feat(agent-presets): add Codex and Claude Code subagent tools

2026-08-10 12:45:05 +08:00
parent e637f95d78
commit b1d67a6935
45 changed files with 1708 additions and 45 deletions
@@ -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/2026-08-03-per-session-agent-presets.md
-2026-08-03-per-session-agent-presets.md: 6f1643c25008c3363cb10adb7fbff7afeea31cbe
-2026-08-03-per-session-agent-presets.zh.md: 7afe9ade5c98fadb96384a7e0acd47531c370e0c
+2026-08-03-per-session-agent-presets.md: ca3e6967504ac62b1d79ec28ef9dbd4bf8383bac
+2026-08-03-per-session-agent-presets.zh.md: 94651d465939d760554ed8637376ae77cfae6812
@@ -18,7 +18,7 @@ Composition splits into two planes, decided by what must be shared rather than b

 | Plane | Instances | Contents |
 |---|---|---|
-| Host | one | The registries themselves (`tools`, `systemPrompt`, `agents`, `agent-loop`, `sessions`), cross-session facilities (persistence, query, projections, storage, settings, credentials, telemetry), and the web host |
+| Host | one | The registries themselves (`tools`, `systemPrompt`, `agents`, `agent-loop`, `sessions`), cross-session facilities (persistence, query, projections, storage, settings, credentials, telemetry), the subagent providers those facilities resolve, and the web host |
 | Agent | one per session | What a single agent contributes to those registries: tool plugins, persona and prompt sections, compaction policy |

 Model routing stays out of presets. `installAgentLlmTarget` is already the per-agent seam for provider, model, and reasoning effort, and an LLM adapter mounted inside a preset would never be resolved by `agent-loop`, which lives in the host plane.
@@ -55,7 +55,7 @@ Which preset an unnamed session gets is a user setting (`agent-presets.default`)

 **Authoring a preset is an RPC, and a privileged one.** A composition is a file, but "edit it on the filesystem" is not a browser affordance, so the roster gained `read`/`write`/`remove` beside `select`. Those three are loopback-pinned: a composition names the plugins a session runs, so reading one is reconnaissance and writing one is arbitrary capability. `list` and `select` deliberately stay ordinary. The roster carries ids and trust only, and a LAN client's picker needs it; and choosing a preset looked like escalation — one of them mounts the toolset that edits the live runtime — but `session.create` already takes an `agentPreset`, so pinning only the switch would have left the same capability one method over. The capability is not the preset's to grant either: the deployment's own default already carries `bash` and the filesystem tools, so any caller that may start a session at all can already run commands as this process. Containment is a property of the id (`[a-z0-9][a-z0-9-]*`), checked before it becomes a directory name rather than by inspecting the joined path afterwards; the text is parsed with the loader's own schema and dialect, so a save cannot leave a file no session could load. Shipped presets are refused for writes and deletes, because the deployment's copy is what a broken local preset is compared against — which also makes "duplicate, then edit" the authoring path rather than an afterthought.

-**A service with a consumer outside the agent plane cannot move into a preset.** The aggressive split moved the `subagents` registry and its spawn/fork backends into the delegation group's entry-local realm, and `dsh web` then failed to boot: `dsh-host-apiproxy` is a HOST row that injects `subagents` to answer the browser's cross-session queries (`listChildren`, `followup`), so it waited forever for a service only sessions now provided. A per-session copy is wrong twice over — a provider name registers once, so the second session would have collided anyway. The registry and its backends are host-plane; the preset contributes the delegation TOOLS, which resolve the host registry. `workflows` stays entry-local because nothing outside an agent reads it. Grepping injectors is what should have caught this and did not: the search has to include the host packages, not just the agent-plane ones.
+**A service with a consumer outside the agent plane cannot move into a preset.** The aggressive split moved the `subagents` registry and its spawn/fork backends into the delegation group's entry-local realm, and `dsh web` then failed to boot: `dsh-host-apiproxy` is a HOST row that injects `subagents` to answer the browser's cross-session queries (`listChildren`, `followup`), so it waited forever for a service only sessions now provided. A per-session copy is wrong twice over — a provider name registers once, so the second session would have collided anyway. The registry and every shared backend, including the fixed Codex and Claude Code product providers, are host-plane; a preset contributes whichever delegation TOOLS its agent should see, and those tools resolve the host registry. `workflows` stays entry-local because nothing outside an agent reads it. Grepping injectors is what should have caught this and did not: the search has to include the host packages, not just the agent-plane ones.

 **A real-composition test that disables a host row cannot audit that row.** The web composition test disabled `api-gateway` — the api-proxy itself — as a row with side effects, which is exactly the row whose pending injection would have named the break. It now boots with the api-proxy enabled and the browse directory picker substituted, so the boot audit covers the whole host-plane injection graph; only the port, the asset tree, and the telemetry exporter stay off.

@@ -78,3 +78,7 @@ Which preset an unnamed session gets is a user setting (`agent-presets.default`)
 **Make the agent's scope key the preset.** Sessions on one preset would share a layer for free, but per-agent registrations — `installAgentLlmTarget`, per-agent tool restrictions — would then collide across sessions.

 **Run each preset as a child process.** [`subagent-dsh-sdk`](../../../../packages/subagent/subagent-dsh-sdk/README.md) already proves a full child harness works, and isolation would be absolute. It also means proxying streaming, approvals, and projections per session, which is a transport project rather than a composition one.
+
+**Give product subagents global enable settings and a separate settings page.** The process-wide value would compete with the preset as owner of model-visible tools and could not express two sessions using different compositions. Product providers stay host-side, while ordinary preset rows independently expose Codex and Claude Code tools.
+
+**Ship one preset for every Codex and Claude Code combination.** Four identities duplicate the full preset composition to represent two independent rows. A copied preset can enable either row directly, so combination presets add roster and maintenance cost without adding a user result.
@@ -18,7 +18,7 @@ Status: implemented

 | 平面 | 实例数 | 内容 |
 |---|---|---|
-| 宿主 | 一份 | 注册表本身（`tools`、`systemPrompt`、`agents`、`agent-loop`、`sessions`）、跨会话设施（持久化、查询、投影、存储、设置、凭据、遥测），以及 web 宿主 |
+| 宿主 | 一份 | 注册表本身（`tools`、`systemPrompt`、`agents`、`agent-loop`、`sessions`）、跨会话设施（持久化、查询、投影、存储、设置、凭据、遥测）、这些设施所解析的 subagent provider，以及 web 宿主 |
 | agent | 每会话一份 | 单个 agent 对这些注册表的贡献：工具插件、人设与提示词段落、压缩策略 |

 模型路由不进 preset。`installAgentLlmTarget` 已经是 provider、model 与 reasoning effort 的按 agent 可替换点；而挂在 preset 内部的 LLM 适配器永远不会被 `agent-loop` 解析到，因为后者位于宿主平面。
@@ -56,7 +56,7 @@ Status: implemented

 **创作 preset 是一次 RPC，而且是特权 RPC。** 组装是一个文件，但“去文件系统里改它”并不是浏览器能提供的操作，因此名单在 `select` 之外新增了 `read`/`write`/`remove`。这三者被固定在环回地址：组装指明了一个会话所运行的插件，因此读取它是侦察，写入它是任意能力。`list` 与 `select` 刻意保持为普通方法。名单只携带 id 与信任级别，而局域网客户端的选择器需要它；至于选择本身，它看起来像提权——其中一个 preset 会挂载可编辑活动运行时的工具集——但 `session.create` 本就接受 `agentPreset`，只固定切换会把同一能力留在隔壁一个方法上。这份能力也不由 preset 授予：部署自带的默认 preset 本就带着 `bash` 与文件系统工具，因此任何被允许开启会话的调用方，早已能以本进程的身份执行命令。约束是 id 自身的性质（`[a-z0-9][a-z0-9-]*`），在它成为目录名之前就检查，而不是事后再去审视拼接出的路径；文本使用 loader 自身的 schema 与方言解析，因此保存不会留下任何会话都无法加载的文件。随部署提供的 preset 拒绝写入与删除，因为部署自带的那一份正是用来对照有问题的本地 preset 的——这也让“先复制、再编辑”成为创作路径本身，而非事后补充。

-**在 agent 平面之外还有消费方的服务，不能搬进 preset。** 激进拆分把 `subagents` 注册表连同 spawn/fork 后端一起搬进了 delegation 组的 entry-local realm，于是 `dsh web` 直接起不来：`dsh-host-apiproxy` 是宿主行，它注入 `subagents` 来回答浏览器的跨会话查询（`listChildren`、`followup`），因而永远等待一个此刻只有会话才提供的服务。按会话各一份在两个层面上都是错的——provider 名只能注册一次，第二个会话本来也会相撞。注册表与后端属于宿主平面；preset 贡献的是委派**工具**，它们解析宿主注册表。`workflows` 保持 entry-local，因为 agent 之外没有任何东西读它。本该拦下它的是「检索注入方」这一步，而它没拦住：检索必须覆盖宿主包，而不只是 agent 平面的包。
+**在 agent 平面之外还有消费方的服务，不能搬进 preset。** 激进拆分把 `subagents` 注册表连同 spawn/fork 后端一起搬进了 delegation 组的 entry-local realm，于是 `dsh web` 直接起不来：`dsh-host-apiproxy` 是宿主行，它注入 `subagents` 来回答浏览器的跨会话查询（`listChildren`、`followup`），因而永远等待一个此刻只有会话才提供的服务。按会话各一份在两个层面上都是错的——provider 名只能注册一次，第二个会话本来也会相撞。注册表与所有共享后端，包括固定的 Codex 与 Claude Code 产品 provider，都属于宿主平面；preset 只贡献自己的 agent 应看见的委派**工具**，这些工具解析宿主注册表。`workflows` 保持 entry-local，因为 agent 之外没有任何东西读它。本该拦下它的是「检索注入方」这一步，而它没拦住：检索必须覆盖宿主包，而不只是 agent 平面的包。

 **真实组装测试若禁用了某个宿主行，就无法审计该行。** web 组装测试把 `api-gateway`——也就是 api-proxy 本身——当作「有外部副作用的行」禁用了，而它恰恰是那个会以 pending 注入点名此次断裂的行。现在它在启用 api-proxy、并替换为 browse 目录选择器的前提下引导，启动审计因此覆盖整个宿主平面的注入图；只有端口、资源目录与遥测导出器仍然关闭。

@@ -79,3 +79,7 @@ Status: implemented
 **把 agent 的 scope 键设为 preset。** 同一 preset 上的会话就能免费共享一层，但按 agent 的注册——`installAgentLlmTarget`、按 agent 的工具限制——会跨会话相撞。

 **把每个 preset 作为子进程运行。** [`subagent-dsh-sdk`](../../../../packages/subagent/subagent-dsh-sdk/README.md) 已经证明完整的子 harness 可行，隔离性也会是绝对的。但这同时意味着要按会话代理流式输出、审批与投影，那是一个传输层项目，而非组装问题。
+
+**给产品 subagent 增加全局启用设置与独立设置页。** 进程级值会与 preset 争夺模型可见工具的所有权，也无法表达两个会话使用不同组装。产品 provider 留在宿主，普通 preset 行分别暴露 Codex 与 Claude Code 工具。
+
+**为 Codex 与 Claude Code 的每种组合交付一份 preset。** 四个身份会复制完整 preset 组装，只为表示两条独立行。复制后的 preset 已能直接启用任一行，因此组合 preset 只增加名单与维护成本，不增加用户结果。
@@ -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/2026-08-04-claude-code-and-codex-subagent-backends.md
-2026-08-04-claude-code-and-codex-subagent-backends.md: e81d1fb14f719331c503dba539d6a5ec0f1eed4f
-2026-08-04-claude-code-and-codex-subagent-backends.zh.md: aa6d80e38c12a2808a27e93bde8aec5d71e6a509
+2026-08-04-claude-code-and-codex-subagent-backends.md: cf41190c2d03965106e817b50fd2edb806e084bf
+2026-08-04-claude-code-and-codex-subagent-backends.zh.md: e0c9cfc96a68fe919a70af64b23ee7ed33ffcea3
@@ -12,7 +12,7 @@ The product integrations must not become second owners for task text, cwd, cance

 ## Decision

-The harness publishes two sibling one-shot providers as independently installable, opt-in packages. A user loads a provider and the existing common subagent tool in their own `cordis.yml`: `subagent_codex` binds `codex`, while `subagent_claude_code` binds `claude-code`. The shipped CLI dependency closure and base, Web, and headless configurations load neither provider. Each tool accepts only a standalone text task; product selection and background execution are not model arguments.
+The harness publishes two sibling one-shot providers in the shared profile host: `codex` and `claude-code`. Loading the host providers starts no product process. An Agent Preset independently contributes ordinary `dsh-tool-subagent` rows when its agent should see `subagent_codex`, `subagent_claude_code`, both, or neither; the shipped full presets carry both rows disabled so copies have one accurate configuration template without changing the default model schema. Each tool accepts only a standalone text task; product selection and background execution are not model arguments.

 Both providers report `inheritsParentContext: false`, advertise no optional start capabilities, and pass the parent Session cwd without copying the parent conversation. Their documented tools disable background execution and use `maxDepth: 'provider-managed'`, leaving recursion policy with the out-of-process product instead of sending a limit the provider cannot enforce. Every call creates a fresh product process and a non-resumable product conversation. The shared subagent service continues to own request resolution, lifecycle events, result settlement, and foreground collection; the shared subprocess service owns credential scrubbing, process-tree termination, and whole-tree exit observation.

@@ -47,7 +47,7 @@ Codex 0.146.0 speaks the Responses protocol, while DeepSeek's public OpenAI-comp

 ## Claude Code provider

-`@deepseek-ai/dsh-subagent-claude-code` registers the fixed `claude-code` provider and invokes `@anthropic-ai/claude-agent-sdk@0.3.220`. The SDK's platform `optionalDependency` supplies the real Claude Code 2.1.220 CLI. The provider uses the official `query()` entrypoint and passes the SDK's `spawnClaudeCodeProcess` command, arguments, cwd, environment, and forwarded signal unchanged to `dsh-subprocess`; its private `SpawnedProcess` adapter exposes only the stream, event, kill, and exit facts the SDK requires.
+`@deepseek-ai/dsh-subagent-claude-code` registers the fixed `claude-code` provider and invokes `@anthropic-ai/claude-agent-sdk@0.3.220`. Before each run, the provider resolves the fixed `claude` name through the host subprocess execution world and passes that exact path as `pathToClaudeCodeExecutable`; the SDK therefore uses the native product that launched DSH rather than selecting its platform `optionalDependency`. The provider uses the official `query()` entrypoint and passes the SDK's `spawnClaudeCodeProcess` arguments, cwd, environment, and forwarded signal to `dsh-subprocess`; its private `SpawnedProcess` adapter exposes only the stream, event, kill, and exit facts the SDK requires.

