 中固定使用 `pnpm@11.7.0`；如果 `pnpm --version` 无法通过 Corepack 解析，请先运行 `corepack enable`。\n- Git 2.26 或更高版本；钩子设置会启用 Git 的 worktree 专属配置扩展。\n- 可选：一个 DeepSeek API key，用于 TUI、headless 和 ACP（Agent Client Protocol）自动化 agent（智能体）演示以及真实 API 的 e2e 测试。\n\n## 首次搭建\n\n在仓库根目录安装依赖：\n\n```sh\npnpm install\n```\n\n安装过程同时会运行根目录的 `postinstall` 脚本，该脚本通过 `scripts/install-lefthook.mjs` 从仓库 dev 依赖安装 lefthook。当 `CI=true` 或 `GITHUB_ACTIONS=true` 时，该脚本会在探测 Git 前返回，因为自动化任务不会使用贡献者钩子。否则，包装脚本要求使用 Git 2.26 或更高版本，并会为当前 worktree 在其自身的 Git 目录下设置显式钩子目录；因此，关联 worktree 会使用各自的 lefthook 二进制文件和配置，而不会改写共用钩子。首次安装会启用 Git 的 worktree 专属配置扩展和仓库格式 1；见 [worktree 本地钩子 Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md)。\n\n如果依赖是从缓存恢复或 `postinstall` 被跳过而导致缺少钩子，请手动安装：\n\n```sh\nnode scripts/install-lefthook.mjs\n```\n\n包装层会拒绝用户自有的 `core.hooksPath` 值。继承自系统、全局或共用仓库配置的路径必须设置 `DSH_LEFTHOOK_ALLOW_HOOKS_PATH_OVERRIDE=1`。当 Git 使用另一个已注册 worktree 中由所有权标记佐证的钩子路径初始化新 worktree 时，包装层会将这个复制值替换为新 worktree 自有的路径；命令作用域和其他 worktree 作用域的路径必须显式集成或移除。\n\n启用 worktree 配置之前，请迁移格式 0 共用配置中直接设置的 `extensions.*`，并迁移直接设置的 `core.worktree` 或 `core.bare=true`，以及任何非空且尚未生效的 `config.worktree`。共用配置和每个 worktree 配置都必须是常规文件，而自有钩子目录只能包含不带别名的常规文件。\n\n检出目录移动后，请重新运行包装层，使其重新定位自有路径并重新生成钩子。对于陈旧或无效的安装程序锁，请先确认没有安装程序正在运行，再移除报告的锁并重试。若安装和钩子路径回滚都失败，请在重试前检查报告的 worktree 配置。完整安全契约由 [worktree 本地钩子 Agent Note](../.agents/notes/implemented/process/2026-07-27-worktree-local-lefthook.md) 统一定义。\n\n新克隆后请先运行一次类型检查：\n\n```sh\npnpm run typecheck\n```\n\n首次类型检查会执行全仓 `tsc -b tsconfig.json` 图：发射每个 package/vendor 的 `lib/types`，并通过下述两个 no-emit 聚合检查示例、测试和脚本。\n\n## TypeScript 项目布局\n\n仓库的 TypeScript 配置只有三种角色；每个 tsconfig 文件恰好扮演其中一种。\n\n| 文件 | 角色 | 是否构成 program？ |\n|---|---|---|\n| `tsconfig.json` | solution 根：`extends` base、`files: []`、引用两个聚合。全仓 `tsc -b tsconfig.json` 图、tsserver 发现入口，并经继承的 `paths` 充当 tsx 运行 `examples/` 与 `scripts/` 时的解析配置（它们最近的 tsconfig 就是此文件）。 | 否 |\n| `tsconfig.host.json` | host 聚合：host 侧各包（经 references）、示例、测试、脚本、website。排除 `packages/client`。 | 是 |\n| `tsconfig.client.json` | client 聚合：`packages/client/*` 各包及其测试、`apps/web`。 | 是 |\n| `tsconfig.base.json` | 共享 compilerOptions 与源码 `paths` 映射。同时是各 vitest 配置让 vite-tsconfig-paths 指向的解析门面：它没有 `include`，因此其 `paths` 适用于任何 importer。 | 否 |\n| `tsconfig.base.client.json` | 浏览器编译形状（`jsx`、DOM lib、`types: []`），由 client 聚合和每个 `packages/client/*` 包 extends。 | 否 |\n\nhost 与 client 保持两个聚合 program，是因为两侧在相同键下以不同服务对 cordis `Context` 接口做声明合并；单一 program 同时看到两份合并会报冲突。这种冲突只存在于 `ts.Program` 内部——模块解析永远不会触发它——所以 solution 可以同时引用两个聚合，一个 paths 门面也可以横跨两侧。由此推出两条纪律：\n\n- `tsconfig.base.json` 永不添加 `include` 或 `files`：它们会泄漏进每个 extends 它的包项目，并收窄门面的全匹配范围。\n- 构造全仓 `ts.Program` 的脚本显式种子 `tsconfig.host.json` 或 `tsconfig.client.json`——永不种子根 solution，因为把两个聚合展平进一个 program 会撞上 `Context` 合并冲突。基于 program 的生成器与门禁（`scripts/ts-project.ts` 的消费者、doc-typecheck standalone 模式）按决策仅覆盖 host 侧；client 侧只在出现真实需求时再获得基于 program 的工具。\n\n静态分析和测试通过 base 的 `paths` 映射把工作区 import 解析到 `src`，且必须在干净树上通过；消费构建产物 `lib/` 的门禁显式声明该依赖。决策记录：[solution-root note](../.agents/notes/implemented/process/2026-07-22-tsconfig-solution-root-two-aggregates.md)；tsc-first 发射管线见 [ts-build-config note](../.agents/notes/implemented/process/2026-06-17-ts-build-config.md)。\n\n如果相关的本地检查需要使用构建后的包产物，请先构建一次：\n\n```sh\npnpm run build\n```\n\n`pnpm run hygiene` 包含 `publint`（用构建出的 `lib/*.js` 文件校验 package 入口点）和 `verify-node-next-types`（用一个临时的 NodeNext 消费方校验构建出的声明文件）。新 worktree 在 `pnpm run build` 运行之前没有打包的 JS 和声明文件；普通提交和推送无需构建，除非所选检查会使用这些产物。\n\n## 环境变量\n\n真实的 DeepSeek 适配器和需要密钥的 agent 演示从环境变量或仓库根目录一个被 gitignore 的 `.env` 文件读取凭证：\n\n```sh\nDEEPSEEK_API_KEY=sk-...\nDEEPSEEK_BASE_URL=https://... # optional\n```\n\n`DEEPSEEK_BASE_URL` 可选，默认为公开 API。请勿提交真实凭证。未设置 `DEEPSEEK_API_KEY` 时，真实 API 的 e2e 套件会自动跳过。\n\n## Git 钩子\n\nlefthook 在 `lefthook.yml` 中配置，作为快速的本地检查点：\n\n- `pre-commit` 运行对暂存文件的 ESLint 修复，检查暂存 diff 中的空白错误，并运行 vendor manifest（元数据清单）守卫；\n- `pre-push` 只运行仓库增量类型检查（对根 solution 执行 `tsc -b`，覆盖 host 与 client 两个聚合）。\n\nvendor manifest 守卫检查 `vendor/*/src` 下的改动是否连同对应的 `vendor/README.md` manifest 更新一起暂存。请在编辑 vendor 代码前先阅读 `vendor/README.md`。\n\n这些钩子有意不运行测试、快照、文档检查、构建或 `hygiene`。贡献者只运行一次[与改动行为相关的检查](../AGENTS.md#run-relevant-checks-locally)；CI 负责全量覆盖率门禁、构建产物冒烟测试，以及 Node 22.19、24 和 26 兼容性矩阵。\n\n贡献者可以选择运行 `pnpm run check:all`，执行全面的本地门禁集。该命令独立于两个 Git 钩子，也不是对 agent 的指令。\n\n## CI 门禁\n\nkeyless [CI 工作流](../.github/workflows/ci.yml) 将独立门禁分组到若干宽粒度 lane，并在受支持的 Node 版本上运行一组较小的兼容性检查。产物消费方在各自 lane 内等待一次 build。单独的真实 API 工作流按其配置的 worker 上限运行 `pnpm run test:e2e`。当前门禁和 job 清单以 [scripts/run-gates.ts](../scripts/run-gates.ts) 和工作流文件为准。\n\n## 日常命令\n\n在仓库根目录使用：\n\n```sh\npnpm run test           # unit tests\npnpm run test:coverage  # unit tests with per-file coverage gates\npnpm run test:e2e       # real-API tests; self-skips without DEEPSEEK_API_KEY\npnpm run check:all      # comprehensive opt-in gate set; not wired to Git hooks\npnpm run typecheck      # tsc -b over the root solution: emits package/vendor lib/types, checks both aggregates\npnpm run lint           # eslint .\npnpm run lint:fix       # eslint . --fix\npnpm run doc-typecheck  # compile checked TypeScript snippets in Markdown docs\npnpm run gen-cordis-catalog     # regenerate docs/cordis-catalog/events.md + services.md from source\npnpm run verify-cordis-catalog  # fail if either cordis catalog is stale\npnpm run verify-export-jsdoc    # fail if a module-level package export lacks complete JSDoc\npnpm run gen-doc-graphs     # regenerate generated relationship docs from source and curated graph definitions\npnpm run verify-doc-graphs  # fail if generated relationship docs are stale\npnpm run verify-md-wrap  # fail on hard-wrapped prose paragraphs in docs/README markdown\npnpm run verify-mermaid  # fail if a ```mermaid diagram has invalid Mermaid syntax\npnpm run verify-type-equiv  # fail if a ```ts type-equiv doc block drifts from its source type\npnpm run verify-doc-budgets  # fail if a budgeted standing doc exceeds its word ceiling\npnpm run gen-translation-brief   # print the minimal-update briefing for out-of-sync translation pairs (--apply splices code-only edits)\npnpm run doc-sync       # all Markdown/doc gates, scheduled concurrently; the doc-sync leaf list in scripts/run-gates.ts is the full list\npnpm run gen-module-graph     # regenerate docs/module-graph.md from package peerDeps\npnpm run verify-module-graph  # fail if docs/module-graph.md is stale\npnpm run build          # emit lib/types intermediates, then bundle lib/index.