 The public configuration contains the same two deployment-owned values as the Codex sibling: an explicit `env` overlay and a positive finite `disposeGraceMs` no greater than the repository's shared `MAX_TIMER_DELAY_MS`. Each run creates its own `AbortController`, sets `persistSession: false`, and disables `AskUserQuestion`. The provider deliberately omits `settingSources`, so the SDK reads the host's normal user, project, and local Claude settings relative to the parent Session cwd. It neither copies nor filters those settings and does not create or modify login state. It supplies no `canUseTool`, elicitation, or dialog callback, so unattended interactions fail through the SDK rather than waiting for a user interface the provider does not own.

@@ -65,7 +65,7 @@ The Codex evidence pins `@openai/codex@0.146.0` and `codex-cli 0.146.0`. Its rea

 The Codex credentialed e2e registers the production provider, starts the same real app-server, and requests one random nonce through the test-private bridge described above. It fixes the external endpoint and model, stores no credential or request payload, requires exactly one completed upstream response, compares the trimmed product answer byte-for-byte with the nonce, and waits for every managed handle to exit.

-The Claude Code evidence pins Agent SDK 0.3.220 and its platform-distributed Claude Code 2.1.220 CLI. Its real-product spec observes the exact `x-api-key`, original task, byte-exact final answer, inherited temporary host-setting marker, process failure, local cancellation, and whole-tree exit. The Loader e2e resolves both product packages by name while neither product command is available and records zero child starts.
+The Claude Code evidence pins Agent SDK 0.3.220 and a native Claude Code installation compatible with its query protocol. Its real-product spec observes the exact `x-api-key`, original task, byte-exact final answer, inherited temporary host-setting marker, process failure, local cancellation, and whole-tree exit. The Loader and shipped-profile evidence resolve both product packages by name while starting neither product, and the provider suite proves that the SDK receives the executable resolved from the host `PATH`.

 The Claude Code credentialed e2e maps the key and fixed official endpoint only in the provider's in-memory environment, uses the documented `deepseek-v4-pro[1m]` and `deepseek-v4-flash` model variables, and traverses the production provider, official SDK, and real CLI. It compares the trimmed result with a random nonce and proves whole-tree exit without calling the Messages API directly from the test.

@@ -79,6 +79,10 @@ The project owner's distribution authorization is scoped to the official `@anthr

 **A model-visible product selector.** Product availability and authentication are deployment facts. Two fixed tools keep each schema and provider binding explicit and avoid adding dynamic selection state to the common service.

+**Global product enable settings and a product-specific Web page.** Those controls make Codex and Claude Code exceptions to the Agent Preset composition that already owns one agent's tool set, and one process-wide choice cannot represent two sessions using different presets. The host always supplies the providers; the preset alone decides which fixed tools its agent receives.
+
+**One shipped preset per product combination.** Four preset variants encode a two-boolean choice in preset identities and multiply every future standard-preset change. Independent ordinary rows express the same result in the user's copied preset without adding a roster taxonomy.
+
 **Product doubles as required evidence.** Doubles cover exhaustive private protocol branches but do not prove package exports, official distributions, authentication, or real process behavior. Required evidence drives each official product against a loopback model fixture.

 **Plugin-managed login, product home, models, settings, or permissions.** Those choices would create another authority beside each product's native configuration and enlarge a one-shot provider into account management. The providers expose only an explicit environment overlay and teardown grace; unattended interaction fails closed.
@@ -87,7 +91,7 @@ The project owner's distribution authorization is scoped to the official `@anthr

 ## Consequences

-Users can install either or both product providers, bind stable foreground tools in their own Cordis configuration, and delegate one self-contained task through the existing subagent contract. Official product integrations preserve native settings and behavior while shared services retain the sole ownership of task settlement and process-tree quiescence.
+Users copy or author an Agent Preset and independently enable either or both stable foreground tools. Every profile host supplies the reusable providers once, while each preset owns only its agent's model-visible tool rows. Official product integrations preserve native settings and behavior while shared services retain the sole ownership of task settlement and process-tree quiescence.

 Every delegation pays for a fresh product process and independent model context, and only final text reaches the parent. Product-native configuration makes behavior depend on the deployment's installed product, account state, and workspace settings. Credentialed e2e runs also spend external API quota and depend on the official DeepSeek endpoint; deterministic protocol, failure, cancellation, and approval coverage remains in the keyless tier. The providers do not resume sessions, stream progress, accept new human interaction, roll back tool or file side effects, or impose a wall-clock timeout.

@@ -12,7 +12,7 @@ Status: implemented

 ## 决策

-harness 将两个一次性兄弟提供方作为可独立安装、选择启用的包交付。用户在自己的 `cordis.yml` 中加载提供方与现有的通用 subagent 工具：`subagent_codex` 绑定 `codex`，`subagent_claude_code` 绑定 `claude-code`。随产品交付的 CLI（命令行界面）依赖闭包，以及基础、Web 与 headless 配置都不会加载任一提供方。每个工具只接受独立文本任务；产品选择与后台执行都不作为模型参数。
+harness 在共享 profile 宿主中交付两个一次性兄弟提供方：`codex` 与 `claude-code`。加载宿主提供方不会启动产品进程。某个 Agent Preset 是否让自己的 agent 看见 `subagent_codex`、`subagent_claude_code`、两者或两者皆无，由该 preset 独立贡献普通的 `dsh-tool-subagent` 行；随附的完整 preset 携带两条默认禁用的行，使复制品拥有一份准确配置模板，同时不改变默认模型 schema。每个工具只接受独立文本任务；产品选择与后台执行都不作为模型参数。

 这两个提供方都报告 `inheritsParentContext: false`，不声明任何可选的启动能力，并传递父会话 cwd，但不会复制父级对话。文档所示的工具会禁用后台执行，并使用 `maxDepth: 'provider-managed'`，将递归策略留给进程外产品，而不是发送提供方无法强制执行的限制。每次调用都会创建一个全新的产品进程和一次不可续接的产品对话。共享 subagent 服务继续负责请求解析、生命周期事件、结果结算和前台收集；共享子进程服务负责凭证清洗、进程树终止以及整棵进程树的退出观测。

@@ -47,7 +47,7 @@ Codex 0.146.0 使用 Responses 协议，而 DeepSeek 的公开 OpenAI 兼容端

 ## Claude Code 提供方

-`@deepseek-ai/dsh-subagent-claude-code` 注册固定的 `claude-code` 提供方，并调用 `@anthropic-ai/claude-agent-sdk@0.3.220`。SDK 的平台 `optionalDependency` 提供真实的 Claude Code 2.1.220 CLI。提供方使用官方 `query()` 入口点，并将 SDK 的 `spawnClaudeCodeProcess` 命令、参数、cwd、环境和转发的信号原样传入 `dsh-subprocess`；其私有 `SpawnedProcess` 适配器只公开 SDK 所需的流、事件、终止和退出事实。
+`@deepseek-ai/dsh-subagent-claude-code` 注册固定的 `claude-code` 提供方，并调用 `@anthropic-ai/claude-agent-sdk@0.3.220`。每次运行前，提供方经宿主 subprocess 执行世界解析固定名称 `claude`，并把准确路径作为 `pathToClaudeCodeExecutable` 交给 SDK；SDK 因此使用启动 DSH 的原生产品，而不是选择自身的 platform `optionalDependency`。提供方使用官方 `query()` 入口点，并将 SDK 的 `spawnClaudeCodeProcess` 参数、cwd、环境和转发的信号交给 `dsh-subprocess`；其私有 `SpawnedProcess` 适配器只公开 SDK 所需的流、事件、终止和退出事实。

 公开配置包含与 Codex 兄弟提供方相同、由部署方负责的两个值：显式的 `env` 覆盖项，以及须为正有限值且不得大于仓库共享 `MAX_TIMER_DELAY_MS` 的 `disposeGraceMs`。每次运行都会创建自己的 `AbortController`，设置 `persistSession: false` 并禁用 `AskUserQuestion`。提供方故意省略 `settingSources`，因此 SDK 会相对于父会话 cwd 读取宿主机常规的用户、项目和本地 Claude 设置。它既不复制也不过滤这些设置，也不会创建或修改登录状态。提供方不设置 `canUseTool`、elicitation 或对话回调，因此无人值守交互会经 SDK 失败，而不会等待本提供方不负责的用户界面。

@@ -65,7 +65,7 @@ Codex 证据锁定 `@openai/codex@0.146.0` 与 `codex-cli 0.146.0`。其真实

 带密钥 Codex e2e 会注册生产提供方，启动同样的真实 app-server，并通过上述测试专用桥接层请求一个随机数。该测试固定外部端点与模型，不存储任何凭据或请求载荷，要求上游恰好完成一次响应，将去除首尾空白后的产品答案与该随机数逐字节比较，并等待所有受管句柄退出。

-Claude Code 证据锁定 Agent SDK 0.3.220 及其平台分发的 Claude Code 2.1.220 CLI。其真实产品测试会观测确切的 `x-api-key`、原始任务、逐字节完全一致的最终回答、继承的临时宿主设置标记、进程失败、本地取消以及整棵进程树退出。Loader e2e 会在两个产品命令均不可用时按名称解析两个产品包，并记录零次子级启动。
+Claude Code 证据锁定 Agent SDK 0.3.220 及与其查询协议兼容的原生 Claude Code 安装。其真实产品测试会观测确切的 `x-api-key`、原始任务、逐字节完全一致的最终回答、继承的临时宿主设置标记、进程失败、本地取消以及整棵进程树退出。Loader 与随附 profile 证据会按名称解析两个产品包且不启动产品，provider 测试则证明 SDK 收到由宿主 `PATH` 解析出的可执行文件。

 带密钥 Claude Code e2e 仅在提供方的内存环境中映射密钥与固定的官方端点，把模型变量设为文档所示的 `deepseek-v4-pro[1m]` 与 `deepseek-v4-flash`，并实际经过生产提供方、官方 SDK 与真实 CLI。它将去除首尾空白后的结果与一个随机数比较，并证明整棵进程树退出，且测试不会直接调用 Messages API。

@@ -79,6 +79,10 @@ Claude Code 证据锁定 Agent SDK 0.3.220 及其平台分发的 Claude Code 2.1

 **面向模型的产品选择器。** 产品可用性和身份验证属于部署事实。两个固定工具使各自的 schema 与提供方绑定保持明确，也避免在通用服务中添加动态选择状态。

+**全局产品启用设置与产品专属 Web 页面。** 这类控制会让 Codex 与 Claude Code 成为 Agent Preset 组装的例外，而后者本就拥有单个 agent 的工具集；一个进程级选择也无法表达两个会话使用不同 preset。宿主始终提供 provider，只有 preset 决定其 agent 获得哪些固定工具。
+
+**为每种产品组合交付一份 preset。** 四个 preset 变体把两个布尔选择编码成 preset 身份，并让未来每次标准 preset 修改都要同步多份副本。用户复制的 preset 中两条独立普通行已经能表达同一结果，无需新增名单分类。
+
 **以产品替身作为强制证据。** 替身可以穷尽覆盖私有协议分支，但无法证明包导出、官方发行版、身份验证或真实进程行为。强制证据会驱动每个官方产品连接回环模型 fixture（测试前置数据）。

 **由插件管理登录、产品主目录、模型、设置或权限。** 这些选择会在每个产品的原生配置之外建立另一套权威来源，并将一次性提供方扩张为账户管理功能。提供方只公开显式环境覆盖项和清理宽限期；无人值守交互会以默认拒绝方式失败。
@@ -87,7 +91,7 @@ Claude Code 证据锁定 Agent SDK 0.3.220 及其平台分发的 Claude Code 2.1

 ## 后果

-用户可以安装任一或两个产品提供方，在自己的 Cordis 配置中绑定稳定的前台工具，并通过现有 subagent 约定委派一项自包含任务。官方产品集成会保留原生设置与行为，而共享服务继续独占任务结算与进程树完全停稳的责任。
+用户可以复制或创作一个 Agent Preset，并分别启用任一或两个稳定前台工具。每个 profile 宿主只提供一次可复用 provider，而每个 preset 只拥有自己 agent 的模型可见工具行。官方产品集成会保留原生设置与行为，而共享服务继续独占任务结算与进程树完全停稳的责任。

 每次委派都要承担新建产品进程和独立模型上下文的开销，且只有最终文本会到达父级。产品原生配置使行为取决于部署环境中安装的产品、账户状态和工作区设置。带密钥 e2e 运行还会消耗外部 API 配额，并依赖 DeepSeek 官方端点；对协议、失败、取消与审批的确定性覆盖仍由无密钥层级承担。提供方不会恢复会话、以流式方式传送进度、接受新的人工交互、回滚工具或文件副作用，也不会施加按实际经过时间触发的超时。

@@ -110,6 +110,10 @@ flowchart LR
  cfg --> plugin_dsh_base_subagent_spawn
  plugin_dsh_base_subagent_fork["subagent-fork<br/>@deepseek-ai/dsh-subagent-fork"]
  cfg --> plugin_dsh_base_subagent_fork
+  plugin_dsh_base_subagent_codex["subagent-codex<br/>@deepseek-ai/dsh-subagent-codex"]
+  cfg --> plugin_dsh_base_subagent_codex
+  plugin_dsh_base_subagent_claude_code["subagent-claude-code<br/>@deepseek-ai/dsh-subagent-claude-code"]
+  cfg --> plugin_dsh_base_subagent_claude_code
  plugin_dsh_base_tool_subagent_control["tool-subagent-control<br/>@deepseek-ai/dsh-tool-subagent-control"]
  cfg --> plugin_dsh_base_tool_subagent_control
  plugin_dsh_base_tool_subagent_list_agents["tool-subagent-list-agents<br/>@deepseek-ai/dsh-tool-subagent-control/list-agents"]
@@ -215,6 +219,8 @@ flowchart LR
 | `subagent` | `@deepseek-ai/dsh-subagent` |
 | `subagent-spawn` | `@deepseek-ai/dsh-subagent-spawn` |
 | `subagent-fork` | `@deepseek-ai/dsh-subagent-fork` |
+| `subagent-codex` | `@deepseek-ai/dsh-subagent-codex` |
+| `subagent-claude-code` | `@deepseek-ai/dsh-subagent-claude-code` |
 | `tool-subagent-control` | `@deepseek-ai/dsh-tool-subagent-control` |
 | `tool-subagent-list-agents` | `@deepseek-ai/dsh-tool-subagent-control/list-agents` |
 | `tool-subagent` | `@deepseek-ai/dsh-tool-subagent` |
@@ -197,6 +197,27 @@
        toolName: subagent_fork
        backgroundMode: continuable

+    # Product providers are host-plane singletons. Copy this preset, then
+    # remove `disabled` from either ordinary tool row to expose that product
+    # only to agents composed from the copy.
+    - id: tool-subagent-codex
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: codex
+        toolName: subagent_codex
+        enableRunInBackground: false
+        maxDepth: provider-managed
+
+    - id: tool-subagent-claude-code
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: claude-code
+        toolName: subagent_claude_code
+        enableRunInBackground: false
+        maxDepth: provider-managed
+
    - id: workflow-workerthread
      name: '@deepseek-ai/dsh-workflow-workerthread'
      config:
@@ -184,6 +184,27 @@
        toolName: subagent_fork
        backgroundMode: continuable

+    # Product providers are host-plane singletons. Copy this preset, then
+    # remove `disabled` from either ordinary tool row to expose that product
+    # only to agents composed from the copy.
+    - id: tool-subagent-codex
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: codex
+        toolName: subagent_codex
+        enableRunInBackground: false
+        maxDepth: provider-managed
+
+    - id: tool-subagent-claude-code
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: claude-code
+        toolName: subagent_claude_code
+        enableRunInBackground: false
+        maxDepth: provider-managed
+
    - id: workflow-workerthread
      name: '@deepseek-ai/dsh-workflow-workerthread'
      config:
@@ -26,6 +26,34 @@ A preset is a directory holding one `agent.cordis.yml`, optionally beside a `pre
 3. **Rewrite `preset.yml`**: give the copy its own `name` and `description`, and drop any `order` the source declared — that field sorts the shipped roster.
 4. **Edit `agent.cordis.yml`** row by row, keeping the plane rule and realm rule above.

+### Native product subagents
+
+Codex and Claude Code providers already live in the host composition. A preset chooses either product by contributing the same ordinary delegation-tool row used for spawn and fork; never move a product provider into the preset and never add a product-specific settings field.
+
+Copy these disabled templates from a shipped full preset and remove `disabled` only for the products the user requested:
+
+```yaml
+- id: tool-subagent-codex
+  name: '@deepseek-ai/dsh-tool-subagent'
+  disabled: true
+  config:
+    provider: codex
+    toolName: subagent_codex
+    enableRunInBackground: false
+    maxDepth: provider-managed
+
+- id: tool-subagent-claude-code
+  name: '@deepseek-ai/dsh-tool-subagent'
+  disabled: true
+  config:
+    provider: claude-code
+    toolName: subagent_claude_code
+    enableRunInBackground: false
+    maxDepth: provider-managed
+```
+
+The two rows are independent. Leaving both disabled preserves the copied preset, enabling one exposes only that product tool, and enabling both exposes both. The host must provide `codex` or `claude` on `PATH`; the preset does not install, authenticate, select a model for, or probe either product.
+
 The shipped preset directories are off-limits: never edit or delete them, and never escalate the sandbox to reach them, even when a change there looks quicker — an upgrade overwrites the install, and corrupting the `cordis` preset disables preset authoring itself. Locally authored presets under the user root are yours to create, edit, and delete.

 ## The rule that catches people
@@ -196,6 +196,27 @@
        toolName: subagent_fork
        backgroundMode: continuable

+    # Product providers are host-plane singletons. Copy this preset, then
+    # remove `disabled` from either ordinary tool row to expose that product
+    # only to agents composed from the copy.
+    - id: tool-subagent-codex
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: codex
+        toolName: subagent_codex
+        enableRunInBackground: false
+        maxDepth: provider-managed
+
+    - id: tool-subagent-claude-code
+      name: '@deepseek-ai/dsh-tool-subagent'
+      disabled: true
+      config:
+        provider: claude-code
+        toolName: subagent_claude_code
+        enableRunInBackground: false
+        maxDepth: provider-managed
+
    - id: workflow-workerthread
      name: '@deepseek-ai/dsh-workflow-workerthread'
      config:
@@ -8,7 +8,7 @@ import { boot, healProfilesModuleFallback, loadOverlayPatches } from '@deepseek-
 import { SessionId } from '@deepseek-ai/dsh-session'
 import type { Agent } from '@deepseek-ai/dsh-agent'
 import type { PatchOptions } from '@cordisjs/plugin-include'
-import { beforeAll, describe, expect, it } from 'vitest'
+import { afterAll, beforeAll, describe, expect, it } from 'vitest'
 import { settingsNamespace } from '@deepseek-ai/dsh-settings'
 import { resolveSessionPreset, SETTINGS_NAMESPACE } from '@deepseek-ai/dsh-agent-presets'
 import { CallId } from '@deepseek-ai/dsh-llm'
@@ -95,6 +95,18 @@ async function bootWeb(settingsFile: string, extra: PatchOptions[] = []): Promis
 const toolNames = (ctx: Context, agent?: Agent): string[] =>
  ctx.tools.schemas(agent).map(schema => schema.name).sort()

+function enablePresetTool(composition: string, id: string): string {
+  const row = `    - id: ${id}\n`
+  const start = composition.indexOf(row)
+  if (start < 0) throw new Error(`missing preset row ${id}`)
+  const end = composition.indexOf('\n    - id:', start + row.length)
+  const disabled = composition.indexOf('      disabled: true\n', start)
+  if (disabled < 0 || (end >= 0 && disabled > end)) {
+    throw new Error(`preset row ${id} is not disabled`)
+  }
+  return composition.slice(0, disabled) + composition.slice(disabled + '      disabled: true\n'.length)
+}
+
 let ctx: Context
 beforeAll(async () => {
  const settingsFile = join(await mkdtemp(join(tmpdir(), 'dsh-web-presets-')), 'settings.yaml')
@@ -120,6 +132,24 @@ describe('the shipped Web composition', () => {
    expect(ctx.agentPresets.defaultId).toBe('standard')
  })

+  it('keeps product providers on the host while shipped presets leave their tools disabled', async () => {
+    expect(ctx.subagents.list()).toEqual(expect.arrayContaining([
+      'spawn', 'fork', 'codex', 'claude-code',
+    ]))
+
+    const handle = await ctx.agents.create({
+      sessionId: SessionId('preset-products-disabled'),
+      setup: agentCtx => ctx.agentPresets.mount(agentCtx, 'standard').then(() => undefined),
+    })
+    try {
+      expect(toolNames(ctx, handle.agent)).not.toEqual(expect.arrayContaining([
+        'subagent_codex', 'subagent_claude_code',
+      ]))
+    } finally {
+      await handle.dispose()
+    }
+  })
+
  it('composes the full agent from `standard`', async () => {
    const handle = await ctx.agents.create({
      sessionId: SessionId('preset-standard'),
@@ -355,6 +385,98 @@ describe('the shipped Web composition', () => {
  })
 })

+describe('product subagent rows in user presets', () => {
+  let productCtx: Context
+  const ids = ['products-none', 'products-codex', 'products-claude', 'products-both'] as const
+
+  beforeAll(async () => {
+    const root = await mkdtemp(join(tmpdir(), 'dsh-product-presets-'))
+    const userRoot = join(root, 'presets')
+    const settingsFile = join(root, 'settings.yaml')
+    const standard = await readFile(join(CONFIG_DIR, 'agent-presets', 'standard', 'agent.cordis.yml'), 'utf8')
+    await writeFile(settingsFile, '{}\n')
+    for (const id of ids) {
+      let composition = standard
+      if (id === 'products-codex' || id === 'products-both') {
+        composition = enablePresetTool(composition, 'tool-subagent-codex')
+      }
+      if (id === 'products-claude' || id === 'products-both') {
+        composition = enablePresetTool(composition, 'tool-subagent-claude-code')
+      }
+      const directory = join(userRoot, id)
+      await mkdir(directory, { recursive: true })
+      await writeFile(join(directory, 'agent.cordis.yml'), composition)
+    }
+    productCtx = await bootWeb(settingsFile, [{
+      id: 'agent-presets',
+      config: {
+        default: 'standard',
+        roots: [
+          { path: join(CONFIG_DIR, 'agent-presets'), trust: 'system' },
+          { path: userRoot, trust: 'user' },
+        ],
+      },
+    }])
+  }, 120_000)
+
+  afterAll(async () => {
+    await productCtx.fiber.dispose()
+  })
+
+  it('composes none, either product, or both without changing the shared host registry', async () => {
+    const expected = new Map<string, string[]>([
+      ['products-none', []],
+      ['products-codex', ['subagent_codex']],
+      ['products-claude', ['subagent_claude_code']],
+      ['products-both', ['subagent_claude_code', 'subagent_codex']],
+    ])
+    expect(productCtx.subagents.list()).toEqual(expect.arrayContaining([
+      'spawn', 'fork', 'codex', 'claude-code',
+    ]))
+
+    for (const [id, productTools] of expected) {
+      const handle = await productCtx.agents.create({
+        sessionId: SessionId(`preset-${id}`),
+        setup: agentCtx => productCtx.agentPresets.mount(agentCtx, id).then(() => undefined),
+      })
+      try {
+        const tools = toolNames(productCtx, handle.agent)
+        expect(tools.filter(name => name === 'subagent_codex' || name === 'subagent_claude_code'))
+          .toEqual(productTools)
+      } finally {
+        await handle.dispose()
+      }
+    }
+  })
+
+  it('applies a product-row edit only to later sessions on the preset', async () => {
+    const preset = await productCtx.agentPresets.resolve('products-none')
+    const original = await readFile(preset.path, 'utf8')
+    const existing = await productCtx.agents.create({
+      sessionId: SessionId('preset-product-generation-existing'),
+      setup: agentCtx => productCtx.agentPresets.mount(agentCtx, 'products-none').then(() => undefined),
+    })
+    try {
+      expect(toolNames(productCtx, existing.agent)).not.toContain('subagent_codex')
+      await writeFile(preset.path, enablePresetTool(original, 'tool-subagent-codex'))
+
+      const later = await productCtx.agents.create({
+        sessionId: SessionId('preset-product-generation-later'),
+        setup: agentCtx => productCtx.agentPresets.mount(agentCtx, 'products-none').then(() => undefined),
+      })
+      try {
+        expect(toolNames(productCtx, existing.agent)).not.toContain('subagent_codex')
+        expect(toolNames(productCtx, later.agent)).toContain('subagent_codex')
+      } finally {
+        await later.dispose()
+      }
+    } finally {
+      await existing.dispose()
+      await writeFile(preset.path, original)
+    }
+  })
+})
+
 describe('a switch survives the session', () => {
  it('records the choice so the log states what the agent runs', async () => {
    const handle = await ctx.agents.create({
@@ -0,0 +1,38 @@
+# Keyless twin of product-subagent-both.cordis.yml: preserve both product
+# tools while replacing only the external model adapter.
+- id: base
+  name: '@cordisjs/plugin-include'
+  config:
+    path: ./cordis.yml
+    patches:
+      - id: llm-deepseek
+        name: '@deepseek-ai/dsh-llm-deepseek'
+        disabled: true
+      - insert:
+          - id: llm-replay
+            name: '@deepseek-ai/dsh-llm-replay'
+            config:
+              providers:
+                - id: deepseek-official
+                  name: DeepSeek
+                  models:
+                    - id: deepseek-v4-flash
+                    - id: deepseek-v4-pro
+          - id: subagent-codex
+            name: '@deepseek-ai/dsh-subagent-codex'
+          - id: subagent-claude-code
+            name: '@deepseek-ai/dsh-subagent-claude-code'
+          - id: tool-subagent-codex
+            name: '@deepseek-ai/dsh-tool-subagent'
+            config:
+              provider: codex
+              toolName: subagent_codex
+              enableRunInBackground: false
+              maxDepth: provider-managed
+          - id: tool-subagent-claude-code
+            name: '@deepseek-ai/dsh-tool-subagent'
+            config:
+              provider: claude-code
+              toolName: subagent_claude_code
+              enableRunInBackground: false
+              maxDepth: provider-managed
@@ -0,0 +1,27 @@
+# Add both native product providers and the same independent foreground tool
+# rows an Agent Preset may contribute. Loading the composition starts neither
+# product; the scenario pins both model-visible schemas.
+- id: base
+  name: '@cordisjs/plugin-include'
+  config:
+    path: ./cordis.yml
+    patches:
+      - insert:
+          - id: subagent-codex
+            name: '@deepseek-ai/dsh-subagent-codex'
+          - id: subagent-claude-code
+            name: '@deepseek-ai/dsh-subagent-claude-code'
+          - id: tool-subagent-codex
+            name: '@deepseek-ai/dsh-tool-subagent'
+            config:
+              provider: codex
+              toolName: subagent_codex
+              enableRunInBackground: false
+              maxDepth: provider-managed
+          - id: tool-subagent-claude-code
+            name: '@deepseek-ai/dsh-tool-subagent'
+            config:
+              provider: claude-code
+              toolName: subagent_claude_code
+              enableRunInBackground: false
+              maxDepth: provider-managed
@@ -0,0 +1,29 @@
+# Keyless twin of product-subagent-codex.cordis.yml: keep the same product
+# provider/tool composition and replace only the external model adapter.
+- id: base
+  name: '@cordisjs/plugin-include'
+  config:
+    path: ./cordis.yml
+    patches:
+      - id: llm-deepseek
+        name: '@deepseek-ai/dsh-llm-deepseek'
+        disabled: true
+      - insert:
+          - id: llm-replay
+            name: '@deepseek-ai/dsh-llm-replay'
+            config:
+              providers:
+                - id: deepseek-official
+                  name: DeepSeek
+                  models:
+                    - id: deepseek-v4-flash
+                    - id: deepseek-v4-pro
+          - id: subagent-codex
+            name: '@deepseek-ai/dsh-subagent-codex'
+          - id: tool-subagent-codex
+            name: '@deepseek-ai/dsh-tool-subagent'
+            config:
+              provider: codex
+              toolName: subagent_codex
+              enableRunInBackground: false
+              maxDepth: provider-managed
@@ -0,0 +1,18 @@
+# Add the native Codex product provider and its preset-shaped foreground tool to
+# the real ACP composition. The model is told not to call it; the scenario pins
+# the assembled request schema without starting Codex.
+- id: base
+  name: '@cordisjs/plugin-include'
+  config:
+    path: ./cordis.yml
+    patches:
+      - insert:
+          - id: subagent-codex
+            name: '@deepseek-ai/dsh-subagent-codex'
+          - id: tool-subagent-codex
+            name: '@deepseek-ai/dsh-tool-subagent'
+            config:
+              provider: codex
+              toolName: subagent_codex
+              enableRunInBackground: false
+              maxDepth: provider-managed
@@ -52,6 +52,8 @@ const WEB_CONFIG = fileURLToPath(new URL('../web.cordis.yml', import.meta.url))
 const FS_SEARCH_CONFIG = fileURLToPath(new URL('./fs-search.cordis.yml', import.meta.url))
 const PARTIAL_LANDLOCK_CONFIG = fileURLToPath(new URL('../partial-landlock.cordis.yml', import.meta.url))
 const PWSH_CONFIG = fileURLToPath(new URL('./pwsh.cordis.yml', import.meta.url))
+const PRODUCT_SUBAGENT_CODEX_CONFIG = fileURLToPath(new URL('../product-subagent-codex.cordis.yml', import.meta.url))
+const PRODUCT_SUBAGENT_BOTH_CONFIG = fileURLToPath(new URL('../product-subagent-both.cordis.yml', import.meta.url))
 const SNAPSHOTS_DIR = join(dirname(fileURLToPath(import.meta.url)), 'snapshots')
 const PACKED_CHUNKS_SOURCE = 'hook-cc-pretool-deny'