* runtime files\npnpm run verify-node-next-types  # fail if built declarations are not NodeNext-consumable\npnpm run hygiene        # knip, publint, workspace constraints, and NodeNext declaration check\n```\n\n修改 package 的公开行为时，请在同一个变更中更新相关 README 或 JSDoc。`pnpm run doc-sync` 能检测到被检查的 TypeScript 片段、生成文档的新鲜度、Markdown 换行/链接漂移、type-equiv、翻译配对、Mermaid 语法和文档预算，但更广泛的行文/API 同步仍需评审把关。\n\n## 演示\n\n单次运行的 Headless coding agent 需要环境变量或仓库根目录 `.env` 中的 `DEEPSEEK_API_KEY`：\n\n```sh\npnpm run demo:headless \"summarize this workspace\"\n```\n\n全屏交互式 coding agent 需要环境变量或仓库根目录 `.env` 中的 `DEEPSEEK_API_KEY`：\n\n```sh\npnpm run demo:tui\n```\n\n自指的 cordis-agent 演示可以检查并修改其实时插件运行时，并需要相同的凭证：\n\n```sh\npnpm run demo:cordis\n```\n\nACP 自动化服务器通过 JSON-RPC stdio 提供全新 agent 会话，同样需要 `DEEPSEEK_API_KEY`：\n\n```sh\npnpm run demo:acp\n```\n\n## TODO 标记\n\n请使用以下三种注释标签之一标记代码中的已知问题，按紧急程度排序：\n\n- `FIXME`：应当阻塞新版本发布的问题。除非评审者明确同意该更改可以合并，否则发布版本不应包含未解决的 `FIXME`；\n- `TODO`：应当尽快修复的问题，等资源到位即可处理；\n- `XXX`：也许某天会修复的问题，优先级最低，不作承诺。\n\n请选择与紧急程度匹配的标签，让浏览代码的人一眼分清「发布阻塞」和「有空再说」。\n\n## 逐字记录类型（`ts type-equiv`）\n\n[核心数据结构](core-data-structures/core.md)文档会把与源码等价的声明及其原始 JSDoc 一并粘贴，让读者看到确切形状和源码契约。为防止粘贴内容在源码变化时漂移，请将其围栏为 ` ```ts type-equiv `（而不是 ` ```ts `），并在 `scripts/type-equiv.manifest.json` 中登记它镜像的源文件和符号：\n\n```json\n{ \"doc\": \"docs/core-data-structures/session.md\", \"symbol\": \"SessionEvent\", \"source\": \"packages/core/session/src/types.ts\" }\n```\n\n`pnpm run verify-type-equiv`（`doc-sync` 的一环）随后通过 TypeScript 解析器从源码提取该符号的声明及其附带的 JSDoc，并断言代码块同时匹配两者。对于不应把实现体写进目录的类，请使用 ` ```ts public-api ` 并设置 `\"projection\": \"public-api\"`；门禁检查的投影会保留公共字段、构造函数、访问器、方法以及类和成员的原始 JSDoc，同时省略实现体和私有或受保护成员。比对会忽略空白和非 JSDoc 注释，但要求保留每条原始 JSDoc（包括成员文档），让读者同时看到源码契约和确切形状。该门禁按文档、符号和投影，在主块与 manifest 条目之间强制 1:1 对应；只有当配对 `.zh.md` 块的完整受跟踪围栏序列与其无后缀兄弟文件按字节一致且顺序相同时，才会复用后者的条目。`doc-typecheck` 对可编译围栏应用同一派生规则，同时跳过两种源码等价围栏的编译，并将其排除在 opt-out 比例之外。当你改动一个已记录的类型声明或其 JSDoc 时，门禁会失败直到你更新粘贴内容；当你增删一个主块时，请在同一个变更里更新 manifest。\n\n## 架构上下文\n\n在修改 `packages/` 目录下的任何内容之前，请先阅读 `docs/architecture.md`。这套代码围绕 Cordis 插件、事件溯源的会话、类型化的服务 seam 与显式扩展点构建。\n"
       },
       {
         "role": "user",
-        "content": "# Bilingual documentation\n\nEnglish | [中文](README.zh.md)\n\nThis repo's documentation is read by people and agents both inside and outside the company, so every document in scope is maintained in English and Simplified Chinese. This page defines the pairing contract, enforcement gate, scope, and exclusions; [translation-rules.md](translation-rules.md) defines how to translate; [terminology.md](terminology.md) is the terminology source of truth. The committed agent workflow lives in [.agents/skills/dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md).\n\n## The pairing contract\n\n- **Both languages carry equal authority.** A document may be authored and reviewed in either language first — a Chinese-first Agent Note is as legitimate as an English-first one — and the counterpart is translated from it. Neither file outranks the other; what binds them is that they must say the same thing.\n- **A pair is three sibling files.** The English `foo.md`, the Chinese `foo.zh.md`, and a consistency record `foo.i18n.yaml`, all in the same directory. No locale directories, no separate translation repo, no interleaved bilingual files. Pairs merge whole: a PR never lands one language without the other two files.\n- **The consistency record.** `foo.i18n.yaml` holds the full git blob hash of each side as of the last time the two were confirmed to say the same thing:\n\n  ```yaml\n  foo.md: 3f786850e387550fdab836ed7e6dc881de23001b\n  foo.zh.md: 89e6c98d92887913cadf06b2adb97f26cde4849b\n  ```\n\n  Blob hashes, not commit hashes, so the record is computable for files edited in the same PR (`git hash-object foo.md`) and consistency is a pure content comparison. The recorded hashes also recover the exact last-confirmed text of either side, so an out-of-sync pair is updated by patching the counterpart minimally against the edited side's diff — never by re-translating whole files. `pnpm run gen-translation-brief <pair>` assembles that update's working set mechanically at the narrowest safely aligned granularity — changed Markdown units, then heading sections, then whole document — with the edited side's diff since last confirmation, each changed span's three-way text, the terminology rows the change touches, and the binding update rules; a change confined to the pair's byte-identical code fences is computed outright, and `--apply` splices it into the counterpart after structural validation ([briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)). After bringing the pair back in line, `pnpm run verify-translation-pairing --write <pair>` re-records both hashes; that yaml diff is the reviewable act of confirming consistency, which is why `--write` requires naming the pairs you confirmed (`--write --all` is the explicit corpus-wide form).\n- **Language switcher.** Both files link to each other immediately after their H1 heading: the English file carries `English | [中文](foo.zh.md)` and the Chinese file carries `[English](foo.md) | 中文`.\n- **Structure mirrors the counterpart.** Heading depths and order, list kinds, ordered-list starts, list item counts, table row and column counts, link targets, and verbatim code blocks match one to one across the pair — see [translation-rules.md](translation-rules.md) for the full preservation rules. Existing Markdown gates apply to `.zh.md` files unchanged (`verify-md-wrap`, `verify-md-links`).\n\n## The gate: verify-translation-pairing\n\n`pnpm run verify-translation-pairing` (part of `doc-sync`, which contributors run locally for documentation changes and CI runs exhaustively) enforces the contract mechanically:\n\n1. Every document in scope has a complete pair. README discovery is case-insensitive on the basename, so `missions/readme.md` is in scope alongside the other documentation roots.\n2. Every pair artifact that exists at all is complete and consistent: all three files present, each side's current blob hash equals the recorded one (editing either side without re-confirming the pair goes red), both sides carry the language switcher, and the structural signatures match in order — heading depths, verbatim code blocks (info string and content), table row and column counts, list kinds, ordered-list starts, item counts, and every link target apart from the switcher.