@@ -122,6 +124,23 @@ const SCENARIOS: Scenario[] = [
  // text-turn is the default header pin and owns the prompt and tool-schema
  // sidecars reused by alternate classes with identical component sequences.
  { name: 'text-turn', hasModelTurn: true, recorded: true, pinsHeader: true },
+  {
+    name: 'product-subagent-codex',
+    hasModelTurn: true,
+    recorded: false,
+    pinsHeader: true,
+    headerClass: 'product-subagent-codex',
+    configPath: PRODUCT_SUBAGENT_CODEX_CONFIG,
+  },
+  {
+    name: 'product-subagent-both',
+    hasModelTurn: true,
+    recorded: false,
+    pinsHeader: true,
+    headerClass: 'product-subagent-both',
+    systemPromptSource: 'product-subagent-codex',
+    configPath: PRODUCT_SUBAGENT_BOTH_CONFIG,
+  },
  {
    name: 'session-title-after-turn',
    hasModelTurn: true,
@@ -0,0 +1,7 @@
+{
+  "steps": [
+    { "op": "initialize" },
+    { "op": "newSession" },
+    { "op": "prompt", "text": "Reply with exactly the word: PONG. Do not use any tools." }
+  ]
+}
@@ -0,0 +1,22 @@
+{"type":"session","version":0,"id":"539aa64c-7f37-40ff-abd8-ed45b717be1b","createdAt":1783600629539,"cwd":"{{cwd}}","delegationDepth":0}
+{"type":"agent/inbox/spliced","seq":0,"time":1785498761270,"data":{"target":"next-turn","start":0,"inserted":[{"content":[{"type":"text","text":"Reply with exactly the word: PONG. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"3e25dc34-48e0-4738-8401-1a8d181d37e5"}]}}
+{"type":"turn/start","seq":1,"time":1785821359466,"data":{"turn":1}}
+{"type":"agent/inbox/spliced","seq":2,"time":1785821359466,"data":{"target":"next-turn","start":0,"removedCount":1,"inserted":[]}}
+{"type":"step/start","seq":3,"time":1783600629542,"data":{"turn":1,"step":1}}
+{"type":"user/message","seq":4,"time":1785498761313,"data":{"content":[{"type":"text","text":"Reply with exactly the word: PONG. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"3e25dc34-48e0-4738-8401-1a8d181d37e5"},"surfaceOp":"append"}
+{"type":"user/message","seq":5,"time":1785730415287,"data":{"content":[{"type":"text","text":"Current runtime context. This snapshot supersedes earlier runtime-context snapshots.\n\nCurrent DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations.\n\nApproval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}],"source":{"kind":"plugin","plugin":"@deepseek-ai/dsh-system-prompt","form":"snapshot","sections":[{"name":"sandbox:policy","text":"Current DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations."},{"name":"approval:policy","text":"Approval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}]},"role":"user","id":"4b8d9730-0b7b-4e14-8a30-3d852f808f0e"},"surfaceOp":"append"}
+{"type":"session/title","seq":6,"time":1785730415287,"data":{"title":"Reply with exactly the word:","messageSeqs":[4],"source":{"kind":"fallback"}}}
+{"type":"request/header","seq":7,"time":1785498761318,"data":{"header":{"config":{"provider":"deepseek-official","model":"deepseek-v4-pro"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}}
+{"type":"request/context","seq":8,"time":1785730415288,"data":{"provider":"deepseek-official","model":"deepseek-v4-pro"}}
+{"type":"assistant/chunk","seq":9,"time":1783600630822,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
+{"type":"reasoning-chunks","seq0":10,"time0":1783600630852,"data":{"turn":1,"step":1,"index":0,"dt":[0,0,0,33,1,40,0,0,0,0,0,18,0,36,0,0,0,0,0],"texts":["The"," user"," wants"," me"," to"," reply"," with"," exactly"," the"," word"," \"","P","ONG","\""," and"," not"," use"," any"," tools","."]}}
+{"type":"assistant/chunk","seq":30,"time":1783600631006,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"text"}}}
+{"type":"assistant/chunk","seq":31,"time":1783600631008,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":1,"text":"P"}}}
+{"type":"assistant/chunk","seq":32,"time":1783600631009,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":1,"text":"ONG"}}}
+{"type":"assistant/chunk","seq":33,"time":1783600631009,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"reasoning","text":"The user wants me to reply with exactly the word \"PONG\" and not use any tools."}}}}
+{"type":"assistant/chunk","seq":34,"time":1783600631009,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"PONG"}}}}
+{"type":"assistant/chunk","seq":35,"time":1785498761338,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":3091,"outputTokens":23,"cacheReadTokens":0,"reasoningTokens":20}}}}
+{"type":"assistant/chunk","seq":36,"time":1785730415297,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
+{"type":"assistant/message","seq":37,"time":1785730415298,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to reply with exactly the word \"PONG\" and not use any tools."},{"type":"text","text":"PONG"}],"source":{"kind":"model","provider":"deepseek-official","model":"deepseek-v4-pro"},"id":"f1418376-f303-4017-acd7-92899c841c8a"},"usage":{"inputTokens":3091,"outputTokens":23,"cacheReadTokens":0,"reasoningTokens":20}},"sourceEventSeqs":[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],"surfaceOp":"append"}
+{"type":"step/end","seq":38,"time":1785730415298,"data":{"turn":1,"step":1}}
+{"type":"turn/end","seq":39,"time":1785730415298,"data":{"turn":1,"reason":{"kind":"completed"}}}
@@ -0,0 +1,4 @@
+{"jsonrpc":"2.0","id":1,"result":{"protocolVersion":1,"agentInfo":{"name":"deepseek-harness-acp","version":"0.0.1"},"agentCapabilities":{"promptCapabilities":{"image":false,"audio":false,"embeddedContext":false}},"authMethods":[]}}
+{"jsonrpc":"2.0","id":2,"result":{"sessionId":"{{sessionId}}"}}
+{"jsonrpc":"2.0","method":"session/update","params":{"sessionId":"{{sessionId}}","update":{"sessionUpdate":"agent_message_chunk","content":{"type":"text","text":"PONG"}}}}
+{"jsonrpc":"2.0","id":3,"result":{"stopReason":"end_turn"}}
@@ -0,0 +1,569 @@
+{
+  "initial": [
+    {
+      "name": "bash",
+      "description": "Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. Attempting a command the sandbox may deny is safe and expected: run it and read the marker rather than assuming the denial. When a command is denied and a wider mode would let it succeed, escalate immediately in the same turn — the one sanctioned exception to a denial: retry the exact same command once with `sandbox_permissions` (the narrowest wider mode that suffices) plus a one-sentence `justification`. Do not detour through chat to ask permission first — the approval prompt raised by that retry is how the user consents. If the session states approval prompts are disabled, there is no exception: a denial is final — do not set `sandbox_permissions`. Never escalate speculatively: ground the request in a real denial — normally the one this command just hit; escalating up front is fine only when this session already denied the same access. A rejected escalation is final for that command — stop and explain, never work around it — but it does not forbid attempting or escalating other commands later.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "command": {
+            "type": "string",
+            "description": "The bash command to execute."
+          },
+          "description": {
+            "type": "string",
+            "description": "Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."
+          },
+          "timeoutMs": {
+            "type": "number",
+            "description": "Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."
+          },
+          "workdir": {
+            "type": "string",
+            "description": "Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."
+          },
+          "run_in_background": {
+            "type": "boolean",
+            "description": "Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."
+          },
+          "sandbox_permissions": {
+            "type": "string",
+            "description": "The wider sandbox mode this command needs. Only valid as a one-shot retry of a command the sandbox just denied; requires justification and user approval.",
+            "enum": [
+              "workspace-write",
+              "danger-full-access"
+            ]
+          },
+          "justification": {
+            "type": "string",
+            "description": "Required with sandbox_permissions: one sentence for the user explaining why this exact command needs the wider access."
+          }
+        },
+        "required": [
+          "command",
+          "description"
+        ]
+      }
+    },
+    {
+      "name": "create_goal",
+      "description": "Create one persisted same-session completion goal when the current direct human request is a long-running objective that should continue across autonomous goal rounds. You may infer that intent without requiring the user to say \"create a goal\". Do not use this for trivial single-turn work. Execution rejects non-human and subagent authority.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "objective": {
+            "type": "string",
+            "description": "The concrete completion objective inferred from the direct human request."
+          },
+          "max_goal_rounds": {
+            "type": "number",
+            "description": "Optional positive safe-integer limit on automatic continuation rounds."
+          }
+        },
+        "required": [
+          "objective"
+        ]
+      }
+    },
+    {
+      "name": "edit",
+      "description": "Edit an existing UTF-8 text file by replacing literal text.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "file_path": {
+            "type": "string",
+            "description": "Path to edit, resolved by the filesystem backend."
+          },
+          "old_string": {
+            "type": "string",
+            "description": "Literal text to replace. Must match exactly."
+          },
+          "new_string": {
+            "type": "string",
+            "description": "Literal replacement text. Use an empty string to delete the match."
+          },
+          "replace_all": {
+            "type": "boolean",
+            "description": "Replace all matches. Defaults to false; when false, old_string must appear exactly once."
+          },
+          "sandbox_permissions": {
+            "type": "string",
+            "description": "The wider sandbox mode this file operation needs. Only valid as a one-shot retry of an operation the sandbox just denied; requires justification and user approval.",
+            "enum": [
+              "workspace-write",
+              "danger-full-access"
+            ]
+          },
+          "justification": {
+            "type": "string",
+            "description": "Required with sandbox_permissions: one sentence for the user explaining why this exact file operation needs the wider access."
+          }
+        },
+        "required": [
+          "file_path",
+          "old_string",
+          "new_string"
+        ]
+      }
+    },
+    {
+      "name": "get_goal",
+      "description": "Read the current same-session goal, including its exact id/revision, objective, phase, completed continuation rounds, round limit, blocker reason when present, and whether another continuation is armed. Call this before updating a goal.",
+      "parameters": {
+        "type": "object",
+        "properties": {}
+      }
+    },
+    {
+      "name": "interrupt_agent",
+      "description": "Request cancellation of a background agent's current turn by its agent id. The target may be your direct child or a deeper agent created under you. Only the current turn stops: messages already queued for the agent stay parked until a later send_message, agents it started keep running, and the agent itself stays available for follow-ups. This call returns as soon as the stop request is accepted, so the target may keep running briefly; interrupting an agent that already finished is an accepted no-op.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "agent_id": {
+            "type": "string",
+            "description": "The agent id of the running agent to interrupt."
+          }
+        },
+        "required": [
+          "agent_id"
+        ]
+      }
+    },
+    {
+      "name": "list_agents",
+      "description": "List your continuable background subagents by durable id and label. Status comes from the live registry: running means the agent is working right now, idle means it is loaded but between turns (it may be waiting on agents it started), and complete means it exists only in storage — a direct child remains a `send_message` candidate in every status. The snapshot is not a delivery promise — `send_message` performs the authoritative check and may still fail. Children that could not be read are reported as diagnostics instead of being silently dropped. Scope `descendants` walks the whole tree below you in stable pre-order, annotating each entry with its durable direct-parent session id and depth. You may use `send_message` only for depth-1 entries; deeper entries are candidates for `interrupt_agent` only.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "scope": {
+            "type": "string",
+            "description": "children (default) lists direct children only; descendants walks the complete tree below you.",
+            "enum": [
+              "children",
+              "descendants"
+            ]
+          }
+        }
+      }
+    },
+    {
+      "name": "ralph",
+      "description": "Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "objective": {
+            "type": "string",
+            "description": "The immutable completion objective for every fresh Ralph round."
+          },
+          "maxRounds": {
+            "type": "number",
+            "description": "Optional positive safe-integer round cap, bounded by the deployment ceiling."
+          }
+        },
+        "required": [
+          "objective"
+        ]
+      }
+    },
+    {
+      "name": "read",
+      "description": "Read a UTF-8 text file and return line-numbered content.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "file_path": {
+            "type": "string",
+            "description": "Path to read, resolved by the filesystem backend."
+          },
+          "offset": {
+            "type": "number",
+            "description": "1-based first line to return. Defaults to 1."
+          },
+          "limit": {
+            "type": "number",
+            "description": "Maximum number of lines to return. Defaults to 2000."
+          }
+        },
+        "required": [
+          "file_path"
+        ]
+      }
+    },
+    {
+      "name": "send_message",
+      "description": "Send a message to a background subagent by its subagent id, continuing the same conversation. It becomes the subagent's next turn: if it is still working, the message waits until its current turn finishes, so it cannot redirect work already underway. This call returns no answer from the subagent — only confirmation that the message was delivered — so use it to give it more work. A failure means the message was NOT delivered.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "subagent_id": {
+            "type": "string",
+            "description": "The subagent id returned when the background subagent was started."
+          },
+          "message": {
+            "type": "string",
+            "description": "The message to deliver to the subagent."
+          }
+        },
+        "required": [
+          "subagent_id",
+          "message"
+        ]
+      }
+    },
+    {
+      "name": "skill",
+      "description": "Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "name": {
+            "type": "string",
+            "description": "The exact skill name from the available skills list."
+          }
+        },
+        "required": [
+          "name"
+        ]
+      }
+    },
+    {
+      "name": "subagent",
+      "description": "Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to start a background subagent that keeps its conversation: you receive only its subagent id, never its result, and it works on its own. Use this for work whose result you do not need returned by this call; `send_message` sends it more work.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."
+          },
+          "run_in_background": {
+            "type": "boolean",
+            "description": "Run as a background subagent that keeps its conversation and return only its subagent id. This call never returns its result; send it more work with send_message."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "subagent_claude_code",
+      "description": "Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "subagent_codex",
+      "description": "Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "subagent_fork",
+      "description": "Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to start a background subagent that keeps its conversation: you receive only its subagent id, never its result, and it works on its own. Use this for work whose result you do not need returned by this call; `send_message` sends it more work.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."
+          },
+          "run_in_background": {
+            "type": "boolean",
+            "description": "Run as a background subagent that keeps its conversation and return only its subagent id. This call never returns its result; send it more work with send_message."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "task_kill",
+      "description": "Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "task_id": {
+            "type": "string",
+            "description": "Task id returned by the tool that started the background work."
+          },
+          "reason": {
+            "type": "string",
+            "description": "Optional short reason, recorded in the log and forwarded to the task."
+          }
+        },
+        "required": [
+          "task_id"
+        ]
+      }
+    },
+    {
+      "name": "task_list",
+      "description": "List your background tasks (running and finished) with their ids, kinds, and statuses.",
+      "parameters": {
+        "type": "object",
+        "properties": {}
+      }
+    },
+    {
+      "name": "task_output",
+      "description": "Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "task_id": {
+            "type": "string",
+            "description": "Task id returned by the tool that started the background work."
+          },
+          "wait": {
+            "type": "boolean",
+            "description": "Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."
+          },
+          "timeout_ms": {
+            "type": "number",
+            "description": "Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."
+          }
+        },
+        "required": [
+          "task_id"
+        ]
+      }
+    },
+    {
+      "name": "todo_write",
+      "description": "Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Mark every todo being actively worked on `in_progress` — several at once when work genuinely runs in parallel (e.g. concurrent subagents or background commands), one for sequential work; while work remains, at least one task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "todos": {
+            "type": "array",
+            "description": "The COMPLETE task list, replacing any previous list.",
+            "items": {
+              "type": "object",
+              "additionalProperties": false,
+              "properties": {
+                "content": {
+                  "type": "string",
+                  "description": "What the task is — a short imperative line."
+                },
+                "status": {
+                  "type": "string",
+                  "description": "pending (not started) | in_progress (now) | completed (done).",
+                  "enum": [
+                    "pending",
+                    "in_progress",
+                    "completed"
+                  ]
+                }
+              },
+              "required": [
+                "content",
+                "status"
+              ]
+            }
+          }
+        },
+        "required": [
+          "todos"
+        ]
+      }
+    },
+    {
+      "name": "update_goal",
+      "description": "Update the exact current goal revision. edit, pause, and resume require a direct top-level human request. During an automatic continuation of the current goal, complete and blocked are also allowed. blocked is rejected before the configured minimum round count; the model remains responsible for judging that the same condition persisted across those rounds and must explain it in blocked_reason.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "goal_id": {
+            "type": "string",
+            "description": "Exact id returned by get_goal."
+          },
+          "revision": {
+            "type": "number",
+            "description": "Exact positive revision returned by get_goal."
+          },
+          "action": {
+            "type": "string",
+            "description": "edit | pause | resume | complete | blocked",
+            "enum": [
+              "edit",
+              "pause",
+              "resume",
+              "complete",
+              "blocked"
+            ]
+          },
+          "objective": {
+            "type": "string",
+            "description": "Replacement objective; valid only with action edit."
+          },
+          "max_goal_rounds": {
+            "type": "number",
+            "description": "Replacement cap; valid only with action edit."
+          },
+          "blocked_reason": {
+            "type": "string",
+            "description": "Concrete blocking condition; required only with action blocked."
+          }
+        },
+        "required": [
+          "goal_id",
+          "revision",
+          "action"
+        ]
+      }
+    },
+    {
+      "name": "workflow",
+      "description": "Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "script": {
+            "type": "string",
+            "description": "The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`)."
+          },
+          "meta": {
+            "type": "object",
+            "description": "The workflow identity block (plain JSON — never code).",
+            "additionalProperties": true,
+            "properties": {
+              "name": {
+                "type": "string",
+                "description": "Short kebab-case workflow name."
+              },
+              "description": {
+                "type": "string",
+                "description": "One-line description of what the workflow does."
+              },
+              "whenToUse": {
+                "type": "string",
+                "description": "Optional guidance on when this workflow applies."
+              },
+              "phases": {
+                "type": "array",
+                "description": "Optional phase declarations matched by phase() calls.",
+                "items": {
+                  "type": "object",
+                  "additionalProperties": true,
+                  "properties": {
+                    "title": {
+                      "type": "string",
+                      "description": "The phase title phase() calls match by exact string."
+                    },
+                    "detail": {
+                      "type": "string",
+                      "description": "Optional one-line description of the phase."
+                    },
+                    "provider": {
+                      "type": "string",
+                      "description": "Optional provider override this phase is expected to use."
+                    },
+                    "model": {
+                      "type": "string",
+                      "description": "Optional model override this phase is expected to use."
+                    }
+                  },
+                  "required": [
+                    "title"
+                  ]
+                }
+              }
+            },
+            "required": [
+              "name",
+              "description"
+            ]
+          },
+          "args": {
+            "type": "object",
+            "description": "Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).",
+            "additionalProperties": true
+          }
+        },
+        "required": [
+          "script",
+          "meta"
+        ]
+      }
+    },
+    {
+      "name": "write",
+      "description": "Create or fully replace a UTF-8 text file.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "file_path": {
+            "type": "string",
+            "description": "Path to write, resolved by the filesystem backend."
+          },
+          "content": {
+            "type": "string",
+            "description": "Full UTF-8 text content to write."
+          },
+          "sandbox_permissions": {
+            "type": "string",
+            "description": "The wider sandbox mode this file operation needs. Only valid as a one-shot retry of an operation the sandbox just denied; requires justification and user approval.",
+            "enum": [
+              "workspace-write",
+              "danger-full-access"
+            ]
+          },
+          "justification": {
+            "type": "string",
+            "description": "Required with sandbox_permissions: one sentence for the user explaining why this exact file operation needs the wider access."
+          }
+        },
+        "required": [
+          "file_path",
+          "content"
+        ]
+      }
+    }
+  ],
+  "changes": []
+}
@@ -0,0 +1,7 @@
+{
+  "steps": [
+    { "op": "initialize" },
+    { "op": "newSession" },
+    { "op": "prompt", "text": "Reply with exactly the word: PONG. Do not use any tools." }
+  ]
+}
@@ -0,0 +1,22 @@
+{"type":"session","version":0,"id":"539aa64c-7f37-40ff-abd8-ed45b717be1b","createdAt":1783600629539,"cwd":"{{cwd}}","delegationDepth":0}
+{"type":"agent/inbox/spliced","seq":0,"time":1785498761270,"data":{"target":"next-turn","start":0,"inserted":[{"content":[{"type":"text","text":"Reply with exactly the word: PONG. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"3e25dc34-48e0-4738-8401-1a8d181d37e5"}]}}
+{"type":"turn/start","seq":1,"time":1785821359466,"data":{"turn":1}}
+{"type":"agent/inbox/spliced","seq":2,"time":1785821359466,"data":{"target":"next-turn","start":0,"removedCount":1,"inserted":[]}}
+{"type":"step/start","seq":3,"time":1783600629542,"data":{"turn":1,"step":1}}
+{"type":"user/message","seq":4,"time":1785498761313,"data":{"content":[{"type":"text","text":"Reply with exactly the word: PONG. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"3e25dc34-48e0-4738-8401-1a8d181d37e5"},"surfaceOp":"append"}
+{"type":"user/message","seq":5,"time":1785730415287,"data":{"content":[{"type":"text","text":"Current runtime context. This snapshot supersedes earlier runtime-context snapshots.\n\nCurrent DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations.\n\nApproval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}],"source":{"kind":"plugin","plugin":"@deepseek-ai/dsh-system-prompt","form":"snapshot","sections":[{"name":"sandbox:policy","text":"Current DSH file policy: danger-full-access. The DSH file sandbox does not restrict file modifications by available operations."},{"name":"approval:policy","text":"Approval prompts are disabled in this session: actions that require approval are rejected automatically — do not request sandbox escalation (do not set `sandbox_permissions`)."}]},"role":"user","id":"4b8d9730-0b7b-4e14-8a30-3d852f808f0e"},"surfaceOp":"append"}
+{"type":"session/title","seq":6,"time":1785730415287,"data":{"title":"Reply with exactly the word:","messageSeqs":[4],"source":{"kind":"fallback"}}}
+{"type":"request/header","seq":7,"time":1785498761318,"data":{"header":{"config":{"provider":"deepseek-official","model":"deepseek-v4-pro"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}}
+{"type":"request/context","seq":8,"time":1785730415288,"data":{"provider":"deepseek-official","model":"deepseek-v4-pro"}}
+{"type":"assistant/chunk","seq":9,"time":1783600630822,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}
+{"type":"reasoning-chunks","seq0":10,"time0":1783600630852,"data":{"turn":1,"step":1,"index":0,"dt":[0,0,0,33,1,40,0,0,0,0,0,18,0,36,0,0,0,0,0],"texts":["The"," user"," wants"," me"," to"," reply"," with"," exactly"," the"," word"," \"","P","ONG","\""," and"," not"," use"," any"," tools","."]}}
+{"type":"assistant/chunk","seq":30,"time":1783600631006,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"text"}}}
+{"type":"assistant/chunk","seq":31,"time":1783600631008,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":1,"text":"P"}}}
+{"type":"assistant/chunk","seq":32,"time":1783600631009,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":1,"text":"ONG"}}}
+{"type":"assistant/chunk","seq":33,"time":1783600631009,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"reasoning","text":"The user wants me to reply with exactly the word \"PONG\" and not use any tools."}}}}
+{"type":"assistant/chunk","seq":34,"time":1783600631009,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":1,"block":{"type":"text","text":"PONG"}}}}
+{"type":"assistant/chunk","seq":35,"time":1785498761338,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":3091,"outputTokens":23,"cacheReadTokens":0,"reasoningTokens":20}}}}
+{"type":"assistant/chunk","seq":36,"time":1785730415297,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"stop"}}}}
+{"type":"assistant/message","seq":37,"time":1785730415298,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to reply with exactly the word \"PONG\" and not use any tools."},{"type":"text","text":"PONG"}],"source":{"kind":"model","provider":"deepseek-official","model":"deepseek-v4-pro"},"id":"c883cf16-01fe-4afc-b37c-d255bb450d21"},"usage":{"inputTokens":3091,"outputTokens":23,"cacheReadTokens":0,"reasoningTokens":20}},"sourceEventSeqs":[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],"surfaceOp":"append"}
+{"type":"step/end","seq":38,"time":1785730415298,"data":{"turn":1,"step":1}}
+{"type":"turn/end","seq":39,"time":1785730415298,"data":{"turn":1,"reason":{"kind":"completed"}}}
@@ -0,0 +1,4 @@
+{"jsonrpc":"2.0","id":1,"result":{"protocolVersion":1,"agentInfo":{"name":"deepseek-harness-acp","version":"0.0.1"},"agentCapabilities":{"promptCapabilities":{"image":false,"audio":false,"embeddedContext":false}},"authMethods":[]}}
+{"jsonrpc":"2.0","id":2,"result":{"sessionId":"{{sessionId}}"}}
+{"jsonrpc":"2.0","method":"session/update","params":{"sessionId":"{{sessionId}}","update":{"sessionUpdate":"agent_message_chunk","content":{"type":"text","text":"PONG"}}}}
+{"jsonrpc":"2.0","id":3,"result":{"stopReason":"end_turn"}}
@@ -0,0 +1,22 @@
+You are an AI agent powered by the DeepSeek Harness SDK.
+
+You are a coding assistant powered by the deepseek-v4-pro model. Your working directory is {{cwd}}. Your bash tool runs under a file sandbox — a `[sandbox: file access denied …]` result is policy, not a command bug.
+
+Verify your work by running the code or tests. Keep answers brief and factual.
+
+
+Use the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.
+
+Use the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.
+
+Use the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.
+
+Check the [exit code: N] marker on every bash result; investigate failures before moving on.
+
+Track every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.
+
+Use goal tools for one long-running completion objective in the current session. create_goal may infer goal intent from a direct human request in any language; do not create a goal for routine single-turn work. Call get_goal before update_goal and copy its exact goal_id and revision. After session resume or fork, an active goal is disarmed: when a human asks to continue or resume in any wording or language, use update_goal action resume to rearm it. Mark complete only when the objective is actually achieved. Mark blocked only after the same blocking condition persists for at least 3 consecutive goal rounds, and report that concrete condition in blocked_reason; difficulty, uncertainty, or useful remaining work is not blocked.
+
+Use the workflow 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.
+
+Use the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.
@@ -0,0 +1,548 @@
+{
+  "initial": [
+    {
+      "name": "bash",
+      "description": "Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. Attempting a command the sandbox may deny is safe and expected: run it and read the marker rather than assuming the denial. When a command is denied and a wider mode would let it succeed, escalate immediately in the same turn — the one sanctioned exception to a denial: retry the exact same command once with `sandbox_permissions` (the narrowest wider mode that suffices) plus a one-sentence `justification`. Do not detour through chat to ask permission first — the approval prompt raised by that retry is how the user consents. If the session states approval prompts are disabled, there is no exception: a denial is final — do not set `sandbox_permissions`. Never escalate speculatively: ground the request in a real denial — normally the one this command just hit; escalating up front is fine only when this session already denied the same access. A rejected escalation is final for that command — stop and explain, never work around it — but it does not forbid attempting or escalating other commands later.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "command": {
+            "type": "string",
+            "description": "The bash command to execute."
+          },
+          "description": {
+            "type": "string",
+            "description": "Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."
+          },
+          "timeoutMs": {
+            "type": "number",
+            "description": "Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."
+          },
+          "workdir": {
+            "type": "string",
+            "description": "Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."
+          },
+          "run_in_background": {
+            "type": "boolean",
+            "description": "Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."
+          },
+          "sandbox_permissions": {
+            "type": "string",
+            "description": "The wider sandbox mode this command needs. Only valid as a one-shot retry of a command the sandbox just denied; requires justification and user approval.",
+            "enum": [
+              "workspace-write",
+              "danger-full-access"
+            ]
+          },
+          "justification": {
+            "type": "string",
+            "description": "Required with sandbox_permissions: one sentence for the user explaining why this exact command needs the wider access."
+          }
+        },
+        "required": [
+          "command",
+          "description"
+        ]
+      }
+    },
+    {
+      "name": "create_goal",
+      "description": "Create one persisted same-session completion goal when the current direct human request is a long-running objective that should continue across autonomous goal rounds. You may infer that intent without requiring the user to say \"create a goal\". Do not use this for trivial single-turn work. Execution rejects non-human and subagent authority.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "objective": {
+            "type": "string",
+            "description": "The concrete completion objective inferred from the direct human request."
+          },
+          "max_goal_rounds": {
+            "type": "number",
+            "description": "Optional positive safe-integer limit on automatic continuation rounds."
+          }
+        },
+        "required": [
+          "objective"
+        ]
+      }
+    },
+    {
+      "name": "edit",
+      "description": "Edit an existing UTF-8 text file by replacing literal text.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "file_path": {
+            "type": "string",
+            "description": "Path to edit, resolved by the filesystem backend."
+          },
+          "old_string": {
+            "type": "string",
+            "description": "Literal text to replace. Must match exactly."
+          },
+          "new_string": {
+            "type": "string",
+            "description": "Literal replacement text. Use an empty string to delete the match."
+          },
+          "replace_all": {
+            "type": "boolean",
+            "description": "Replace all matches. Defaults to false; when false, old_string must appear exactly once."
+          },
+          "sandbox_permissions": {
+            "type": "string",
+            "description": "The wider sandbox mode this file operation needs. Only valid as a one-shot retry of an operation the sandbox just denied; requires justification and user approval.",
+            "enum": [
+              "workspace-write",
+              "danger-full-access"
+            ]
+          },
+          "justification": {
+            "type": "string",
+            "description": "Required with sandbox_permissions: one sentence for the user explaining why this exact file operation needs the wider access."
+          }
+        },
+        "required": [
+          "file_path",
+          "old_string",
+          "new_string"
+        ]
+      }
+    },
+    {
+      "name": "get_goal",
+      "description": "Read the current same-session goal, including its exact id/revision, objective, phase, completed continuation rounds, round limit, blocker reason when present, and whether another continuation is armed. Call this before updating a goal.",
+      "parameters": {
+        "type": "object",
+        "properties": {}
+      }
+    },
+    {
+      "name": "interrupt_agent",
+      "description": "Request cancellation of a background agent's current turn by its agent id. The target may be your direct child or a deeper agent created under you. Only the current turn stops: messages already queued for the agent stay parked until a later send_message, agents it started keep running, and the agent itself stays available for follow-ups. This call returns as soon as the stop request is accepted, so the target may keep running briefly; interrupting an agent that already finished is an accepted no-op.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "agent_id": {
+            "type": "string",
+            "description": "The agent id of the running agent to interrupt."
+          }
+        },
+        "required": [
+          "agent_id"
+        ]
+      }
+    },
+    {
+      "name": "list_agents",
+      "description": "List your continuable background subagents by durable id and label. Status comes from the live registry: running means the agent is working right now, idle means it is loaded but between turns (it may be waiting on agents it started), and complete means it exists only in storage — a direct child remains a `send_message` candidate in every status. The snapshot is not a delivery promise — `send_message` performs the authoritative check and may still fail. Children that could not be read are reported as diagnostics instead of being silently dropped. Scope `descendants` walks the whole tree below you in stable pre-order, annotating each entry with its durable direct-parent session id and depth. You may use `send_message` only for depth-1 entries; deeper entries are candidates for `interrupt_agent` only.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "scope": {
+            "type": "string",
+            "description": "children (default) lists direct children only; descendants walks the complete tree below you.",
+            "enum": [
+              "children",
+              "descendants"
+            ]
+          }
+        }
+      }
+    },
+    {
+      "name": "ralph",
+      "description": "Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "objective": {
+            "type": "string",
+            "description": "The immutable completion objective for every fresh Ralph round."
+          },
+          "maxRounds": {
+            "type": "number",
+            "description": "Optional positive safe-integer round cap, bounded by the deployment ceiling."
+          }
+        },
+        "required": [
+          "objective"
+        ]
+      }
+    },
+    {
+      "name": "read",
+      "description": "Read a UTF-8 text file and return line-numbered content.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "file_path": {
+            "type": "string",
+            "description": "Path to read, resolved by the filesystem backend."
+          },
+          "offset": {
+            "type": "number",
+            "description": "1-based first line to return. Defaults to 1."
+          },
+          "limit": {
+            "type": "number",
+            "description": "Maximum number of lines to return. Defaults to 2000."
+          }
+        },
+        "required": [
+          "file_path"
+        ]
+      }
+    },
+    {
+      "name": "send_message",
+      "description": "Send a message to a background subagent by its subagent id, continuing the same conversation. It becomes the subagent's next turn: if it is still working, the message waits until its current turn finishes, so it cannot redirect work already underway. This call returns no answer from the subagent — only confirmation that the message was delivered — so use it to give it more work. A failure means the message was NOT delivered.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "subagent_id": {
+            "type": "string",
+            "description": "The subagent id returned when the background subagent was started."
+          },
+          "message": {
+            "type": "string",
+            "description": "The message to deliver to the subagent."
+          }
+        },
+        "required": [
+          "subagent_id",
+          "message"
+        ]
+      }
+    },
+    {
+      "name": "skill",
+      "description": "Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "name": {
+            "type": "string",
+            "description": "The exact skill name from the available skills list."
+          }
+        },
+        "required": [
+          "name"
+        ]
+      }
+    },
+    {
+      "name": "subagent",
+      "description": "Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to start a background subagent that keeps its conversation: you receive only its subagent id, never its result, and it works on its own. Use this for work whose result you do not need returned by this call; `send_message` sends it more work.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."
+          },
+          "run_in_background": {
+            "type": "boolean",
+            "description": "Run as a background subagent that keeps its conversation and return only its subagent id. This call never returns its result; send it more work with send_message."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "subagent_codex",
+      "description": "Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "subagent_fork",
+      "description": "Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to start a background subagent that keeps its conversation: you receive only its subagent id, never its result, and it works on its own. Use this for work whose result you do not need returned by this call; `send_message` sends it more work.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "description": {
+            "type": "string",
+            "description": "A short (3-5 word) description of the delegated task, for display."
+          },
+          "prompt": {
+            "type": "string",
+            "description": "The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."
+          },
+          "run_in_background": {
+            "type": "boolean",
+            "description": "Run as a background subagent that keeps its conversation and return only its subagent id. This call never returns its result; send it more work with send_message."
+          }
+        },
+        "required": [
+          "description",
+          "prompt"
+        ]
+      }
+    },
+    {
+      "name": "task_kill",
+      "description": "Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "task_id": {
+            "type": "string",
+            "description": "Task id returned by the tool that started the background work."
+          },
+          "reason": {
+            "type": "string",
+            "description": "Optional short reason, recorded in the log and forwarded to the task."
+          }
+        },
+        "required": [
+          "task_id"
+        ]
+      }
+    },
+    {
+      "name": "task_list",
+      "description": "List your background tasks (running and finished) with their ids, kinds, and statuses.",
+      "parameters": {
+        "type": "object",
+        "properties": {}
+      }
+    },
+    {
+      "name": "task_output",
+      "description": "Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "task_id": {
+            "type": "string",
+            "description": "Task id returned by the tool that started the background work."
+          },
+          "wait": {
+            "type": "boolean",
+            "description": "Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."
+          },
+          "timeout_ms": {
+            "type": "number",
+            "description": "Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."
+          }
+        },
+        "required": [
+          "task_id"
+        ]
+      }
+    },
+    {
+      "name": "todo_write",
+      "description": "Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Mark every todo being actively worked on `in_progress` — several at once when work genuinely runs in parallel (e.g. concurrent subagents or background commands), one for sequential work; while work remains, at least one task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "todos": {
+            "type": "array",
+            "description": "The COMPLETE task list, replacing any previous list.",
+            "items": {
+              "type": "object",
+              "additionalProperties": false,
+              "properties": {
+                "content": {
+                  "type": "string",
+                  "description": "What the task is — a short imperative line."
+                },
+                "status": {
+                  "type": "string",
+                  "description": "pending (not started) | in_progress (now) | completed (done).",
+                  "enum": [
+                    "pending",
+                    "in_progress",
+                    "completed"
+                  ]
+                }
+              },
+              "required": [
+                "content",
+                "status"
+              ]
+            }
+          }
+        },
+        "required": [
+          "todos"
+        ]
+      }
+    },
+    {
+      "name": "update_goal",
+      "description": "Update the exact current goal revision. edit, pause, and resume require a direct top-level human request. During an automatic continuation of the current goal, complete and blocked are also allowed. blocked is rejected before the configured minimum round count; the model remains responsible for judging that the same condition persisted across those rounds and must explain it in blocked_reason.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "goal_id": {
+            "type": "string",
+            "description": "Exact id returned by get_goal."
+          },
+          "revision": {
+            "type": "number",
+            "description": "Exact positive revision returned by get_goal."
+          },
+          "action": {
+            "type": "string",
+            "description": "edit | pause | resume | complete | blocked",
+            "enum": [
+              "edit",
+              "pause",
+              "resume",
+              "complete",
+              "blocked"
+            ]
+          },
+          "objective": {
+            "type": "string",
+            "description": "Replacement objective; valid only with action edit."
+          },
+          "max_goal_rounds": {
+            "type": "number",
+            "description": "Replacement cap; valid only with action edit."
+          },
+          "blocked_reason": {
+            "type": "string",
+            "description": "Concrete blocking condition; required only with action blocked."
+          }
+        },
+        "required": [
+          "goal_id",
+          "revision",
+          "action"
+        ]
+      }
+    },
+    {
+      "name": "workflow",
+      "description": "Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "script": {
+            "type": "string",
+            "description": "The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`)."
+          },
+          "meta": {
+            "type": "object",
+            "description": "The workflow identity block (plain JSON — never code).",
+            "additionalProperties": true,
+            "properties": {
+              "name": {
+                "type": "string",
+                "description": "Short kebab-case workflow name."
+              },
+              "description": {
+                "type": "string",
+                "description": "One-line description of what the workflow does."
+              },
+              "whenToUse": {
+                "type": "string",
+                "description": "Optional guidance on when this workflow applies."
+              },
+              "phases": {
+                "type": "array",
+                "description": "Optional phase declarations matched by phase() calls.",
+                "items": {
+                  "type": "object",
+                  "additionalProperties": true,
+                  "properties": {
+                    "title": {
+                      "type": "string",
+                      "description": "The phase title phase() calls match by exact string."
+                    },
+                    "detail": {
+                      "type": "string",
+                      "description": "Optional one-line description of the phase."
+                    },
+                    "provider": {
+                      "type": "string",
+                      "description": "Optional provider override this phase is expected to use."
+                    },
+                    "model": {
+                      "type": "string",
+                      "description": "Optional model override this phase is expected to use."
+                    }
+                  },
+                  "required": [
+                    "title"
+                  ]
+                }
+              }
+            },
+            "required": [
+              "name",
+              "description"
+            ]
+          },
+          "args": {
+            "type": "object",
+            "description": "Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).",
+            "additionalProperties": true
+          }
+        },
+        "required": [
+          "script",
+          "meta"
+        ]
+      }
+    },
+    {
+      "name": "write",
+      "description": "Create or fully replace a UTF-8 text file.",
+      "parameters": {
+        "type": "object",
+        "properties": {
+          "file_path": {
+            "type": "string",
+            "description": "Path to write, resolved by the filesystem backend."
+          },
+          "content": {
+            "type": "string",
+            "description": "Full UTF-8 text content to write."
+          },
+          "sandbox_permissions": {
+            "type": "string",
+            "description": "The wider sandbox mode this file operation needs. Only valid as a one-shot retry of an operation the sandbox just denied; requires justification and user approval.",
+            "enum": [
+              "workspace-write",
+              "danger-full-access"
+            ]
+          },
+          "justification": {
+            "type": "string",
+            "description": "Required with sandbox_permissions: one sentence for the user explaining why this exact file operation needs the wider access."
+          }
+        },
+        "required": [
+          "file_path",
+          "content"
+        ]
+      }
+    }
+  ],
+  "changes": []
+}
@@ -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/bundle/base/README.md
-README.md: fb003908a262dc21edd3c9d49c972e487534f367
-README.zh.md: 13e64db6d34374fac63bf9bfd60544fc46b86f35
+README.md: aea28a7412fd1123e5bbc593fb7978b2e4595248
+README.zh.md: f11745061c340d23401729ff326d277a0eade428
@@ -2,7 +2,7 @@