\n3. Files listed as `excluded` have no `.zh.md` and no `.i18n.yaml` at all. Frozen Agent Notes under `.agents/notes/archived/` are outside this evolving gate; their dedicated verifier requires and seals the complete existing triplet instead.\n\nSource-oriented code gates consume an exact `.zh.md` fence sequence as a derivative of its unsuffixed sibling instead of compiling or manifesting the same code twice. The sequence must match in length, order, fence kind, and byte-exact body; otherwise both copies remain independently checked and the pairing gate reports the structural mismatch.\n\n`pnpm run verify-translation-pairing --list` prints the current pairing state of every document in scope — missing, out-of-sync, or ok. It never fails; `missing` and `out-of-sync` rows identify violations that the normal check rejects.\n\n`pnpm run verify-translation-pairing <pair...>` checks just the named pairs — any of a pair's three files (or its bare stem) names it — so an update loop verifies its own pair in seconds instead of re-scanning the corpus. The no-argument corpus-wide form is what `doc-sync` and CI run; a scoped green never substitutes for it at PR level.\n\nThe practical rule this gate creates: **when a PR edits either side of a paired document, the same PR updates the counterpart and re-records the pair** (run the [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) skill, then `--write <pair>`), exactly like the repo's existing doc-sync rule for code and READMEs. A PR that leaves a pair out of sync goes red in CI.\n\nThe gate's limit, stated plainly: **a green gate means the pair was confirmed consistent at these exact contents, not that the confirmation was sound.** It checks hashes and shape; it cannot judge whether the two sides actually say the same thing, or whether the wording is accurate, well-termed, and natural — that is the reviewer's half of the contract, per [translation-rules.md](translation-rules.md). A re-recorded pair with a sloppy counterpart passes the gate; it must not pass review.\n\n## Scope and exclusions\n\n**Scope**: every non-vendor README, plus every active document under `.agents/notes/**`, `docs/**`, and `python/**`. README matching is case-insensitive on the basename and covers future directories without another manifest edit. Dependency and ignored build-output trees and the frozen `.agents/notes/archived/` tree are discovery exclusions, not evolving translation source.\n\n**Excluded** (never paired, and the gate rejects a `.zh.md` or `.i18n.yaml` for them):\n\n- `docs/cordis-catalog/`, `docs/tool-catalog/`, `docs/config-catalog.md`, `docs/persistence-catalog.md`, `docs/module-graph.md`, `docs/agent-lifecycle.md`, `docs/capability-seams.md`, `docs/event-producer-consumer.md`, `docs/graph-atlas.md`, and `docs/tool-execution-pipeline.md` — generated files; their generators emit English only today, so a hand-written translation would go stale on every regeneration. The planned follow-up is to teach the generators to emit Chinese alongside English, at which point these leave the exclusion list.\n- `docs/AGENTS.md`, `.agents/notes/**/AGENTS.md`, and their `CLAUDE.md` instruction symlinks — agent instructions, maintained in English only like the root `AGENTS.md`.\n- `docs/i18n/terminology.md` and [style-samples.md](style-samples.md) — both are bilingual by construction.\n- [translation-prompt.md](translation-prompt.md) — the automated pipeline's prompt template; its body is machine-consumed verbatim, so a paired translation would change pipeline behavior.\n- `.agents/notes/archived/` — frozen historical triplets. [`verify-archived-agent-notes`](../../scripts/verify-archived-agent-notes.ts) validates their completeness and content seals; translation maintenance must never rewrite them.\n\n**Universal requirement**: every current or future document in scope must merge as a complete bilingual pair. [scripts/translation-pairing.manifest.json](../../scripts/translation-pairing.manifest.json) contains only explicit exclusions; there is no per-file rollout list, date cutoff, or README-specific policy class.\n\n## Division of labor\n\nCounterparts here are produced by an agent running [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) and reviewed by a human — inference is cheap here, review attention is the scarce resource. The gate checks pair completeness, recorded hashes, switchers, and its documented structural signature. Review still owns translation quality, terminology, and structural requirements that the signature does not encode. The prompt contract is executable: [scripts/translation-prompt.ts](../../scripts/translation-prompt.ts) renders the committed template (terminology injected; the template carries its own calibrated rules) into either direction and parses the three-section response, while `verify-translation-prompt` exercises both render directions and the checked-in example in `doc-sync`.\n"
+        "content": "# Bilingual documentation\n\nEnglish | [中文](README.zh.md)\n\nThis repo's documentation is read by people and agents both inside and outside the company, so every document in scope is maintained in English and Simplified Chinese. This page defines the pairing contract, enforcement gate, scope, and exclusions; [translation-rules.md](translation-rules.md) defines how to translate; [terminology.md](terminology.md) is the terminology source of truth. The committed agent workflow lives in [.agents/skills/dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md).\n\n## The pairing contract\n\n- **Both languages carry equal authority.** A document may be authored and reviewed in either language first — a Chinese-first Agent Note is as legitimate as an English-first one — and the counterpart is translated from it. Neither file outranks the other; what binds them is that they must say the same thing.\n- **A pair is three sibling files.** The English `foo.md`, the Chinese `foo.zh.md`, and a consistency record `foo.i18n.yaml`, all in the same directory. No locale directories, no separate translation repo, no interleaved bilingual files. Pairs merge whole: a PR never lands one language without the other two files.\n- **The consistency record.** `foo.i18n.yaml` holds the full git blob hash of each side as of the last time the two were confirmed to say the same thing:\n\n  ```yaml\n  foo.md: 3f786850e387550fdab836ed7e6dc881de23001b\n  foo.zh.md: 89e6c98d92887913cadf06b2adb97f26cde4849b\n  ```\n\n  Blob hashes, not commit hashes, so the record is computable for files edited in the same PR (`git hash-object foo.md`) and consistency is a pure content comparison. `--write` stores those snapshots in the local Git object database before recording them, including uncommitted working-tree contents, and pins every distinct stored blob under a content-addressed `refs/dsh/translation-pairing/snapshots/` ref so garbage collection cannot invalidate a recorded recovery pointer. The recorded hashes therefore recover the exact last-confirmed text of either side, so an out-of-sync pair is updated by patching the counterpart minimally against the edited side's diff — never by re-translating whole files. `pnpm run gen-translation-brief <pair>` assembles that update's working set mechanically at the narrowest safely aligned granularity — changed Markdown units, then heading sections, then whole document — with the edited side's diff since last confirmation, each changed span's three-way text, the terminology rows the change touches, and the binding update rules; a change confined to the pair's byte-identical code fences is computed outright, and `--apply` splices it into the counterpart after structural validation ([briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)). After bringing the pair back in line, `pnpm run verify-translation-pairing --write <pair>` re-records both hashes; that yaml diff is the reviewable act of confirming consistency, which is why `--write` requires naming the pairs you confirmed (`--write --all` is the explicit corpus-wide form).\n- **Language switcher.** Both files link to each other immediately after their H1 heading: the English file carries `English | [中文](foo.zh.md)` and the Chinese file carries `[English](foo.md) | 中文`.\n- **Structure mirrors the counterpart.** Heading depths and order, list kinds, ordered-list starts, list item counts, table row and column counts, link targets, and verbatim code blocks match one to one across the pair — see [translation-rules.md](translation-rules.md) for the full preservation rules. Existing Markdown gates apply to `.zh.md` files unchanged (`verify-md-wrap`, `verify-md-links`).\n\n## The gate: verify-translation-pairing\n\n`pnpm run verify-translation-pairing` (part of `doc-sync`, which contributors run locally for documentation changes and CI runs exhaustively) enforces the contract mechanically:\n\n1. Every document in scope has a complete pair. README discovery is case-insensitive on the basename, so `missions/readme.md` is in scope alongside the other documentation roots.\n2. Every pair artifact that exists at all is complete and consistent: all three files present, each side's current blob hash equals the recorded one (editing either side without re-confirming the pair goes red), both sides carry the language switcher, and the structural signatures match in order — heading depths, verbatim code blocks (info string and content), table row and column counts, list kinds, ordered-list starts, item counts, and every link target apart from the switcher.\n3. Files listed as `excluded` have no `.zh.md` and no `.i18n.yaml` at all. Frozen Agent Notes under `.agents/notes/archived/` are outside this evolving gate; their dedicated verifier requires and seals the complete existing triplet instead.\n\nSource-oriented code gates consume an exact `.zh.md` fence sequence as a derivative of its unsuffixed sibling instead of compiling or manifesting the same code twice. The sequence must match in length, order, fence kind, and byte-exact body; otherwise both copies remain independently checked and the pairing gate reports the structural mismatch.\n\n`pnpm run verify-translation-pairing --list` prints the current pairing state of every document in scope — missing, out-of-sync, or ok. It never fails; `missing` and `out-of-sync` rows identify violations that the normal check rejects.\n\n`pnpm run verify-translation-pairing <pair...>` checks just the named pairs — any of a pair's three files (or its bare stem) names it — so an update loop verifies its own pair in seconds instead of re-scanning the corpus. The no-argument corpus-wide form is what `doc-sync` and CI run; a scoped green never substitutes for it at PR level.\n\nThe practical rule this gate creates: **when a PR edits either side of a paired document, the same PR updates the counterpart and re-records the pair** (run the [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) skill, then `--write <pair>`), exactly like the repo's existing doc-sync rule for code and READMEs. A PR that leaves a pair out of sync goes red in CI.\n\nThe gate's limit, stated plainly: **a green gate means the pair was confirmed consistent at these exact contents, not that the confirmation was sound.** It checks hashes and shape; it cannot judge whether the two sides actually say the same thing, or whether the wording is accurate, well-termed, and natural — that is the reviewer's half of the contract, per [translation-rules.md](translation-rules.md). A re-recorded pair with a sloppy counterpart passes the gate; it must not pass review.\n\n## Scope and exclusions\n\n**Scope**: every non-vendor README, plus every active document under `.agents/notes/**`, `docs/**`, and `python/**`. README matching is case-insensitive on the basename and covers future directories without another manifest edit. Dependency and ignored build-output trees and the frozen `.agents/notes/archived/` tree are discovery exclusions, not evolving translation source.\n\n**Excluded** (never paired, and the gate rejects a `.zh.md` or `.i18n.yaml` for them):\n\n- `docs/cordis-catalog/`, `docs/tool-catalog/`, `docs/config-catalog.md`, `docs/persistence-catalog.md`, `docs/module-graph.md`, `docs/agent-lifecycle.md`, `docs/capability-seams.md`, `docs/event-producer-consumer.md`, `docs/graph-atlas.md`, and `docs/tool-execution-pipeline.md` — generated files; their generators emit English only today, so a hand-written translation would go stale on every regeneration. The planned follow-up is to teach the generators to emit Chinese alongside English, at which point these leave the exclusion list.\n- `docs/AGENTS.md`, `.agents/notes/**/AGENTS.md`, and their `CLAUDE.md` instruction symlinks — agent instructions, maintained in English only like the root `AGENTS.md`.