 English | [中文](README.zh.md)

-The shared dsh core as a profile bundle: [`cordis.patch.yml`](cordis.patch.yml) inserts every base plugin row — model adapters, the shared [`agent-default-model`](../../core/agent-default-model/README.md) selection, tools, persistence, policy, settings/credentials, repository Plugins, telemetry — over the empty profile root, as the first layer of every profile's `dsh.profile.bundles` list. Later bundle layers (e.g. [`dsh-web-app`](../web-app/README.md)) and the user's profile `cordis.patch.yml` override these rows by id; a patch replaces a row's whole `config`, so mode-specific values live in mode bundles, not here. The package has no runtime API; the profile composer resolves the patch through the `dsh.bundle.patch` manifest field, never through code.
+The shared dsh core as a profile bundle: [`cordis.patch.yml`](cordis.patch.yml) inserts every base plugin row — model adapters, the shared [`agent-default-model`](../../core/agent-default-model/README.md) selection, tools, persistence, policy, settings/credentials, repository Plugins, telemetry, and host-level subagent providers — over the empty profile root, as the first layer of every profile's `dsh.profile.bundles` list. Codex and Claude Code providers load dormant; Agent Presets independently decide whether their agent contributes either model-facing delegation tool. Later bundle layers (e.g. [`dsh-web-app`](../web-app/README.md)) and the user's profile `cordis.patch.yml` override these rows by id; a patch replaces a row's whole `config`, so mode-specific values live in mode bundles, not here. The package has no runtime API; the profile composer resolves the patch through the `dsh.bundle.patch` manifest field, never through code.

 The row set and its rationale are documented inline in the patch file; the [generated composition graph](../../../apps/cli/composition.md) renders it.

@@ -2,7 +2,7 @@

 [English](README.md) | 中文

-以 profile 组合包形式交付的共享 dsh 核心：[`cordis.patch.yml`](cordis.patch.yml) 在空的 profile 根之上插入全部基础插件行——模型适配器、共享的 [`agent-default-model`](../../core/agent-default-model/README.md) 选择、工具、持久化、策略、settings／credentials、repository 插件、遥测——作为每个 profile 的 `dsh.profile.bundles` 列表中的第一层。后续的组合包层（例如 [`dsh-web-app`](../web-app/README.md)）和用户 profile 的 `cordis.patch.yml` 按 id 覆盖这些行；patch 会替换目标行的整个 `config`，因此模式专属的值放在各模式组合包中，而不是这里。该包没有运行时 API；profile 组合器通过 manifest（元数据清单）的 `dsh.bundle.patch` 字段解析 patch，绝不通过代码。
+以 profile 组合包形式交付的共享 dsh 核心：[`cordis.patch.yml`](cordis.patch.yml) 在空的 profile 根之上插入全部基础插件行——模型适配器、共享的 [`agent-default-model`](../../core/agent-default-model/README.md) 选择、工具、持久化、策略、settings／credentials、repository 插件、遥测与宿主级 subagent provider——作为每个 profile 的 `dsh.profile.bundles` 列表中的第一层。Codex 与 Claude Code provider 以休眠状态加载；Agent Preset 分别决定自己的 agent 是否贡献任一面向模型的委派工具。后续的组合包层（例如 [`dsh-web-app`](../web-app/README.md)）和用户 profile 的 `cordis.patch.yml` 按 id 覆盖这些行；patch 会替换目标行的整个 `config`，因此模式专属的值放在各模式组合包中，而不是这里。该包没有运行时 API；profile 组合器通过 manifest（元数据清单）的 `dsh.bundle.patch` 字段解析 patch，绝不通过代码。