\n- `docs/i18n/terminology.md` and [style-samples.md](style-samples.md) — both are bilingual by construction.\n- [translation-prompt.md](translation-prompt.md) — the automated pipeline's prompt template; its body is machine-consumed verbatim, so a paired translation would change pipeline behavior.\n- `.agents/notes/archived/` — frozen historical triplets. [`verify-archived-agent-notes`](../../scripts/verify-archived-agent-notes.ts) validates their completeness and content seals; translation maintenance must never rewrite them.\n\n**Universal requirement**: every current or future document in scope must merge as a complete bilingual pair. [scripts/translation-pairing.manifest.json](../../scripts/translation-pairing.manifest.json) contains only explicit exclusions; there is no per-file rollout list, date cutoff, or README-specific policy class.\n\n## Division of labor\n\nCounterparts here are produced by an agent running [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) and reviewed by a human — inference is cheap here, review attention is the scarce resource. The gate checks pair completeness, recorded hashes, switchers, and its documented structural signature. Review still owns translation quality, terminology, and structural requirements that the signature does not encode. The prompt contract is executable: [scripts/translation-prompt.ts](../../scripts/translation-prompt.ts) renders the committed template (terminology injected; the template carries its own calibrated rules) into either direction and parses the three-section response, while `verify-translation-prompt` exercises both render directions and the checked-in example in `doc-sync`.\n"
       },
       {
         "role": "assistant",
-        "content": "# 双语文档\n\n[English](README.md) | 中文\n\n本仓库的文档会被公司内外的人和 agent（智能体）阅读，因此范围内的每篇文档都以英文和简体中文维护。本页定义配对契约、强制门禁、范围与排除规则；[translation-rules.md](translation-rules.md) 定义如何翻译；[terminology.md](terminology.md) 是术语真源。仓库内置的 agent 工作流见 [.agents/skills/dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md)。\n\n## 配对契约\n\n- **两种语言同权。** 一篇文档可以先用任一语言撰写和评审（先写中文的 Agent Note 与先写英文的一样正当），另一侧由它翻译而来。两个文件谁也不高于谁；约束它们的是二者必须说同样的话。\n- **一对文档是三个同目录文件。** 英文 `foo.md`、中文 `foo.zh.md`，加一份一致性记录 `foo.i18n.yaml`，都在同一目录。不用语言目录，不用独立翻译仓库，不用中英混排的单文件。配对必须整体合并：PR（Pull Request）永远不会只带一种语言而缺其余两个文件。\n- **一致性记录。**`foo.i18n.yaml` 保存两侧文件在上一次被确认「说同样的话」时各自的完整 git blob hash：\n\n  ```yaml\n  foo.md: 3f786850e387550fdab836ed7e6dc881de23001b\n  foo.zh.md: 89e6c98d92887913cadf06b2adb97f26cde4849b\n  ```\n\n  用 blob hash 而不是 commit hash，这样同一个 PR 里改动的文件也能算出记录（`git hash-object foo.md`），一致性是纯内容比较。记录的 hash 还能还原任一侧上次确认时的确切文本，所以失去同步的配对是「按被改一侧的 diff 最小化地修补另一侧」，从不整篇重译。`pnpm run gen-translation-brief <pair>` 会以能安全对齐的最窄粒度——先是有改动的 Markdown 单元，再是标题小节，最后是整篇文档——机械地汇集这次更新的工作集：被改一侧自上次确认以来的 diff、每个改动块的三方文本、改动触及的术语表行，以及有约束力的更新规则；仅落在配对中逐字节一致的围栏代码块内的改动可以直接算出，`--apply` 则经结构签名校验后把它拼接进对侧文件（[briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)）。两侧对齐后，`pnpm run verify-translation-pairing --write <pair>` 重新记录两个 hash；那份 yaml diff 就是「确认一致」这个动作本身，可以被评审，也正因如此，`--write` 要求点名你确认过的配对（`--write --all` 是显式的全语料形式）。\n- **语言切换行。** 两个文件在各自 H1 标题之后立即互链：英文文件带 `English | [中文](foo.zh.md)`，中文文件带 `[English](foo.md) | 中文`。\n- **结构与另一侧一一对应。** 标题深度与顺序、列表类型、有序列表起始编号、列表项数量、表格行列数、链接目标与逐字节一致的代码块在配对两侧一一对应；完整保持规则见 [translation-rules.md](translation-rules.md)。既有 Markdown 门禁对 `.zh.md` 文件原样生效（`verify-md-wrap`、`verify-md-links`）。\n\n## 门禁：verify-translation-pairing\n\n`pnpm run verify-translation-pairing`（`doc-sync`（文档同步门禁）的一环，贡献者会针对文档变更在本地运行，CI 则会完整运行）机械地强制执行这份契约：\n\n1. 范围内的每篇文档都有完整配对。发现 README 时，basename 不区分大小写，因此 `missions/readme.md` 与其他文档根一样属于范围。\n2. 任何已存在的配对产物都完整且一致：三个文件齐全、每一侧的当前 blob hash 等于记录值（改了任一侧而没重新确认配对就变红）、双方都带语言切换行、结构签名按序一致：标题深度、逐字节一致的代码块（信息字符串与内容）、表格行列数、列表类型、有序列表起始编号、列表项数量，以及除切换行之外的每个链接目标。\n3. 列为 `excluded` 的文件完全没有 `.zh.md`，也没有 `.i18n.yaml`。`.agents/notes/archived/` 下冻结的 Agent Note 不受这个持续演进的门禁约束；专用校验器会要求其现有的三个配对文件完整，并将其封存。\n\n面向源码的代码门禁会把精确的 `.zh.md` 围栏序列视为其无后缀兄弟文件的派生内容，而不会再次编译相同代码或在 manifest 中重复登记。该序列必须在长度、顺序、围栏类型和按字节精确的正文上一致；否则两份副本仍会独立受检，配对门禁也会报告结构不匹配。\n\n`pnpm run verify-translation-pairing --list` 打印范围内每篇文档的当前配对状态（missing、out-of-sync 或 ok）。它从不失败；其中 missing 与 out-of-sync 行指出普通检查会拒绝的违规。\n\n`pnpm run verify-translation-pairing <pair...>` 只检查被点名的配对——配对的三个文件中的任意一个（或其裸词干）都能点名它——因此更新循环几秒内就能验证自己的配对，而不必重新扫描全语料。`doc-sync` 与 CI 运行的是无参数的全语料形式；限定范围的绿灯在 PR 层面永远不能替代它。\n\n这个门禁带来的实际规则是：**当一个 PR 修改了已配对文档的任一侧时，同一个 PR 更新另一侧并重新记录配对**（运行 [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) skill（技能），再 `--write <pair>`），与本仓库既有的代码与 README 的 doc-sync 规则完全一致。留下失去同步的配对的 PR 会在 CI 变红。\n\n把门禁的边界说白：**门禁通过意味着这组文档在当前内容上的一致性得到了确认，不代表确认本身正确可靠。** 它检查记录的 hash 与结构签名；它无法判断两侧是否真的在说同样的话，也无法判断措辞是否准确、术语是否得当、行文是否自然；这部分契约由评审者把关，见 [translation-rules.md](translation-rules.md)。重新记录了 hash 但另一侧翻得潦草的配对能通过门禁；它不得通过评审。\n\n## 范围与排除\n\n**范围**：除 vendor 源码外的全部 README，以及 `.agents/notes/**`、`docs/**` 与 `python/**` 下的全部活跃文档。匹配 README 时只看文件名且不区分大小写，因此今后新增的目录无需再修改 manifest。依赖目录、被忽略的构建产物目录以及冻结的 `.agents/notes/archived/` 目录树只在发现阶段排除，不属于持续演进的翻译源文档。\n\n**排除**（永不配对，门禁拒绝为它们建 `.zh.md` 或 `.i18n.yaml`）：\n\n- `docs/cordis-catalog/`、`docs/tool-catalog/`、`docs/config-catalog.md`、`docs/persistence-catalog.md`、`docs/module-graph.md`、`docs/agent-lifecycle.md`、`docs/capability-seams.md`、`docs/event-producer-consumer.md`、`docs/graph-atlas.md` 与 `docs/tool-execution-pipeline.md`：生成文件；生成器目前只输出英文，手写译文在每次重新生成时必然陈旧。计划中的后续工作是让生成器同时输出中文，届时这些文件移出排除清单。\n- `docs/AGENTS.md`、`.agents/notes/**/AGENTS.md` 以及指向它们的 `CLAUDE.md` 指令符号链接：agent 指令，与根 `AGENTS.md` 一样只以英文维护。\n- `docs/i18n/terminology.md` 与 [style-samples.md](style-samples.md)：二者本身即为中英对照文档。\n- [translation-prompt.md](translation-prompt.md)：自动翻译流水线的提示词模板；正文逐字进入模型请求，配对翻译会改变流水线行为。\n- `.agents/notes/archived/`：冻结的历史三文件配对。[`verify-archived-agent-notes`](../../scripts/verify-archived-agent-notes.ts) 校验其完整性和内容封存记录；翻译维护绝不能重写这些文件。\n\n**统一要求**：当前及今后纳入范围的每篇文档，合并时都必须构成完整的双语配对。[scripts/translation-pairing.manifest.json](../../scripts/translation-pairing.manifest.json) 只包含显式排除项；不存在逐文件推进清单、日期分界或 README 专用政策类别。\n\n## 分工\n\n这里的对侧文件由运行 [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) 的 agent 生成，再由人评审：在这里推理（inference）很便宜，评审注意力才是稀缺资源。门禁负责检查配对是否完整、记录的 hash、语言切换行以及本文列出的结构签名；翻译质量、术语和签名未涵盖的结构要求仍由评审把关。提示词契约也有可执行实现：[scripts/translation-prompt.ts](../../scripts/translation-prompt.ts) 会把仓库内置的模板（注入术语表；模板自带经人工校准的规则）渲染为英译中或中译英两个方向的提示词，并解析三段式响应；`doc-sync` 中的 `verify-translation-prompt` 会检查两个渲染方向与仓库内示例。\n"
+        "content": "# 双语文档\n\n[English](README.md) | 中文\n\n本仓库的文档会被公司内外的人和 agent（智能体）阅读，因此范围内的每篇文档都以英文和简体中文维护。本页定义配对契约、强制门禁、范围与排除规则；[translation-rules.md](translation-rules.md) 定义如何翻译；[terminology.md](terminology.md) 是术语真源。仓库内置的 agent 工作流见 [.agents/skills/dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md)。\n\n## 配对契约\n\n- **两种语言同权。** 一篇文档可以先用任一语言撰写和评审（先写中文的 Agent Note 与先写英文的一样正当），另一侧由它翻译而来。两个文件谁也不高于谁；约束它们的是二者必须说同样的话。\n- **一对文档是三个同目录文件。** 英文 `foo.md`、中文 `foo.zh.md`，加一份一致性记录 `foo.i18n.yaml`，都在同一目录。不用语言目录，不用独立翻译仓库，不用中英混排的单文件。配对必须整体合并：PR（Pull Request）永远不会只带一种语言而缺其余两个文件。\n- **一致性记录。**`foo.i18n.yaml` 保存两侧文件在上一次被确认「说同样的话」时各自的完整 git blob hash：\n\n  ```yaml\n  foo.md: 3f786850e387550fdab836ed7e6dc881de23001b\n  foo.zh.md: 89e6c98d92887913cadf06b2adb97f26cde4849b\n  ```\n\n  用 blob hash 而不是 commit hash，这样同一个 PR 里改动的文件也能算出记录（`git hash-object foo.md`），一致性是纯内容比较。`--write` 会先把这些快照存入本地 Git 对象库再写下记录，未提交的工作树内容也不例外；它还会在内容寻址的 `refs/dsh/translation-pairing/snapshots/` ref 下固定每个不同的已存 blob，使垃圾回收无法让已记录的恢复指针失效。因此记录的 hash 能还原任一侧上次确认时的确切文本，所以失去同步的配对是「按被改一侧的 diff 最小化地修补另一侧」，从不整篇重译。`pnpm run gen-translation-brief <pair>` 会以能安全对齐的最窄粒度——先是有改动的 Markdown 单元，再是标题小节，最后是整篇文档——机械地汇集这次更新的工作集：被改一侧自上次确认以来的 diff、每个改动块的三方文本、改动触及的术语表行，以及有约束力的更新规则；仅落在配对中逐字节一致的围栏代码块内的改动可以直接算出，`--apply` 则经结构签名校验后把它拼接进对侧文件（[briefed-updates Agent Note](../../.agents/notes/implemented/process/2026-07-26-briefed-minimal-translation-updates.md)）。两侧对齐后，`pnpm run verify-translation-pairing --write <pair>` 重新记录两个 hash；那份 yaml diff 就是「确认一致」这个动作本身，可以被评审，也正因如此，`--write` 要求点名你确认过的配对（`--write --all` 是显式的全语料形式）。\n- **语言切换行。** 两个文件在各自 H1 标题之后立即互链：英文文件带 `English | [中文](foo.zh.md)`，中文文件带 `[English](foo.md) | 中文`。\n- **结构与另一侧一一对应。** 标题深度与顺序、列表类型、有序列表起始编号、列表项数量、表格行列数、链接目标与逐字节一致的代码块在配对两侧一一对应；完整保持规则见 [translation-rules.md](translation-rules.md)。既有 Markdown 门禁对 `.zh.md` 文件原样生效（`verify-md-wrap`、`verify-md-links`）。\n\n## 门禁：verify-translation-pairing\n\n`pnpm run verify-translation-pairing`（`doc-sync`（文档同步门禁）的一环，贡献者会针对文档变更在本地运行，CI 则会完整运行）机械地强制执行这份契约：\n\n1. 范围内的每篇文档都有完整配对。发现 README 时，basename 不区分大小写，因此 `missions/readme.md` 与其他文档根一样属于范围。\n2. 任何已存在的配对产物都完整且一致：三个文件齐全、每一侧的当前 blob hash 等于记录值（改了任一侧而没重新确认配对就变红）、双方都带语言切换行、结构签名按序一致：标题深度、逐字节一致的代码块（信息字符串与内容）、表格行列数、列表类型、有序列表起始编号、列表项数量，以及除切换行之外的每个链接目标。\n3. 列为 `excluded` 的文件完全没有 `.zh.md`，也没有 `.i18n.yaml`。`.agents/notes/archived/` 下冻结的 Agent Note 不受这个持续演进的门禁约束；专用校验器会要求其现有的三个配对文件完整，并将其封存。\n\n面向源码的代码门禁会把精确的 `.zh.md` 围栏序列视为其无后缀兄弟文件的派生内容，而不会再次编译相同代码或在 manifest 中重复登记。该序列必须在长度、顺序、围栏类型和按字节精确的正文上一致；否则两份副本仍会独立受检，配对门禁也会报告结构不匹配。\n\n`pnpm run verify-translation-pairing --list` 打印范围内每篇文档的当前配对状态（missing、out-of-sync 或 ok）。它从不失败；其中 missing 与 out-of-sync 行指出普通检查会拒绝的违规。\n\n`pnpm run verify-translation-pairing <pair...>` 只检查被点名的配对——配对的三个文件中的任意一个（或其裸词干）都能点名它——因此更新循环几秒内就能验证自己的配对，而不必重新扫描全语料。`doc-sync` 与 CI 运行的是无参数的全语料形式；限定范围的绿灯在 PR 层面永远不能替代它。\n\n这个门禁带来的实际规则是：**当一个 PR 修改了已配对文档的任一侧时，同一个 PR 更新另一侧并重新记录配对**（运行 [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) skill（技能），再 `--write <pair>`），与本仓库既有的代码与 README 的 doc-sync 规则完全一致。留下失去同步的配对的 PR 会在 CI 变红。\n\n把门禁的边界说白：**门禁通过意味着这组文档在当前内容上的一致性得到了确认，不代表确认本身正确可靠。** 它检查记录的 hash 与结构签名；它无法判断两侧是否真的在说同样的话，也无法判断措辞是否准确、术语是否得当、行文是否自然；这部分契约由评审者把关，见 [translation-rules.md](translation-rules.md)。重新记录了 hash 但另一侧翻得潦草的配对能通过门禁；它不得通过评审。\n\n## 范围与排除\n\n**范围**：除 vendor 源码外的全部 README，以及 `.agents/notes/**`、`docs/**` 与 `python/**` 下的全部活跃文档。匹配 README 时只看文件名且不区分大小写，因此今后新增的目录无需再修改 manifest。依赖目录、被忽略的构建产物目录以及冻结的 `.agents/notes/archived/` 目录树只在发现阶段排除，不属于持续演进的翻译源文档。\n\n**排除**（永不配对，门禁拒绝为它们建 `.zh.md` 或 `.i18n.yaml`）：\n\n- `docs/cordis-catalog/`、`docs/tool-catalog/`、`docs/config-catalog.md`、`docs/persistence-catalog.md`、`docs/module-graph.md`、`docs/agent-lifecycle.md`、`docs/capability-seams.md`、`docs/event-producer-consumer.md`、`docs/graph-atlas.md` 与 `docs/tool-execution-pipeline.md`：生成文件；生成器目前只输出英文，手写译文在每次重新生成时必然陈旧。计划中的后续工作是让生成器同时输出中文，届时这些文件移出排除清单。\n- `docs/AGENTS.md`、`.agents/notes/**/AGENTS.md` 以及指向它们的 `CLAUDE.md` 指令符号链接：agent 指令，与根 `AGENTS.md` 一样只以英文维护。\n- `docs/i18n/terminology.md` 与 [style-samples.md](style-samples.md)：二者本身即为中英对照文档。\n- [translation-prompt.md](translation-prompt.md)：自动翻译流水线的提示词模板；正文逐字进入模型请求，配对翻译会改变流水线行为。\n- `.agents/notes/archived/`：冻结的历史三文件配对。[`verify-archived-agent-notes`](../../scripts/verify-archived-agent-notes.ts) 校验其完整性和内容封存记录；翻译维护绝不能重写这些文件。\n\n**统一要求**：当前及今后纳入范围的每篇文档，合并时都必须构成完整的双语配对。[scripts/translation-pairing.manifest.json](../../scripts/translation-pairing.manifest.json) 只包含显式排除项；不存在逐文件推进清单、日期分界或 README 专用政策类别。\n\n## 分工\n\n这里的对侧文件由运行 [dsh-translate-docs](../../.agents/skills/dsh-translate-docs/SKILL.md) 的 agent 生成，再由人评审：在这里推理（inference）很便宜，评审注意力才是稀缺资源。门禁负责检查配对是否完整、记录的 hash、语言切换行以及本文列出的结构签名；翻译质量、术语和签名未涵盖的结构要求仍由评审把关。提示词契约也有可执行实现：[scripts/translation-prompt.ts](../../scripts/translation-prompt.ts) 会把仓库内置的模板（注入术语表；模板自带经人工校准的规则）渲染为英译中或中译英两个方向的提示词，并解析三段式响应；`doc-sync` 中的 `verify-translation-prompt` 会检查两个渲染方向与仓库内示例。\n"
       },
       {
         "role": "user",
diff --git a/scripts/translation-pairing-git.ts b/scripts/translation-pairing-git.ts
new file mode 100644
index 0000000000..992f0abbb9
--- /dev/null
+++ b/scripts/translation-pairing-git.ts
@@ -0,0 +1,50 @@
+/** Git-blob operations owned by the bilingual pairing workflow. */
+
+import { spawnSync } from 'node:child_process'
+import { createHash } from 'node:crypto'
+
+const SNAPSHOT_REF_PREFIX = 'refs/dsh/translation-pairing/snapshots'
+
+/** Full SHA-1 Git blob hash (the 40-hex format used by pairing records). */
+export function gitBlobHash(content: Buffer): string {
+  const hash = createHash('sha1')
+  hash.update(`blob ${content.byteLength}\0`)
+  hash.update(content)
+  return hash.digest('hex')
+}
+
+function runGit(root: string, args: string[], operation: string, input?: Buffer): Buffer {
+  const result = spawnSync('git', ['-C', root, ...args], {
+    input,
+    maxBuffer: 1 << 26,
+  })
+  if (result.error) {
+    throw new Error(`${operation} failed: ${result.error.message}`, { cause: result.error })
+  }
+  if (result.status !== 0) {
+    throw new Error(`${operation} failed with status ${String(result.status)}: ${result.stderr.toString('utf8').trim()}`)
+  }
+  return result.stdout
+}
+
+/**
+ * Persist exact working-tree bytes so a pairing record can later recover them
+ * with `git cat-file`, even when they have never appeared in the index or a
+ * commit. The returned object ID is checked against the pairing format's own
+ * content hash before the caller writes a sidecar.