 行集合及其设计依据以行内注释写在 patch 文件里；[生成的组合图](../../../apps/cli/composition.md)负责渲染它。

@@ -288,6 +288,15 @@
      config:
        providerName: fork

+    # Product providers stay on the host plane because the registry is a
+    # process singleton. Agent presets decide whether their own model sees the
+    # matching delegation tools; loading either provider starts no product.
+    - id: subagent-codex
+      name: '@deepseek-ai/dsh-subagent-codex'
+
+    - id: subagent-claude-code
+      name: '@deepseek-ai/dsh-subagent-claude-code'
+
    # Continuable background children are selected per delegation tool. The
    # separately loaded follow-up tool registers the one global `send_message`.
    - id: tool-subagent-control
@@ -76,6 +76,8 @@
    "@deepseek-ai/dsh-spill-local": "workspace:^",
    "@deepseek-ai/dsh-spill-policy": "workspace:^",
    "@deepseek-ai/dsh-subagent": "workspace:^",
+    "@deepseek-ai/dsh-subagent-claude-code": "workspace:^",
+    "@deepseek-ai/dsh-subagent-codex": "workspace:^",
    "@deepseek-ai/dsh-subagent-fork": "workspace:^",
    "@deepseek-ai/dsh-subagent-spawn": "workspace:^",
    "@deepseek-ai/dsh-subprocess-local": "workspace:^",
@@ -13,7 +13,10 @@ import { entryListSchema } from '@cordisjs/plugin-include'
 describe('dsh-base bundle', () => {
  it('declares a parseable patch list through the dsh.bundle.patch manifest field', () => {
    const root = fileURLToPath(new URL('..', import.meta.url))
-    const manifest = JSON.parse(readFileSync(resolve(root, 'package.json'), 'utf8')) as { dsh?: { bundle?: { patch?: string } } }
+    const manifest = JSON.parse(readFileSync(resolve(root, 'package.json'), 'utf8')) as {
+      dependencies?: Record<string, string>
+      dsh?: { bundle?: { patch?: string } }
+    }
    expect(manifest.dsh?.bundle?.patch).toBe('./cordis.patch.yml')
    const parsed = yaml.load(readFileSync(resolve(root, manifest.dsh!.bundle!.patch!), 'utf8'), { schema: entryListSchema })
    expect(Array.isArray(parsed)).toBe(true)
@@ -21,5 +24,11 @@ describe('dsh-base bundle', () => {
    const rows = (parsed as { insert?: { id?: string }[] }[]).flatMap(patch => patch.insert ?? [])
    expect(rows.length).toBeGreaterThan(50)
    expect(rows.some(row => row.id === 'agent-loop')).toBe(true)
+    expect(rows.filter(row => row.id === 'subagent-codex')).toHaveLength(1)
+    expect(rows.filter(row => row.id === 'subagent-claude-code')).toHaveLength(1)
+    expect(manifest.dependencies).toMatchObject({
+      '@deepseek-ai/dsh-subagent-codex': 'workspace:^',
+      '@deepseek-ai/dsh-subagent-claude-code': 'workspace:^',
+    })
  })
 })
@@ -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/subagent/subagent-claude-code/README.md
-README.md: 222cf796f71f8dc0dc2c06f7f32bab70ded6ab43
-README.zh.md: 9334820a591cbfcb8dc2046dc3a78201ba193aab
+README.md: 5e3138b9211b01de9096fa1b8e8b68321aad0c7d
+README.zh.md: cc536d13d58e08efc77f4f7b4374c8a1af8c5caa
@@ -2,7 +2,7 @@

 English | [中文](README.zh.md)

-This package registers the fixed `claude-code` subagent provider. Each accepted run invokes the official Claude Agent SDK in the delegating Session's workspace, starts the SDK-distributed Claude Code CLI through the shared subprocess service, submits one self-contained text task, and returns only the final answer through the shared [`dsh-subagent`](../subagent/README.md) result contract.
+This package registers the fixed `claude-code` subagent provider. Each accepted run invokes the official Claude Agent SDK in the delegating Session's workspace, resolves the native `claude` executable through the shared subprocess service, submits one self-contained text task, and returns only the final answer through the shared [`dsh-subagent`](../subagent/README.md) result contract.

 ## Start and ownership

@@ -29,9 +29,9 @@ The provider advertises no optional start-time capabilities and reports `inherit
 | `env` | `{}` | Explicit SDK/CLI environment layered over the shared credential-scrubbed parent environment. |
 | `disposeGraceMs` | `3000` | Positive finite grace in milliseconds, no greater than [`MAX_TIMER_DELAY_MS`](../../util/timeout/README.md), between the shared process-tree owner's termination tiers; disposal then waits for whole-tree exit. |

-Production uses the Claude Code CLI supplied by `@anthropic-ai/claude-agent-sdk` and the host's native settings and authentication. The plugin does not install another CLI, select a model, create a product home, log in, or probe an account. Credential-shaped ambient variables are removed before the explicit `env` overlay is applied, so an API key or token intended for the child must be supplied there. Non-credential endpoint variables such as `ANTHROPIC_BASE_URL`, along with ordinary ambient values such as `PATH` and `HOME`, remain inherited unless overridden.
+Production resolves `claude` from the subprocess execution world's credential-scrubbed `PATH`, with explicit `env` entries applied, and passes the resulting path to the SDK as `pathToClaudeCodeExecutable`. Native settings and authentication remain authoritative. The plugin does not install another CLI, select a model, create a product home, log in, or probe an account. Credential-shaped ambient variables are removed before the explicit `env` overlay is applied, so an API key or token intended for the child must be supplied there. Non-credential endpoint variables such as `ANTHROPIC_BASE_URL`, along with ordinary ambient values such as `PATH` and `HOME`, remain inherited unless overridden.

-Install this package and add the following rows to your own `cordis.yml`. Shipped CLI configurations do not load this provider or expose `subagent_claude_code` by default.
+Shipped profiles load this provider once on the host and start no Claude process until a tool call. Full Agent Presets carry the tool row below with `disabled: true`; copy a preset and remove that field to expose `subagent_claude_code` only to agents composed from the copy. A custom host composition can still use both rows directly.

 ```yaml
 - id: subagent-claude-code
@@ -42,6 +42,7 @@ Install this package and add the following rows to your own `cordis.yml`. Shippe

 - id: tool-subagent-claude-code
  name: '@deepseek-ai/dsh-tool-subagent'
+  disabled: true
  config:
    provider: claude-code
    toolName: subagent_claude_code
@@ -51,7 +52,7 @@ Install this package and add the following rows to your own `cordis.yml`. Shippe

 ## Product compatibility and evidence

-The runtime dependency is pinned to `@anthropic-ai/claude-agent-sdk@0.3.220`, whose platform optional dependency supplies Claude Code 2.1.220. Required evidence exercises that official distribution through a keyless loopback product path and a credentialed DeepSeek path, while Loader composition proves that both opt-in product packages coexist without starting either product.
+The runtime dependency is pinned to `@anthropic-ai/claude-agent-sdk@0.3.220`. Production runs the native `claude` installation; the SDK's platform optional payload remains in the current installation closure and is tracked as a separate distribution follow-up. Required evidence exercises the compatible native product through a keyless loopback path and a credentialed DeepSeek path, while Loader composition proves that both product packages coexist without starting either product.

 The project owner's identity-scoped distribution authorization covers the official SDK and the official CLI/platform payloads declared by each SDK version. [`THIRD_PARTY_NOTICES.md`](../../../THIRD_PARTY_NOTICES.md) discloses the current optional payload closure without classifying its declared terms as permissive; unrelated non-permissive runtime dependencies continue to fail the notices gate.

@@ -89,7 +90,7 @@ Append-only: the new tool result follows the reusable parent request prefix.

 - **One fresh query and process per run** — there is no continuation, resume, pooling, progress stream, or product-session persistence.
 - **Host settings are intentionally authoritative** — project and user settings can change model, tools, and behavior; the provider does not provide a filtered or hermetic production mode.
- **Product installation and account state remain native** — an incompatible SDK payload, configuration error, or authentication failure is surfaced as a startup or run error; the plugin provides no installer or login flow.
+- **Product installation and account state remain native** — a missing or incompatible `claude`, configuration error, or authentication failure is surfaced as a startup or run error; the plugin provides no installer or login flow.
 - **No human interaction path** — `AskUserQuestion` is disabled and other interactive callbacks are absent, so tasks requiring new approval or input fail instead of suspending.
 - **Final text only** — reasoning, intermediate messages, tool traffic, usage, stderr, and workspace diffs remain product-local.
 - **No optional shared capabilities** — output schemas, child personas, tool filtering, and harness depth enforcement are rejected by the shared service for this provider.
@@ -2,7 +2,7 @@

 [English](README.md) | 中文

-本包（package）注册固定的 `claude-code` subagent 提供方。每次接受运行请求后，它都会在发起委托的会话工作区中调用官方 Claude Agent SDK，通过共享子进程服务启动 SDK 分发的 Claude Code CLI，提交一个自包含的文本任务，并通过共享的 [`dsh-subagent`](../subagent/README.md) 结果约定仅返回最终答案。
+本包（package）注册固定的 `claude-code` subagent 提供方。每次接受运行请求后，它都会在发起委托的会话工作区中调用官方 Claude Agent SDK，通过共享子进程服务解析原生 `claude` 可执行文件，提交一个自包含的文本任务，并通过共享的 [`dsh-subagent`](../subagent/README.md) 结果约定仅返回最终答案。

 ## 启动与所有权

@@ -29,9 +29,9 @@ SDK 接收由文本块原样拼接成的任务。提供方会完整迭代 SDK
 | `env` | `{}` | 显式指定的 SDK/CLI 环境，叠加在由共享机制清除凭证后的父环境之上。 |
 | `disposeGraceMs` | `3000` | 共享进程树责任方各终止层级之间的宽限期，单位为毫秒且须为正有限值，并不得大于仓库共享的 [`MAX_TIMER_DELAY_MS`](../../util/timeout/README.md)；随后资源释放会等待整棵进程树退出。 |

-生产环境使用 `@anthropic-ai/claude-agent-sdk` 提供的 Claude Code CLI，以及宿主机原生设置与身份验证。本插件不安装另一份 CLI、不选择模型、不创建产品主目录、不执行登录，也不探测账户。具有凭证特征的环境变量会在显式 `env` 覆盖生效前被清除，因此供子进程使用的 API 密钥或 token 必须在该配置中显式提供。除非被覆盖，`ANTHROPIC_BASE_URL` 等非凭证端点变量以及 `PATH` 和 `HOME` 等普通环境变量仍会被继承。
+生产环境从子进程执行世界清除凭证后的 `PATH` 解析 `claude`，再应用显式 `env` 条目，并把所得路径作为 `pathToClaudeCodeExecutable` 交给 SDK。原生设置与身份验证继续是权威来源。本插件不安装另一份 CLI、不选择模型、不创建产品主目录、不执行登录，也不探测账户。具有凭证特征的环境变量会在显式 `env` 覆盖生效前被清除，因此供子进程使用的 API 密钥或 token 必须在该配置中显式提供。除非被覆盖，`ANTHROPIC_BASE_URL` 等非凭证端点变量以及 `PATH` 和 `HOME` 等普通环境变量仍会被继承。

-请安装此包，并将以下配置项添加到你自己的 `cordis.yml`。正式 CLI 配置默认不会加载此提供方，也不会暴露 `subagent_claude_code`。
+随附 profile 会在宿主上加载一次该提供方，而且在工具被调用前不会启动 Claude 进程。完整 Agent Preset 携带下列工具行并设置 `disabled: true`；复制一个 preset 后删除该字段，即可只向由该副本组装的 agent 暴露 `subagent_claude_code`。自定义宿主组装仍可直接使用两条配置行。

 ```yaml
 - id: subagent-claude-code
@@ -42,6 +42,7 @@ SDK 接收由文本块原样拼接成的任务。提供方会完整迭代 SDK

 - id: tool-subagent-claude-code
  name: '@deepseek-ai/dsh-tool-subagent'
+  disabled: true
  config:
    provider: claude-code
    toolName: subagent_claude_code
@@ -51,7 +52,7 @@ SDK 接收由文本块原样拼接成的任务。提供方会完整迭代 SDK

 ## 产品兼容性与证据

-运行时依赖精确锁定为 `@anthropic-ai/claude-agent-sdk@0.3.220`，其平台可选依赖提供 Claude Code 2.1.220。强制证据会通过无密钥回环产品路径与带密钥 DeepSeek 路径运行该官方发行版，而 Loader 组合则证明两个选择启用的产品包能够共存，且不会启动任一产品。
+运行时依赖精确锁定为 `@anthropic-ai/claude-agent-sdk@0.3.220`。生产运行使用原生 `claude` 安装；SDK 的平台可选载荷仍处于当前安装闭包，并作为独立分发后续项跟踪。强制证据会通过无密钥回环路径与带密钥 DeepSeek 路径运行兼容的原生产品，而 Loader 组合则证明两个产品包能够共存且不会启动任一产品。

 项目所有者按身份范围授权分发官方 SDK 及每个 SDK 版本声明的官方 CLI／平台载荷。[`THIRD_PARTY_NOTICES.md`](../../../THIRD_PARTY_NOTICES.md) 会披露当前可选载荷闭包，但不会把其声明条款归类为宽松许可证；其他无关的非宽松运行时依赖仍会使第三方声明门禁失败。

@@ -89,7 +90,7 @@ Claude Code 子任务会在一个全新的 SDK query 中接收独立文本任务

 - **每次运行均新建一个 query 和一个进程**：不支持续接、恢复、池化、进度流或产品会话持久化。
 - **宿主设置有意保持权威**：项目和用户设置可以改变模型、工具与行为；本提供方不提供经过筛选或与宿主环境隔离的生产模式。
- **产品安装与账户状态仍由原生机制管理**：不兼容的 SDK 载荷、配置错误或身份验证失败都会呈现为启动错误或运行错误；本插件不提供安装程序或登录流程。
+- **产品安装与账户状态仍由原生机制管理**：`claude` 缺失或不兼容、配置错误或身份验证失败都会呈现为启动错误或运行错误；本插件不提供安装程序或登录流程。
 - **没有人工交互路径**：`AskUserQuestion` 被禁用，其他交互回调也不存在，因此需要新审批或输入的任务会失败而不会挂起。
 - **仅返回最终文本**：推理、中间消息、工具通信、用量信息、stderr 和工作区差异仍只保留在产品内部。
 - **没有可选的共享能力**：对于本提供方，共享服务会拒绝输出 schema、子任务角色设定、工具筛选和 harness 深度强制约束。
@@ -59,19 +59,25 @@ class ClaudeCodeProvider implements SubagentProvider {
    private readonly config: ResolvedConfig,
  ) {}