+ */
+export function storeGitBlob(root: string, content: Buffer): string {
+  const expected = gitBlobHash(content)
+  const stored = runGit(root, ['hash-object', '-w', '--stdin'], 'git hash-object -w --stdin', content)
+    .toString('utf8')
+    .trim()
+  if (stored !== expected) {
+    throw new Error(`git hash-object -w --stdin returned unexpected object ID ${JSON.stringify(stored)}; expected ${expected}`)
+  }
+  runGit(
+    root,
+    ['update-ref', `${SNAPSHOT_REF_PREFIX}/${stored}`, stored],
+    'git update-ref for translation snapshot',
+  )
+  return stored
+}
diff --git a/scripts/translation-pairing.spec.ts b/scripts/translation-pairing.spec.ts
index da33dfdd6b..4d44266887 100644
--- a/scripts/translation-pairing.spec.ts
+++ b/scripts/translation-pairing.spec.ts
@@ -1,6 +1,11 @@
-/** Regression tests for the bilingual corpus scope and structural signature. */
+/** Regression tests for bilingual snapshots, corpus scope, and structure. */
 
+import { execFileSync, spawnSync } from 'node:child_process'
+import { mkdtempSync, rmSync } from 'node:fs'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
 import { describe, expect, it } from 'vitest'
+import { gitBlobHash, storeGitBlob } from './translation-pairing-git.ts'
 import {
   isTranslationScopeFile,
   pairAnchorOfArgument,
@@ -15,6 +20,71 @@ function signature(markdown: string) {
   return translationStructureSignature(parseTranslationMarkdown(markdown), 'counterpart.zh.md')
 }
 
+function gitSupportsObjectFormat(format: 'sha256'): boolean {
+  const root = mkdtempSync(join(tmpdir(), 'dsh-git-object-format-'))
+  try {
+    return spawnSync('git', ['init', '--quiet', `--object-format=${format}`, root], {
+      stdio: 'ignore',
+    }).status === 0
+  } finally {
+    rmSync(root, { recursive: true, force: true })
+  }
+}
+
+const supportsSha256ObjectFormat = gitSupportsObjectFormat('sha256')
+
+describe('translation pairing snapshots', () => {
+  it('stores exact uncommitted bytes for later recovery by object ID', () => {
+    const root = mkdtempSync(join(tmpdir(), 'dsh-translation-pairing-'))
+    try {
+      execFileSync('git', ['init', '--quiet', root], {
+        env: { ...process.env, GIT_DEFAULT_HASH: 'sha1' },
+      })
+      const content = Buffer.from([0x75, 0x6e, 0x63, 0x6f, 0x6d, 0x6d, 0x69, 0x74, 0x74, 0x65, 0x64, 0x0a, 0xff])
+
+      const objectId = storeGitBlob(root, content)
+
+      expect(objectId).toBe(gitBlobHash(content))
+      expect(execFileSync('git', [
+        '-C', root, 'rev-parse', `refs/dsh/translation-pairing/snapshots/${objectId}`,
+      ], { encoding: 'utf8' }).trim()).toBe(objectId)
+      execFileSync('git', ['-C', root, 'gc', '--prune=now'])
+      expect(execFileSync('git', ['-C', root, 'cat-file', '-p', objectId])).toEqual(content)
+    } finally {
+      rmSync(root, { recursive: true, force: true })
+    }
+  })
+
+  it('fails before a sidecar can reference an unavailable object', () => {
+    const root = mkdtempSync(join(tmpdir(), 'dsh-translation-pairing-'))
+    try {
+      expect(() => storeGitBlob(root, Buffer.from('snapshot'))).toThrow('git hash-object -w --stdin failed')
+    } finally {
+      rmSync(root, { recursive: true, force: true })
+    }
+  })
+
+  it('fails clearly when Git cannot be started', () => {
+    const previousPath = process.env.PATH
+    try {
+      process.env.PATH = ''
+      expect(() => storeGitBlob('.', Buffer.from('snapshot'))).toThrow('git hash-object -w --stdin failed')
+    } finally {
+      process.env.PATH = previousPath
+    }
+  })
+
+  it.skipIf(!supportsSha256ObjectFormat)('rejects an object format that pairing records cannot represent', () => {
+    const root = mkdtempSync(join(tmpdir(), 'dsh-translation-pairing-'))
+    try {
+      execFileSync('git', ['init', '--quiet', '--object-format=sha256', root])
+      expect(() => storeGitBlob(root, Buffer.from('snapshot'))).toThrow('returned unexpected object ID')
+    } finally {
+      rmSync(root, { recursive: true, force: true })
+    }
+  })
+})
+
 describe('translation pairing manifest', () => {
   it('accepts an exclusions-only manifest', () => {
     expect(parseTranslationPairingManifest(JSON.stringify({
diff --git a/scripts/tspath-loader.ts b/scripts/tspath-loader.ts
deleted file mode 100644
index adeecb6b64..0000000000
--- a/scripts/tspath-loader.ts
+++ /dev/null
@@ -1,14 +0,0 @@
-/** Register source-only tsconfig paths resolution before a TypeScript entry loads. */
-
-import { register } from 'node:module'
-import { resolve } from 'node:path'
-import { fileURLToPath } from 'node:url'
-
-const tsconfigPath = process.env.TSX_TSCONFIG_PATH === undefined
-  ? fileURLToPath(new URL('../tsconfig.json', import.meta.url))
-  : resolve(process.env.TSX_TSCONFIG_PATH)
-
-register(new URL('../apps/cli/src/tsconfig-paths-loader.ts', import.meta.url), {
-  parentURL: import.meta.url,
-  data: { tsconfigPath },
-})
diff --git a/scripts/verify-translation-pairing.ts b/scripts/verify-translation-pairing.ts
index afa458f8a5..4ebd664cb7 100644
--- a/scripts/verify-translation-pairing.ts
+++ b/scripts/verify-translation-pairing.ts
@@ -9,9 +9,9 @@
  * See `docs/i18n/README.md` for the owning contract.