-  start(request: ResolvedSubagentStartRequest) {
+  async start(request: ResolvedSubagentStartRequest) {
    const parentCwd = request.parent.session.header.cwd
    if (parentCwd === undefined) {
      throw new Error(
        'subagent-claude-code: no working directory for the child — delegate from a parent session that has one',
      )
    }
+    const executable = await this.ctx.subprocess.resolveExecutable(
+      'claude',
+      this.config.env,
+      request.signal,
+    )
    const spec: ClaudeCodeRunSpec = {
      cwd: resolveChildCwd(
        'subagent-claude-code',
        undefined,
        parentCwd,
      ),
+      executable,
      env: this.config.env,
      disposeGraceMs: this.config.disposeGraceMs,
      spawn: spawnSpec => this.ctx.subprocess.spawn(spawnSpec),
@@ -6,6 +6,7 @@
 */

 import { EventEmitter } from 'node:events'
+import { extname } from 'node:path'
 import type {
  SpawnedProcess,
  SpawnOptions,
@@ -40,17 +41,23 @@ export function sdkEnvironmentOverlay(
 * Translate one official SDK spawn request to the shared process owner.
 * @param options - command, arguments, workspace, environment, and forwarded signal from the SDK.
 * @param graceMs - process-tree termination grace.
+ * @param platform - host platform selecting the Windows batch-shim boundary.
 * @returns the fully explicit shared subprocess request.
 */
 export function claudeSpawnSpec(
  options: SpawnOptions,
  graceMs: number,
+  platform: NodeJS.Platform = process.platform,
 ): SubprocessSpawnSpec {
  if (options.cwd === undefined || options.cwd.length === 0) {
    throw new Error('subagent-claude-code: SDK spawn request omitted its workspace')
  }
+  const extension = extname(options.command).toLowerCase()
+  const argv = platform === 'win32' && (extension === '.cmd' || extension === '.bat')
+    ? ['cmd.exe', '/d', '/s', '/c', options.command, ...options.args]
+    : [options.command, ...options.args]
  return {
-    argv: [options.command, ...options.args],
+    argv,
    cwd: options.cwd,
    stdio: { stdin: 'pipe', stdout: 'pipe', stderr: 'inherit' },
    graceMs,
@@ -44,6 +44,8 @@ export const DEFAULT_DISPOSE_GRACE_MS = 3_000
 export interface ClaudeCodeRunSpec {
  /** Parent Session workspace supplied to the SDK and real CLI. */
  readonly cwd: string
+  /** Exact native Claude Code executable resolved from the host PATH. */
+  readonly executable: string
  /** Explicit deployment/test environment layered after shared scrubbing. */
  readonly env: Record<string, string>
  /** Subprocess termination grace passed to the shared process-tree owner. */
@@ -180,6 +182,7 @@ export function claudeQueryOptions(
  return {
    abortController: controller,
    cwd: spec.cwd,
+    pathToClaudeCodeExecutable: spec.executable,
    env: { ...scrubbedParentEnv(), ...spec.env },
    persistSession: false,
    disallowedTools: ['AskUserQuestion'],
@@ -7,7 +7,7 @@ import {
  rmSync,
 } from 'node:fs'
 import { tmpdir } from 'node:os'
-import { dirname, join, resolve } from 'node:path'
+import { delimiter, dirname, join, resolve } from 'node:path'
 import { fileURLToPath } from 'node:url'
 import { promisify } from 'node:util'
 import { Context } from 'cordis'
@@ -87,6 +87,7 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)(
      ]) mkdirSync(directory)

      const env = {
+        PATH: `${dirname(claudeBin)}${delimiter}${process.env.PATH ?? ''}`,
        ANTHROPIC_AUTH_TOKEN: apiKey,
        ANTHROPIC_BASE_URL: `${deepSeekBaseUrl()}/anthropic`,
        ANTHROPIC_MODEL: 'deepseek-v4-pro[1m]',
@@ -1,13 +1,15 @@
 import { execFile } from 'node:child_process'
 import {
+  copyFileSync,
  mkdirSync,
  mkdtempSync,
  readFileSync,
  rmSync,
+  symlinkSync,
  writeFileSync,
 } from 'node:fs'
 import { tmpdir } from 'node:os'
-import { dirname, join, resolve } from 'node:path'
+import { delimiter, dirname, join, resolve } from 'node:path'
 import { fileURLToPath } from 'node:url'
 import { promisify } from 'node:util'
 import type {
@@ -19,7 +21,7 @@ import { Context } from 'cordis'
 import { afterEach, describe, expect, it, vi } from 'vitest'
 import type { Agent } from '@deepseek-ai/dsh-agent'
 import SubagentService from '@deepseek-ai/dsh-subagent'
-import type { SubprocessHandle } from '@deepseek-ai/dsh-subprocess'
+import type { SubprocessHandle, SubprocessSpawnSpec } from '@deepseek-ai/dsh-subprocess'
 import LocalSubprocessService from '@deepseek-ai/dsh-subprocess-local'
 import * as claudeCode from '../src/index.ts'
 import {
@@ -98,9 +100,11 @@ afterEach(async () => {
 interface RealHarness {
  readonly ctx: Context
  readonly handles: SubprocessHandle[]
+  readonly spawnSpecs: SubprocessSpawnSpec[]
  readonly parent: Agent
  readonly workspace: string
  readonly env: Record<string, string>
+  readonly executable: string
 }

 async function realHarness(behavior: MessagesBehavior): Promise<{
@@ -112,9 +116,14 @@ async function realHarness(behavior: MessagesBehavior): Promise<{
  const workspace = join(root, 'workspace')
  const claudeConfig = join(root, 'claude-config')
  const xdgConfig = join(root, 'xdg')
+  const nativeBin = join(root, 'native-bin')
  mkdirSync(workspace)
  mkdirSync(claudeConfig)
  mkdirSync(xdgConfig)
+  mkdirSync(nativeBin)
+  const executable = join(nativeBin, process.platform === 'win32' ? 'claude.exe' : 'claude')
+  if (process.platform === 'win32') copyFileSync(claudeBin, executable)
+  else symlinkSync(claudeBin, executable)
  writeFileSync(
    join(claudeConfig, 'settings.json'),
    `${JSON.stringify({ model: settingsModel }, null, 2)}\n`,
@@ -122,6 +131,7 @@ async function realHarness(behavior: MessagesBehavior): Promise<{
  const fixture = await startMessagesFixture(behavior)
  fixtures.push(fixture)
  const env = {
+    PATH: `${nativeBin}${delimiter}${process.env.PATH ?? ''}`,
    ANTHROPIC_API_KEY: fakeKey,
    ANTHROPIC_BASE_URL: fixture.baseUrl,
    CLAUDE_CONFIG_DIR: claudeConfig,
@@ -141,8 +151,10 @@ async function realHarness(behavior: MessagesBehavior): Promise<{
  await ctx.plugin(SubagentService)
  await ctx.plugin(LocalSubprocessService)
  const handles: SubprocessHandle[] = []
+  const spawnSpecs: SubprocessSpawnSpec[] = []
  const spawn = ctx.subprocess.spawn.bind(ctx.subprocess)
  vi.spyOn(ctx.subprocess, 'spawn').mockImplementation((spec) => {
+    spawnSpecs.push(spec)
    const handle = spawn(spec)
    handles.push(handle)
    return handle
@@ -153,7 +165,7 @@ async function realHarness(behavior: MessagesBehavior): Promise<{
    session: { header: { cwd: workspace } },
  } as unknown as Agent
  return {
-    harness: { ctx, handles, parent, workspace, env },
+    harness: { ctx, handles, spawnSpecs, parent, workspace, env, executable },
    fixture,
  }
 }
@@ -195,7 +207,7 @@ describe('real Claude Agent SDK 0.3.220 and Claude Code 2.1.220', {
    expect(sdkPackage.version).toBe('0.3.220')
    expect(sdkPackage.claudeCodeVersion).toBe('2.1.220')
    expect(sdkPackage.optionalDependencies[platformPackage]).toBe('0.3.220')
-    const version = await execFileAsync(claudeBin, ['--version'], {
+    const version = await execFileAsync(harness.executable, ['--version'], {
      env: { ...process.env, ...harness.env },
    })
    expect(version.stdout.trim()).toBe('2.1.220 (Claude Code)')
@@ -212,6 +224,7 @@ describe('real Claude Agent SDK 0.3.220 and Claude Code 2.1.220', {
        message.type === 'system' && message.subtype === 'init',
    )
    expect(initMessage?.claude_code_version).toBe('2.1.220')
+    expect(harness.spawnSpecs[0]?.argv[0]).toBe(harness.executable)

    expect(fixture.requests).toHaveLength(1)
    const recorded = fixture.requests[0]!
@@ -248,6 +248,7 @@ function fakeRun(
  const options: FakeRun['options'] = []
  const spec: ClaudeCodeRunSpec = {
    cwd: '/workspace',
+    executable: '/native/claude',
    env: { ANTHROPIC_API_KEY: 'fake-key' },
    disposeGraceMs: 5,
    spawn: (spawnSpec) => {
@@ -331,6 +332,8 @@ describe('task admission and package contracts', () => {
    const child = fakeChild()
    const spawn = vi.spyOn(ctx.subprocess, 'spawn')
      .mockImplementation(() => child.handle)
+    const resolveExecutable = vi.spyOn(ctx.subprocess, 'resolveExecutable')
+      .mockResolvedValue('/native/claude')
    const warn = vi.spyOn(ctx.logger, 'warn').mockImplementation(() => {})
    await ctx.plugin(claudeCode, {
      env: {
@@ -352,6 +355,11 @@ describe('task admission and package contracts', () => {
    )
    expect(queryMock).not.toHaveBeenCalled()

+    resolveExecutable.mockRejectedValueOnce(new Error('claude missing from PATH'))
+    await expect(ctx.subagents.start('claude-code', request()))
+      .rejects.toThrow('claude missing from PATH')
+    expect(queryMock).not.toHaveBeenCalled()
+
    const run = await ctx.subagents.start('claude-code', request())
    child.settle({ exitCode: 9, signal: null })
    child.stdout.end()
@@ -362,6 +370,13 @@ describe('task admission and package contracts', () => {
    expect(warn).toHaveBeenCalledWith(expect.stringContaining(
      'subagent-claude-code: child run failed (error):',
    ))
+    expect(resolveExecutable).toHaveBeenCalledWith(
+      'claude',
+      expect.objectContaining({ ANTHROPIC_API_KEY: 'provider-fake-key' }),
+      expect.any(AbortSignal),
+    )
+    expect(queryMock.mock.calls[0]?.[0].options.pathToClaudeCodeExecutable)
+      .toBe('/native/claude')
    expect(spawn).toHaveBeenCalledWith(expect.objectContaining({
      cwd: process.cwd(),
      graceMs: 29,
@@ -441,6 +456,19 @@ describe('official spawn projection', () => {
    )).toThrow('SDK spawn request omitted its workspace')
  })

+  it.each(['cmd', 'bat'])('routes a Windows .%s shim through cmd.exe', (extension) => {
+    const command = String.raw`C:\Program Files\Claude\claude.${extension}`
+    const spec = claudeSpawnSpec(sdkSpawnOptions({
+      command,
+      args: ['--output-format', 'stream-json'],
+    }), 7, 'win32')
+
+    expect(spec.argv).toEqual([
+      'cmd.exe', '/d', '/s', '/c', command,
+      '--output-format', 'stream-json',
+    ])
+  })
+
  it('projects streams, exit facts, listeners, and idempotent tree termination', async () => {
    const child = fakeChild({ exitOnTerminate: false })
    const process = new ManagedClaudeCodeProcess(child.handle)
@@ -508,6 +536,7 @@ describe('query options and result mapping', () => {
    const captured: SubprocessHandle[] = []
    const spec: ClaudeCodeRunSpec = {
      cwd: '/workspace',
+      executable: '/native/claude',
      env: {
        HOST_VISIBLE: 'overridden',
        ANTHROPIC_API_KEY: 'explicit-fake-key',
@@ -523,6 +552,7 @@ describe('query options and result mapping', () => {
    expect(options).toMatchObject({
      abortController: controller,
      cwd: '/workspace',
+      pathToClaudeCodeExecutable: '/native/claude',
      persistSession: false,
      disallowedTools: ['AskUserQuestion'],
    })
@@ -670,6 +700,7 @@ describe('run publication, cancellation, and settlement', () => {
    let index = 0
    const spec: ClaudeCodeRunSpec = {
      cwd: '/workspace',
+      executable: '/native/claude',
      env: {},
      disposeGraceMs: 5,
      spawn: () => children[index++]!.handle,
@@ -720,6 +751,7 @@ describe('run publication, cancellation, and settlement', () => {
      request(undefined, parentAbort.signal),
      {
        cwd: '/workspace',
+        executable: '/native/claude',
        env: {},
        disposeGraceMs: 5,
        spawn: () => child.handle,
@@ -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/subagent/subagent-codex/README.md
-README.md: c75a98550894f180d0f37e9cdd135b961488f726
-README.zh.md: 0dd3280b363c1a26250aa00cfeba5398114a2e3e
+README.md: f10ebe0448b2942e2cad8efecb6be4681cf601a6
+README.zh.md: ef107577afcdc81a64ea46b2e996d46a562d1505
@@ -27,7 +27,7 @@ The provider advertises no optional start-time capabilities and reports `inherit

 Production resolves `codex` from `PATH` and uses the host's native Codex configuration and authentication. The plugin does not install Codex, select a model, create `CODEX_HOME`, log in, or probe a version. Credential-shaped ambient variables are removed by the subprocess seam, so an API key intended for the child must be supplied explicitly in `env`; ordinary ambient values such as `PATH` and `HOME` remain available unless overridden.

-Install this package and add the following rows to your own `cordis.yml`. Shipped CLI configurations do not load this provider or expose `subagent_codex` by default.
+Shipped profiles load this provider once on the host and start no Codex process until a tool call. Full Agent Presets carry the tool row below with `disabled: true`; copy a preset and remove that field to expose `subagent_codex` only to agents composed from the copy. A custom host composition can still use both rows directly.

 ```yaml
 - id: subagent-codex
@@ -38,6 +38,7 @@ Install this package and add the following rows to your own `cordis.yml`. Shippe

 - id: tool-subagent-codex
  name: '@deepseek-ai/dsh-tool-subagent'
+  disabled: true
  config:
    provider: codex
    toolName: subagent_codex
@@ -27,7 +27,7 @@

 生产环境会从 `PATH` 中解析 `codex`，并使用宿主机原生的 Codex 配置与身份验证。本插件不安装 Codex、不选择模型、不创建 `CODEX_HOME`、不执行登录，也不探测版本。子进程 seam 会移除具有凭证特征的环境变量，因此供子进程使用的 API 密钥必须在 `env` 中显式提供；除非被覆盖，`PATH` 和 `HOME` 等普通环境变量值仍然可用。

-请安装此包，并将以下配置项添加到你自己的 `cordis.yml`。正式 CLI 配置默认不会加载此提供方，也不会暴露 `subagent_codex`。
+随附 profile 会在宿主上加载一次该提供方，而且在工具被调用前不会启动 Codex 进程。完整 Agent Preset 携带下列工具行并设置 `disabled: true`；复制一个 preset 后删除该字段，即可只向由该副本组装的 agent 暴露 `subagent_codex`。自定义宿主组装仍可直接使用两条配置行。

 ```yaml
 - id: subagent-codex
@@ -38,6 +38,7 @@

 - id: tool-subagent-codex
  name: '@deepseek-ai/dsh-tool-subagent'
+  disabled: true
  config:
    provider: codex
    toolName: subagent_codex
@@ -1226,6 +1226,12 @@ importers:
      '@deepseek-ai/dsh-subagent':
        specifier: workspace:^
        version: link:../../subagent/subagent
+      '@deepseek-ai/dsh-subagent-claude-code':
+        specifier: workspace:^
+        version: link:../../subagent/subagent-claude-code
+      '@deepseek-ai/dsh-subagent-codex':
+        specifier: workspace:^
+        version: link:../../subagent/subagent-codex
      '@deepseek-ai/dsh-subagent-fork':
        specifier: workspace:^
        version: link:../../subagent/subagent-fork