  */
 
-import { createHash } from 'node:crypto'
 import { existsSync, globSync, readFileSync, writeFileSync } from 'node:fs'
 import { basename, join, resolve, sep } from 'node:path'
+import { gitBlobHash, storeGitBlob } from './translation-pairing-git.ts'
 import {
   linksTo,
   parseTranslationMarkdown,
@@ -54,14 +54,6 @@ function isExcluded(file: string): boolean {
   return manifest.excluded.some(entry => (entry.endsWith('/') ? file.startsWith(entry) : file === entry))
 }
 
-/** Full git blob hash (what `git hash-object` prints). */
-function blobHash(content: Buffer): string {
-  const hash = createHash('sha1')
-  hash.update(`blob ${content.byteLength}\0`)
-  hash.update(content)
-  return hash.digest('hex')
-}
-
 /** The three paths of a pair, derived from the English-file path. */
 function pairPaths(source: string): { zh: string; meta: string } {
   return { zh: source.replace(/\.md$/, '.zh.md'), meta: source.replace(/\.md$/, '.i18n.yaml') }
@@ -148,7 +140,12 @@ if (writeMode) {
       }
       continue
     }
-    const record = renderMeta(source, blobHash(readFileSync(join(root, source))), zh, blobHash(readFileSync(join(root, zh))))
+    const sourceContent = readFileSync(join(root, source))
+    const zhContent = readFileSync(join(root, zh))
+    // A consistency record is also a recovery pointer for the briefing
+    // generator. Persist both snapshots even when the sidecar text is already
+    // current, because the bytes may exist only in this working tree.
+    const record = renderMeta(source, storeGitBlob(root, sourceContent), zh, storeGitBlob(root, zhContent))
     if (existsSync(join(root, meta)) && readFileSync(join(root, meta), 'utf8') === record) continue
     writeFileSync(join(root, meta), record)
     console.log(`verify-translation-pairing: recorded ${meta}`)
@@ -203,7 +200,7 @@ for (const source of [...pairAnchors].sort()) {
 
   let consistent = true
   for (const [file, content] of [[source, sourceContent], [zh, zhContent]] as const) {
-    const current = blobHash(content)
+    const current = gitBlobHash(content)
     if (record.get(basename(file)) !== current) {
       errors.push(`${file}: out of sync — content no longer matches the pair's last confirmed-consistent state in ${meta} (bring the other side along, then re-record with --write)`)
       consistent = false
diff --git a/tsconfig.base.json b/tsconfig.base.json
index 00c19c4b8c..fc2f2c1ee4 100644
--- a/tsconfig.base.json
+++ b/tsconfig.base.json
@@ -86,6 +86,7 @@
         "./packages/hooks/*/src/invariant.ts",
         "./packages/session-persistence/*/src/invariant.ts",
         "./packages/session-projection/*/src/invariant.ts",
+        "./packages/session-title/*/src/invariant.ts",
         "./packages/session-query/*/src/invariant.ts",
         "./packages/telemetry/*/src/invariant.ts",
         "./packages/acp/*/src/invariant.ts",
@@ -104,7 +105,7 @@
       "@deepseek-ai/dsh-host-*/invariant": ["./packages/host/*/src/invariant.ts"],
       "@deepseek-ai/dsh-client-*/invariant": ["./packages/client/*/src/invariant.ts"],
       "@deepseek-ai/dsh-client-*/client": ["./packages/client/*/src/client"],
-      // One wildcard maps every @deepseek-ai/dsh-<name> to its source. Package
+       // One wildcard maps every @deepseek-ai/dsh-<name> to its source. Package
       // dir names are unique across groups, so first-on-disk-wins resolution is
       // unambiguous; adding a package under an existing group needs no edit
       // here. The aggregates' project references (tsconfig.host.json /
@@ -135,6 +136,8 @@
       "@deepseek-ai/dsh-client-ui-slash": ["./packages/client/ui-slash/src"],
       "@deepseek-ai/dsh-client-ui-command": ["./packages/client/ui-command/src"],
       "@deepseek-ai/dsh-client-ui-model": ["./packages/client/ui-model/src"],
+      "@deepseek-ai/dsh-client-ui-goal": ["./packages/client/ui-goal/src"],
+      "@deepseek-ai/dsh-client-ui-permission": ["./packages/client/ui-permission/src"],
       "@deepseek-ai/dsh-client-ui-skill": ["./packages/client/ui-skill/src"],
       "@deepseek-ai/dsh-client-ui-subagent": ["./packages/client/ui-subagent/src"],
       "@deepseek-ai/dsh-client-ui-plan": ["./packages/client/ui-plan/src"],
diff --git a/tsconfig.client.json b/tsconfig.client.json
index 5a52f59bb0..aca5772ec0 100644
--- a/tsconfig.client.json
+++ b/tsconfig.client.json
@@ -24,6 +24,7 @@
     "packages/host/directory-picker-native/tests/**/*.ts",
     "packages/host/directory-picker-native/tests/**/*.tsx",
     "packages/client/tsdown.client.ts",
+    "scripts/client-bundle-css.spec.ts",
     "scripts/client-bundle-purity.spec.ts"
   ],
   "references": [
diff --git a/tsconfig.host.json b/tsconfig.host.json
index e2112b7f6a..51f6ad0fa7 100644
--- a/tsconfig.host.json
+++ b/tsconfig.host.json
@@ -15,6 +15,7 @@
     "apps/web/tests/steering.e2e.ts",
     "apps/web/tests/navigation-panes.e2e.ts",
     "apps/web/tests/lifecycle-chrome.e2e.ts",
+    "apps/web/tests/details-session-lifecycle.e2e.ts",
     "apps/web/tests/settings-chrome.e2e.ts",
     "apps/web/tests/workspace-management.e2e.ts",
     "apps/web/tests/replay-round-trip.e2e.ts",
@@ -22,6 +23,7 @@
     "apps/web/tests/sidebar-scrollbar.e2e.ts",
     "apps/web/tests/code-mode-round.e2e.ts",
     "apps/web/tests/cordis-tool-round.e2e.ts",
+    "apps/web/tests/message-actions.e2e.ts",
     "apps/cli/tests/**/*.ts",
     "examples/*/src/**/*.ts",
     "examples/*/start.ts",
@@ -35,6 +37,7 @@
     "packages/client/**",
     "packages/host/directory-picker-browse/**",
     "packages/host/directory-picker-native/**",
+    "scripts/client-bundle-css.spec.ts",
     "scripts/client-bundle-purity.spec.ts"
   ],
   "references": [
@@ -84,6 +87,7 @@
     { "path": "./packages/goal/goal-session" },
     { "path": "./packages/goal/command-goal" },
     { "path": "./packages/context/time-context" },
+    { "path": "./packages/context/tmux-context" },
     { "path": "./packages/context/session-reference" },
     { "path": "./packages/ui/user-interaction" },
     { "path": "./packages/ui/user-approval" },
diff --git a/vitest.config.ts b/vitest.config.ts
index 55f172416f..8f9b9ed50e 100644
--- a/vitest.config.ts
+++ b/vitest.config.ts
@@ -102,6 +102,8 @@ export default defineConfig({
         // branches need a browser-grade harness the jsdom lane doesn't cover
         // yet. TODO(gui): cover and remove as the client test lane matures.
         'packages/client/ui-trajectory/src/*',
+        // Trajectory's compact Markdown projection retains deferred branch coverage.
+        'packages/client/ui-primitives/src/markdown/plain-text.ts',
         'packages/client/ui-question/src/client/QuestionComposer.tsx',
         'packages/client/ui-primitives/src/Menu.tsx',
         'packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx',