From bdff8573b686773fc5d82ab71eb047e8cb7a48c8 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 12:44:00 +0800
Subject: [PATCH 01/27] ci: run coverage on in-house vm-backup pool

Coverage does not gate merges, so move it off the metered
dsh-enterprise-ubuntu-24-04-32core-test pool onto the in-house
self-hosted pool (vm-backup label, 64-core).

Also switch the pnpm store cache path to ~ so it resolves under
both /home/runner (hosted) and self-hosted home directories.

Verified on the self-hosted pool: the full coverage job (including
prepare-ci-bubblewrap and the exhaustive suite) completed green in
~5 min.
---
 .github/workflows/ci.yml | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 2eceefa114..a2e70cab6f 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -77,7 +77,9 @@ jobs:
 
   node-24-coverage:
     if: github.event_name == 'pull_request'
-    runs-on: dsh-enterprise-ubuntu-24-04-32core-test
+    # Coverage does not gate merges, so it runs on the in-house pool
+    # (self-hosted, 64-core) instead of the metered enterprise pool.
+    runs-on: [self-hosted, linux, x64, vm-backup]
     name: node 24 / coverage
     env:
       DSH_COVERAGE_MAX_WORKERS: '24'
@@ -89,7 +91,8 @@ jobs:
 
       - uses: actions/cache/restore@v4
         with:
-          path: /home/runner/.local/share/pnpm/store/v11
+          # ~ resolves on both hosted (/home/runner) and self-hosted homes
+          path: ~/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
           restore-keys: |
             ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-

From 81890d7a994ab791c7db8bc93667caf21fc38f45 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 16:37:59 +0800
Subject: [PATCH 02/27] =?UTF-8?q?ci:=20address=20review=20=E2=80=94=20same?=
 =?UTF-8?q?-repo=20guard,=20keep=20cache=20path=20identical?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Restrict node-24-coverage to same-repo PRs so fork-originated code can
  never reach the self-hosted runner (defense in depth; the repo is
  private with forking disabled today).
- Revert the pnpm cache path to the literal /home/runner/... save-side
  path: actions/cache hashes the path into the cache version, so the ~
  variant could never match the cache saved by the master lane. On
  self-hosted the persistent local pnpm store covers warm installs.
- Drop the incorrect 'does not gate merges' claim: node-24-coverage is
  needed by all-checks-passed. Pool capacity notes moved into comments.
---
 .github/workflows/ci.yml | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index a2e70cab6f..dd60593a85 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -76,9 +76,14 @@ jobs:
           compression-level: 0
 
   node-24-coverage:
-    if: github.event_name == 'pull_request'
-    # Coverage does not gate merges, so it runs on the in-house pool
-    # (self-hosted, 64-core) instead of the metered enterprise pool.
+    # Same-repo PRs only: this lane runs on an in-house self-hosted runner,
+    # so fork-originated code must never land here. The repo is currently
+    # private with forking disabled; this guard keeps that invariant explicit
+    # if either setting ever changes.
+    if: github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository
+    # Runs on the in-house pool (self-hosted, 64-core) instead of the metered
+    # enterprise pool. The pool holds 4 always-on instances plus 4 registered
+    # spares; the runner service is systemd-managed and self-healing.
     runs-on: [self-hosted, linux, x64, vm-backup]
     name: node 24 / coverage
     env:
@@ -91,8 +96,12 @@ jobs:
 
       - uses: actions/cache/restore@v4
         with:
-          # ~ resolves on both hosted (/home/runner) and self-hosted homes
-          path: ~/.local/share/pnpm/store/v11
+          # Path must stay byte-identical to the save-side path in the master
+          # lane: actions/cache hashes the literal path into the cache version,
+          # so any variation (e.g. ~) would never match the saved cache. On
+          # self-hosted this restore simply misses and the persistent local
+          # pnpm store covers warm installs instead.
+          path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
           restore-keys: |
             ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-

From 5818fd62242f8799484fbf166c11f1fc8434bf48 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 23:00:56 +0800
Subject: [PATCH 03/27] =?UTF-8?q?ci:=20address=20second=20review=20round?=
 =?UTF-8?q?=20=E2=80=94=20dependabot=20lane,=20drop=20dead=20restore,=20up?=
 =?UTF-8?q?date=20topology=20note?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Route untrusted PRs (forks + Dependabot, same author test as e2e.yml)
  back to the hosted enterprise pool via a runs-on expression: Dependabot
  PRs are same-repo, so the previous head.repo guard admitted
  dependency-supplied code onto the persistent self-hosted VM. A single
  job with pool selection keeps all-checks-passed free of skips.
- Drop the pnpm-store cache restore from this lane: on self-hosted the
  hosted-path cache actually HIT (Linux key) and spent ~52 s pulling
  181 MB into a path pnpm never reads; the persistent local store
  already serves warm installs in seconds.
- Update the larger-hosted-runners Agent Note (en/zh + i18n pairing
  record) so the decision record describes the shipped topology:
  coverage on the in-house vm-backup pool for trusted PRs, hosted
  Ubuntu 24.04 32-core retained for untrusted PRs.
---
 ...ence-based-larger-hosted-runners.i18n.yaml |  4 +-
 ...22-evidence-based-larger-hosted-runners.md |  2 +-
 ...evidence-based-larger-hosted-runners.zh.md |  2 +-
 .github/workflows/ci.yml                      | 39 +++++++++----------
 4 files changed, 23 insertions(+), 24 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 9d87cb9ad3..360395102e 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: aaeab4ed9ae9687598f9f1d4a862120405697672
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: 72b69c85908990a9f35b60f4c0a2ce213f9c8134
+2026-07-22-evidence-based-larger-hosted-runners.md: c3e6344ae61669da4810090e558589875ca7536e
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: e5b322673b7a1eb004eb15b3784d21f500e83719
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index aaeab4ed9a..c3e6344ae6 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -12,7 +12,7 @@ Larger runners make it possible to pay setup once and parallelize inside the rep
 
 ## Decision
 
-The enterprise keeps repo-restricted x64 larger-runner pools for Ubuntu and Windows. Ordinary pull requests name three 32-core pools directly: Ubuntu 24.04 for exhaustive coverage, Ubuntu latest for the remaining primary Node 24 inventory, and Windows 2025 for blocking Windows contracts. Public IPs are disabled, and workflow concurrency remains bounded because an autoscaling ceiling neither allocates idle machines nor makes repository work scale without limit.
+The enterprise keeps repo-restricted x64 larger-runner pools for Ubuntu and Windows. Ordinary pull requests name two 32-core hosted pools directly: Ubuntu latest for the remaining primary Node 24 inventory and Windows 2025 for blocking Windows contracts. Exhaustive coverage moved off the metered Ubuntu 24.04 32-core pool onto the in-house self-hosted pool (`vm-backup` label: a 64-core VM running four always-on systemd-managed runner instances plus four registered spares) for trusted same-repo PRs; untrusted PRs — forks and Dependabot — keep coverage on the hosted Ubuntu 24.04 32-core pool so dependency-supplied code never reaches the persistent VM. Public IPs are disabled, and workflow concurrency remains bounded because an autoscaling ceiling neither allocates idle machines nor makes repository work scale without limit.
 
 The required primary path depends on those enterprise pools. Standard GitHub-hosted jobs retain the Node 22.19, Node 26, and Python SDK compatibility contracts, while the [portable recovery boundary](2026-07-23-portable-required-pull-request-ci.md) and [serial reference](2026-07-21-serial-cross-platform-ci-reference.md) keep complete standard-runner evidence available on `master`. `suite=larger-runner-benchmark` compares isolated critical lanes across provisioned sizes, and `suite=consolidated-runner-benchmark` compares whole aggregates. Each benchmark reports its observed processor and memory capacity before running repository work.
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index 72b69c8590..e5b322673b 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -12,7 +12,7 @@ Status: implemented
 
 ## 决策
 
-企业保留仅限本仓库使用的 Ubuntu 和 Windows x64 大型运行器池。普通拉取请求直接指定 3 个 32 核运行器池：Ubuntu 24.04 用于完整覆盖率，Ubuntu latest 用于其余主 Node 24 清单，Windows 2025 用于阻塞性 Windows 契约。公网 IP 已禁用；工作流并发仍设有边界，因为自动扩缩容上限既不会分配闲置机器，也不意味着仓库工作可以无限扩展。
+企业保留仅限本仓库使用的 Ubuntu 和 Windows x64 大型运行器池。普通拉取请求直接指定 2 个 32 核托管运行器池：Ubuntu latest 用于其余主 Node 24 清单，Windows 2025 用于阻塞性 Windows 契约。完整覆盖率已从计费的 Ubuntu 24.04 32 核池迁移至公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位），仅面向可信的同仓库拉取请求；不可信的拉取请求——fork 与 Dependabot——的覆盖率仍在托管的 Ubuntu 24.04 32 核池上运行，确保依赖方提供的代码永远不会进入持久化虚拟机。公网 IP 已禁用；工作流并发仍设有边界，因为自动扩缩容上限既不会分配闲置机器，也不意味着仓库工作可以无限扩展。
 
 必需主路径依赖这些企业级运行器池。GitHub 标准托管作业保留 Node 22.19、Node 26 和 Python SDK 兼容性契约，而[可移植恢复边界](2026-07-23-portable-required-pull-request-ci.md)与[串行参考流程](2026-07-21-serial-cross-platform-ci-reference.md)则在 `master` 上持续提供完整的标准运行器证据。`suite=larger-runner-benchmark` 比较已预配规格上相互独立的关键通道，`suite=consolidated-runner-benchmark` 则比较完整聚合流程。每项基准测试都会先报告实测的处理器和内存容量，再运行仓库工作。
 
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index dd60593a85..dc45bc9a7a 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -76,15 +76,19 @@ jobs:
           compression-level: 0
 
   node-24-coverage:
-    # Same-repo PRs only: this lane runs on an in-house self-hosted runner,
-    # so fork-originated code must never land here. The repo is currently
-    # private with forking disabled; this guard keeps that invariant explicit
-    # if either setting ever changes.
-    if: github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository
-    # Runs on the in-house pool (self-hosted, 64-core) instead of the metered
-    # enterprise pool. The pool holds 4 always-on instances plus 4 registered
-    # spares; the runner service is systemd-managed and self-healing.
-    runs-on: [self-hosted, linux, x64, vm-backup]
+    if: github.event_name == 'pull_request'
+    # Trusted same-repo PRs run on the in-house pool (self-hosted, 64-core;
+    # 4 always-on systemd-managed instances plus 4 registered spares) instead
+    # of the metered enterprise pool. Untrusted PRs — forks and Dependabot
+    # (same-repo but dependency-supplied code; same author test as e2e.yml) —
+    # stay on the hosted enterprise pool so no untrusted code reaches the
+    # persistent self-hosted VM. Selecting the pool via runs-on keeps this a
+    # single job, so the all-checks-passed aggregate never sees a skip.
+    runs-on: >-
+      ${{ (github.event.pull_request.head.repo.full_name != github.repository
+           || github.event.pull_request.user.login == 'dependabot[bot]')
+          && 'dsh-enterprise-ubuntu-24-04-32core-test'
+          || fromJSON('["self-hosted", "linux", "x64", "vm-backup"]') }}
     name: node 24 / coverage
     env:
       DSH_COVERAGE_MAX_WORKERS: '24'
@@ -94,17 +98,12 @@ jobs:
         with:
           persist-credentials: false
 
-      - uses: actions/cache/restore@v4
-        with:
-          # Path must stay byte-identical to the save-side path in the master
-          # lane: actions/cache hashes the literal path into the cache version,
-          # so any variation (e.g. ~) would never match the saved cache. On
-          # self-hosted this restore simply misses and the persistent local
-          # pnpm store covers warm installs instead.
-          path: /home/runner/.local/share/pnpm/store/v11
-          key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
-          restore-keys: |
-            ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-
+      # No pnpm-store cache restore in this lane: on the self-hosted pool
+      # pnpm's persistent store lives outside /home/runner, so restoring the
+      # hosted cache here downloads ~180 MB into a path pnpm never reads
+      # (measured: 52 s restore, then a 2.8 s install straight from the
+      # persistent store). The rare hosted (untrusted-PR) run just does a
+      # cold install.
 
       - uses: actions/setup-node@v6
         with:

From e532c9ccc245a2360df74bb6d4795ea1f3c13162 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 23:12:20 +0800
Subject: [PATCH 04/27] ci: restore pnpm cache on the hosted leg only

Keep the cache restore for the ephemeral hosted (untrusted-PR) leg where
it is a genuine speedup, gated by the same expression as the runs-on
pool selector; the self-hosted leg skips it and installs from the
persistent local store.
---
 .github/workflows/ci.yml | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index dc45bc9a7a..d0d51fde9b 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -98,12 +98,21 @@ jobs:
         with:
           persist-credentials: false
 
-      # No pnpm-store cache restore in this lane: on the self-hosted pool
-      # pnpm's persistent store lives outside /home/runner, so restoring the
-      # hosted cache here downloads ~180 MB into a path pnpm never reads
-      # (measured: 52 s restore, then a 2.8 s install straight from the
-      # persistent store). The rare hosted (untrusted-PR) run just does a
-      # cold install.
+      # Restore the pnpm-store cache only on the hosted (untrusted-PR) leg,
+      # where the VM is ephemeral and the same-region download is fast. On
+      # the self-hosted leg pnpm's persistent store lives outside
+      # /home/runner, so this restore would spend ~52 s pulling ~180 MB into
+      # a path pnpm never reads (measured; install then took 2.8 s straight
+      # from the persistent store). Condition mirrors the runs-on selector.
+      - uses: actions/cache/restore@v4
+        if: >-
+          github.event.pull_request.head.repo.full_name != github.repository
+          || github.event.pull_request.user.login == 'dependabot[bot]'
+        with:
+          path: /home/runner/.local/share/pnpm/store/v11
+          key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
+          restore-keys: |
+            ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-
 
       - uses: actions/setup-node@v6
         with:

From 8d53d44b6055ce37aecdd22be1eb9d1429a96cae Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 23:49:58 +0800
Subject: [PATCH 05/27] docs(ci): reconcile every present-tense topology
 description with the coverage lane move
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Sweep all remaining sources that still described coverage as an
enterprise 32-core job: the ci.yml jobs preamble, the three-job
paragraph of the larger-hosted-runners note, and the required-pool
sentence of the portable-recovery note — English and Chinese sides of
both notes, with their i18n pairing records re-recorded.
---
 ...026-07-22-evidence-based-larger-hosted-runners.i18n.yaml | 4 ++--
 .../2026-07-22-evidence-based-larger-hosted-runners.md      | 2 +-
 .../2026-07-22-evidence-based-larger-hosted-runners.zh.md   | 2 +-
 .../2026-07-23-portable-required-pull-request-ci.i18n.yaml  | 4 ++--
 .../process/2026-07-23-portable-required-pull-request-ci.md | 2 +-
 .../2026-07-23-portable-required-pull-request-ci.zh.md      | 2 +-
 .github/workflows/ci.yml                                    | 6 ++++--
 7 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 360395102e..4d781caa54 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: c3e6344ae61669da4810090e558589875ca7536e
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: e5b322673b7a1eb004eb15b3784d21f500e83719
+2026-07-22-evidence-based-larger-hosted-runners.md: 88b9e6d83777172d8afb6a391512e5f293b81171
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: b1105f00cd08b1af633d258ea4ff28a835ce6074
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index c3e6344ae6..88b9e6d837 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -18,7 +18,7 @@ The required primary path depends on those enterprise pools. Standard GitHub-hos
 
 The former gate-level and coarse primary shard jobs are absent from the workflow. Their static, lint, coverage, snapshot, and scenario shard selectors are also absent from the repository, so an unused diagnostic path cannot preserve a second CI architecture.
 
-Linux primary work uses three independent 32-core jobs. Coverage runs alone with its own worker bound, and the static scheduler runs alone so its result has no post-build consumer tail. After static gates finish, that job publishes its emitted `apps/*/lib`, `packages/*/*/lib`, and `vendor/*/lib` tree as a run-scoped artifact. The third job restores that exact tree, then starts lint, Node 24 runtime compatibility, build-backed snapshots, and all artifact consumers without repeating the build. Generated NodeNext consumer directories are excluded from ESLint discovery because the artifact check removes them while these processes overlap. The pnpm store and ESLint cache are restored without putting cache uploads on the pull-request critical path. Performance reports use each job's `startedAt` to `completedAt` interval; runner queue delay is capacity evidence, not repository execution time.
+Linux primary work uses three independent jobs: static gates and the consumer tail on hosted 32-core pools, and coverage on the in-house self-hosted 64-core pool for trusted PRs (hosted 32-core for untrusted ones). Coverage runs alone with its own worker bound, and the static scheduler runs alone so its result has no post-build consumer tail. After static gates finish, that job publishes its emitted `apps/*/lib`, `packages/*/*/lib`, and `vendor/*/lib` tree as a run-scoped artifact. The third job restores that exact tree, then starts lint, Node 24 runtime compatibility, build-backed snapshots, and all artifact consumers without repeating the build. Generated NodeNext consumer directories are excluded from ESLint discovery because the artifact check removes them while these processes overlap. The pnpm store and ESLint cache are restored without putting cache uploads on the pull-request critical path. Performance reports use each job's `startedAt` to `completedAt` interval; runner queue delay is capacity evidence, not repository execution time.
 
 Windows shares one 32-core setup across the blocking build and production site plus observational built-artifact contracts. Linux owns the duplicate lint, coverage, and snapshot inventories because running those observational copies on Windows extends the paid critical path without adding a blocking platform claim.
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index e5b322673b..b1105f00cd 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -18,7 +18,7 @@ Status: implemented
 
 原有的门禁级和粗粒度主流程分片作业已从工作流中移除。相应的静态、lint、覆盖率、快照和场景分片选择器也已从仓库中移除，因此未使用的诊断路径无法继续维系第二套 CI 架构。
 
-Linux 主流程使用 3 个相互独立的 32 核作业。覆盖率单独运行，并设有自己的工作线程上限；静态调度器也单独运行，因此构建后的消费方不会拖延其结果。静态门禁完成后，该作业将其生成的 `apps/*/lib`、`packages/*/*/lib` 和 `vendor/*/lib` 目录树作为仅供本次运行使用的产物发布。第三个作业恢复完全相同的目录树，再让 lint、Node 24 运行时兼容性、依赖构建产物的快照和所有产物消费方基于构建完成后的工作树启动，而不重复构建。生成的 NodeNext 消费方目录不会纳入 ESLint 的文件发现范围，因为这些进程重叠执行时，产物检查会删除这些目录。pnpm store 和 ESLint 缓存会得到恢复，但缓存上传不会进入拉取请求关键路径。性能报告采用每个作业从 `startedAt` 到 `completedAt` 的区间；运行器排队延迟是容量证据，而非仓库执行时间。
+Linux 主流程使用 3 个相互独立的作业：静态门禁与消费方尾部作业运行在托管 32 核池上，覆盖率对可信拉取请求运行在公司自有的自托管 64 核池上（不可信请求仍用托管 32 核池）。覆盖率单独运行，并设有自己的工作线程上限；静态调度器也单独运行，因此构建后的消费方不会拖延其结果。静态门禁完成后，该作业将其生成的 `apps/*/lib`、`packages/*/*/lib` 和 `vendor/*/lib` 目录树作为仅供本次运行使用的产物发布。第三个作业恢复完全相同的目录树，再让 lint、Node 24 运行时兼容性、依赖构建产物的快照和所有产物消费方基于构建完成后的工作树启动，而不重复构建。生成的 NodeNext 消费方目录不会纳入 ESLint 的文件发现范围，因为这些进程重叠执行时，产物检查会删除这些目录。pnpm store 和 ESLint 缓存会得到恢复，但缓存上传不会进入拉取请求关键路径。性能报告采用每个作业从 `startedAt` 到 `completedAt` 的区间；运行器排队延迟是容量证据，而非仓库执行时间。
 
 Windows 以一次 32 核环境设置同时承载阻塞性构建、生产网站和观测性的构建产物契约。重复的 lint、覆盖率和快照清单由 Linux 承担，因为在 Windows 上运行这些观测性副本会延长付费关键路径，却不会新增任何阻塞性平台契约。
 
diff --git a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml
index f8b54b0ec5..ed97fe08a7 100644
--- a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-23-portable-required-pull-request-ci.md: 9cf8d97016300c5258c075879176aa6abd64e59e
-2026-07-23-portable-required-pull-request-ci.zh.md: c6839a133d0c3fe7a699362f6168e17d827a5b61
+2026-07-23-portable-required-pull-request-ci.md: 29b2cfa3f431a4a8be4aaa685b16cffdb4bf2593
+2026-07-23-portable-required-pull-request-ci.zh.md: 8b6d067637ce83c09529977f463f16dfa4af5a8b
diff --git a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md
index 9cf8d97016..29b2cfa3f4 100644
--- a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md
+++ b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md
@@ -12,7 +12,7 @@ Billing health, a runner definition's `Ready` state, and a large autoscaling cei
 
 ## Decision
 
-[CI](../../../../.github/workflows/ci.yml) runs the required primary Node 24 and Windows jobs on repo-restricted enterprise 32-core pools. Standard `ubuntu-latest` jobs retain Node 22.19, Node 26, and Python SDK compatibility, and `master` runs complete serial Linux, macOS, and Windows references. Those standard-hosted jobs keep the portable execution boundary observable without duplicating the primary inventory on every pull request.
+[CI](../../../../.github/workflows/ci.yml) runs the required primary Node 24 and Windows jobs on repo-restricted enterprise 32-core pools, except exhaustive coverage, which runs on the in-house self-hosted 64-core pool for trusted same-repo pull requests (hosted 32-core for forks and Dependabot). Standard `ubuntu-latest` jobs retain Node 22.19, Node 26, and Python SDK compatibility, and `master` runs complete serial Linux, macOS, and Windows references. Those standard-hosted jobs keep the portable execution boundary observable without duplicating the primary inventory on every pull request.
 
 The two Linux primary jobs, Node compatibility, Python SDK, and `windows node 24 / complete` remain dependencies of `all checks passed`; branch protection continues to require `e2e` and `all checks passed`. There is no automatic fallback when an enterprise label cannot allocate: the standard jobs continue to report their own contracts, but they cannot manufacture the missing required result.
 
diff --git a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md
index c6839a133d..8b6d067637 100644
--- a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md
+++ b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md
@@ -12,7 +12,7 @@ Status: implemented
 
 ## 决策
 
-[CI](../../../../.github/workflows/ci.yml) 在仅限本仓库使用的企业级 32 核运行器池上运行必需的主 Node 24 作业和 Windows 作业。标准 `ubuntu-latest` 作业保留 Node 22.19、Node 26 和 Python SDK 兼容性，`master` 则运行完整的 Linux、macOS 和 Windows 串行参考流程。这些标准托管作业让可移植执行边界保持可观测，而不必在每个拉取请求中重复主清单。
+[CI](../../../../.github/workflows/ci.yml) 在仅限本仓库使用的企业级 32 核运行器池上运行必需的主 Node 24 作业和 Windows 作业；唯一例外是完整覆盖率——可信的同仓库拉取请求在公司自有的自托管 64 核池上运行（fork 与 Dependabot 仍用托管 32 核池）。标准 `ubuntu-latest` 作业保留 Node 22.19、Node 26 和 Python SDK 兼容性，`master` 则运行完整的 Linux、macOS 和 Windows 串行参考流程。这些标准托管作业让可移植执行边界保持可观测，而不必在每个拉取请求中重复主清单。
 
 两项 Linux 主作业、Node 兼容性、Python SDK 和 `windows node 24 / complete` 继续作为 `all checks passed` 的依赖项；分支保护继续要求 `e2e` 和 `all checks passed`。企业级运行器标签无法分配运行器时没有自动后备机制：标准作业会继续报告各自的契约，但无法产出缺失的必需结果。
 
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index d0d51fde9b..a21086754b 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -27,8 +27,10 @@ env:
 
 jobs:
 
-  # Three enterprise jobs isolate coverage, static analysis, and the
-  # build-backed consumer tail. The static job publishes its exact build so
+  # Three independent Linux jobs isolate coverage, static analysis, and the
+  # build-backed consumer tail: static and consumers on hosted enterprise
+  # 32-core pools; coverage on the in-house self-hosted pool for trusted PRs
+  # (hosted for forks/Dependabot). The static job publishes its exact build so
   # consumers do not repeat the longest part of their critical path.
   node-24:
     if: github.event_name == 'pull_request'

From 1a5d892ec53beb5f1b7212decfc2a10bd9ea2741 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sat, 25 Jul 2026 00:45:31 +0800
Subject: [PATCH 06/27] ci: halve coverage workers on the shared self-hosted
 leg
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The hosted 32-core runner is exclusive to one job, but the vm-backup
pool shares one 64-core VM across four runner instances; concurrent
PRs could stack 4×24 = 96 Vitest workers and re-trigger the documented
aggregate-contention failures in the timing-sensitive process suites.
Bound the self-hosted leg at 12 workers per job (48 host-wide fully
loaded) and keep 24 on the hosted leg, selected by the same expression
as the pool.
---
 .github/workflows/ci.yml | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index a21086754b..dc5ad98ec4 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -93,7 +93,16 @@ jobs:
           || fromJSON('["self-hosted", "linux", "x64", "vm-backup"]') }}
     name: node 24 / coverage
     env:
-      DSH_COVERAGE_MAX_WORKERS: '24'
+      # Worker bound is per-leg: the hosted 32-core runner is exclusive to
+      # one job, but the self-hosted pool shares one 64-core VM across four
+      # runner instances, so concurrent PRs would otherwise stack up to
+      # 4×24 = 96 workers and re-trigger the aggregate-contention failures
+      # documented for the timing-sensitive process suites. 12 per job caps
+      # the shared host at 48 workers even fully loaded.
+      DSH_COVERAGE_MAX_WORKERS: >-
+        ${{ (github.event.pull_request.head.repo.full_name != github.repository
+             || github.event.pull_request.user.login == 'dependabot[bot]')
+            && '24' || '12' }}
       DSH_GATE_CONCURRENCY: '8'
     steps:
       - uses: actions/checkout@v6

From f09539581d33a5110c97d81cfe2778c74337690e Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sat, 25 Jul 2026 00:54:35 +0800
Subject: [PATCH 07/27] docs(ci): record disabled forking as an explicit
 precondition of the self-hosted lane
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The pool selector is defense-in-depth only — pull_request executes the
PR's own workflow definition, so YAML cannot enforce runner trust. Make
the actual enforcement boundary explicit in the decision record:
org-side disabled forking (the public release is an isolated read-only
mirror under a separate org), with migration to a repo-restricted
org-level runner group with base-branch workflow pinning as a hard
gate before forking could ever be enabled.
---
 .../2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml | 4 ++--
 .../2026-07-22-evidence-based-larger-hosted-runners.md        | 2 +-
 .../2026-07-22-evidence-based-larger-hosted-runners.zh.md     | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 4d781caa54..ea3a57e072 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: 88b9e6d83777172d8afb6a391512e5f293b81171
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: b1105f00cd08b1af633d258ea4ff28a835ce6074
+2026-07-22-evidence-based-larger-hosted-runners.md: 497c6f297d79245fb40cd30457e4b1d1e36db651
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: bcb0c6e9f11081b2cff696a9b6b425a40ee4aeb4
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 88b9e6d837..497c6f297d 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -12,7 +12,7 @@ Larger runners make it possible to pay setup once and parallelize inside the rep
 
 ## Decision
 
-The enterprise keeps repo-restricted x64 larger-runner pools for Ubuntu and Windows. Ordinary pull requests name two 32-core hosted pools directly: Ubuntu latest for the remaining primary Node 24 inventory and Windows 2025 for blocking Windows contracts. Exhaustive coverage moved off the metered Ubuntu 24.04 32-core pool onto the in-house self-hosted pool (`vm-backup` label: a 64-core VM running four always-on systemd-managed runner instances plus four registered spares) for trusted same-repo PRs; untrusted PRs — forks and Dependabot — keep coverage on the hosted Ubuntu 24.04 32-core pool so dependency-supplied code never reaches the persistent VM. Public IPs are disabled, and workflow concurrency remains bounded because an autoscaling ceiling neither allocates idle machines nor makes repository work scale without limit.
+The enterprise keeps repo-restricted x64 larger-runner pools for Ubuntu and Windows. Ordinary pull requests name two 32-core hosted pools directly: Ubuntu latest for the remaining primary Node 24 inventory and Windows 2025 for blocking Windows contracts. Exhaustive coverage moved off the metered Ubuntu 24.04 32-core pool onto the in-house self-hosted pool (`vm-backup` label: a 64-core VM running four always-on systemd-managed runner instances plus four registered spares) for trusted same-repo PRs; untrusted PRs — forks and Dependabot — keep coverage on the hosted Ubuntu 24.04 32-core pool so dependency-supplied code never reaches the persistent VM. **Precondition: repository forking stays disabled.** The workflow's pool selector is defense-in-depth only — `pull_request` executes the PR's own workflow definition, so YAML cannot enforce runner trust against a fork that edits it. Disabled forking (org-side, not PR-editable) is the enforcement boundary; the planned public release is an isolated read-only mirror under a separate org, preserving this. Before forking is ever enabled, the runners must first move into an org-level runner group restricted to this repository with base-branch workflow pinning — that migration is the gate, not a follow-up. Public IPs are disabled, and workflow concurrency remains bounded because an autoscaling ceiling neither allocates idle machines nor makes repository work scale without limit.
 
 The required primary path depends on those enterprise pools. Standard GitHub-hosted jobs retain the Node 22.19, Node 26, and Python SDK compatibility contracts, while the [portable recovery boundary](2026-07-23-portable-required-pull-request-ci.md) and [serial reference](2026-07-21-serial-cross-platform-ci-reference.md) keep complete standard-runner evidence available on `master`. `suite=larger-runner-benchmark` compares isolated critical lanes across provisioned sizes, and `suite=consolidated-runner-benchmark` compares whole aggregates. Each benchmark reports its observed processor and memory capacity before running repository work.
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index b1105f00cd..bcb0c6e9f1 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -12,7 +12,7 @@ Status: implemented
 
 ## 决策
 
-企业保留仅限本仓库使用的 Ubuntu 和 Windows x64 大型运行器池。普通拉取请求直接指定 2 个 32 核托管运行器池：Ubuntu latest 用于其余主 Node 24 清单，Windows 2025 用于阻塞性 Windows 契约。完整覆盖率已从计费的 Ubuntu 24.04 32 核池迁移至公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位），仅面向可信的同仓库拉取请求；不可信的拉取请求——fork 与 Dependabot——的覆盖率仍在托管的 Ubuntu 24.04 32 核池上运行，确保依赖方提供的代码永远不会进入持久化虚拟机。公网 IP 已禁用；工作流并发仍设有边界，因为自动扩缩容上限既不会分配闲置机器，也不意味着仓库工作可以无限扩展。
+企业保留仅限本仓库使用的 Ubuntu 和 Windows x64 大型运行器池。普通拉取请求直接指定 2 个 32 核托管运行器池：Ubuntu latest 用于其余主 Node 24 清单，Windows 2025 用于阻塞性 Windows 契约。完整覆盖率已从计费的 Ubuntu 24.04 32 核池迁移至公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位），仅面向可信的同仓库拉取请求；不可信的拉取请求——fork 与 Dependabot——的覆盖率仍在托管的 Ubuntu 24.04 32 核池上运行，确保依赖方提供的代码永远不会进入持久化虚拟机。**前置条件：仓库必须保持禁用 fork。**工作流中的运行器池选择表达式仅是纵深防御——`pull_request` 执行的是拉取请求自带的工作流定义，因此 YAML 无法对能修改它的 fork 实施运行器信任约束。真正的强制边界是组织侧（拉取请求无法修改）的 fork 禁用设置；规划中的开源发布采用独立组织下的只读镜像仓库，正是为了保持这一边界。将来若要启用 fork，必须先把运行器迁入组织级 runner group（限定本仓库并绑定基线分支工作流）——该迁移是启用 fork 的先决门槛，而非事后跟进项。公网 IP 已禁用；工作流并发仍设有边界，因为自动扩缩容上限既不会分配闲置机器，也不意味着仓库工作可以无限扩展。
 
 必需主路径依赖这些企业级运行器池。GitHub 标准托管作业保留 Node 22.19、Node 26 和 Python SDK 兼容性契约，而[可移植恢复边界](2026-07-23-portable-required-pull-request-ci.md)与[串行参考流程](2026-07-21-serial-cross-platform-ci-reference.md)则在 `master` 上持续提供完整的标准运行器证据。`suite=larger-runner-benchmark` 比较已预配规格上相互独立的关键通道，`suite=consolidated-runner-benchmark` 则比较完整聚合流程。每项基准测试都会先报告实测的处理器和内存容量，再运行仓库工作。
 

From 310a387b144526354bec79ab8f913cd419fcf570 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 00:08:26 +0800
Subject: [PATCH 08/27] =?UTF-8?q?ci:=20pivot=20=E2=80=94=20keep=20coverage?=
 =?UTF-8?q?=20hosted,=20add=20self-hosted=20serial=20standby=20lane?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Direction change after review discussion. Moving a REQUIRED check onto
a single in-house VM traded merge-path availability for modest savings
and accumulated trust/contention caveats (six review rounds' worth).
Revert every coverage-lane change: coverage stays on the enterprise
Ubuntu 24.04 32-core pool exactly as on master.

Instead, add serial-linux-selfhosted: on every master push the in-house
pool (vm-backup) runs the complete unsharded primary aggregate as a
hot-standby drill. It blocks nothing, yet continuously proves the
environment end to end, so any hosted-pool outage can be answered with
a one-line runs-on retarget onto continuously verified capacity.

Push-triggered lanes execute the base branch's own workflow definition,
so no PR-editable path selects these runners — the entire fork-trust
discussion is structurally moot for this lane. Topology notes (en/zh +
pairing records) describe the standby lane and the switch play.
---
 ...ence-based-larger-hosted-runners.i18n.yaml |  4 +-
 ...22-evidence-based-larger-hosted-runners.md |  6 +-
 ...evidence-based-larger-hosted-runners.zh.md |  6 +-
 ...ortable-required-pull-request-ci.i18n.yaml |  4 +-
 ...07-23-portable-required-pull-request-ci.md |  2 +-
 ...23-portable-required-pull-request-ci.zh.md |  2 +-
 .github/workflows/ci.yml                      | 78 ++++++++++---------
 7 files changed, 57 insertions(+), 45 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index ea3a57e072..1b14f5b689 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: 497c6f297d79245fb40cd30457e4b1d1e36db651
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: bcb0c6e9f11081b2cff696a9b6b425a40ee4aeb4
+2026-07-22-evidence-based-larger-hosted-runners.md: 6654f5eb3e21b48c6d33fd9d74ebd23cf3065d54
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: 3fe5715b20d3b881f8fb439b61900bdb84e5a588
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 497c6f297d..6654f5eb3e 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -12,13 +12,13 @@ Larger runners make it possible to pay setup once and parallelize inside the rep
 
 ## Decision
 
-The enterprise keeps repo-restricted x64 larger-runner pools for Ubuntu and Windows. Ordinary pull requests name two 32-core hosted pools directly: Ubuntu latest for the remaining primary Node 24 inventory and Windows 2025 for blocking Windows contracts. Exhaustive coverage moved off the metered Ubuntu 24.04 32-core pool onto the in-house self-hosted pool (`vm-backup` label: a 64-core VM running four always-on systemd-managed runner instances plus four registered spares) for trusted same-repo PRs; untrusted PRs — forks and Dependabot — keep coverage on the hosted Ubuntu 24.04 32-core pool so dependency-supplied code never reaches the persistent VM. **Precondition: repository forking stays disabled.** The workflow's pool selector is defense-in-depth only — `pull_request` executes the PR's own workflow definition, so YAML cannot enforce runner trust against a fork that edits it. Disabled forking (org-side, not PR-editable) is the enforcement boundary; the planned public release is an isolated read-only mirror under a separate org, preserving this. Before forking is ever enabled, the runners must first move into an org-level runner group restricted to this repository with base-branch workflow pinning — that migration is the gate, not a follow-up. Public IPs are disabled, and workflow concurrency remains bounded because an autoscaling ceiling neither allocates idle machines nor makes repository work scale without limit.
+The enterprise keeps repo-restricted x64 larger-runner pools for Ubuntu and Windows. Ordinary pull requests name three 32-core pools directly: Ubuntu 24.04 for exhaustive coverage, Ubuntu latest for the remaining primary Node 24 inventory, and Windows 2025 for blocking Windows contracts. Public IPs are disabled, and workflow concurrency remains bounded because an autoscaling ceiling neither allocates idle machines nor makes repository work scale without limit.
 
 The required primary path depends on those enterprise pools. Standard GitHub-hosted jobs retain the Node 22.19, Node 26, and Python SDK compatibility contracts, while the [portable recovery boundary](2026-07-23-portable-required-pull-request-ci.md) and [serial reference](2026-07-21-serial-cross-platform-ci-reference.md) keep complete standard-runner evidence available on `master`. `suite=larger-runner-benchmark` compares isolated critical lanes across provisioned sizes, and `suite=consolidated-runner-benchmark` compares whole aggregates. Each benchmark reports its observed processor and memory capacity before running repository work.
 
 The former gate-level and coarse primary shard jobs are absent from the workflow. Their static, lint, coverage, snapshot, and scenario shard selectors are also absent from the repository, so an unused diagnostic path cannot preserve a second CI architecture.
 
-Linux primary work uses three independent jobs: static gates and the consumer tail on hosted 32-core pools, and coverage on the in-house self-hosted 64-core pool for trusted PRs (hosted 32-core for untrusted ones). Coverage runs alone with its own worker bound, and the static scheduler runs alone so its result has no post-build consumer tail. After static gates finish, that job publishes its emitted `apps/*/lib`, `packages/*/*/lib`, and `vendor/*/lib` tree as a run-scoped artifact. The third job restores that exact tree, then starts lint, Node 24 runtime compatibility, build-backed snapshots, and all artifact consumers without repeating the build. Generated NodeNext consumer directories are excluded from ESLint discovery because the artifact check removes them while these processes overlap. The pnpm store and ESLint cache are restored without putting cache uploads on the pull-request critical path. Performance reports use each job's `startedAt` to `completedAt` interval; runner queue delay is capacity evidence, not repository execution time.
+Linux primary work uses three independent 32-core jobs. Coverage runs alone with its own worker bound, and the static scheduler runs alone so its result has no post-build consumer tail. After static gates finish, that job publishes its emitted `apps/*/lib`, `packages/*/*/lib`, and `vendor/*/lib` tree as a run-scoped artifact. The third job restores that exact tree, then starts lint, Node 24 runtime compatibility, build-backed snapshots, and all artifact consumers without repeating the build. Generated NodeNext consumer directories are excluded from ESLint discovery because the artifact check removes them while these processes overlap. The pnpm store and ESLint cache are restored without putting cache uploads on the pull-request critical path. Performance reports use each job's `startedAt` to `completedAt` interval; runner queue delay is capacity evidence, not repository execution time.
 
 Windows shares one 32-core setup across the blocking build and production site plus observational built-artifact contracts. Linux owns the duplicate lint, coverage, and snapshot inventories because running those observational copies on Windows extends the paid critical path without adding a blocking platform claim.
 
@@ -48,6 +48,8 @@ The process-bound coverage project contains exactly five suite files. Thirty-two
 
 Complete serial Linux, macOS, and Windows references run only when `master` moves. Pull requests use the enterprise required path plus standard-hosted compatibility jobs, while other larger-runner sizes run only by manual dispatch.
 
+An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with four always-on systemd-managed runner instances plus four registered spares) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate, so if the enterprise pools degrade, a required lane can be retargeted with a one-line `runs-on` change onto an environment with continuously verified evidence. Because the lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
+
 ## Alternatives considered
 
 **Keep the three coarse primary Linux lanes.** The core, CPU, and production-site jobs met the latency targets, but they paid three setup waves and left primary Node work sharded after larger runners were available. The all-size trace showed that one unnecessary dependency, not a lack of host capacity, kept the single-box aggregate above one minute.
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index bcb0c6e9f1..3fe5715b20 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -12,13 +12,13 @@ Status: implemented
 
 ## 决策
 
-企业保留仅限本仓库使用的 Ubuntu 和 Windows x64 大型运行器池。普通拉取请求直接指定 2 个 32 核托管运行器池：Ubuntu latest 用于其余主 Node 24 清单，Windows 2025 用于阻塞性 Windows 契约。完整覆盖率已从计费的 Ubuntu 24.04 32 核池迁移至公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位），仅面向可信的同仓库拉取请求；不可信的拉取请求——fork 与 Dependabot——的覆盖率仍在托管的 Ubuntu 24.04 32 核池上运行，确保依赖方提供的代码永远不会进入持久化虚拟机。**前置条件：仓库必须保持禁用 fork。**工作流中的运行器池选择表达式仅是纵深防御——`pull_request` 执行的是拉取请求自带的工作流定义，因此 YAML 无法对能修改它的 fork 实施运行器信任约束。真正的强制边界是组织侧（拉取请求无法修改）的 fork 禁用设置；规划中的开源发布采用独立组织下的只读镜像仓库，正是为了保持这一边界。将来若要启用 fork，必须先把运行器迁入组织级 runner group（限定本仓库并绑定基线分支工作流）——该迁移是启用 fork 的先决门槛，而非事后跟进项。公网 IP 已禁用；工作流并发仍设有边界，因为自动扩缩容上限既不会分配闲置机器，也不意味着仓库工作可以无限扩展。
+企业保留仅限本仓库使用的 Ubuntu 和 Windows x64 大型运行器池。普通拉取请求直接指定 3 个 32 核运行器池：Ubuntu 24.04 用于完整覆盖率，Ubuntu latest 用于其余主 Node 24 清单，Windows 2025 用于阻塞性 Windows 契约。公网 IP 已禁用；工作流并发仍设有边界，因为自动扩缩容上限既不会分配闲置机器，也不意味着仓库工作可以无限扩展。
 
 必需主路径依赖这些企业级运行器池。GitHub 标准托管作业保留 Node 22.19、Node 26 和 Python SDK 兼容性契约，而[可移植恢复边界](2026-07-23-portable-required-pull-request-ci.md)与[串行参考流程](2026-07-21-serial-cross-platform-ci-reference.md)则在 `master` 上持续提供完整的标准运行器证据。`suite=larger-runner-benchmark` 比较已预配规格上相互独立的关键通道，`suite=consolidated-runner-benchmark` 则比较完整聚合流程。每项基准测试都会先报告实测的处理器和内存容量，再运行仓库工作。
 
 原有的门禁级和粗粒度主流程分片作业已从工作流中移除。相应的静态、lint、覆盖率、快照和场景分片选择器也已从仓库中移除，因此未使用的诊断路径无法继续维系第二套 CI 架构。
 
-Linux 主流程使用 3 个相互独立的作业：静态门禁与消费方尾部作业运行在托管 32 核池上，覆盖率对可信拉取请求运行在公司自有的自托管 64 核池上（不可信请求仍用托管 32 核池）。覆盖率单独运行，并设有自己的工作线程上限；静态调度器也单独运行，因此构建后的消费方不会拖延其结果。静态门禁完成后，该作业将其生成的 `apps/*/lib`、`packages/*/*/lib` 和 `vendor/*/lib` 目录树作为仅供本次运行使用的产物发布。第三个作业恢复完全相同的目录树，再让 lint、Node 24 运行时兼容性、依赖构建产物的快照和所有产物消费方基于构建完成后的工作树启动，而不重复构建。生成的 NodeNext 消费方目录不会纳入 ESLint 的文件发现范围，因为这些进程重叠执行时，产物检查会删除这些目录。pnpm store 和 ESLint 缓存会得到恢复，但缓存上传不会进入拉取请求关键路径。性能报告采用每个作业从 `startedAt` 到 `completedAt` 的区间；运行器排队延迟是容量证据，而非仓库执行时间。
+Linux 主流程使用 3 个相互独立的 32 核作业。覆盖率单独运行，并设有自己的工作线程上限；静态调度器也单独运行，因此构建后的消费方不会拖延其结果。静态门禁完成后，该作业将其生成的 `apps/*/lib`、`packages/*/*/lib` 和 `vendor/*/lib` 目录树作为仅供本次运行使用的产物发布。第三个作业恢复完全相同的目录树，再让 lint、Node 24 运行时兼容性、依赖构建产物的快照和所有产物消费方基于构建完成后的工作树启动，而不重复构建。生成的 NodeNext 消费方目录不会纳入 ESLint 的文件发现范围，因为这些进程重叠执行时，产物检查会删除这些目录。pnpm store 和 ESLint 缓存会得到恢复，但缓存上传不会进入拉取请求关键路径。性能报告采用每个作业从 `startedAt` 到 `completedAt` 的区间；运行器排队延迟是容量证据，而非仓库执行时间。
 
 Windows 以一次 32 核环境设置同时承载阻塞性构建、生产网站和观测性的构建产物契约。重复的 lint、覆盖率和快照清单由 Linux 承担，因为在 Windows 上运行这些观测性副本会延长付费关键路径，却不会新增任何阻塞性平台契约。
 
@@ -48,6 +48,8 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程，因此当企业池发生故障时，只需一行 `runs-on` 修改即可把必需通道切换到一个具有持续验证证据的环境上。该通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+
 ## 曾考虑的替代方案
 
 **保留 3 个粗粒度 Linux 主流程通道。** 核心、CPU 和生产网站作业均达到延迟目标，但它们需要 3 轮设置，而且在大型运行器已经可用后仍对主 Node 工作进行分片。全规格运行轨迹表明，让单机聚合流程超过 1 分钟的是一项不必要的依赖，而非主机容量不足。
diff --git a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml
index ed97fe08a7..f8b54b0ec5 100644
--- a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-23-portable-required-pull-request-ci.md: 29b2cfa3f431a4a8be4aaa685b16cffdb4bf2593
-2026-07-23-portable-required-pull-request-ci.zh.md: 8b6d067637ce83c09529977f463f16dfa4af5a8b
+2026-07-23-portable-required-pull-request-ci.md: 9cf8d97016300c5258c075879176aa6abd64e59e
+2026-07-23-portable-required-pull-request-ci.zh.md: c6839a133d0c3fe7a699362f6168e17d827a5b61
diff --git a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md
index 29b2cfa3f4..9cf8d97016 100644
--- a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md
+++ b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.md
@@ -12,7 +12,7 @@ Billing health, a runner definition's `Ready` state, and a large autoscaling cei
 
 ## Decision
 
-[CI](../../../../.github/workflows/ci.yml) runs the required primary Node 24 and Windows jobs on repo-restricted enterprise 32-core pools, except exhaustive coverage, which runs on the in-house self-hosted 64-core pool for trusted same-repo pull requests (hosted 32-core for forks and Dependabot). Standard `ubuntu-latest` jobs retain Node 22.19, Node 26, and Python SDK compatibility, and `master` runs complete serial Linux, macOS, and Windows references. Those standard-hosted jobs keep the portable execution boundary observable without duplicating the primary inventory on every pull request.
+[CI](../../../../.github/workflows/ci.yml) runs the required primary Node 24 and Windows jobs on repo-restricted enterprise 32-core pools. Standard `ubuntu-latest` jobs retain Node 22.19, Node 26, and Python SDK compatibility, and `master` runs complete serial Linux, macOS, and Windows references. Those standard-hosted jobs keep the portable execution boundary observable without duplicating the primary inventory on every pull request.
 
 The two Linux primary jobs, Node compatibility, Python SDK, and `windows node 24 / complete` remain dependencies of `all checks passed`; branch protection continues to require `e2e` and `all checks passed`. There is no automatic fallback when an enterprise label cannot allocate: the standard jobs continue to report their own contracts, but they cannot manufacture the missing required result.
 
diff --git a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md
index 8b6d067637..c6839a133d 100644
--- a/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md
+++ b/.agents/notes/implemented/process/2026-07-23-portable-required-pull-request-ci.zh.md
@@ -12,7 +12,7 @@ Status: implemented
 
 ## 决策
 
-[CI](../../../../.github/workflows/ci.yml) 在仅限本仓库使用的企业级 32 核运行器池上运行必需的主 Node 24 作业和 Windows 作业；唯一例外是完整覆盖率——可信的同仓库拉取请求在公司自有的自托管 64 核池上运行（fork 与 Dependabot 仍用托管 32 核池）。标准 `ubuntu-latest` 作业保留 Node 22.19、Node 26 和 Python SDK 兼容性，`master` 则运行完整的 Linux、macOS 和 Windows 串行参考流程。这些标准托管作业让可移植执行边界保持可观测，而不必在每个拉取请求中重复主清单。
+[CI](../../../../.github/workflows/ci.yml) 在仅限本仓库使用的企业级 32 核运行器池上运行必需的主 Node 24 作业和 Windows 作业。标准 `ubuntu-latest` 作业保留 Node 22.19、Node 26 和 Python SDK 兼容性，`master` 则运行完整的 Linux、macOS 和 Windows 串行参考流程。这些标准托管作业让可移植执行边界保持可观测，而不必在每个拉取请求中重复主清单。
 
 两项 Linux 主作业、Node 兼容性、Python SDK 和 `windows node 24 / complete` 继续作为 `all checks passed` 的依赖项；分支保护继续要求 `e2e` 和 `all checks passed`。企业级运行器标签无法分配运行器时没有自动后备机制：标准作业会继续报告各自的契约，但无法产出缺失的必需结果。
 
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index dc5ad98ec4..2666c93b8c 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -27,10 +27,8 @@ env:
 
 jobs:
 
-  # Three independent Linux jobs isolate coverage, static analysis, and the
-  # build-backed consumer tail: static and consumers on hosted enterprise
-  # 32-core pools; coverage on the in-house self-hosted pool for trusted PRs
-  # (hosted for forks/Dependabot). The static job publishes its exact build so
+  # Three enterprise jobs isolate coverage, static analysis, and the
+  # build-backed consumer tail. The static job publishes its exact build so
   # consumers do not repeat the longest part of their critical path.
   node-24:
     if: github.event_name == 'pull_request'
@@ -79,46 +77,17 @@ jobs:
 
   node-24-coverage:
     if: github.event_name == 'pull_request'
-    # Trusted same-repo PRs run on the in-house pool (self-hosted, 64-core;
-    # 4 always-on systemd-managed instances plus 4 registered spares) instead
-    # of the metered enterprise pool. Untrusted PRs — forks and Dependabot
-    # (same-repo but dependency-supplied code; same author test as e2e.yml) —
-    # stay on the hosted enterprise pool so no untrusted code reaches the
-    # persistent self-hosted VM. Selecting the pool via runs-on keeps this a
-    # single job, so the all-checks-passed aggregate never sees a skip.
-    runs-on: >-
-      ${{ (github.event.pull_request.head.repo.full_name != github.repository
-           || github.event.pull_request.user.login == 'dependabot[bot]')
-          && 'dsh-enterprise-ubuntu-24-04-32core-test'
-          || fromJSON('["self-hosted", "linux", "x64", "vm-backup"]') }}
+    runs-on: dsh-enterprise-ubuntu-24-04-32core-test
     name: node 24 / coverage
     env:
-      # Worker bound is per-leg: the hosted 32-core runner is exclusive to
-      # one job, but the self-hosted pool shares one 64-core VM across four
-      # runner instances, so concurrent PRs would otherwise stack up to
-      # 4×24 = 96 workers and re-trigger the aggregate-contention failures
-      # documented for the timing-sensitive process suites. 12 per job caps
-      # the shared host at 48 workers even fully loaded.
-      DSH_COVERAGE_MAX_WORKERS: >-
-        ${{ (github.event.pull_request.head.repo.full_name != github.repository
-             || github.event.pull_request.user.login == 'dependabot[bot]')
-            && '24' || '12' }}
+      DSH_COVERAGE_MAX_WORKERS: '24'
       DSH_GATE_CONCURRENCY: '8'
     steps:
       - uses: actions/checkout@v6
         with:
           persist-credentials: false
 
-      # Restore the pnpm-store cache only on the hosted (untrusted-PR) leg,
-      # where the VM is ephemeral and the same-region download is fast. On
-      # the self-hosted leg pnpm's persistent store lives outside
-      # /home/runner, so this restore would spend ~52 s pulling ~180 MB into
-      # a path pnpm never reads (measured; install then took 2.8 s straight
-      # from the persistent store). Condition mirrors the runs-on selector.
       - uses: actions/cache/restore@v4
-        if: >-
-          github.event.pull_request.head.repo.full_name != github.repository
-          || github.event.pull_request.user.login == 'dependabot[bot]'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
@@ -396,6 +365,45 @@ jobs:
           DSH_SNAPSHOT_MAX_CONCURRENCY: '1'
         run: pnpm run check:ci
 
+  # Hot-standby drill for the in-house self-hosted pool: every master move
+  # re-runs the complete unsharded aggregate on the persistent 64-core VM,
+  # continuously proving that environment can take over a required lane if
+  # the hosted pools degrade (the switch is then a one-line runs-on change).
+  # Push-triggered, so it always executes the base branch's own workflow
+  # definition — no PR-editable path selects these runners. Non-blocking for
+  # pull requests; no cache steps because the VM's persistent pnpm store and
+  # tool caches make them redundant (and saving here would poison the hosted
+  # cache namespace with self-hosted paths).
+  serial-linux-selfhosted:
+    if: github.event_name == 'push' && github.ref == 'refs/heads/master'
+    name: serial / linux (self-hosted standby)
+    runs-on: [self-hosted, linux, x64, vm-backup]
+    steps:
+      - uses: actions/checkout@v6
+
+      - uses: actions/setup-node@v6
+        with:
+          node-version: ${{ env.PRIMARY_NODE_VERSION }}
+
+      - name: Enable corepack (pnpm)
+        run: corepack enable
+
+      - name: Install (immutable)
+        run: pnpm install --frozen-lockfile
+
+      - name: Prepare bubblewrap (unrestrict userns)
+        run: bash scripts/prepare-ci-bubblewrap.sh
+
+      - name: Run complete unsharded primary Node CI serially
+        env:
+          DSH_COVERAGE_MAX_WORKERS: '1'
+          DSH_E2E_MAX_WORKERS: '1'
+          DSH_ESLINT_CACHE: '1'
+          DSH_GATE_CONCURRENCY: '1'
+          DSH_PUBLINT_CONCURRENCY: '1'
+          DSH_SNAPSHOT_MAX_CONCURRENCY: '1'
+        run: pnpm run check:ci
+
   serial-macos:
     if: github.event_name == 'push' && github.ref == 'refs/heads/master'
     name: serial / macos

From 0fd6dc8924a087db5c3a8190a2f1783766660e8b Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 00:34:53 +0800
Subject: [PATCH 09/27] ci: pre-wire admin-only failover from hosted pools to
 the in-house pool
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The three required Linux jobs now resolve their pool through the
DSH_CI_FAILOVER repository variable. Unset, everything runs exactly as
today on the hosted enterprise pools. Setting it to 'selfhosted'
(repo-admin-only, not PR-editable, no merge required — a merge would be
deadlocked behind the failing checks themselves) retargets all three
onto the vm-backup pool, halves the coverage worker bound and snapshot
concurrency for the shared VM, and skips the hosted-path cache restores.

Adds a bilingual failover runbook (switch, capacity via the four
registered spare instances, switch-back, trust boundary) and links it
from the topology note. The push-triggered standby lane remains the
continuous proof that the failover target works.
---
 ...ence-based-larger-hosted-runners.i18n.yaml |  4 +-
 ...22-evidence-based-larger-hosted-runners.md |  2 +-
 ...evidence-based-larger-hosted-runners.zh.md |  2 +-
 .../process/ci-failover-runbook.i18n.yaml     |  6 +++
 .../process/ci-failover-runbook.md            | 33 +++++++++++++++
 .../process/ci-failover-runbook.zh.md         | 33 +++++++++++++++
 .github/workflows/ci.yml                      | 40 ++++++++++++++++---
 7 files changed, 111 insertions(+), 9 deletions(-)
 create mode 100644 .agents/notes/implemented/process/ci-failover-runbook.i18n.yaml
 create mode 100644 .agents/notes/implemented/process/ci-failover-runbook.md
 create mode 100644 .agents/notes/implemented/process/ci-failover-runbook.zh.md

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 1b14f5b689..cd3ae7a181 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: 6654f5eb3e21b48c6d33fd9d74ebd23cf3065d54
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: 3fe5715b20d3b881f8fb439b61900bdb84e5a588
+2026-07-22-evidence-based-larger-hosted-runners.md: dd07280092565257f4b5324f997d5efd4c9c51cc
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: a9c034b643da0cb9148d08c0300f3e142eec31e6
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 6654f5eb3e..dd07280092 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -48,7 +48,7 @@ The process-bound coverage project contains exactly five suite files. Thirty-two
 
 Complete serial Linux, macOS, and Windows references run only when `master` moves. Pull requests use the enterprise required path plus standard-hosted compatibility jobs, while other larger-runner sizes run only by manual dispatch.
 
-An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with four always-on systemd-managed runner instances plus four registered spares) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate, so if the enterprise pools degrade, a required lane can be retargeted with a one-line `runs-on` change onto an environment with continuously verified evidence. Because the lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
+An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with four always-on systemd-managed runner instances plus four registered spares) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the admin-only `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index 3fe5715b20..a9c034b643 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -48,7 +48,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
-另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程，因此当企业池发生故障时，只需一行 `runs-on` 修改即可把必需通道切换到一个具有持续验证证据的环境上。该通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](ci-failover-runbook.zh.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
 
 ## 曾考虑的替代方案
 
diff --git a/.agents/notes/implemented/process/ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/ci-failover-runbook.i18n.yaml
new file mode 100644
index 0000000000..294ed38ddf
--- /dev/null
+++ b/.agents/notes/implemented/process/ci-failover-runbook.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write
+ci-failover-runbook.md: d22c93fbedd3216e71bc24101dfa06dc606521c2
+ci-failover-runbook.zh.md: d7d26287191165cd3cb2666de4b1c7d6217ba71d
diff --git a/.agents/notes/implemented/process/ci-failover-runbook.md b/.agents/notes/implemented/process/ci-failover-runbook.md
new file mode 100644
index 0000000000..d22c93fbed
--- /dev/null
+++ b/.agents/notes/implemented/process/ci-failover-runbook.md
@@ -0,0 +1,33 @@
+# Agent Note: CI failover runbook — hosted pools → in-house pool
+
+Status: implemented
+
+English | [中文](ci-failover-runbook.zh.md)
+
+## What this is
+
+The three required Linux jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) resolve their runner pool through the `DSH_CI_FAILOVER` repository variable. Normally the variable is unset and they run on the hosted enterprise 32-core pools. When the hosted pools are degraded (jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails), a repository admin can retarget all three onto the in-house self-hosted pool without merging anything — merging would itself be blocked by the very checks that are failing.
+
+The in-house pool (`vm-backup`: one 64-core VM, four always-on systemd-managed runner instances, four registered spares) is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push. Check its latest run before switching: green standby = verified-yesterday capacity.
+
+## Switch (repo admin, ~1 minute, no merge)
+
+1. Repository **Settings → Secrets and variables → Actions → Variables → New repository variable**: name `DSH_CI_FAILOVER`, value `selfhosted`.
+2. Re-run the failed/queued required jobs (Re-run failed jobs on affected PRs, or let new pushes pick it up).
+3. That is the entire switch. Under failover the workflow also, automatically: halves `DSH_COVERAGE_MAX_WORKERS` to 12 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 16 (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
+
+## Capacity during failover
+
+Four always-on instances absorb normal PR traffic. If queues build, bring the four registered spares online on the VM (no token needed — they are already registered):
+
+```bash
+for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
+```
+
+## Switch back
+
+Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhosted`). New runs resolve back to the hosted enterprise pools. Stop the spare instances if they were started.
+
+## Trust boundary
+
+The variable is repository-admin-only state: a pull request can neither set it nor read a different value into effect, and the expressions live in the base branch's workflow definition. This failover path therefore adds no PR-editable route to the self-hosted pool. (Runner-side enforcement — an org-level runner group restricting these runners to the master-ref workflow — is tracked separately and composes with this mechanism.)
diff --git a/.agents/notes/implemented/process/ci-failover-runbook.zh.md b/.agents/notes/implemented/process/ci-failover-runbook.zh.md
new file mode 100644
index 0000000000..d7d2628719
--- /dev/null
+++ b/.agents/notes/implemented/process/ci-failover-runbook.zh.md
@@ -0,0 +1,33 @@
+# Agent Note: CI 故障切换手册 — 托管池 → 自有池
+
+Status: implemented
+
+[English](ci-failover-runbook.md) | 中文
+
+## 这是什么
+
+[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。正常情况下该变量不存在，作业运行在托管的企业级 32 核池上。当托管池发生故障（作业无限排队、企业标签消失或 GitHub 侧容量故障）时，仓库管理员无需合并任何代码即可把三个作业整体切换到公司自有的自托管池——此时合并本身正被这些失败的检查阻塞，任何"先合 PR 再切换"的方案都是死锁。
+
+自有池（`vm-backup`：一台 64 核虚拟机，4 个常驻 systemd 管理的运行器实例，另有 4 个已注册备用位）由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。切换前先看该通道最近一次运行：绿色 = 这套环境昨天刚被全量验证过。
+
+## 切换步骤（仓库管理员，约 1 分钟，无需合并）
+
+1. 仓库 **Settings → Secrets and variables → Actions → Variables → New repository variable**：名称 `DSH_CI_FAILOVER`，值 `selfhosted`。
+2. 对受影响 PR 的失败/排队作业点 Re-run failed jobs（或等新推送自然触发）。
+3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 12、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 16（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
+
+## 切换期间的容量
+
+4 个常驻实例可承接正常 PR 流量。若出现排队，在虚拟机上把 4 个已注册的备用位拉起（无需 token——它们已注册）：
+
+```bash
+for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
+```
+
+## 切回
+
+删除 `DSH_CI_FAILOVER` 变量（或改为 `selfhosted` 以外的任何值），新的运行即解析回托管企业池。若启动过备用实例，将其停止。
+
+## 信任边界
+
+该变量是仅限仓库管理员的状态：拉取请求既不能设置它，也不能让不同的值生效，且表达式存在于基线分支的工作流定义中。因此这条故障切换路径没有增加任何可由 PR 编辑的自托管池访问途径。（运行器侧的强制约束——通过组织级 runner group 把这批运行器限定到 master 引用的工作流——另行跟踪，与本机制互补。）
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 2666c93b8c..88722804f9 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -30,9 +30,22 @@ jobs:
   # Three enterprise jobs isolate coverage, static analysis, and the
   # build-backed consumer tail. The static job publishes its exact build so
   # consumers do not repeat the longest part of their critical path.
+  #
+  # FAILOVER: each Linux enterprise job resolves its pool through the
+  # DSH_CI_FAILOVER repository variable. Unset (normal), the expressions
+  # pick the hosted enterprise pools below. Setting the variable to
+  # 'selfhosted' (repo Settings → Actions → Variables; admin-only, not
+  # PR-editable, no merge required) retargets all three onto the in-house
+  # vm-backup pool and re-running the failed jobs is the entire switch —
+  # see .agents/notes/implemented/process/ci-failover-runbook.md. The
+  # in-house pool's readiness is re-proven on every master push by the
+  # serial-linux-selfhosted standby lane below.
   node-24:
     if: github.event_name == 'pull_request'
-    runs-on: dsh-enterprise-ubuntu-latest-32core-test
+    runs-on: >-
+      ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
+          || 'dsh-enterprise-ubuntu-latest-32core-test' }}
     name: node 24 / static
     env:
       DSH_GATE_CONCURRENCY: '8'
@@ -77,17 +90,28 @@ jobs:
 
   node-24-coverage:
     if: github.event_name == 'pull_request'
-    runs-on: dsh-enterprise-ubuntu-24-04-32core-test
+    runs-on: >-
+      ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
+          || 'dsh-enterprise-ubuntu-24-04-32core-test' }}
     name: node 24 / coverage
     env:
-      DSH_COVERAGE_MAX_WORKERS: '24'
+      # Failover halves the worker bound: the hosted 32-core runner is
+      # exclusive to one job, but the failover pool shares one 64-core VM
+      # across four runner instances, and the timing-sensitive process
+      # suites have documented aggregate-contention failures.
+      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '12' || '24' }}
       DSH_GATE_CONCURRENCY: '8'
     steps:
       - uses: actions/checkout@v6
         with:
           persist-credentials: false
 
+      # Skipped under failover: the self-hosted VM's persistent pnpm store
+      # serves warm installs directly, and this hosted-path restore would
+      # spend ~52 s pulling ~180 MB into a path pnpm never reads there.
       - uses: actions/cache/restore@v4
+        if: vars.DSH_CI_FAILOVER != 'selfhosted'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
@@ -118,7 +142,10 @@ jobs:
   node-24-consumers:
     needs: node-24
     if: github.event_name == 'pull_request'
-    runs-on: dsh-enterprise-ubuntu-latest-32core-test
+    runs-on: >-
+      ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
+          || 'dsh-enterprise-ubuntu-latest-32core-test' }}
     name: node 24 / snapshots and artifacts
     env:
       DSH_ESLINT_CACHE: '1'
@@ -126,7 +153,8 @@ jobs:
       DSH_GATE_CONCURRENCY: '8'
       DSH_NODE_COMPAT_SKIP_TYPECHECK: '1'
       DSH_PUBLINT_CONCURRENCY: '8'
-      DSH_SNAPSHOT_MAX_CONCURRENCY: '32'
+      # Failover halves snapshot concurrency for the shared 64-core VM.
+      DSH_SNAPSHOT_MAX_CONCURRENCY: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '16' || '32' }}
     steps:
       - uses: actions/checkout@v6
         with:
@@ -140,7 +168,9 @@ jobs:
       - name: Restore built tree
         run: tar -xzf "$RUNNER_TEMP/node-24-built-tree.tar.gz"
 
+      # Skipped under failover — see the coverage lane's identical rationale.
       - uses: actions/cache/restore@v4
+        if: vars.DSH_CI_FAILOVER != 'selfhosted'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}

From 68e280ce4ff86629ea0443a012d7c7080289ce4d Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 05:22:28 +0800
Subject: [PATCH 10/27] docs(ci): make the failover runbook a conforming dated
 Agent Note

The failover runbook landed as .agents/notes/implemented/process/ci-failover-runbook.md,
which fails three doc-sync gates: the classification/format gates require a
yyyy-mm-dd-topic.md filename and the implemented Agent Note skeleton
(Problem/Decision/Alternatives/Consequences), and the bilingual pairing gate
requires cross-note link targets to match between the two language sides.

Rename to 2026-07-26-ci-failover-runbook.md/.zh.md, reshape both sides into the
implemented skeleton (the runbook steps live in bespoke sections under Decision),
point the sibling topology note and the ci.yml comment at the dated filename, and
make both sides link the canonical .md per the bilingual convention. Re-recorded
the i18n pairing records.
---
 ...ence-based-larger-hosted-runners.i18n.yaml |  4 +-
 ...22-evidence-based-larger-hosted-runners.md |  2 +-
 ...evidence-based-larger-hosted-runners.zh.md |  2 +-
 ... 2026-07-26-ci-failover-runbook.i18n.yaml} |  4 +-
 .../process/2026-07-26-ci-failover-runbook.md | 49 +++++++++++++++++++
 .../2026-07-26-ci-failover-runbook.zh.md      | 49 +++++++++++++++++++
 .../process/ci-failover-runbook.md            | 33 -------------
 .../process/ci-failover-runbook.zh.md         | 33 -------------
 .github/workflows/ci.yml                      |  2 +-
 9 files changed, 105 insertions(+), 73 deletions(-)
 rename .agents/notes/implemented/process/{ci-failover-runbook.i18n.yaml => 2026-07-26-ci-failover-runbook.i18n.yaml} (65%)
 create mode 100644 .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
 create mode 100644 .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
 delete mode 100644 .agents/notes/implemented/process/ci-failover-runbook.md
 delete mode 100644 .agents/notes/implemented/process/ci-failover-runbook.zh.md

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index b2d20fb999..84a10e5ab9 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: 6e989a908b1faa363d04746e4efaa1a77358be9d
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: 02c2ab405ec10dd381b051581d72662ec342e21e
+2026-07-22-evidence-based-larger-hosted-runners.md: 21e602b2b5850176df981dcf448f4f827b756719
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: ba49ff18ac304f4078d4c8ebfd00bb1a85ada0b3
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 6e989a908b..21e602b2b5 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -52,7 +52,7 @@ The process-bound coverage project contains exactly five suite files. Thirty-two
 
 Complete serial Linux, macOS, and Windows references run only when `master` moves. Pull requests use the enterprise required path plus standard-hosted compatibility jobs, while other larger-runner sizes run only by manual dispatch.
 
-An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with four always-on systemd-managed runner instances plus four registered spares) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the admin-only `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
+An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with four always-on systemd-managed runner instances plus four registered spares) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the admin-only `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index 02c2ab405e..ba49ff18ac 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -52,7 +52,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
-另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](ci-failover-runbook.zh.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
 
 ## 曾考虑的替代方案
 
diff --git a/.agents/notes/implemented/process/ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
similarity index 65%
rename from .agents/notes/implemented/process/ci-failover-runbook.i18n.yaml
rename to .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index 294ed38ddf..7a65ff5479 100644
--- a/.agents/notes/implemented/process/ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-ci-failover-runbook.md: d22c93fbedd3216e71bc24101dfa06dc606521c2
-ci-failover-runbook.zh.md: d7d26287191165cd3cb2666de4b1c7d6217ba71d
+2026-07-26-ci-failover-runbook.md: 9100cf226467d06835478b13c41904bc50270b78
+2026-07-26-ci-failover-runbook.zh.md: 4ec80ae36411335a378f7979b9bca704c17732d0
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
new file mode 100644
index 0000000000..9100cf2264
--- /dev/null
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -0,0 +1,49 @@
+# Agent Note: CI failover runbook — hosted pools → in-house pool
+
+Status: implemented
+
+English | [中文](2026-07-26-ci-failover-runbook.zh.md)
+
+## Problem
+
+The three required Linux jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) run on the hosted enterprise 32-core pools. When those pools degrade — jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. An outage therefore needs a switch a repository admin can throw without merging anything.
+
+## Decision
+
+Each of the three required Linux jobs resolves its runner pool through the `DSH_CI_FAILOVER` repository variable. Unset (normal), they run on the hosted enterprise pools. Set to `selfhosted` by a repository admin, all three retarget onto the in-house self-hosted `vm-backup` pool, coverage and snapshot concurrency drop to shared-VM bounds, and the hosted-path pnpm cache restores are skipped. The switch is admin-only repository state, not a merge, so it works while every check is red. The in-house pool's readiness is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push.
+
+### What the in-house pool is
+
+`vm-backup`: one 64-core VM, four always-on systemd-managed runner instances, four registered spares. Check the latest `serial / linux (self-hosted standby)` run before switching: a green standby is verified-yesterday capacity.
+
+### Switch (repo admin, ~1 minute, no merge)
+
+1. Repository **Settings → Secrets and variables → Actions → Variables → New repository variable**: name `DSH_CI_FAILOVER`, value `selfhosted`.
+2. Re-run the failed/queued required jobs (Re-run failed jobs on affected PRs, or let new pushes pick it up).
+3. That is the entire switch. Under failover the workflow also, automatically: halves `DSH_COVERAGE_MAX_WORKERS` to 12 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 16 (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
+
+### Capacity during failover
+
+Four always-on instances absorb normal PR traffic. If queues build, bring the four registered spares online on the VM (no token needed — they are already registered):
+
+```bash
+for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
+```
+
+### Switch back
+
+Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhosted`). New runs resolve back to the hosted enterprise pools. Stop the spare instances if they were started.
+
+### Trust boundary
+
+The variable is repository-admin-only state: a pull request can neither set it nor read a different value into effect, and the expressions live in the base branch's workflow definition. This failover path therefore adds no PR-editable route to the self-hosted pool. Runner-side enforcement — an org-level runner group restricting these runners to the master-ref workflow — is tracked separately and composes with this mechanism.
+
+## Alternatives considered
+
+**Merge a workflow change to switch pools.** Rejected because the outage that motivates the switch is exactly the state in which no PR can merge: the required checks are the ones failing. A repository variable is admin-controlled state that takes effect on re-run without a merge.
+
+**Keep the self-hosted pool always in the required path.** Rejected because it trades hosted-pool availability for the in-house VM's, moving a single point of failure rather than adding a fallback. The variable keeps the hosted pools primary and the self-hosted pool a proven, one-action standby.
+
+## Consequences
+
+Recovering from a hosted-pool outage is a single admin variable plus a re-run, with no merge on the critical path. The cost is a second runner topology to keep working: the standby lane exercises it on every master push so the failover target never goes stale, and the concurrency and cache-restore branches in `ci.yml` carry a `selfhosted` leg that must stay in step with the hosted leg.
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
new file mode 100644
index 0000000000..4ec80ae364
--- /dev/null
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -0,0 +1,49 @@
+# Agent Note: CI 故障切换手册 — 托管池 → 自有池
+
+Status: implemented
+
+[English](2026-07-26-ci-failover-runbook.md) | 中文
+
+## 问题
+
+[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）运行在托管的企业级 32 核池上。当这些托管池发生故障——作业无限排队、企业标签消失或 GitHub 侧容量故障——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。因此故障需要一个仓库管理员无需合并任何代码即可触发的开关。
+
+## 决策
+
+三个必需的 Linux 作业各自通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。变量不存在（正常）时它们运行在托管企业池上；由仓库管理员设为 `selfhosted` 时，三者全部切换到公司自有的自托管 `vm-backup` 池，coverage 与 snapshot 的并发降到共享虚拟机上限，并跳过托管路径的 pnpm 缓存恢复。这个开关是仅限管理员的仓库状态而非一次合并，因此在所有检查都是红色时仍然有效。自有池的就绪状态由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。
+
+### 自有池是什么
+
+`vm-backup`：一台 64 核虚拟机，4 个常驻 systemd 管理的运行器实例，另有 4 个已注册备用位。切换前先看 `serial / linux (self-hosted standby)` 最近一次运行：绿色 = 这套环境昨天刚被全量验证过。
+
+### 切换步骤（仓库管理员，约 1 分钟，无需合并）
+
+1. 仓库 **Settings → Secrets and variables → Actions → Variables → New repository variable**：名称 `DSH_CI_FAILOVER`，值 `selfhosted`。
+2. 对受影响 PR 的失败/排队作业点 Re-run failed jobs（或等新推送自然触发）。
+3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 12、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 16（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
+
+### 切换期间的容量
+
+4 个常驻实例可承接正常 PR 流量。若出现排队，在虚拟机上把 4 个已注册的备用位拉起（无需 token——它们已注册）：
+
+```bash
+for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
+```
+
+### 切回
+
+删除 `DSH_CI_FAILOVER` 变量（或改为 `selfhosted` 以外的任何值），新的运行即解析回托管企业池。若启动过备用实例，将其停止。
+
+### 信任边界
+
+该变量是仅限仓库管理员的状态：拉取请求既不能设置它，也不能让不同的值生效，且表达式存在于基线分支的工作流定义中。因此这条故障切换路径没有增加任何可由 PR 编辑的自托管池访问途径。运行器侧的强制约束——通过组织级 runner group 把这批运行器限定到 master 引用的工作流——另行跟踪，与本机制互补。
+
+## 曾考虑的替代方案
+
+**通过合并一次工作流改动来切换池。** 否决，因为触发切换的故障状态恰恰是任何 PR 都无法合并的状态：必需检查正是失败的那些。仓库变量是管理员控制的状态，重跑即生效，无需合并。
+
+**让自托管池长期处于必需路径中。** 否决，因为这是拿托管池的可用性去换自有虚拟机的可用性，只是搬移了单点故障而非增加回退。该变量让托管池保持主路径，自托管池作为一个经过验证、一步即可启用的热备。
+
+## 后果
+
+从托管池故障中恢复只需一个管理员变量加一次重跑，关键路径上没有合并。代价是要维护第二套运行器拓扑：热备通道在每次 master 推送时都运行它，使故障切换目标永不失效；而 `ci.yml` 中的并发与缓存恢复分支带有一条 `selfhosted` 支路，必须与托管支路保持同步。
diff --git a/.agents/notes/implemented/process/ci-failover-runbook.md b/.agents/notes/implemented/process/ci-failover-runbook.md
deleted file mode 100644
index d22c93fbed..0000000000
--- a/.agents/notes/implemented/process/ci-failover-runbook.md
+++ /dev/null
@@ -1,33 +0,0 @@
-# Agent Note: CI failover runbook — hosted pools → in-house pool
-
-Status: implemented
-
-English | [中文](ci-failover-runbook.zh.md)
-
-## What this is
-
-The three required Linux jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) resolve their runner pool through the `DSH_CI_FAILOVER` repository variable. Normally the variable is unset and they run on the hosted enterprise 32-core pools. When the hosted pools are degraded (jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails), a repository admin can retarget all three onto the in-house self-hosted pool without merging anything — merging would itself be blocked by the very checks that are failing.
-
-The in-house pool (`vm-backup`: one 64-core VM, four always-on systemd-managed runner instances, four registered spares) is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push. Check its latest run before switching: green standby = verified-yesterday capacity.
-
-## Switch (repo admin, ~1 minute, no merge)
-
-1. Repository **Settings → Secrets and variables → Actions → Variables → New repository variable**: name `DSH_CI_FAILOVER`, value `selfhosted`.
-2. Re-run the failed/queued required jobs (Re-run failed jobs on affected PRs, or let new pushes pick it up).
-3. That is the entire switch. Under failover the workflow also, automatically: halves `DSH_COVERAGE_MAX_WORKERS` to 12 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 16 (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
-
-## Capacity during failover
-
-Four always-on instances absorb normal PR traffic. If queues build, bring the four registered spares online on the VM (no token needed — they are already registered):
-
-```bash
-for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
-```
-
-## Switch back
-
-Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhosted`). New runs resolve back to the hosted enterprise pools. Stop the spare instances if they were started.
-
-## Trust boundary
-
-The variable is repository-admin-only state: a pull request can neither set it nor read a different value into effect, and the expressions live in the base branch's workflow definition. This failover path therefore adds no PR-editable route to the self-hosted pool. (Runner-side enforcement — an org-level runner group restricting these runners to the master-ref workflow — is tracked separately and composes with this mechanism.)
diff --git a/.agents/notes/implemented/process/ci-failover-runbook.zh.md b/.agents/notes/implemented/process/ci-failover-runbook.zh.md
deleted file mode 100644
index d7d2628719..0000000000
--- a/.agents/notes/implemented/process/ci-failover-runbook.zh.md
+++ /dev/null
@@ -1,33 +0,0 @@
-# Agent Note: CI 故障切换手册 — 托管池 → 自有池
-
-Status: implemented
-
-[English](ci-failover-runbook.md) | 中文
-
-## 这是什么
-
-[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。正常情况下该变量不存在，作业运行在托管的企业级 32 核池上。当托管池发生故障（作业无限排队、企业标签消失或 GitHub 侧容量故障）时，仓库管理员无需合并任何代码即可把三个作业整体切换到公司自有的自托管池——此时合并本身正被这些失败的检查阻塞，任何"先合 PR 再切换"的方案都是死锁。
-
-自有池（`vm-backup`：一台 64 核虚拟机，4 个常驻 systemd 管理的运行器实例，另有 4 个已注册备用位）由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。切换前先看该通道最近一次运行：绿色 = 这套环境昨天刚被全量验证过。
-
-## 切换步骤（仓库管理员，约 1 分钟，无需合并）
-
-1. 仓库 **Settings → Secrets and variables → Actions → Variables → New repository variable**：名称 `DSH_CI_FAILOVER`，值 `selfhosted`。
-2. 对受影响 PR 的失败/排队作业点 Re-run failed jobs（或等新推送自然触发）。
-3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 12、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 16（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
-
-## 切换期间的容量
-
-4 个常驻实例可承接正常 PR 流量。若出现排队，在虚拟机上把 4 个已注册的备用位拉起（无需 token——它们已注册）：
-
-```bash
-for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
-```
-
-## 切回
-
-删除 `DSH_CI_FAILOVER` 变量（或改为 `selfhosted` 以外的任何值），新的运行即解析回托管企业池。若启动过备用实例，将其停止。
-
-## 信任边界
-
-该变量是仅限仓库管理员的状态：拉取请求既不能设置它，也不能让不同的值生效，且表达式存在于基线分支的工作流定义中。因此这条故障切换路径没有增加任何可由 PR 编辑的自托管池访问途径。（运行器侧的强制约束——通过组织级 runner group 把这批运行器限定到 master 引用的工作流——另行跟踪，与本机制互补。）
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 88722804f9..0be7655193 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -37,7 +37,7 @@ jobs:
   # 'selfhosted' (repo Settings → Actions → Variables; admin-only, not
   # PR-editable, no merge required) retargets all three onto the in-house
   # vm-backup pool and re-running the failed jobs is the entire switch —
-  # see .agents/notes/implemented/process/ci-failover-runbook.md. The
+  # see .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md. The
   # in-house pool's readiness is re-proven on every master push by the
   # serial-linux-selfhosted standby lane below.
   node-24:

From 498df1d8de66d3f17ed52ec93d7ffa863604cde8 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 05:44:35 +0800
Subject: [PATCH 11/27] ci: gate static lane's cache restore under failover;
 fix runbook recovery steps

Review round on the pivoted design:

- node-24 (static) kept an unconditional hosted pnpm cache restore while the
  coverage and consumers lanes skip it under failover. On the self-hosted VM
  that restore downloads ~180 MB into /home/runner, a path pnpm never reads
  there, adding latency and contention during an outage. Gate it with the same
  `vars.DSH_CI_FAILOVER != 'selfhosted'` condition so all three lanes match.

- Runbook switch step 2 said "Re-run failed jobs", but the documented
  indefinite-queue outage leaves jobs queued (not failed), which cannot be
  re-run in place and do not retarget on variable change. Correct both language
  sides to cancel the run and re-run all jobs, or push a new commit.

- The standby-lane comment still described the switch as a one-line runs-on
  change; it is now setting the admin-only DSH_CI_FAILOVER variable.
---
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml       | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.md  | 2 +-
 .../process/2026-07-26-ci-failover-runbook.zh.md           | 2 +-
 .github/workflows/ci.yml                                   | 7 +++++--
 4 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index 7a65ff5479..a2725da1b2 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-26-ci-failover-runbook.md: 9100cf226467d06835478b13c41904bc50270b78
-2026-07-26-ci-failover-runbook.zh.md: 4ec80ae36411335a378f7979b9bca704c17732d0
+2026-07-26-ci-failover-runbook.md: db8e0676ecc6eeaea16438e7868ccf9ac43887cc
+2026-07-26-ci-failover-runbook.zh.md: b3b4149f460784e88ce03458fc556f402c38fa2f
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index 9100cf2264..db8e0676ec 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -19,7 +19,7 @@ Each of the three required Linux jobs resolves its runner pool through the `DSH_
 ### Switch (repo admin, ~1 minute, no merge)
 
 1. Repository **Settings → Secrets and variables → Actions → Variables → New repository variable**: name `DSH_CI_FAILOVER`, value `selfhosted`.
-2. Re-run the failed/queued required jobs (Re-run failed jobs on affected PRs, or let new pushes pick it up).
+2. Retrigger the required jobs so they re-resolve their pool. Jobs already **queued** for the hosted labels do not retarget and cannot be re-run in place, so for the documented indefinite-queue outage, cancel the stuck run and re-run all jobs, or push a new commit; "Re-run failed jobs" only helps once a job has actually failed rather than queued.
 3. That is the entire switch. Under failover the workflow also, automatically: halves `DSH_COVERAGE_MAX_WORKERS` to 12 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 16 (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
 
 ### Capacity during failover
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index 4ec80ae364..b3b4149f46 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -19,7 +19,7 @@ Status: implemented
 ### 切换步骤（仓库管理员，约 1 分钟，无需合并）
 
 1. 仓库 **Settings → Secrets and variables → Actions → Variables → New repository variable**：名称 `DSH_CI_FAILOVER`，值 `selfhosted`。
-2. 对受影响 PR 的失败/排队作业点 Re-run failed jobs（或等新推送自然触发）。
+2. 重新触发必需作业，使其重新解析运行器池。已经为托管标签**排队**的作业不会重定向，也无法原地 re-run，因此对于本手册所述的无限排队故障，应取消卡住的运行并 re-run all jobs，或推送一个新提交；“Re-run failed jobs”只有在作业真正失败（而非仍在排队）时才有用。
 3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 12、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 16（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
 
 ### 切换期间的容量
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 0be7655193..95b54b44f9 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -55,8 +55,10 @@ jobs:
           persist-credentials: false
 
       # Pull requests consume the default-branch cache but do not put cache
-      # compression and upload on the paid latency-critical path.
+      # compression and upload on the paid latency-critical path. Skipped
+      # under failover — see the coverage lane's identical rationale.
       - uses: actions/cache/restore@v4
+        if: vars.DSH_CI_FAILOVER != 'selfhosted'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
@@ -398,7 +400,8 @@ jobs:
   # Hot-standby drill for the in-house self-hosted pool: every master move
   # re-runs the complete unsharded aggregate on the persistent 64-core VM,
   # continuously proving that environment can take over a required lane if
-  # the hosted pools degrade (the switch is then a one-line runs-on change).
+  # the hosted pools degrade (the switch is then setting the admin-only
+  # DSH_CI_FAILOVER variable — see the failover runbook, no merge required).
   # Push-triggered, so it always executes the base branch's own workflow
   # definition — no PR-editable path selects these runners. Non-blocking for
   # pull requests; no cache steps because the VM's persistent pnpm store and

From 187cf6f804bfef9800f3e07e1629207af92eac0c Mon Sep 17 00:00:00 2001
From: NI0317 <stniii317@gmail.com>
Date: Mon, 27 Jul 2026 12:38:11 +0800
Subject: [PATCH 12/27] feat(web): delete workspace registrations

---
 ...-07-25-workspace-ui-product-flow.i18n.yaml |   6 +-
 .../2026-07-25-workspace-ui-product-flow.md   |   8 +-
 ...2026-07-25-workspace-ui-product-flow.zh.md |   8 +-
 ...-workspace-registration-deletion.i18n.yaml |   6 +
 ...6-07-27-workspace-registration-deletion.md |  53 +++++++++
 ...7-27-workspace-registration-deletion.zh.md |  53 +++++++++
 ...-domain-kv-storage-and-workspace.i18n.yaml |   6 +-
 ...6-07-24-domain-kv-storage-and-workspace.md |  14 ++-
 ...7-24-domain-kv-storage-and-workspace.zh.md |  14 ++-
 apps/web/tests/workspace-management.e2e.ts    | 103 ++++++++++++++++--
 docs/cordis-catalog/services.md               |  10 ++
 .../client/connection/src/client/fixture.ts   |  15 +++
 packages/client/connection/tests/fake-api.ts  |   1 +
 .../client/connection/tests/fixture.spec.ts   |  25 +++++
 packages/client/runtime/README.i18n.yaml      |   6 +-
 packages/client/runtime/README.md             |   4 +-
 packages/client/runtime/README.zh.md          |   4 +-
 .../runtime/src/client/workspaces/manager.ts  |  46 +++++++-
 .../runtime/src/client/workspaces/service.ts  |  10 ++
 packages/client/runtime/tests/fake-api.ts     |   4 +
 .../runtime/tests/workspaces-service.spec.ts  |  58 ++++++++++
 packages/client/ui-workspace/README.i18n.yaml |   6 +-
 packages/client/ui-workspace/README.md        |   4 +-
 packages/client/ui-workspace/README.zh.md     |   4 +-
 .../src/client/WorkspaceBrowser.module.css    |  10 ++
 .../src/client/WorkspaceBrowser.tsx           |  74 ++++++++++++-
 .../ui-workspace/src/client/contract/slots.ts |   2 +
 .../client/ui-workspace/src/client/index.ts   |   1 +
 .../ui-workspace/src/client/rows/Rows.tsx     |  14 +--
 .../client/ui-workspace/tests/rows.spec.tsx   |   8 +-
 .../tests/workspace-browser.spec.tsx          |  69 ++++++++++++
 .../cordis/tool-cordis/src/api-catalog.ts     |   4 +
 packages/host/apiproxy/README.i18n.yaml       |   6 +-
 packages/host/apiproxy/README.md              |   2 +-
 packages/host/apiproxy/README.zh.md           |   2 +-
 packages/host/apiproxy/src/api-proxy.ts       |  23 +++-
 .../host/apiproxy/src/api/events.schema.ts    |   3 +-
 packages/host/apiproxy/src/api/events.ts      |   5 +-
 packages/host/apiproxy/src/api/rpc-map.ts     |   1 +
 .../host/apiproxy/src/api/workspace.schema.ts |  10 ++
 packages/host/apiproxy/src/api/workspace.ts   |   8 ++
 packages/host/apiproxy/src/fetch/client.ts    |   4 +
 packages/host/apiproxy/src/fetch/handler.ts   |   2 +
 .../tests/api-proxy-workspace.spec.ts         |  27 +++++
 .../apiproxy/tests/client-handler.spec.ts     |   5 +-
 .../host/apiproxy/tests/fetch-carrier.spec.ts |   3 +
 .../host/apiproxy/tests/rpc-schemas.spec.ts   |  13 +++
 packages/workspace/README.i18n.yaml           |   6 +-
 packages/workspace/README.md                  |   4 +-
 packages/workspace/README.zh.md               |   4 +-
 packages/workspace/workspace/README.i18n.yaml |   6 +-
 packages/workspace/workspace/README.md        |   3 +-
 packages/workspace/workspace/README.zh.md     |   3 +-
 packages/workspace/workspace/src/index.ts     |  39 +++++++
 packages/workspace/workspace/src/invariant.ts |   5 +-
 .../workspace/tests/invariant.spec.ts         |   2 +-
 .../workspace/tests/workspace.spec.ts         |  34 ++++++
 57 files changed, 786 insertions(+), 84 deletions(-)
 create mode 100644 .agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
 create mode 100644 .agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
 create mode 100644 .agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md

diff --git a/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.i18n.yaml b/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.i18n.yaml
index 3295a845f3..b8266cdd49 100644
--- a/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-25-workspace-ui-product-flow.md: a02087235a36f2c257de407facf2dc02ed072f3b
-2026-07-25-workspace-ui-product-flow.zh.md: 8ccbf5b98401bef9c3fd40e948d35ec5f0818202
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md
+2026-07-25-workspace-ui-product-flow.md: b8e1ec1efe19127cad8a12405dddeec38a4ff91e
+2026-07-25-workspace-ui-product-flow.zh.md: b80b75a80671e9aa2ab59ff72c44c18a8ec5c16e
diff --git a/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md b/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md
index a02087235a..b8e1ec1efe 100644
--- a/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md
+++ b/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.md
@@ -21,10 +21,11 @@ The Host provides the following GUI wiring on the Workspace entity:
 | `workspace.list` | Returns persistent Workspaces in order and filters out Session ids that fail header validation |
 | `workspace.create({ name })` | Creates a directory and Workspace at `workspaceRoot/name`; fails on a display-name conflict |
 | `workspace.create({ path })` | Adopts an existing directory and does not create an arbitrary path |
+| `workspace.delete({ workspaceId })` | Removes the Workspace registration while retaining its directory and session logs; its Sessions become Ungrouped |
 | `session.create({ workspaceId, sessionId? })` | Resolves cwd from the Workspace, idempotently creates a Session with an optional preallocated id, and attaches it |
 | `session.create({ cwd })` | Remains available to non-Workspace callers and creates an Ungrouped Session |
 
-`workspaceRoot` is an independent Host setting that falls back to the Host cwd when unset; it is unrelated to `storageRoot`, which stores Workspace domain data. The Host stream pushes Workspace and Session deltas, and the Client refreshes the `workspace.list` and `session.list` baselines separately after reconnecting.
+`workspaceRoot` is an independent Host setting that falls back to the Host cwd when unset; it is unrelated to `storageRoot`, which stores Workspace domain data. The Host stream pushes Workspace and Session deltas, including `host/workspace-removed`, and the Client refreshes the `workspace.list` and `session.list` baselines separately after reconnecting. Registration-deletion ownership and safety are defined in the [Workspace registration deletion Agent Note](2026-07-27-workspace-registration-deletion.md).
 
 A Workspace's `sessionIds` is an ordered candidate index. A membership projection requires both that an id appear in the index and that the corresponding canonicalized `SessionHeader.cwd` equal the Workspace path; SessionHeader does not gain a `workspaceId`. A Session whose cwd matches but whose id is absent from the index remains Ungrouped, while an indexed id is filtered out if its header is missing, its cwd is invalid, or its cwd does not match. Two Workspace indexes claiming the same Session is corrupt state and fails loudly.
 
@@ -51,7 +52,7 @@ When no Workspace exists, the page creates a frontend Workspace object named `wo
 
 Top-level New Session, the plus button on a Workspace row, and the Workspace picker all invoke the same New Session action. An explicit Workspace id becomes the target directly; when none is specified, the action uses the most recent Workspace, or the Workspace Intent if no real Workspace exists. The Workspace picker's Use an existing folder and Create a new workspace actions immediately create a real Workspace when the user confirms, then retarget the frontend Session to it; an explicitly created empty Workspace remains even if the user sends no message.
 
-Create a new workspace temporarily uses the same input as both the directory name and display name. The UI prevents duplicate confirmation based on current Workspace titles, while the Host continues to reject same-name requests that bypass the UI or race concurrently. Rename, Delete, moving across Workspaces, drag-and-drop ordering, manual adoption from Ungrouped, and separate display-name and directory-name inputs are outside this iteration's scope.
+Create a new workspace temporarily uses the same input as both the directory name and display name. The UI prevents duplicate confirmation based on current Workspace titles, while the Host continues to reject same-name requests that bypass the UI or race concurrently. Moving Sessions across Workspaces, manual adoption from Ungrouped, and separate display-name and directory-name inputs remain outside this flow.
 
 ### First send and recovery
 
@@ -75,6 +76,8 @@ A frontend Session Intent appears as a “New session” row and temporarily cou
 
 Real Sessions that cannot be assigned to any Workspace appear under Ungrouped. Host `session-added` and `workspace-changed` events may arrive in either order; list merging does not depend on frame order.
 
+Deleting a Workspace registration removes its group without deleting or closing any Session. Its accounted Sessions immediately join Ungrouped, including the current Session; a reload reconstructs the same result from the independent Workspace and Session baselines.
+
 ### React and slot boundaries
 
 React components only consume `useSessions`, `useWorkspaces`, and session-scoped hooks; they do not own entity lifecycles. The Zustand store retains only layout, the current view, composer text for ordinary real Sessions, and other purely presentational state. Session and Workspace Intents, materialization phases, errors, and retained prompts reside in the React-free runtime object layer.
@@ -106,6 +109,7 @@ The Sidebar and conversation empty hero receive standardized actions through slo
 - The initial default target is determined exactly once after both baselines are ready; Workspace groups are not reordered as a whole by hydration or Session activity, and an active Session moves only itself to the front.
 - A frontend Session under a real Workspace temporarily counts toward the sidebar total, while a Workspace Intent remains hidden; neither publication nor refresh leaves duplicate rows or counts.
 - Both the UI and Host reject duplicate Workspace names; cwd-only Sessions, Sessions with invalid historical cwd values, and unattached Sessions remain Ungrouped.
+- Confirmed Workspace deletion removes only the registration, retains the current Session, directory, files, and session log, and survives reload; package tests pin unary/frame/baseline races and failure rollback.
 - Keyless runnable snapshots cover the zero state, explicit creation, and the first send; package-level tests cover bootstrap, membership validation, ordering, idempotency, failure recovery, and arbitrary frame order.
 
 ## Consequences
diff --git a/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.zh.md b/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.zh.md
index 8ccbf5b984..b80b75a806 100644
--- a/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-25-workspace-ui-product-flow.zh.md
@@ -21,10 +21,11 @@ Host 在 Workspace entity 上提供以下 GUI 接线：
 | `workspace.list` | 返回持久有序的 Workspace，并过滤未通过 header 校验的 Session id |
 | `workspace.create({ name })` | 在 `workspaceRoot/name` 创建目录和 Workspace；显示名冲突时失败 |
 | `workspace.create({ path })` | 收编已经存在的目录，不为任意路径创建目录 |
+| `workspace.delete({ workspaceId })` | 移除 Workspace 注册记录，同时保留目录和会话日志；相关 Session 进入 Ungrouped |
 | `session.create({ workspaceId, sessionId? })` | 从 Workspace 解析 cwd，以可选预分配 id 幂等创建 Session 并 attach |
 | `session.create({ cwd })` | 保留给非 Workspace 调用方，创建 Ungrouped Session |
 
-`workspaceRoot` 是独立 Host 配置，未配置时回退到 Host cwd；它与保存 Workspace domain 数据的 `storageRoot` 无关。Host stream 推送 Workspace 与 Session 增量，Client 重连后分别刷新 `workspace.list` 与 `session.list` 基线。
+`workspaceRoot` 是独立 Host 配置，未配置时回退到 Host cwd；它与保存 Workspace domain 数据的 `storageRoot` 无关。Host stream 推送 Workspace 与 Session 增量，包括 `host/workspace-removed`；Client 重连后分别刷新 `workspace.list` 与 `session.list` 基线。删除注册记录的所有权与安全边界由 [Workspace 注册记录删除 Agent Note](2026-07-27-workspace-registration-deletion.md)定义。
 
 Workspace 的 `sessionIds` 是有序候选索引。成员投影同时要求 id 位于索引且对应 `SessionHeader.cwd` canonical 后等于 Workspace path；SessionHeader 不增加 `workspaceId`。cwd 匹配但未入索引的 Session 保持 Ungrouped，索引命中但 header 缺失、cwd 无效或 cwd 不匹配的 id 被过滤。同一 Session 被两个 Workspace 索引占用属于损坏状态并 fail loud。
 
@@ -51,7 +52,7 @@ Session 自己持有首条输入并驱动一条内部流水线：必要时以预
 
 顶部 New Session、Workspace 行内加号和 Workspace picker 最终都调用同一 New Session 动作：显式 Workspace id 直接成为目标，未指定时使用最近 Workspace，没有真实 Workspace 时使用 Workspace Intent。Workspace picker 的 Use an existing folder 与 Create a new workspace 会在用户确认时立即创建真实 Workspace，再把前端 Session 定位到该 Workspace；即使用户不发送消息，显式创建的空 Workspace 也保留。
 
-Create a new workspace 暂时用同一个输入作为目录名和显示名。UI 根据当前 Workspace title 禁止重复确认，Host 继续拒绝绕过 UI 或并发产生的同名请求。Rename、Delete、跨 Workspace 移动、拖拽排序、Ungrouped 手动收编和显示名/目录名双输入不在本期范围。
+Create a new workspace 暂时用同一个输入作为目录名和显示名。UI 根据当前 Workspace title 禁止重复确认，Host 继续拒绝绕过 UI 或并发产生的同名请求。跨 Workspace 移动 Session、从 Ungrouped 手动收编以及分别输入显示名和目录名仍不在此动线范围内。
 
 ### 首次发送与恢复
 
@@ -75,6 +76,8 @@ Workspace 组严格使用 Host 返回的持久顺序。Bootstrap 一次性确定
 
 无法归入任何 Workspace 的真实 Session 进入 Ungrouped。Host `session-added` 与 `workspace-changed` 可以任意顺序到达，列表合并不依赖 frame 顺序。
 
+删除 Workspace 注册记录会移除其分组，但不会删除或关闭任何 Session。已记账的 Session（包括当前 Session）会立即进入 Ungrouped；刷新后，独立的 Workspace 与 Session 基线会重建出相同结果。
+
 ### React 与 slot 边界
 
 React 组件只消费 `useSessions`、`useWorkspaces` 与 session-scoped hooks，不拥有实体生命周期。Zustand store 只保留布局、当前 view、普通真实 Session 的 composer 文本和其他纯呈现状态；Session/Workspace Intent、materialize phase、错误和 retained prompt 位于 React-free runtime 对象层。
@@ -106,6 +109,7 @@ Sidebar 与 conversation empty hero 通过 slot 获得标准化动作：`startSe
 - 初始默认目标只在两份基线 ready 后确定一次；Workspace 组不因 hydration 或 Session 活跃整体重排，单个活跃 Session 只前移自身。
 - 真实 Workspace 下的前端 Session 临时计入 sidebar 数量，Workspace Intent 保持隐藏，发布与刷新都不会留下重复行或重复计数。
 - UI 与 Host 两层拒绝同名 Workspace；cwd-only Session、无效历史 cwd 和未 attach Session 保持 Ungrouped。
+- 经确认的 Workspace 删除只移除注册记录，保留当前 Session、目录、文件和会话日志，并在刷新后保持该状态；包级测试固定一元响应／帧／基线竞态和失败回滚行为。
 - keyless runnable snapshot 覆盖零态、显式创建和首次发送；包级测试覆盖 bootstrap、成员校验、排序、幂等、失败恢复及任意 frame 顺序。
 
 ## Consequences
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
new file mode 100644
index 0000000000..847b040457
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
+2026-07-27-workspace-registration-deletion.md: cae01d529bc6fd97da6fb61839bd5ec8e21557e2
+2026-07-27-workspace-registration-deletion.zh.md: 76377ebc5e93101e1e3efce1d29c3c654df032c2
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
new file mode 100644
index 0000000000..cae01d529b
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
@@ -0,0 +1,53 @@
+# Agent Note: Workspace Registration Deletion
+
+Status: implemented
+
+English | [中文](2026-07-27-workspace-registration-deletion.zh.md)
+
+## Problem
+
+A Workspace registers an existing code directory so the GUI can name it and order its Sessions. That record has no reliable provenance proving that Harness created or owns the directory, and the Session log is an independent persistence object. Treating the row's Delete action as recursive source deletion or Session deletion would destroy data outside the record's ownership boundary.
+
+The existing visual-only menu row also left deletion semantics undefined across durable order, the Workspace table, Host streams, concurrent browser tabs, reconnect baselines, and a list request racing the mutation.
+
+## Decision
+
+`ctx.workspace.delete(id)` deletes only the Workspace registration: its id leaves durable `workspaceIds`, its `workspaces` table row and entity-cache entry disappear, and its ordered `sessionIds` account disappears with that row. It never calls filesystem removal or `SessionPersistence`; the directory, every user file, every live Session, and every persisted Session log remain. Because sidebar grouping is the complement of all surviving Workspace accounts, those Sessions immediately appear under Ungrouped, including the current Session.
+
+Unknown ids return `false` at the domain seam. `workspace.delete({ workspaceId })` maps that distinction to `workspace-not-found`; success returns `{ deleted: true }`. `workspace.list` remains the reconnect baseline.
+
+## Durable commit and publication
+
+Registry operations serialize create and delete. Deletion first writes the Workspace order without the id, then removes the entity from the cache, then deletes the table row. The table deletion is the notification commit point: the package invariant accepts it only after the cache stopped publishing the entity, and the Host emits `host/workspace-removed` only from that committed deletion. A table-write failure restores the cache and prior durable order; no removal frame is published.
+
+The Host stream keeps its committed-id set through the preceding global-order write and removes the id only on the table deletion. Create rollback therefore emits no false removal, while every connected tab receives exactly the id needed to delete its projection.
+
+## Client convergence
+
+`WorkspaceManager` treats both `host/workspace-changed` and `host/workspace-removed` as ordered deltas replayed over an in-flight `workspace.list` response. A successful unary delete removes the row immediately instead of waiting for its own stream echo. Removal is idempotent, and a process-local tombstone rejects late changed frames or stale baseline rows for the never-reused Workspace id. A reconnect still refreshes from `workspace.list`; Session state is never pruned by a Workspace delta.
+
+## Confirmation interaction
+
+The existing Workspace row menu opens a shared `Modal` before deletion. The text states all three consequences: the Workspace leaves the list, the folder and session logs remain, and its Sessions appear under Ungrouped. While the request is pending, the confirm and Cancel controls are disabled, duplicate confirmation is ignored, and Escape or Close cannot dismiss the operation. Failure keeps the Modal open with the error; Cancel, Escape, and Close before submission never delete.
+
+The menu, Modal, and buttons retain their existing structure and design tokens. Session deletion remains visual-only and outside this decision.
+
+## Alternatives considered
+
+**Cascade-delete Sessions.** Rejected because Workspace registration does not own Session persistence and the product requirement is to preserve histories under Ungrouped. Session deletion needs its own lifecycle, running checks, descendant semantics, and explicit UI.
+
+**Move the folder to Trash.** Rejected because the record cannot prove directory ownership. A future destructive filesystem action must be separately named, separately confirmed, and enforce explicit safety boundaries.
+
+**Delete the table row and repair order later.** Rejected because a crash or write failure would leave an initialized registry whose order and table disagree. The registry updates both under one serialized operation and restores the prior order on table failure.
+
+**Refetch both lists after success.** Rejected because the committed removal frame plus immediate unary echo is sufficient, preserves the current Session object, and avoids turning a local mutation into two list requests. Reconnect baselines remain the repair path.
+
+## Verification
+
+Workspace package tests pin successful metadata-only deletion, unknown-id idempotence, table-failure rollback, and cache/table invariant behavior. Apiproxy and carrier tests pin the schema, handler, `workspace-not-found`, retained Session/folder, and committed `host/workspace-removed` frame. Client tests pin unary direct echo, duplicate removal, late changed frames, and deletion racing an in-flight baseline. Component tests pin confirmation, pending-state duplicate suppression, success, failure, Cancel, Escape, and Close.
+
+The assembled keyless Web scenario registers an existing temporary project directory, accounts a persisted Session, makes that Session current, confirms deletion in Chromium, and verifies the Workspace group disappears while Ungrouped retains the current Session. It checks the user file and JSONL log before and after deletion and repeats the UI, directory, and log assertions after reload.
+
+## Consequences
+
+Deleting a Workspace is intentionally reversible by registering the same directory again, although its prior manual Session order is gone; re-registration does not automatically re-adopt existing Sessions after bootstrap. The operation gives up a one-click cleanup of Session histories or source directories in exchange for a deletion boundary that matches what the record actually owns.
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
new file mode 100644
index 0000000000..76377ebc5e
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
@@ -0,0 +1,53 @@
+# Agent Note（agent 决策记录）：删除 Workspace 注册记录
+
+Status: implemented
+
+[English](2026-07-27-workspace-registration-deletion.md) | 中文
+
+## Problem
+
+Workspace 注册已有代码目录，使 GUI 能够为目录命名，并对其会话排序。该记录没有可靠的来源信息来证明 Harness 创建或拥有该目录，会话日志也是独立的持久化对象。若将行内 Delete 操作视为递归删除源码或删除会话，就会破坏该记录所有权边界之外的数据。
+
+现有菜单行仅提供视觉效果，因此持久顺序、Workspace 表、Host 流、并发浏览器标签页、重连基线，以及列表请求与变更并发时的删除语义也没有定义。
+
+## Decision
+
+`ctx.workspace.delete(id)` 只删除 Workspace 注册记录：其 id 会从持久 `workspaceIds` 中移除，`workspaces` 表行与实体缓存条目会消失，有序 `sessionIds` 账本也随该行一并消失。它绝不调用文件系统移除操作或 `SessionPersistence`；目录、所有用户文件、所有实时会话和所有持久化会话日志都会保留。侧边栏分组是所有存续 Workspace 账本的补集，因此这些会话（包括当前会话）会立即出现在 Ungrouped 下。
+
+未知 id 在 domain seam 返回 `false`。`workspace.delete({ workspaceId })` 将该结果映射为 `workspace-not-found`；成功时返回 `{ deleted: true }`。`workspace.list` 仍是重连基线。
+
+## 持久提交与发布
+
+注册表操作会串行执行创建与删除。删除时先写入移除该 id 后的 Workspace 顺序，再从缓存中移除实体，最后删除表行。表删除是通知提交点：只有缓存停止发布该实体后，包不变量才接受该删除；Host 也只根据这次已提交的删除发出 `host/workspace-removed`。表写入失败时，系统会恢复缓存和此前的持久顺序，且不会发布移除帧。
+
+Host 流在前一笔全局顺序写入期间继续保留其已提交 id 集合，只在删除表行时移除该 id。因此，创建回滚不会发出错误的移除帧，而每个已连接标签页都能收到从自身投影中删除该记录所需的准确 id。
+
+## 客户端收敛
+
+`WorkspaceManager` 将 `host/workspace-changed` 与 `host/workspace-removed` 都视为有序增量，并在进行中的 `workspace.list` 响应之上回放。成功的一元删除会立即移除行，无需等待本次操作自己的流回显。移除操作具有幂等性；由于 Workspace id 永不复用，进程本地删除标记会拒绝延迟到达的 changed 帧或陈旧基线行。重连仍从 `workspace.list` 刷新；Workspace 增量绝不会剪除会话状态。
+
+## 确认交互
+
+现有 Workspace 行菜单会在删除前打开共享 `Modal`。文案明确说明三项后果：Workspace 会从列表中移除，文件夹和会话日志会保留，相关会话会出现在 Ungrouped 下。请求待处理期间，确认与 Cancel 控件均被禁用，重复确认会被忽略，Escape 或 Close 也无法关闭此次操作。失败时 `Modal` 保持打开并显示错误；提交前使用 Cancel、Escape 或 Close 绝不会触发删除。
+
+菜单、`Modal` 和按钮保留现有结构与设计 token。会话删除仍仅提供视觉效果，不在本决策范围内。
+
+## Alternatives considered
+
+**级联删除会话。** 不予采纳，因为 Workspace 注册记录不拥有会话持久化，且产品需求是将历史记录保留在 Ungrouped 下。会话删除需要自己的生命周期、运行状态检查、后代对象的处理语义和明确 UI。
+
+**将文件夹移到废纸篓。** 不予采纳，因为该记录无法证明目录所有权。未来的破坏性文件系统操作必须使用单独名称、单独确认，并实施明确的安全边界。
+
+**先删除表行，之后再修复顺序。** 不予采纳，因为崩溃或写入失败会使已初始化注册表的顺序与表不一致。注册表会在同一串行操作内更新二者，并在表操作失败时恢复此前顺序。
+
+**成功后重新拉取两个列表。** 不予采纳，因为已提交的移除帧与即时一元回显已足够，既能保留当前会话对象，也避免将局部变更扩大为两次列表请求。重连基线仍是修复路径。
+
+## Verification
+
+Workspace 包测试固定了仅删除元数据的成功路径、未知 id 的幂等行为、表操作失败回滚，以及缓存／表不变量行为。Apiproxy 与载体测试固定了 schema、处理器、`workspace-not-found`、保留会话／文件夹，以及已提交的 `host/workspace-removed` 帧。客户端测试固定了一元直接回显、重复移除、延迟到达的 changed 帧，以及删除与进行中基线并发的行为。组件测试固定了确认交互、待处理状态下抑制重复提交、成功、失败、Cancel、Escape 与 Close。
+
+组装后的无密钥 Web 场景会注册一个已有临时项目目录，将持久化会话计入账本，把该会话设为当前会话，在 Chromium 中确认删除，并验证 Workspace 分组消失，而 Ungrouped 保留当前会话。该场景在删除前后检查用户文件和 JSONL 日志，并在刷新后重复验证 UI、目录与日志。
+
+## Consequences
+
+删除 Workspace 后仍可重新注册同一目录，因此该操作有意设计为可逆；但此前的手动会话顺序会丢失，重新注册后，系统也不会在 bootstrap 结束后自动重新收编现有会话。该操作放弃一键清理会话历史或源码目录，以换取与记录实际所有权一致的删除边界。
diff --git a/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.i18n.yaml b/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.i18n.yaml
index 5ab7a8229d..6e5f8b8391 100644
--- a/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.i18n.yaml
+++ b/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-24-domain-kv-storage-and-workspace.md: cd666a47a3cba4dea8846cd0f1373224e6fc456f
-2026-07-24-domain-kv-storage-and-workspace.zh.md: 81adf1eb6bc32aa3ca8b9ef4c352fb94f95ace91
+#   pnpm run verify-translation-pairing --write .agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md
+2026-07-24-domain-kv-storage-and-workspace.md: 230877628428dc88dbddeecfe5f4353cf15e151d
+2026-07-24-domain-kv-storage-and-workspace.zh.md: 050f72cd3327f83e2c3f3cefcab63c01e8f112ee
diff --git a/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md b/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md
index cd666a47a3..2308776284 100644
--- a/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md
+++ b/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md
@@ -11,7 +11,9 @@ The host's only persistence surface is the session event log (`packages/session-
 - **The workspace entity.** The GUI needs workspace as a real object: path, title, and the list of owned sessions. Ownership belongs to the workspace — "which sessions belong to this workspace" is not any single session's fact, so writing it into the session log is semantically wrong. Until now workspace was only a sidebar visual grouping derived from cwd, with no entity (that conclusion has been overturned).
 - **Dynamic session metadata** (the foreseeable second consumer). Cold session listings read only the first log line (an immutable creation-time snapshot); title, terminal status, and anything that evolves with the session is unavailable. The fix direction is a sidecar metadata table — exactly a KV table with high-frequency per-key updates.
 
-Separately, workspace deletion will eventually need to delete its owned sessions, and `SessionPersistence` has no delete primitive nor does the host expose a `session.delete` endpoint — that gap's design is settled in this note, but its implementation is marked future work: this phase touches no session-side code.
+Separately, Session deletion needs a `SessionPersistence` delete primitive and a `session.delete` endpoint. That gap's design is settled in this note, but its implementation remains future work.
+
+The later [Workspace registration deletion decision](../../implemented/feature/2026-07-27-workspace-registration-deletion.md) supersedes only that coupling: deleting a Workspace registration preserves its Sessions and their logs, while Session deletion remains separate future work. The cascade design below is therefore not the Workspace GUI delete semantic.
 
 ## Proposal
 
@@ -234,14 +236,14 @@ export class WorkspaceRegistry extends Service {
   get(id: WorkspaceId): Workspace | undefined
   list(): Workspace[]
   resolveByPath(path: string): Promise<Workspace | undefined> // 同 realpath 口径，故 async
-  // delete：future work（与 session 级联删一起做，见下）；本期不提供任何删除入口
+  delete(id: WorkspaceId): Promise<boolean>      // 只删注册记录；目录与 session 日志保留
 }
 ```
 
 - **Path canon**: the stored value = `fs.realpath(input)` (trailing slashes, `..`, and symlinks all resolved); uniqueness = string equality after normalization (a symlink resolving to the same directory counts as a collision). A missing directory makes create reject outright (realpath fails — a workspace must point at an existing directory; "Create new = make the directory" is upper-layer interaction: mkdir first, then create). The session cwd in attach checks follows the same canon. Single-valued cwd + unique path ⇒ one session structurally belongs to at most one workspace; double bookkeeping is impossible on the write side.
 - **Title**: a display name, defaults to `basename(path)`, mutable, duplicates allowed. Ownership is never derived from cwd as a fallback — cwd cannot express ordering, and ownership is a workspace-side fact; sessions started headless belong to no workspace.
 - Consumers see only the `Workspace` interface; `WorkspaceEntity` stays inside the package (a single implementation does not pre-split a seam). Entities are unique per id (registry cache); the record snapshot is swapped in place after each write, and the outside sees getters only. Every write funnels through the entity's internal `mutate(fn)` → `table.update`, with `updatedAt` refreshed inside mutate. Domain objects never cross RPC; next phase the wire layer projects records into zod wire schemas.
-- **Workspace deletion is future work as a whole** (settled 2026-07-24): the registry ships no delete method this phase — the half-measure "delete the record, keep the sessions" is not exposed; deletion and the session cascade (`recursive` parameter, running checks, bottom-up order, crash-rerun convergence) land as one complete semantic together with the session delete primitive; the order then is delete sessions one by one → prune the ledger → delete the workspace record.
+- **Session deletion remains future work.** The later [Workspace registration deletion decision](../../implemented/feature/2026-07-27-workspace-registration-deletion.md) ships `ctx.workspace.delete(id)` as a metadata-only operation that preserves Sessions and logs. Recursive Session deletion, running checks, and crash-rerun convergence belong to a separate `session.delete` capability.
 
 Consistency doctrine (the ledger = the only ownership authority; the implementation and test baseline):
 
@@ -285,7 +287,7 @@ Snapshots: no model-visible or assembly surface this phase, none added; next pha
 
 | Not doing | Trigger | Rework point | Groundwork |
 | --- | --- | --- | --- |
-| The full deletion suite (`SessionPersistence.delete`, the deleted event, `registry.delete` cascade, recursive delete, running checks) | future work starts (before the GUI needs delete interactions) | implement per the future-work section above: the session primitive + `registry.delete(id, { recursive? })` land as one | orchestration rules and rejection table settled in this note; no deletion entry exists this phase, so no half-semantics to stay compatible with |
+| Session deletion (`SessionPersistence.delete`, the deleted event, recursive delete, running checks) | a destructive Session-delete product flow starts | implement the session primitive plus `session.delete`; keep it independent from Workspace registration deletion | orchestration rules and rejection table above remain groundwork; Workspace deletion preserves Sessions and logs |
 | The `log` facet and the session-backend migration | any phase after this one | sink the medium operations (the reuse audit table is the work list) | the facet structure is in place; both backends' medium code is organized in sinkable shape already |
 | Multi-process write protection | two host processes writing one medium | JSON backend file locks; SQLite WAL is natively multi-process | all writes already funnel through the domain's single point; locking touches backends only |
 | Cross-process change observation | GUI reconnect awareness | the revision pattern (copy session-persistence) | `domain/changed` already exists in-process |
@@ -296,7 +298,7 @@ Snapshots: no model-visible or assembly surface this phase, none added; next pha
 | Cross-table atomic transactions | one business operation touching two tables of one domain atomically | `domain.transact(fn)`; JSON whole-unit rewrite is naturally atomic, SQLite wraps a transaction | — |
 | Secondary indexes / conditional queries | in-memory filtering stops scaling (tens of thousands of records) | SQLite JSON1 over the value column, a read-only query facet on the seam | the JSON backend does not follow |
 | Moving a session across workspaces | a product need appears | relax the attach check into a "detach first, then attach" orchestration | — |
-| RPC/GUI/boot | next phase | `workspace.*` + `session.delete` endpoints, wire schemas, boot mounting, sidebar on real data | this phase's model and semantics are the direct source of the wire projection |
+| Session-delete RPC/GUI | a destructive Session-delete product flow starts | `session.delete` endpoint, wire schema, and explicit confirmation UI | Workspace RPC/GUI is shipped separately; no cascade coupling remains |
 
 ## Alternatives considered
 
@@ -317,7 +319,7 @@ Snapshots: no model-visible or assembly surface this phase, none added; next pha
 ## Acceptance criteria
 
 - This phase's four test suites all green: the shared backend contract suite on both json/sqlite, registry/mount disposer semantics, the domain layer (including the six open steps and fail-loud routing), and full workspace semantics (create/attach checks/consistency doctrine).
-- `ctx.workspace` completes the create → attach → list lifecycle under a test assembly (deletion is future work).
+- `ctx.workspace` completes the create → attach → list → metadata-only delete lifecycle under a test assembly.
 - Zero diff in the session-persistence packages (the acceptance line for not touching the session side this phase).
 - No new snapshots this phase (no model-visible or assembly surface); added next phase with the RPC wiring.
 
diff --git a/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md b/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md
index 81adf1eb6b..050f72cd33 100644
--- a/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md
+++ b/.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md
@@ -11,7 +11,9 @@ host 侧唯一的持久化面是 session 事件日志（`packages/session-persis
 - **workspace 实体**。GUI 要把 workspace 做成真实对象：路径、标题、关联 session 清单。归属关系由 workspace 持有——"哪些 session 属于这个 workspace"不是任何单个 session 自己的事实，塞进 session log 语义不成立。此前 workspace 只是 sidebar 上按 cwd 分组的视觉概念，没有实体（该结论已被推翻）。
 - **session 动态元信息**（可预见的第二个消费者）。冷会话列表只读日志首行 header（创建时的不可变快照），title、结束状态这类随会话推进变化的信息拿不到；补齐方向是 sidecar 元数据表——正是一张按 key 高频点更新的 KV 表。
 
-另外，workspace 删除最终需要删除其关联 session，而 `SessionPersistence` 没有删除原语，host 也没有 `session.delete` 端点——该空白的设计随本 Note 定案，但实施标记为 future work：本期不动 session 侧任何代码。
+另外，Session 删除需要 `SessionPersistence` 删除原语和 `session.delete` 端点。该空白的设计随本 Note 定案，但实现仍属未来工作。
+
+后续的 [Workspace 注册记录删除决策](../../implemented/feature/2026-07-27-workspace-registration-deletion.md)取代的仅是上述耦合关系：删除 Workspace 注册记录会保留相关 Session 及其日志，Session 删除仍是独立的未来工作。因此，下文的级联设计并不是 Workspace GUI 的删除语义。
 
 ## Proposal
 
@@ -234,14 +236,14 @@ export class WorkspaceRegistry extends Service {
   get(id: WorkspaceId): Workspace | undefined
   list(): Workspace[]
   resolveByPath(path: string): Promise<Workspace | undefined> // 同 realpath 口径，故 async
-  // delete：future work（与 session 级联删一起做，见下）；本期不提供任何删除入口
+  delete(id: WorkspaceId): Promise<boolean>      // 只删注册记录；目录与 session 日志保留
 }
 ```
 
 - **path 规范**：落盘值 = `fs.realpath(输入)`（尾斜杠、`..`、符号链接全解析）；唯一性 = 规范化后字符串相等（符号链接指向同一目录算撞）。目录不存在时 create 直接 reject（realpath 失败——workspace 必须指向存在目录；"Create new = 建目录"是上层交互，先 mkdir 再 create）。attach 校验的 session cwd 同口径。cwd 单值 + path 唯一 ⇒ 一个 session 结构上最多归属一个 workspace，双重记账写侧不可能。
 - **title**：显示名，默认 `basename(path)`，可改，允许重复。归属不用 cwd 派生兜底——cwd 表达不了排序，归属是 workspace 侧事实；headless 直开的 session 不属于任何 workspace。
 - 消费者只见 `Workspace` 接口，`WorkspaceEntity` 不出包（单实现不预拆 seam）；实体按 id 唯一（registry 缓存），记录快照写后原地换新，外部只见 getter；所有写收敛到实体内 `mutate(fn)` → `table.update`，`updatedAt` 在 mutate 内统一刷。领域对象不过 RPC，下期 wire 层把记录投影成 zod wire schema。
-- **workspace 删除整体为 future work**（2026-07-24 拍板）：本期 registry 不提供 delete 方法——半截的"只删记录留 session"语义不对外暴露，删除与 session 级联（`recursive` 参数、运行中检查、自底向上、崩溃重跑收敛）作为一个完整语义随 session 删除原语一起落地；届时顺序为逐个删 session → 摘账 → 删记录。
+- **Session 删除仍属未来工作。** 后续的 [Workspace 注册记录删除决策](../../implemented/feature/2026-07-27-workspace-registration-deletion.md)已将 `ctx.workspace.delete(id)` 作为仅删除元数据、保留 Session 与日志的操作交付。递归删除 Session、运行中检查和崩溃重跑收敛属于独立的 `session.delete` 能力。
 
 一致性口径（账 = 归属唯一依据；实现与测试基准）：
 
@@ -285,7 +287,7 @@ export class WorkspaceRegistry extends Service {
 
 | 不做 | 触发条件 | 返工点 | 预埋 |
 | --- | --- | --- | --- |
-| 删除全套（`SessionPersistence.delete`、deleted 事件、`registry.delete` 级联、递归删、运行中检查） | future work 启动（GUI 需要删除交互前） | 按上文 future work 节实施：session 原语 + `registry.delete(id, { recursive? })` 一体落地 | 编排规则/拒绝清单已定案在本 Note；本期无任何删除入口，无半截语义要兼容 |
+| Session 删除（`SessionPersistence.delete`、deleted 事件、递归删除、运行中检查） | 破坏性的 Session 删除产品流启动 | 实现 Session 原语及 `session.delete`；与 Workspace 注册记录删除保持独立 | 上文编排规则和拒绝清单仍是基础；Workspace 删除会保留 Session 与日志 |
 | `log` facet 与 session 后端迁移 | 本期后任意期启动 | 介质操作下沉（复用审计表即施工清单） | facet 结构已留位；两后端介质代码本期即按可下沉形状组织 |
 | 多进程并发写保护 | 两 host 进程同写一介质 | JSON 后端文件锁；SQLite WAL 天然多进程 | 写全经 domain 单点串行，加锁只动后端 |
 | 跨进程变更观测 | GUI 断线重连感知 | revision 模式（抄 session-persistence） | 进程内已有 `domain/changed` |
@@ -296,7 +298,7 @@ export class WorkspaceRegistry extends Service {
 | 跨表原子事务 | 同域两表一次原子操作需求 | `domain.transact(fn)`；JSON 天然原子，SQLite 包事务 | — |
 | 二级索引/条件查询 | 内存过滤不动（万级记录） | SQLite JSON1 查 value 列，加只读 query 面 | JSON 后端不陪跑 |
 | session 跨 workspace 移动 | 产品需求出现 | attach 校验放宽为"先 detach 后 attach"编排 | — |
-| RPC/GUI/boot | 下期 | `workspace.*` + `session.delete` 端点、wire schema、boot 挂载、sidebar 接真数据 | 本期模型与语义即 wire 投影的直接来源 |
+| Session 删除 RPC／GUI | 破坏性的 Session 删除产品流启动 | `session.delete` 端点、wire schema 与明确的确认 UI | Workspace RPC／GUI 已独立交付，不再存在级联耦合 |
 
 ## Alternatives considered
 
@@ -317,7 +319,7 @@ export class WorkspaceRegistry extends Service {
 ## Acceptance criteria
 
 - 测试矩阵本期四套件全绿：backend 契约共享套件在 json/sqlite 双端、registry/mount disposer 语义、domain 层（含 open 六步与路由 fail-loud）、workspace 全语义（create/attach 校验/一致性口径）。
-- `ctx.workspace` 可在测试组装下完成 create → attach → list 生命周期（删除为 future work）。
+- `ctx.workspace` 可在测试组装下完成 create → attach → list → 仅删除元数据的 delete 生命周期。
 - session-persistence 包零 diff（本期不动 session 侧的验收线）。
 - 本期无新快照（无模型可见面与组装面）；下期 RPC 接线时补。
 
diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts
index a19211c6bf..4dbcca36ae 100644
--- a/apps/web/tests/workspace-management.e2e.ts
+++ b/apps/web/tests/workspace-management.e2e.ts
@@ -5,12 +5,13 @@
 // calls: workspace.create/rename are host RPCs with no model involvement,
 // and the one session row the flat/hover scenarios need comes from a seeded
 // fixture (the seeded-history seed reused verbatim — no new recording).
-import { mkdir, readFile, writeFile } from 'node:fs/promises'
+import { mkdir, readFile, stat, writeFile } from 'node:fs/promises'
 import { fileURLToPath } from 'node:url'
 import { join } from 'node:path'
 import type { Browser, Page } from 'playwright'
 import { chromium } from 'playwright'
 import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest'
+import { SessionId } from '@deepseek-ai/dsh-session'
 import {
   acknowledgeReloadConnectionLoss, assertFixtureInventory, launchWebScaffold, seedSession, watchConsole,
   webSnapshotMode, type WebScaffold,
@@ -105,6 +106,86 @@ describe('web e2e: workspace management (create / rename / flat view / hover car
     expect(tripwire.pageErrors).toEqual([])
   }, 90_000)
 
+  it('deletes only the Workspace registration and keeps its current Session, folder, and log', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-ws-delete'))
+    // Register the scaffold's existing project directory through the real UI.
+    await page.getByRole('button', { name: 'Create workspace' }).click()
+    await page.getByRole('menuitem', { name: 'Create workspace' }).hover()
+    await page.getByRole('menuitem', { name: 'Use an existing folder' }).click()
+    const useFolder = page.getByRole('dialog', { name: 'Use an existing folder' })
+    await useFolder.getByLabel('Existing folder path').fill(scaffold.workspaceCwd)
+    await useFolder.getByRole('button', { name: 'Use folder' }).click()
+    await expect.poll(() => useFolder.count(), { timeout: 10_000 }).toBe(0)
+
+    const workspace = await scaffold.ctx.workspace.resolveByPath(scaffold.workspaceCwd)
+    if (workspace === undefined) throw new Error('GUI did not register the existing project directory')
+    await workspace.attachSession(SessionId(SEED_ID))
+    const header = (await scaffold.ctx.sessionPersistence.list())
+      .find(candidate => candidate.id === SEED_ID)
+    if (header === undefined) throw new Error('seeded Session log disappeared before deletion')
+    const logLocation = scaffold.ctx.sessionPersistence.locate(header)
+    if (logLocation === undefined) throw new Error('JSONL persistence did not expose the seeded log path')
+    expect(await readFile(join(scaffold.workspaceCwd, 'workspace', 'a.txt'), 'utf8')).toBe('alpha\n')
+    await stat(logLocation.path)
+
+    // Open the seeded (first/accounted) Session so deletion must preserve the
+    // current selection while it moves into Ungrouped.
+    const groupRow = page.locator('[role="treeitem"]').filter({ hasText: workspace.title }).first()
+    await groupRow.waitFor({ timeout: 10_000 })
+    const groupSection = groupRow.locator('..')
+    if (await groupSection.locator('[role="treeitem"]').count() < 2) await groupRow.click()
+    await expect.poll(
+      () => groupSection.locator('[role="treeitem"]').count(),
+      { timeout: 10_000 },
+    ).toBeGreaterThanOrEqual(2)
+    const seededRow = groupSection.locator('[role="treeitem"]').nth(1)
+    await seededRow.click()
+    await expect.poll(() => seededRow.getAttribute('aria-selected'), { timeout: 10_000 }).toBe('true')
+
+    await groupRow.hover()
+    await page.getByRole('button', { name: `Workspace actions for ${workspace.title}` }).click()
+    await page.getByRole('menuitem', { name: 'Delete workspace' }).click()
+    const dialog = page.getByRole('dialog', { name: 'Delete workspace' })
+    await dialog.waitFor({ timeout: 10_000 })
+    const copy = await dialog.textContent()
+    expect(copy).toContain('workspace list')
+    expect(copy).toContain('folder and session logs will be kept')
+    expect(copy).toContain('sessions will appear under Ungrouped')
+    await dialog.getByRole('button', { name: 'Delete workspace' }).click()
+    await expect.poll(() => dialog.count(), { timeout: 10_000 }).toBe(0)
+
+    expect(scaffold.ctx.workspace.get(workspace.id)).toBeUndefined()
+    await expect.poll(
+      () => page.getByRole('button', { name: `Workspace actions for ${workspace.title}` }).count(),
+      { timeout: 10_000 },
+    ).toBe(0)
+    await expect.poll(() => page.getByText('Ungrouped', { exact: true }).count(), { timeout: 10_000 })
+      .toBeGreaterThanOrEqual(1)
+    await expect.poll(
+      () => page.locator('[role="treeitem"][aria-selected="true"]').count(),
+      { timeout: 10_000 },
+    ).toBe(1)
+    expect(await readFile(join(scaffold.workspaceCwd, 'workspace', 'a.txt'), 'utf8')).toBe('alpha\n')
+    await stat(logLocation.path)
+    expect((await scaffold.ctx.sessionPersistence.inspect(SessionId(SEED_ID))).events.length).toBeGreaterThan(0)
+
+    const warningStart = tripwire.warnings.length
+    await page.reload({ waitUntil: 'load' })
+    await page.waitForSelector('[class*="frame"]', { timeout: 30_000 })
+    acknowledgeReloadConnectionLoss(tripwire, warningStart)
+    await expect.poll(() => page.getByText('Ungrouped', { exact: true }).count(), { timeout: 15_000 })
+      .toBeGreaterThanOrEqual(1)
+    await expect.poll(
+      () => page.locator('[role="treeitem"][aria-selected="true"]').count(),
+      { timeout: 15_000 },
+    ).toBe(1)
+    expect(scaffold.ctx.workspace.get(workspace.id)).toBeUndefined()
+    expect(await readFile(join(scaffold.workspaceCwd, 'workspace', 'a.txt'), 'utf8')).toBe('alpha\n')
+    await stat(logLocation.path)
+    expect((await scaffold.ctx.sessionPersistence.inspect(SessionId(SEED_ID))).events.length).toBeGreaterThan(0)
+    expect(tripwire.pageErrors).toEqual([])
+  }, 90_000)
+
   it('switches to the flat "In one list" view and persists the preference', async () => {
     onTestFailed(() => saveFailureShot(page, 'web-e2e-ws-flat'))
     // Grouped default: workspace group rows render (the seeded session sits
@@ -134,12 +215,20 @@ describe('web e2e: workspace management (create / rename / flat view / hover car
     onTestFailed(() => saveFailureShot(page, 'web-e2e-ws-hover'))
     // Expand Ungrouped to reveal the seeded session row, then dwell on it
     // (the card opens after a 500ms hover delay, portaled to body).
-    await page.getByText('Ungrouped', { exact: true }).click()
-    // A cold summary carries no durable title, so the row falls back to a
-    // cwd-derived display title — anchored on the run-local workspace-root
-    // basename rather than a literal.
-    const wsBase = scaffold.workspaceCwd.split('/').pop()!
-    const sessionRow = page.locator('[role="treeitem"]').filter({ hasText: wsBase }).first()
+    const ungroupedRow = page.getByText('Ungrouped', { exact: true }).locator('..').locator('..')
+    const ungroupedSection = ungroupedRow.locator('..')
+    // Initial-current auto-expansion can race this following test's gesture;
+    // converge on expanded rather than assuming which update wins first.
+    await expect.poll(async () => {
+      if (await ungroupedRow.getAttribute('aria-expanded') !== 'true') {
+        await page.getByText('Ungrouped', { exact: true }).click()
+        await page.waitForTimeout(50)
+      }
+      return await ungroupedRow.getAttribute('aria-expanded')
+    }, { timeout: 5_000 }).toBe('true')
+    // The only visible child is the non-blank persisted Session; the blank
+    // Session created while adopting the Workspace remains hidden.
+    const sessionRow = ungroupedSection.locator('[role="treeitem"]').nth(1)
     await sessionRow.waitFor({ timeout: 10_000 })
     await sessionRow.hover()
     // Card content: the full title plus the Idle status line (display-only
diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md
index cc21210f3b..2c72632c2f 100644
--- a/docs/cordis-catalog/services.md
+++ b/docs/cordis-catalog/services.md
@@ -2015,6 +2015,16 @@ get(id: WorkspaceId): Workspace | undefined
  */
 list(): Workspace[]
 
+/**
+ * Delete one workspace registration while retaining its directory and every
+ * session log. The durable order is updated before the table deletion; a
+ * failed table write restores the prior order and keeps the entity
+ * published. Unknown ids are an idempotent no-op for domain callers.
+ * @param id - Workspace registration to remove.
+ * @returns `true` when a record was deleted, `false` when it was unknown.
+ */
+delete(id: WorkspaceId): Promise<boolean>
+
 /**
  * Resolve by canonical directory path without creating or mutating a
  * workspace. A missing path rejects during `realpath`; an existing unowned
diff --git a/packages/client/connection/src/client/fixture.ts b/packages/client/connection/src/client/fixture.ts
index eaab0a43f9..fcd04f2460 100644
--- a/packages/client/connection/src/client/fixture.ts
+++ b/packages/client/connection/src/client/fixture.ts
@@ -723,6 +723,20 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy {
         }
         return ok(request, { workspace: { ...workspace } })
       },
+      delete: (request) => {
+        const { workspaceId } = request.payload
+        const index = workspaces.findIndex(workspace => workspace.workspaceId === workspaceId)
+        if (index === -1) {
+          return err(request, {
+            code: 'workspace-not-found',
+            message: `no workspace ${workspaceId}`,
+            details: { workspaceId },
+          })
+        }
+        workspaces.splice(index, 1)
+        emitHost({ type: 'host/workspace-removed', workspaceId })
+        return ok(request, { deleted: true as const })
+      },
       insertSessionBefore: (request) => {
         const { workspaceId, sessionId, beforeSessionId } = request.payload
         const workspace = workspaces.find(w => w.workspaceId === workspaceId)
@@ -914,6 +928,7 @@ export class FixtureApiClient extends AbstractApiClient {
       case 'workspace.list': return this.api.workspace.list(request)
       case 'workspace.create': return this.api.workspace.create(request)
       case 'workspace.rename': return this.api.workspace.rename(request)
+      case 'workspace.delete': return this.api.workspace.delete(request)
       case 'workspace.insertSessionBefore': return this.api.workspace.insertSessionBefore(request)
       case 'command.list': return this.api.commands.list(request)
       // The in-memory execute never blocks, so a never-aborting signal is faithful here.
diff --git a/packages/client/connection/tests/fake-api.ts b/packages/client/connection/tests/fake-api.ts
index bf7295cc50..1c58e604ab 100644
--- a/packages/client/connection/tests/fake-api.ts
+++ b/packages/client/connection/tests/fake-api.ts
@@ -81,6 +81,7 @@ export class FakeApiClient implements IApiClient {
     rename: (payload: unknown) => this.record('workspace.rename', payload, Promise.resolve(ok({
       workspace: { workspaceId: 'fk-ws' as never, path: '/f/ws', title: 'ws', sessionIds: [], createdAt: '0', updatedAt: '0' },
     }))),
+    delete: (payload: unknown) => this.record('workspace.delete', payload, Promise.resolve(ok({ deleted: true as const }))),
     insertSessionBefore: (payload: unknown) => this.record('workspace.insertSessionBefore', payload, Promise.resolve(ok({
       workspace: { workspaceId: 'fk-ws' as never, path: '/f/ws', title: 'ws', sessionIds: [], createdAt: '0', updatedAt: '0' },
     }))),
diff --git a/packages/client/connection/tests/fixture.spec.ts b/packages/client/connection/tests/fixture.spec.ts
index 0374f92b3b..7200291229 100644
--- a/packages/client/connection/tests/fixture.spec.ts
+++ b/packages/client/connection/tests/fixture.spec.ts
@@ -365,6 +365,31 @@ describe('createFixtureApi', () => {
     expect(noop.result.value.workspace.updatedAt).toBe(before)
   })
 
+  it('workspace.delete removes only the Workspace row and emits the removal frame', async () => {
+    const api = createFixtureApi()
+    const abort = new AbortController()
+    const seen: HostFrame[] = []
+    const consuming = (async () => {
+      for await (const envelope of api.events.host(req({}), abort.signal)) {
+        seen.push(envelope.payload)
+        abort.abort()
+      }
+    })()
+    await new Promise(resolve => setTimeout(resolve, 10))
+    const missing = await api.workspace.delete(req({ workspaceId: 'fx-ws-void' as WorkspaceId }))
+    expect(missing.result).toMatchObject({ ok: false, error: { code: 'workspace-not-found' } })
+    const deleted = await api.workspace.delete(req({ workspaceId: 'fx-ws-fixture' as WorkspaceId }))
+    expect(deleted.result).toEqual({ ok: true, value: { deleted: true } })
+    await consuming
+    expect(seen).toEqual([{ type: 'host/workspace-removed', workspaceId: 'fx-ws-fixture' }])
+    const list = await api.workspace.list(req({}))
+    if (!list.result.ok) throw new Error('workspace list failed')
+    expect(list.result.value.items.some(workspace => workspace.workspaceId === 'fx-ws-fixture')).toBe(false)
+    const sessions = await api.sessions.list(req({}))
+    if (!sessions.result.ok) throw new Error('session list failed')
+    expect(sessions.result.value.items.map(session => session.sessionId)).toContain('fx-alpha')
+  })
+
   it('session.create({workspaceId}) lands on the account and unknown ids error', async () => {
     const api = createFixtureApi()
     const abort = new AbortController()
diff --git a/packages/client/runtime/README.i18n.yaml b/packages/client/runtime/README.i18n.yaml
index 5e73979b0b..bb4a2d4d2b 100644
--- a/packages/client/runtime/README.i18n.yaml
+++ b/packages/client/runtime/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 4724ebc75d441252245a0e811a4ae34f8b529a98
-README.zh.md: 6a0076742efccaf946910c77c77a9b74194b9dc5
+#   pnpm run verify-translation-pairing --write packages/client/runtime/README.md
+README.md: d2a10b3d97837ac859c52c206afab06913ea222e
+README.zh.md: f23f8cb184242edbd6d19aeff5823f1efbee8eba
diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md
index 4724ebc75d..d2a10b3d97 100644
--- a/packages/client/runtime/README.md
+++ b/packages/client/runtime/README.md
@@ -6,7 +6,9 @@ Client cordis boot and React-free object services: SlotsService wraps SlotCore a
 
 ## Workspace and Session lists
 
-Workspace and Session lists have independent monotone `pending` → `ready` baseline phases and separate refresh activity/error state. Incremental frames arriving during a list request replay over its response. The first successful baseline establishes Host order; later refreshes update rows and membership without changing the relative order of identities already shown. Workspace recency is derived only after both baselines are ready and never changes Workspace list order.
+Workspace and Session lists have independent monotone `pending` → `ready` baseline phases and separate refresh activity/error state. Incremental upsert/removal frames and unary mutation echoes arriving during a list request replay over its response. The first successful baseline establishes Host order; later refreshes update rows and membership without changing the relative order of identities already shown. Removed Workspace ids retain process-local tombstones so late changed frames cannot resurrect them; reconnect still takes `workspace.list` as the baseline. Workspace recency is derived only after both baselines are ready and never changes Workspace list order.
+
+`WorkspacesService.delete(workspaceId)` removes the registration from the client projection after the successful unary response; the matching `host/workspace-removed` frame is idempotent and synchronizes other tabs. Session state and the current Session selection are independent, so accounted Sessions immediately project under Ungrouped after their Workspace disappears.
 
 SlotsService gives the renderer separate bare observables for `useSessions` and `useWorkspaces`; web-react creates the hooks. Workspace business state does not enter `SessionListState` or an entry store.
 
diff --git a/packages/client/runtime/README.zh.md b/packages/client/runtime/README.zh.md
index 6a0076742e..f23f8cb184 100644
--- a/packages/client/runtime/README.zh.md
+++ b/packages/client/runtime/README.zh.md
@@ -6,7 +6,9 @@
 
 ## Workspace 与 Session 列表
 
-Workspace 和 Session 列表各自具有单调的 `pending` → `ready` 基线阶段，也有各自的刷新活动／错误状态。列表请求期间到达的增量帧会在其响应之上回放。第一次成功的基线建立 Host 顺序；后续刷新更新行和成员关系，但不改变已经显示的标识之间的相对顺序。Workspace 新近程度只在两条基线都 ready 后派生，且绝不改变 Workspace 列表顺序。
+Workspace 和 Session 列表各自具有单调的 `pending` → `ready` 基线阶段，也有各自的刷新活动／错误状态。列表请求期间到达的增量更新／移除帧与一元变更回显会在其响应之上回放。第一次成功的基线建立 Host 顺序；后续刷新更新行和成员关系，但不改变已经显示的标识之间的相对顺序。已移除的 Workspace id 会保留进程本地删除标记，避免延迟到达的 changed 帧将其复活；重连仍以 `workspace.list` 作为基线。Workspace 新近程度只在两条基线都 ready 后派生，且绝不改变 Workspace 列表顺序。
+
+`WorkspacesService.delete(workspaceId)` 在一元响应成功后从客户端投影中移除注册记录；对应的 `host/workspace-removed` 帧具有幂等性，并负责同步其他标签页。Session 状态与当前 Session selection 相互独立，因此 Workspace 消失后，其已记账的 Session 会立即投影到 Ungrouped 下。
 
 SlotsService 分别为 renderer 提供 `useSessions` 与 `useWorkspaces` 的裸 observable；web-react 创建 hook。Workspace 业务状态不会进入 `SessionListState` 或配置项 store。
 
diff --git a/packages/client/runtime/src/client/workspaces/manager.ts b/packages/client/runtime/src/client/workspaces/manager.ts
index e7caecfe82..83275e9a2a 100644
--- a/packages/client/runtime/src/client/workspaces/manager.ts
+++ b/packages/client/runtime/src/client/workspaces/manager.ts
@@ -19,6 +19,10 @@ export interface WorkspaceListSnapshot {
   error: RpcError | null
 }
 
+type WorkspaceDelta =
+  | { type: 'upsert'; workspace: WorkspaceView }
+  | { type: 'remove'; workspaceId: WorkspaceId }
+
 /** Workspace object cluster driven by one list baseline and changed-frame upserts. */
 export class WorkspaceManager {
   private items: Workspace[] = []
@@ -28,7 +32,8 @@ export class WorkspaceManager {
   private phase: WorkspaceListPhase = 'pending'
   private error: RpcError | null = null
   private inflight: Promise<void> | null = null
-  private refreshFrames: WorkspaceView[] | null = null
+  private refreshFrames: WorkspaceDelta[] | null = null
+  private readonly removedIds = new Set<WorkspaceId>()
   private snapshotCache: WorkspaceListSnapshot
   private readonly notifier = new Notifier(() => {
     this.snapshotCache = this.buildSnapshot()
@@ -51,7 +56,7 @@ export class WorkspaceManager {
     this.state = 'loading'
     this.error = null
     const established = this.itemViews()
-    const frames: WorkspaceView[] = []
+    const frames: WorkspaceDelta[] = []
     this.refreshFrames = frames
     this.notifier.markDirty()
     this.inflight = (async () => {
@@ -61,7 +66,8 @@ export class WorkspaceManager {
           let items = this.phase === 'pending'
             ? result.value.items
             : mergeOrderedBaseline(established, result.value.items, workspace => workspace.workspaceId)
-          for (const workspace of frames) items = upsertWorkspace(items, workspace)
+          items = items.filter(workspace => !this.removedIds.has(workspace.workspaceId))
+          for (const delta of frames) items = applyWorkspaceDelta(items, delta)
           this.installViews(items)
           this.state = 'idle'
           this.phase = 'ready'
@@ -111,6 +117,18 @@ export class WorkspaceManager {
     return result
   }
 
+  /**
+   * Delete a Workspace registration and remove its local projection from the
+   * unary response without waiting for the Host frame.
+   * @param workspaceId - target workspace.
+   * @returns the wire result.
+   */
+  async delete(workspaceId: WorkspaceId): Promise<RpcResult<{ deleted: true }>> {
+    const { result } = await this.api.workspace.delete({ workspaceId })
+    if (result.ok) this.remove(workspaceId)
+    return result
+  }
+
   /**
    * Move a session within its Workspace's manual order, then publish the
    * returned snapshot without waiting for the changed frame.
@@ -139,6 +157,7 @@ export class WorkspaceManager {
    */
   handleHostEnvelope(envelope: RpcRequest<HostFrame>): void {
     if (envelope.payload.type === 'host/workspace-changed') this.upsert(envelope.payload.workspace)
+    else if (envelope.payload.type === 'host/workspace-removed') this.remove(envelope.payload.workspaceId)
   }
 
   /** Re-pull the baseline after each connection generation. */
@@ -175,7 +194,8 @@ export class WorkspaceManager {
 
   /** Upsert one Host view, optionally retaining the local object that materialized it. */
   private upsert(view: WorkspaceView, identity?: Workspace): void {
-    this.refreshFrames?.push(view)
+    if (this.removedIds.has(view.workspaceId)) return
+    this.refreshFrames?.push({ type: 'upsert', workspace: view })
     const index = this.items.findIndex(item => item.getSnapshot().view?.workspaceId === view.workspaceId)
     // Mutation responses and changed frames race (two carriers, no ordering):
     // reject a snapshot strictly older than the installed projection so a
@@ -195,6 +215,17 @@ export class WorkspaceManager {
     this.notifier.markDirty()
   }
 
+  /** Remove one id idempotently and retain a tombstone against late echoes. */
+  private remove(workspaceId: WorkspaceId): void {
+    this.refreshFrames?.push({ type: 'remove', workspaceId })
+    this.removedIds.add(workspaceId)
+    const items = this.items.filter(item =>
+      item.getSnapshot().view?.workspaceId !== workspaceId)
+    if (items.length === this.items.length) return
+    this.items = items
+    this.notifier.markDirty()
+  }
+
   private installViews(views: readonly WorkspaceView[]): void {
     const existing = new Map(
       this.items.flatMap((workspace) => {
@@ -234,3 +265,10 @@ function upsertWorkspace(items: readonly WorkspaceView[], workspace: WorkspaceVi
     ? [workspace, ...items]
     : items.map((item, position) => position === index ? workspace : item)
 }
+
+
+function applyWorkspaceDelta(items: readonly WorkspaceView[], delta: WorkspaceDelta): WorkspaceView[] {
+  return delta.type === 'upsert'
+    ? upsertWorkspace(items, delta.workspace)
+    : items.filter(workspace => workspace.workspaceId !== delta.workspaceId)
+}
diff --git a/packages/client/runtime/src/client/workspaces/service.ts b/packages/client/runtime/src/client/workspaces/service.ts
index 1e281ca792..4cb26aa220 100644
--- a/packages/client/runtime/src/client/workspaces/service.ts
+++ b/packages/client/runtime/src/client/workspaces/service.ts
@@ -174,6 +174,16 @@ export class WorkspacesService {
     return result.value.workspace
   }
 
+  /**
+   * Delete one Workspace registration. Sessions, session logs, and the
+   * directory remain Host-owned outside this operation.
+   * @param workspaceId - target workspace.
+   */
+  async delete(workspaceId: WorkspaceId): Promise<void> {
+    const result = await this.manager.delete(workspaceId)
+    if (!result.ok) throw new Error(`workspace delete failed: ${result.error.code}: ${result.error.message}`)
+  }
+
   /**
    * Move a session within its Workspace's manual order (DOM-insertBefore-like).
    * @param workspaceId - owning workspace.
diff --git a/packages/client/runtime/tests/fake-api.ts b/packages/client/runtime/tests/fake-api.ts
index dcb334f6ea..9f6147cdd3 100644
--- a/packages/client/runtime/tests/fake-api.ts
+++ b/packages/client/runtime/tests/fake-api.ts
@@ -96,6 +96,9 @@ export class FakeApiClient implements IApiClient {
   onWorkspaceRename: (payload: unknown) => Promise<RpcResponse<{ workspace: WorkspaceView }>> =
     () => Promise.resolve(ok({ workspace: fakeWorkspace('fk-ws') }))
 
+  onWorkspaceDelete: (payload: unknown) => Promise<RpcResponse<{ deleted: true }>> =
+    () => Promise.resolve(ok({ deleted: true }))
+
   onWorkspaceInsertSessionBefore: (payload: unknown) => Promise<RpcResponse<{ workspace: WorkspaceView }>> =
     () => Promise.resolve(ok({ workspace: fakeWorkspace('fk-ws') }))
 
@@ -103,6 +106,7 @@ export class FakeApiClient implements IApiClient {
     list: (payload: unknown) => this.record('workspace.list', payload, this.onWorkspaceList(payload)),
     create: (payload: unknown) => this.record('workspace.create', payload, this.onWorkspaceCreate(payload)),
     rename: (payload: unknown) => this.record('workspace.rename', payload, this.onWorkspaceRename(payload)),
+    delete: (payload: unknown) => this.record('workspace.delete', payload, this.onWorkspaceDelete(payload)),
     insertSessionBefore: (payload: unknown) =>
       this.record('workspace.insertSessionBefore', payload, this.onWorkspaceInsertSessionBefore(payload)),
   }
diff --git a/packages/client/runtime/tests/workspaces-service.spec.ts b/packages/client/runtime/tests/workspaces-service.spec.ts
index d020b74fec..210c04d896 100644
--- a/packages/client/runtime/tests/workspaces-service.spec.ts
+++ b/packages/client/runtime/tests/workspaces-service.spec.ts
@@ -76,6 +76,48 @@ describe('WorkspaceManager', () => {
       ok: false, error: { code: 'internal', message: 'create transport' },
     })
   })
+
+  it('replays removal over an in-flight baseline and ignores duplicate or late updates', async () => {
+    const api = new FakeApiClient()
+    const gate = deferred<Awaited<ReturnType<FakeApiClient['onWorkspaceList']>>>()
+    api.onWorkspaceList = () => gate.promise
+    const manager = new WorkspaceManager(api)
+    const hydration = manager.refresh()
+    manager.handleHostEnvelope({
+      rpcId: 'removed' as never,
+      payload: { type: 'host/workspace-removed', workspaceId: wid('gone') },
+    })
+    gate.resolve(ok({ items: [workspace('gone'), workspace('kept')] as never[] }))
+    await hydration
+    expect(manager.getSnapshot().items.map(item => item.workspaceId)).toEqual(['kept'])
+
+    manager.handleHostEnvelope({
+      rpcId: 'late-change' as never,
+      payload: { type: 'host/workspace-changed', workspace: workspace('gone') },
+    })
+    manager.handleHostEnvelope({
+      rpcId: 'duplicate-remove' as never,
+      payload: { type: 'host/workspace-removed', workspaceId: wid('gone') },
+    })
+    expect(manager.getSnapshot().items.map(item => item.workspaceId)).toEqual(['kept'])
+  })
+
+  it('removes from the unary delete echo while a refresh is in flight', async () => {
+    const api = new FakeApiClient()
+    api.onWorkspaceList = () => Promise.resolve(ok({ items: [workspace('gone')] as never[] }))
+    const manager = new WorkspaceManager(api)
+    await manager.refresh()
+    const gate = deferred<Awaited<ReturnType<FakeApiClient['onWorkspaceList']>>>()
+    api.onWorkspaceList = () => gate.promise
+    const refresh = manager.refresh()
+
+    await expect(manager.delete(wid('gone'))).resolves.toMatchObject({ ok: true })
+    expect(api.callsOf('workspace.delete')).toEqual([{ workspaceId: 'gone' }])
+    expect(manager.getSnapshot().items).toEqual([])
+    gate.resolve(ok({ items: [workspace('gone')] as never[] }))
+    await refresh
+    expect(manager.getSnapshot().items).toEqual([])
+  })
 })
 
 describe('WorkspacesService', () => {
@@ -175,4 +217,20 @@ describe('WorkspacesService', () => {
     }))
     await expect(workspaces.create({ path: '/missing' })).rejects.toThrow(/workspace-invalid-path: missing/)
   })
+
+  it('deletes a Workspace or preserves it when the Host rejects deletion', async () => {
+    const ctx = new Context()
+    const api = new FakeApiClient()
+    const sessions = new SessionsService(ctx, api)
+    const workspaces = new WorkspacesService(ctx, api, sessions)
+    api.onWorkspaceList = () => Promise.resolve(ok({ items: [workspace('alpha')] as never[] }))
+    await workspaces.refresh()
+    await expect(workspaces.delete(wid('alpha'))).resolves.toBeUndefined()
+    expect(workspaces.list.getSnapshot().items).toEqual([])
+
+    api.onWorkspaceDelete = () => Promise.resolve(err({
+      code: 'workspace-not-found', message: 'gone', details: { workspaceId: 'ghost' },
+    }))
+    await expect(workspaces.delete(wid('ghost'))).rejects.toThrow(/workspace-not-found: gone/)
+  })
 })
diff --git a/packages/client/ui-workspace/README.i18n.yaml b/packages/client/ui-workspace/README.i18n.yaml
index d0f2d0a20a..88638b9e35 100644
--- a/packages/client/ui-workspace/README.i18n.yaml
+++ b/packages/client/ui-workspace/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: e0247b3e26f617f86e9c0094afa1cbc920f02d33
-README.zh.md: 92ef463faab4b1ccda85d7f3cec1678a338d4010
+#   pnpm run verify-translation-pairing --write packages/client/ui-workspace/README.md
+README.md: b5a78c30ddae5e12612bb8cced65b5fe95f7e259
+README.zh.md: 904543a48f1609e23ba80cf240be965d0654a951
diff --git a/packages/client/ui-workspace/README.md b/packages/client/ui-workspace/README.md
index e0247b3e26..b5a78c30dd 100644
--- a/packages/client/ui-workspace/README.md
+++ b/packages/client/ui-workspace/README.md
@@ -4,7 +4,7 @@ English | [中文](README.zh.md)
 
 Shared Workspace picker plugin. `WorkspacePicker` is registered into the sidebar's `sidebar.workspace` slot and the page-local Session Intent hero's `conversation.empty.workspace` slot, so both surfaces use the same menu and creation modals.
 
-The picker lists real Host Workspace entities through the global `useWorkspaces` hook. Selecting a Workspace invokes the slot owner's `onPick` callback to retarget the frontend Session object; the existing-folder and create-new actions first create a real Workspace through the object layer, then select it. Create-new disables names already present in that list, while the Host remains authoritative for concurrent or non-UI callers. The runtime Session and Workspace services own materialization.
+The picker lists real Host Workspace entities through the global `useWorkspaces` hook. Selecting a Workspace invokes the slot owner's `onPick` callback to retarget the frontend Session object; the existing-folder and create-new actions first create a real Workspace through the object layer, then select it. Create-new disables names already present in that list, while the Host remains authoritative for concurrent or non-UI callers. The runtime Session and Workspace services own materialization. The Workspace row's Delete action opens a confirmation that states the retention boundary, blocks duplicate submission, and keeps failures open; success removes the group while its Sessions remain under Ungrouped.
 
 Both target slots are declared by other plugins, so `apply` registers through declaration-aware deferral and re-registers after a declaring slot is restored.
 
@@ -18,5 +18,5 @@ None; this package neither assembles nor sends a provider request.
 
 ## Known Limitations and Deferred Work
 
-- **No Workspace rename/delete controls** — the picker supports selection and creation only.
+- **No Session deletion control** — the existing Session menu row remains visual-only; Workspace registration deletion does not delete Sessions.
 - **Existing-folder entry is manual path input only** — Host creation failures are shown in the modal.
diff --git a/packages/client/ui-workspace/README.zh.md b/packages/client/ui-workspace/README.zh.md
index 92ef463faa..904543a48f 100644
--- a/packages/client/ui-workspace/README.zh.md
+++ b/packages/client/ui-workspace/README.zh.md
@@ -4,7 +4,7 @@
 
 共享 Workspace 选择器插件。`WorkspacePicker` 注册到侧边栏的 `sidebar.workspace` slot，以及页面局部 Session Intent 主视觉区的 `conversation.empty.workspace` slot，因此两个表层使用同一菜单和创建模态框。
 
-该选择器通过全局 `useWorkspaces` hook 列出真实的 Host Workspace 实体。选择 Workspace 会调用 slot owner 的 `onPick` 回调，重新定位前端 Session 对象；使用现有文件夹和新建操作时，系统会先通过对象层创建真实 Workspace，再将其选中。新建操作会禁用列表中已有的名称，而 Host 对并发或非 UI 调用方仍具有最终决定权。运行时 Session 与 Workspace 服务负责物化。
+该选择器通过全局 `useWorkspaces` hook 列出真实的 Host Workspace 实体。选择 Workspace 会调用 slot owner 的 `onPick` 回调，重新定位前端 Session 对象；使用现有文件夹和新建操作时，系统会先通过对象层创建真实 Workspace，再将其选中。新建操作会禁用列表中已有的名称，而 Host 对并发或非 UI 调用方仍具有最终决定权。运行时 Session 与 Workspace 服务负责物化。Workspace 行内的 Delete 操作会打开确认框，说明保留边界、阻止重复提交，并在失败时保持打开；成功后，该分组会被移除，其 Session 则留在 Ungrouped 下。
 
 两个目标 slot 都由其他插件声明，因此 `apply` 通过声明感知的延迟机制完成注册，并在声明该 slot 的插件恢复后重新注册。
 
@@ -18,5 +18,5 @@
 
 ## 已知限制与暂缓事项
 
-- **没有 Workspace 重命名／删除控件**：选择器仅支持选择和创建。
+- **没有 Session 删除控件**：现有 Session 菜单行仍仅提供视觉效果；删除 Workspace 注册记录不会删除 Session。
 - **现有文件夹入口仅支持手动输入路径**：Host 创建失败会显示在模态框中。
diff --git a/packages/client/ui-workspace/src/client/WorkspaceBrowser.module.css b/packages/client/ui-workspace/src/client/WorkspaceBrowser.module.css
index d6375cb698..c03d511c92 100644
--- a/packages/client/ui-workspace/src/client/WorkspaceBrowser.module.css
+++ b/packages/client/ui-workspace/src/client/WorkspaceBrowser.module.css
@@ -258,6 +258,16 @@
   color: var(--dsw-alias-state-error-primary);
 }
 
+.deleteAction:not(:disabled) {
+  color: var(--dsw-alias-state-error-primary);
+}
+
+.deleteStatus {
+  font-size: 12px;
+  line-height: 18px;
+  color: var(--dsw-alias-label-secondary);
+}
+
 @media (prefers-reduced-motion: reduce) {
   .wide {
     animation: none;
diff --git a/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx b/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
index 0dc6485929..0090164928 100644
--- a/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
+++ b/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
@@ -87,10 +87,15 @@ type SessionTreeProps = Pick<
   query: string
   /** Open the browser-owned rename dialog for a real Workspace group. */
   onRenameRequest: (workspaceId: WorkspaceId, currentTitle: string) => void
+  /** Open the browser-owned delete-confirmation dialog for a real Workspace group. */
+  onDeleteRequest: (workspaceId: WorkspaceId, currentTitle: string) => void
 }
 
 /** The scrolling session tree; unmounting at collapse settle drops the sessions subscription and expansion state. */
-function SessionTree({ useSessions, startSession, open, workspaces, query, onRenameRequest, insertSessionBefore }: SessionTreeProps) {
+function SessionTree({
+  useSessions, startSession, open, workspaces, query,
+  onRenameRequest, onDeleteRequest, insertSessionBefore,
+}: SessionTreeProps) {
   const list = useSessions((s) => s)
   const current = list.current
   const [expandedProjects, setExpandedProjects] = useState<string[]>([])
@@ -128,11 +133,17 @@ function SessionTree({ useSessions, startSession, open, workspaces, query, onRen
               onCreate={() => {
                 if (group.workspaceId !== undefined) startSession(group.workspaceId)
               }}
-              onRename={group.workspaceId === undefined
+              actions={group.workspaceId === undefined
                 ? undefined
-                : () => {
-                    /* v8 ignore next -- narrowing guard: the closure is only created for real-workspace groups. */
-                    if (group.workspaceId !== undefined) onRenameRequest(group.workspaceId, group.label)
+                : {
+                    rename: () => {
+                      /* v8 ignore next -- narrowing guard: the actions object exists only for real-workspace groups. */
+                      if (group.workspaceId !== undefined) onRenameRequest(group.workspaceId, group.label)
+                    },
+                    delete: () => {
+                      /* v8 ignore next -- narrowing guard: the actions object exists only for real-workspace groups. */
+                      if (group.workspaceId !== undefined) onDeleteRequest(group.workspaceId, group.label)
+                    },
                   }}
             />
             {group.sessions.map((node, index) => {
@@ -236,6 +247,7 @@ export function WorkspaceBrowser({
   startSession,
   open,
   renameWorkspace,
+  deleteWorkspace,
   insertSessionBefore,
   createWorkspace,
 }: WorkspaceBrowserProps) {
@@ -291,6 +303,30 @@ export function WorkspaceBrowser({
     })
   }
 
+  // Delete dialog is separate from the row so a successful removal can
+  // unmount that row without tearing down the in-flight confirmation state.
+  const [deleteTarget, setDeleteTarget] = useState<{ workspaceId: WorkspaceId; title: string } | null>(null)
+  const [deleting, setDeleting] = useState(false)
+  const [deleteError, setDeleteError] = useState<string | null>(null)
+  const closeDelete = () => {
+    if (deleting) return
+    setDeleteTarget(null)
+    setDeleteError(null)
+  }
+  const confirmDelete = () => {
+    /* v8 ignore next -- the Modal is absent without a target and its button is disabled while deleting. */
+    if (deleting || deleteTarget === null) return
+    setDeleting(true)
+    setDeleteError(null)
+    deleteWorkspace(deleteTarget.workspaceId).then(() => {
+      setDeleting(false)
+      setDeleteTarget(null)
+    }).catch((reason: unknown) => {
+      setDeleting(false)
+      setDeleteError(reason instanceof Error ? reason.message : String(reason))
+    })
+  }
+
   return (
     <div className={clsx(css.root, !wide && css.rail)}>
       <div className={css.sectionHeader}>
@@ -382,6 +418,10 @@ export function WorkspaceBrowser({
                   setRenameDraft(currentTitle)
                   setRenameError(null)
                 }}
+                onDeleteRequest={(workspaceId, title) => {
+                  setDeleteTarget({ workspaceId, title })
+                  setDeleteError(null)
+                }}
               />
             ))}
       </div>
@@ -416,6 +456,30 @@ export function WorkspaceBrowser({
         )}
         {renameError !== null && <div className={css.renameError} role="alert">{renameError}</div>}
       </Modal>
+      <Modal
+        open={deleteTarget !== null}
+        onClose={closeDelete}
+        title="Delete workspace"
+        {...deleteTarget === null
+          ? {}
+          : { description: `This removes “${deleteTarget.title}” from the workspace list. The folder and session logs will be kept. Its sessions will appear under Ungrouped.` }}
+        footer={(
+          <>
+            <Button variant="outline" disabled={deleting} onClick={closeDelete}>Cancel</Button>
+            <Button
+              variant="outline"
+              className={css.deleteAction!}
+              disabled={deleting}
+              onClick={confirmDelete}
+            >
+              Delete workspace
+            </Button>
+          </>
+        )}
+      >
+        {deleting && <div className={css.deleteStatus} role="status">Deleting workspace…</div>}
+        {deleteError !== null && <div className={css.renameError} role="alert">{deleteError}</div>}
+      </Modal>
     </div>
   )
 }
diff --git a/packages/client/ui-workspace/src/client/contract/slots.ts b/packages/client/ui-workspace/src/client/contract/slots.ts
index 6008da553f..2e4c88dd6b 100644
--- a/packages/client/ui-workspace/src/client/contract/slots.ts
+++ b/packages/client/ui-workspace/src/client/contract/slots.ts
@@ -32,6 +32,8 @@ export type WorkspaceBrowserInjected = {
   open: (sessionId: SessionId) => void
   /** Rename a Host Workspace (rejects on name conflict; resolves on durability). */
   renameWorkspace: (workspaceId: WorkspaceId, title: string) => Promise<void>
+  /** Delete only a Host Workspace registration; directory and Session logs remain. */
+  deleteWorkspace: (workspaceId: WorkspaceId) => Promise<void>
   /**
    * Reorder a session inside its Workspace account (DOM-insertBefore
    * semantics: omitted anchor appends to the end). The view refreshes from
diff --git a/packages/client/ui-workspace/src/client/index.ts b/packages/client/ui-workspace/src/client/index.ts
index a444464441..98adfecce3 100644
--- a/packages/client/ui-workspace/src/client/index.ts
+++ b/packages/client/ui-workspace/src/client/index.ts
@@ -39,6 +39,7 @@ export function apply(ctx: ClientContext): void {
     startSession: (workspaceId) => { ctx.workspaces.startSession(workspaceId) },
     open: (sessionId) => { ctx.sessions.open(sessionId) },
     renameWorkspace: async (workspaceId, title) => { await ctx.workspaces.rename(workspaceId, title) },
+    deleteWorkspace: async (workspaceId) => { await ctx.workspaces.delete(workspaceId) },
     insertSessionBefore: async (workspaceId, sessionId, beforeSessionId) => {
       await ctx.workspaces.insertSessionBefore(workspaceId, sessionId, beforeSessionId)
     },
diff --git a/packages/client/ui-workspace/src/client/rows/Rows.tsx b/packages/client/ui-workspace/src/client/rows/Rows.tsx
index a100da83e9..ae866f58cf 100644
--- a/packages/client/ui-workspace/src/client/rows/Rows.tsx
+++ b/packages/client/ui-workspace/src/client/rows/Rows.tsx
@@ -3,7 +3,7 @@
  * all data and callbacks arrive via props. Hover swaps (folder->chevron,
  * time->ellipsis, action buttons) are CSS-only. Row ... menus are visual-only
  * except workspace Rename; the session hover card is suppressed while a menu
- * is open.
+ * is open. Workspace Rename/Delete are wired; session actions remain visual-only.
  */
 import { useState } from 'react'
 import clsx from 'clsx'
@@ -39,12 +39,12 @@ const WORKSPACE_MENU_ITEMS = [
  * @param props.onCreate - start a frontend Session inside this Workspace.
  * @returns the row element.
  */
-export function ProjectRowItem({ group, onToggle, onCreate, onRename }: {
+export function ProjectRowItem({ group, onToggle, onCreate, actions }: {
   group: GroupNode
   onToggle: () => void
   onCreate: () => void
-  /** Open the rename dialog; absent for the ungrouped bucket (no menu shown). */
-  onRename?: (() => void) | undefined
+  /** Real-Workspace actions; absent for the ungrouped bucket (no menu shown). */
+  actions?: { rename: () => void; delete: () => void } | undefined
 }) {
   const row = group
   const active = group.expanded && group.containsCurrent
@@ -68,15 +68,15 @@ export function ProjectRowItem({ group, onToggle, onCreate, onRename }: {
         <span className={css.meta}>{count}</span>
       </span>
       <span className={css.rowActions}>
-        {onRename !== undefined && (
+        {actions !== undefined && (
           <Menu
             open={menuOpen}
             onClose={() => { setMenuOpen(false) }}
             items={WORKSPACE_MENU_ITEMS}
             onSelect={(id) => {
               setMenuOpen(false)
-              if (id === 'rename') onRename()
-              // Delete is visual-only for now.
+              if (id === 'rename') actions.rename()
+              else actions.delete()
             }}
             portal
             closeOnPointerLeave
diff --git a/packages/client/ui-workspace/tests/rows.spec.tsx b/packages/client/ui-workspace/tests/rows.spec.tsx
index 70cfb36940..6ccd923793 100644
--- a/packages/client/ui-workspace/tests/rows.spec.tsx
+++ b/packages/client/ui-workspace/tests/rows.spec.tsx
@@ -98,12 +98,16 @@ describe('workspace browser rows', () => {
 
   it('workspace row menu opens on the ellipsis, renames, and shows the danger delete row', () => {
     const onRename = vi.fn()
+    const onDelete = vi.fn()
     const onToggle = vi.fn()
     const group: GroupNode = {
       key: 'project', workspaceId: wid('project'), cwd: '/projects/project', label: 'Project',
       sessionCount: 0, expanded: false, containsCurrent: false, sessions: [],
     }
-    render(<ProjectRowItem group={group} onToggle={onToggle} onCreate={vi.fn()} onRename={onRename} />)
+    render(<ProjectRowItem
+      group={group} onToggle={onToggle} onCreate={vi.fn()}
+      actions={{ rename: onRename, delete: onDelete }}
+    />)
     fireEvent.click(screen.getByRole('button', { name: 'Workspace actions for Project' }))
     // Opening the menu neither toggles the group nor renames yet.
     expect(onToggle).not.toHaveBeenCalled()
@@ -111,11 +115,11 @@ describe('workspace browser rows', () => {
     fireEvent.click(screen.getByRole('menuitem', { name: 'Rename' }))
     expect(onRename).toHaveBeenCalledOnce()
     expect(screen.queryByRole('menu')).toBeNull()
-    // Delete stays visual-only: selecting it just closes the menu.
     fireEvent.click(screen.getByRole('button', { name: 'Workspace actions for Project' }))
     fireEvent.click(screen.getByRole('menuitem', { name: 'Delete workspace' }))
     expect(screen.queryByRole('menu')).toBeNull()
     expect(onRename).toHaveBeenCalledOnce()
+    expect(onDelete).toHaveBeenCalledOnce()
     // Escape closes without selecting (Menu onClose path).
     fireEvent.click(screen.getByRole('button', { name: 'Workspace actions for Project' }))
     fireEvent.keyDown(document, { key: 'Escape' })
diff --git a/packages/client/ui-workspace/tests/workspace-browser.spec.tsx b/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
index e9b55e7b76..1dbe895b74 100644
--- a/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
+++ b/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
@@ -54,6 +54,7 @@ function mount(overrides: Partial<WorkspaceBrowserProps> = {}) {
     startSession: vi.fn(),
     open: vi.fn(),
     renameWorkspace: vi.fn(async () => {}),
+    deleteWorkspace: vi.fn(async () => {}),
     insertSessionBefore: vi.fn(async () => {}),
     createWorkspace: vi.fn(async () => workspace('created', [])),
     ...overrides,
@@ -457,6 +458,74 @@ describe('WorkspaceBrowser', () => {
     await waitFor(() => { expect(screen.getByRole('alert').textContent).toBe('denied') })
   })
 
+  it('confirms Workspace deletion, explains retention, and blocks duplicate submission', async () => {
+    let resolveDelete!: () => void
+    const deleteWorkspace = vi.fn(() => new Promise<void>((resolve) => { resolveDelete = resolve }))
+    mount({
+      useWorkspaces: hook(workspaceState([workspace('alpha', ['session'], 'Alpha')])),
+      deleteWorkspace,
+    })
+    fireEvent.click(screen.getByRole('button', { name: 'Workspace actions for Alpha' }))
+    fireEvent.click(screen.getByRole('menuitem', { name: 'Delete workspace' }))
+    const dialog = screen.getByRole('dialog', { name: 'Delete workspace' })
+    expect(dialog.textContent).toContain('removes “Alpha” from the workspace list')
+    expect(dialog.textContent).toContain('folder and session logs will be kept')
+    expect(dialog.textContent).toContain('sessions will appear under Ungrouped')
+
+    const confirm = screen.getByRole('button', { name: 'Delete workspace' }) as HTMLButtonElement
+    fireEvent.click(confirm)
+    fireEvent.click(confirm)
+    expect(deleteWorkspace).toHaveBeenCalledOnce()
+    expect(deleteWorkspace).toHaveBeenCalledWith(wid('alpha'))
+    expect(confirm.disabled).toBe(true)
+    expect((screen.getByRole('button', { name: 'Cancel' }) as HTMLButtonElement).disabled).toBe(true)
+    expect(screen.getByRole('status').textContent).toBe('Deleting workspace…')
+    fireEvent.keyDown(document, { key: 'Escape' })
+    fireEvent.click(screen.getByRole('button', { name: 'Close' }))
+    expect(screen.getByRole('dialog', { name: 'Delete workspace' })).toBeTruthy()
+    await act(async () => { resolveDelete() })
+    expect(screen.queryByRole('dialog', { name: 'Delete workspace' })).toBeNull()
+  })
+
+  it('keeps the delete dialog open on failure and allows retry or cancellation', async () => {
+    const deleteWorkspace = vi.fn()
+      .mockRejectedValueOnce(new Error('storage unavailable'))
+      .mockRejectedValueOnce('denied')
+    mount({
+      useWorkspaces: hook(workspaceState([workspace('alpha', [], 'Alpha')])),
+      deleteWorkspace,
+    })
+    fireEvent.click(screen.getByRole('button', { name: 'Workspace actions for Alpha' }))
+    fireEvent.click(screen.getByRole('menuitem', { name: 'Delete workspace' }))
+    fireEvent.click(screen.getByRole('button', { name: 'Delete workspace' }))
+    await waitFor(() => { expect(screen.getByRole('alert').textContent).toBe('storage unavailable') })
+    expect(screen.getByRole('dialog', { name: 'Delete workspace' })).toBeTruthy()
+    fireEvent.click(screen.getByRole('button', { name: 'Delete workspace' }))
+    await waitFor(() => { expect(screen.getByRole('alert').textContent).toBe('denied') })
+    fireEvent.click(screen.getByRole('button', { name: 'Cancel' }))
+    expect(screen.queryByRole('dialog', { name: 'Delete workspace' })).toBeNull()
+  })
+
+  it('Cancel, Escape, and Close dismiss deletion without calling the action', () => {
+    const deleteWorkspace = vi.fn(async () => {})
+    mount({
+      useWorkspaces: hook(workspaceState([workspace('alpha', [], 'Alpha')])),
+      deleteWorkspace,
+    })
+    const open = () => {
+      fireEvent.click(screen.getByRole('button', { name: 'Workspace actions for Alpha' }))
+      fireEvent.click(screen.getByRole('menuitem', { name: 'Delete workspace' }))
+    }
+    open()
+    fireEvent.click(screen.getByRole('button', { name: 'Cancel' }))
+    open()
+    fireEvent.keyDown(document, { key: 'Escape' })
+    open()
+    fireEvent.click(screen.getByRole('button', { name: 'Close' }))
+    expect(deleteWorkspace).not.toHaveBeenCalled()
+    expect(screen.queryByRole('dialog', { name: 'Delete workspace' })).toBeNull()
+  })
+
   it('search hides drag affordances (rows are not draggable during search)', () => {
     const sessions = sessionState([summary('needle-a', 2, { displayTitle: 'Needle A' })])
     mount({
diff --git a/packages/cordis/tool-cordis/src/api-catalog.ts b/packages/cordis/tool-cordis/src/api-catalog.ts
index 224e8300ab..f33722a1fc 100644
--- a/packages/cordis/tool-cordis/src/api-catalog.ts
+++ b/packages/cordis/tool-cordis/src/api-catalog.ts
@@ -942,6 +942,10 @@ export const SERVICE_API: readonly ServiceApiEntry[] = [
         signature: 'list(): Workspace[]',
         jsDoc: '/**\n * Synchronous workspace projection in durable registry order. Every\n * entity\'s `sessionIds` getter is already filtered by the startup/live\n * canonical-cwd header index; this method performs no persistence reads.\n * @returns a fresh ordered array of workspace entities.\n */',
       },
+      {
+        signature: 'delete(id: WorkspaceId): Promise<boolean>',
+        jsDoc: '/**\n * Delete one workspace registration while retaining its directory and every\n * session log. The durable order is updated before the table deletion; a\n * failed table write restores the prior order and keeps the entity\n * published. Unknown ids are an idempotent no-op for domain callers.\n * @param id - Workspace registration to remove.\n * @returns `true` when a record was deleted, `false` when it was unknown.\n */',
+      },
       {
         signature: 'async resolveByPath(path: string): Promise<Workspace | undefined>',
         jsDoc: '/**\n * Resolve by canonical directory path without creating or mutating a\n * workspace. A missing path rejects during `realpath`; an existing unowned\n * directory returns `undefined`.\n * @param path - Existing directory path in any spelling.\n * @returns the workspace owning the canonical path, when one exists.\n */',
diff --git a/packages/host/apiproxy/README.i18n.yaml b/packages/host/apiproxy/README.i18n.yaml
index eb06e14d2d..653906ba08 100644
--- a/packages/host/apiproxy/README.i18n.yaml
+++ b/packages/host/apiproxy/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 43ad70fa8b865b0b80496bbb67013f24e9e3a33f
-README.zh.md: cc95a7512fb872add816bf0456a93dfcf7b84c10
+#   pnpm run verify-translation-pairing --write packages/host/apiproxy/README.md
+README.md: dc29abdc10f536463358db92a7ac25c1579f2a50
+README.zh.md: a69d51de086dfbc692dccf3f5f88ce7e36c74e9c
diff --git a/packages/host/apiproxy/README.md b/packages/host/apiproxy/README.md
index 43ad70fa8b..dc29abdc10 100644
--- a/packages/host/apiproxy/README.md
+++ b/packages/host/apiproxy/README.md
@@ -12,7 +12,7 @@ The layering/protocol decisions are recorded in the [GUI layering and RPC protoc
 
 The mux stream projects the latest log-backed title as a validated `session/title` control frame after each attached-session subscription baseline and immediately after the corresponding live raw title event. This projection does not add titles to `session.list`; cold sessions remain metadata-only there until opening or resuming attaches their logs.
 
-Workspace and Session lists are separate reconnect baselines. `workspace.create` creates a unique name or adopts an existing directory, `session.create` accepts an optional preallocated Session id, and `host/workspace-changed` plus `host/session-added` carry committed increments in either arrival order. `SessionSummary.blank` and the `host/session-added` frame carry the derived zero-events bit: clients hide blank sessions and reuse them per workspace, flip blank on the first `host/session-status(running:true)`, and treat `session.list` as the reconnect authority; cold summaries are never blank because lazy persistence keeps never-appended sessions out of `list()`.
+Workspace and Session lists are separate reconnect baselines. `workspace.create` creates a unique name or adopts an existing directory, `workspace.delete` removes only the Workspace registration, `session.create` accepts an optional preallocated Session id, and `host/workspace-changed`, `host/workspace-removed`, plus `host/session-added` carry committed increments in either arrival order. Registration deletion preserves the directory and session logs; its Sessions remain in `session.list` and become Ungrouped. `SessionSummary.blank` and the `host/session-added` frame carry the derived zero-events bit: clients hide blank sessions and reuse them per workspace, flip blank on the first `host/session-status(running:true)`, and treat `session.list` as the reconnect authority; cold summaries are never blank because lazy persistence keeps never-appended sessions out of `list()`.
 
 The `command.*` and `skill.*` domains expose the host command registry and skill catalog to clients. Every method addresses one session's agent by `sessionId` (a served session always has an Agent; `command.*` resumes cold sessions through the same path as `session.*`, while `skill.list` resolves the project root from the session header without touching the Agent registry). `command.execute` runs a slash-command line host-side and returns a detached result; the carrier's request signal cancels the running handler. `host/commands-changed` is the catalog invalidation frame: clients refetch `command.list` instead of diffing.
 
diff --git a/packages/host/apiproxy/README.zh.md b/packages/host/apiproxy/README.zh.md
index cc95a7512f..a69d51de08 100644
--- a/packages/host/apiproxy/README.zh.md
+++ b/packages/host/apiproxy/README.zh.md
@@ -12,7 +12,7 @@
 
 mux 流会在每个已附加会话的订阅基线之后，以及对应的实时原始标题事件之后，立即把基于日志的最新标题投影为经过校验的 `session/title` 控制帧。该投影不会把标题加入 `session.list`；冷会话在其中仍只有元数据，直到打开或恢复操作附加其日志。
 
-Workspace 列表与 Session 列表是相互独立的重连基线。`workspace.create` 会创建唯一名称或接纳现有目录，`session.create` 接受可选的预分配 Session id，`host/workspace-changed` 与 `host/session-added` 则以任意到达顺序携带已提交的增量。`SessionSummary.blank` 与 `host/session-added` 帧携带派生的零事件位：客户端隐藏空白会话并按 workspace 复用它们，在首个 `host/session-status(running:true)` 时翻转 blank，并以 `session.list` 作为重连权威；冷会话摘要永远不是空白——惰性持久化让从未追加过事件的会话根本不出现在 `list()` 中。
+Workspace 列表与 Session 列表是相互独立的重连基线。`workspace.create` 会创建唯一名称或接纳现有目录，`workspace.delete` 只移除 Workspace 注册记录，`session.create` 接受可选的预分配 Session id，`host/workspace-changed`、`host/workspace-removed` 与 `host/session-added` 则以任意到达顺序携带已提交的增量。删除注册记录会保留目录和会话日志；相关 Session 仍留在 `session.list` 中，并进入 Ungrouped。`SessionSummary.blank` 与 `host/session-added` 帧携带派生的零事件位：客户端隐藏空白会话并按 workspace 复用它们，在首个 `host/session-status(running:true)` 时翻转 blank，并以 `session.list` 作为重连权威；冷会话摘要永远不是空白：惰性持久化让从未追加过事件的会话根本不出现在 `list()` 中。
 
 `command.*` 与 `skill.*` 领域向客户端暴露宿主命令注册表和技能目录。每个方法都通过 `sessionId` 寻址一个会话的 Agent（被服务的会话必有 Agent；`command.*` 经由与 `session.*` 相同的路径恢复冷会话，而 `skill.list` 从会话头解析项目根目录，不触碰 Agent 注册表）。`command.execute` 在宿主侧运行一条斜杠命令行并返回脱耦结果；载体的请求信号可取消正在运行的处理器。`host/commands-changed` 是目录失效帧：客户端重新拉取 `command.list` 而不是做差分。
 
diff --git a/packages/host/apiproxy/src/api-proxy.ts b/packages/host/apiproxy/src/api-proxy.ts
index f81f5c9be8..34e67edef9 100644
--- a/packages/host/apiproxy/src/api-proxy.ts
+++ b/packages/host/apiproxy/src/api-proxy.ts
@@ -764,6 +764,15 @@ export function createApiProxy(ctx: Context, defaults: ApiProxyDefaults): ApiPro
         return ok(request, { workspace: workspaceView(workspace) })
       },
 
+      async delete(request) {
+        const { workspaceId } = request.payload
+        const operation = workspaceCreationChain.then(() =>
+          ctx.workspace.delete(brandWorkspaceId(workspaceId)))
+        workspaceCreationChain = operation.then(() => undefined, () => undefined)
+        if (!await operation) return workspaceNotFound(request, workspaceId)
+        return ok(request, { deleted: true as const })
+      },
+
       async insertSessionBefore(request) {
         const { payload } = request
         const workspace = ctx.workspace.get(brandWorkspaceId(payload.workspaceId))
@@ -977,8 +986,9 @@ export function createApiProxy(ctx: Context, defaults: ApiProxyDefaults): ApiPro
             queue.push(frame({ type: 'host/agent-error', sessionId: agent.id, message: String(error) }))
           }),
           ctx.on('domain/changed', (change) => {
-            if (change.domain !== 'workspace' || change.operation !== 'put') return
+            if (change.domain !== 'workspace') return
             if (change.table === '') {
+              if (change.operation !== 'put') return
               const state = workspaceDomainState.parse(change.value)
               for (const workspaceId of state.workspaceIds) {
                 if (committedWorkspaceIds.has(workspaceId)) continue
@@ -991,7 +1001,16 @@ export function createApiProxy(ctx: Context, defaults: ApiProxyDefaults): ApiPro
               }
               return
             }
-            if (change.table !== 'workspaces' || !committedWorkspaceIds.has(change.key)) return
+            if (change.table !== 'workspaces') return
+            if (change.operation === 'deleted') {
+              if (!committedWorkspaceIds.delete(change.key)) return
+              queue.push(frame({
+                type: 'host/workspace-removed',
+                workspaceId: change.key as WorkspaceId,
+              }))
+              return
+            }
+            if (!committedWorkspaceIds.has(change.key)) return
             // Existing-entity table writes are complete attach/touch commits.
             // A new entity's first put waits for the global registry write above.
             queue.push(frame({
diff --git a/packages/host/apiproxy/src/api/events.schema.ts b/packages/host/apiproxy/src/api/events.schema.ts
index e95b371c54..973db5a91e 100644
--- a/packages/host/apiproxy/src/api/events.schema.ts
+++ b/packages/host/apiproxy/src/api/events.schema.ts
@@ -11,7 +11,7 @@ import type { Wire } from './rpc.schema.ts'
 import { rpcErrorSchema, rpcIdSchema } from './rpc.schema.ts'
 import { approvalRequestIdSchema } from './approvals.schema.ts'
 import { contentBlockSchema, sessionEventSchema, sessionIdSchema, toolEventViewSchema } from './sessions.schema.ts'
-import { workspaceViewSchema } from './workspace.schema.ts'
+import { workspaceIdSchema, workspaceViewSchema } from './workspace.schema.ts'
 
 /** Question shape validated strictly against core dsh-user-interaction. */
 export const askUserQuestionItemSchema = z.object({
@@ -47,6 +47,7 @@ export const hostFrameSchema = z.discriminatedUnion('type', [
   z.object({ type: z.literal('host/session-status'), sessionId: sessionIdSchema, running: z.boolean() }),
   z.object({ type: z.literal('host/agent-error'), sessionId: sessionIdSchema, message: z.string() }),
   z.object({ type: z.literal('host/workspace-changed'), workspace: workspaceViewSchema }),
+  z.object({ type: z.literal('host/workspace-removed'), workspaceId: workspaceIdSchema }),
   z.object({ type: z.literal('host/commands-changed') }),
   z.object({ type: z.literal('stream/error'), error: rpcErrorSchema }),
 ]) as unknown as z.ZodType<HostFrame>
diff --git a/packages/host/apiproxy/src/api/events.ts b/packages/host/apiproxy/src/api/events.ts
index db572215cb..bf66cbf76b 100644
--- a/packages/host/apiproxy/src/api/events.ts
+++ b/packages/host/apiproxy/src/api/events.ts
@@ -85,7 +85,9 @@ export type MuxFrame =
  * agent-error is the only outlet for live failures with no turn position;
  * workspace-changed pushes the full new snapshot after every durable
  * workspace mutation (create/attach/order change — the client upserts, while
- * `workspace.list` provides the reconnect baseline).
+ * `workspace.list` provides the reconnect baseline); workspace-removed is the
+ * committed registration-deletion increment and never implies directory or
+ * session-log deletion.
  */
 export type HostFrame =
   | { type: 'host/session-added'; sessionId: SessionId; blank: boolean; parentSessionId?: SessionId; cwd?: string }
@@ -93,6 +95,7 @@ export type HostFrame =
   | { type: 'host/session-status'; sessionId: SessionId; running: boolean }
   | { type: 'host/agent-error'; sessionId: SessionId; message: string }
   | { type: 'host/workspace-changed'; workspace: WorkspaceView }
+  | { type: 'host/workspace-removed'; workspaceId: WorkspaceView['workspaceId'] }
   /**
    * The command registry changed (`commands/change` passthrough). Pure
    * invalidation signal, no payload: clients refetch `command.list` in the
diff --git a/packages/host/apiproxy/src/api/rpc-map.ts b/packages/host/apiproxy/src/api/rpc-map.ts
index abe992584c..68ccc9ec89 100644
--- a/packages/host/apiproxy/src/api/rpc-map.ts
+++ b/packages/host/apiproxy/src/api/rpc-map.ts
@@ -26,6 +26,7 @@ export interface RpcMethodMap {
   'workspace.list': WorkspaceApi['list']
   'workspace.create': WorkspaceApi['create']
   'workspace.rename': WorkspaceApi['rename']
+  'workspace.delete': WorkspaceApi['delete']
   'workspace.insertSessionBefore': WorkspaceApi['insertSessionBefore']
   'command.list': CommandsApi['list']
   'command.execute': CommandsApi['execute']
diff --git a/packages/host/apiproxy/src/api/workspace.schema.ts b/packages/host/apiproxy/src/api/workspace.schema.ts
index 47c3ae6d59..e16e5339da 100644
--- a/packages/host/apiproxy/src/api/workspace.schema.ts
+++ b/packages/host/apiproxy/src/api/workspace.schema.ts
@@ -59,6 +59,16 @@ export const workspaceRenameValueSchema = z.object({
   workspace: workspaceViewSchema,
 }) satisfies z.ZodType<Wire<ResponseValue<'workspace.rename'>>>
 
+/** workspace.delete request payload. */
+export const workspaceDeleteRequestSchema = z.object({
+  workspaceId: workspaceIdSchema,
+}) satisfies z.ZodType<Wire<RequestPayload<'workspace.delete'>>>
+
+/** workspace.delete response value. */
+export const workspaceDeleteValueSchema = z.object({
+  deleted: z.literal(true),
+}) satisfies z.ZodType<Wire<ResponseValue<'workspace.delete'>>>
+
 /** workspace.insertSessionBefore request payload (anchor omitted = append to end). */
 export const workspaceInsertSessionBeforeRequestSchema = z.object({
   workspaceId: workspaceIdSchema,
diff --git a/packages/host/apiproxy/src/api/workspace.ts b/packages/host/apiproxy/src/api/workspace.ts
index 6ec636126b..ff22d845fb 100644
--- a/packages/host/apiproxy/src/api/workspace.ts
+++ b/packages/host/apiproxy/src/api/workspace.ts
@@ -65,6 +65,14 @@ export interface WorkspaceApi {
   rename(request: RpcRequest<{ workspaceId: WorkspaceId; title: string }>):
   Promise<RpcResponse<{ workspace: WorkspaceView }>>
 
+  /**
+   * Removes one Workspace registration. The directory, every user file, and
+   * every session log remain untouched; those Sessions consequently become
+   * ungrouped. An unknown id fails with `workspace-not-found`.
+   */
+  delete(request: RpcRequest<{ workspaceId: WorkspaceId }>):
+  Promise<RpcResponse<{ deleted: true }>>
+
   /**
    * Moves an accounted session within its workspace's manual order,
    * DOM-insertBefore-like: with `beforeSessionId` the session is inserted
diff --git a/packages/host/apiproxy/src/fetch/client.ts b/packages/host/apiproxy/src/fetch/client.ts
index 0424ba7a4f..8762670cd7 100644
--- a/packages/host/apiproxy/src/fetch/client.ts
+++ b/packages/host/apiproxy/src/fetch/client.ts
@@ -23,6 +23,7 @@ import {
 } from '../api/sessions.schema.ts'
 import {
   workspaceCreateValueSchema,
+  workspaceDeleteValueSchema,
   workspaceInsertSessionBeforeValueSchema,
   workspaceListValueSchema,
   workspaceRenameValueSchema,
@@ -60,6 +61,7 @@ export interface IApiClient {
     list(payload: RequestPayload<'workspace.list'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'workspace.list'>>>
     create(payload: RequestPayload<'workspace.create'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'workspace.create'>>>
     rename(payload: RequestPayload<'workspace.rename'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'workspace.rename'>>>
+    delete(payload: RequestPayload<'workspace.delete'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'workspace.delete'>>>
     insertSessionBefore(payload: RequestPayload<'workspace.insertSessionBefore'>, signal?: AbortSignal): Promise<RpcResponse<ResponseValue<'workspace.insertSessionBefore'>>>
   }
   commands: {
@@ -91,6 +93,7 @@ const UNARY_VALUE_SCHEMAS: { [K in keyof RpcMethodMap]: z.ZodType<Wire<ResponseV
   'workspace.list': workspaceListValueSchema,
   'workspace.create': workspaceCreateValueSchema,
   'workspace.rename': workspaceRenameValueSchema,
+  'workspace.delete': workspaceDeleteValueSchema,
   'workspace.insertSessionBefore': workspaceInsertSessionBeforeValueSchema,
   'command.list': commandListValueSchema,
   'command.execute': commandExecuteValueSchema,
@@ -285,6 +288,7 @@ export abstract class AbstractApiClient implements IApiClient {
     list: (payload, signal) => this.callUnary('workspace.list', payload, signal),
     create: (payload, signal) => this.callUnary('workspace.create', payload, signal),
     rename: (payload, signal) => this.callUnary('workspace.rename', payload, signal),
+    delete: (payload, signal) => this.callUnary('workspace.delete', payload, signal),
     insertSessionBefore: (payload, signal) => this.callUnary('workspace.insertSessionBefore', payload, signal),
   }
 
diff --git a/packages/host/apiproxy/src/fetch/handler.ts b/packages/host/apiproxy/src/fetch/handler.ts
index b79980d63e..3bbcbffba1 100644
--- a/packages/host/apiproxy/src/fetch/handler.ts
+++ b/packages/host/apiproxy/src/fetch/handler.ts
@@ -24,6 +24,7 @@ import {
 import { hostDescribeRequestSchema } from '../api/host.schema.ts'
 import {
   workspaceCreateRequestSchema,
+  workspaceDeleteRequestSchema,
   workspaceInsertSessionBeforeRequestSchema,
   workspaceListRequestSchema,
   workspaceRenameRequestSchema,
@@ -57,6 +58,7 @@ const UNARY_ROUTES: UnaryRoutes = {
   'workspace.list': { schema: workspaceListRequestSchema, invoke: (api, r) => api.workspace.list(r) },
   'workspace.create': { schema: workspaceCreateRequestSchema, invoke: (api, r) => api.workspace.create(r) },
   'workspace.rename': { schema: workspaceRenameRequestSchema, invoke: (api, r) => api.workspace.rename(r) },
+  'workspace.delete': { schema: workspaceDeleteRequestSchema, invoke: (api, r) => api.workspace.delete(r) },
   'workspace.insertSessionBefore': { schema: workspaceInsertSessionBeforeRequestSchema, invoke: (api, r) => api.workspace.insertSessionBefore(r) },
   'command.list': { schema: commandListRequestSchema, invoke: (api, r) => api.commands.list(r) },
   'command.execute': { schema: commandExecuteRequestSchema, invoke: (api, r, signal) => api.commands.execute(r, signal) },
diff --git a/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts b/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts
index 11cdf5795c..d5ba628590 100644
--- a/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts
+++ b/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts
@@ -244,4 +244,31 @@ describe('Host Workspace increments', () => {
     abort.abort()
     expect(await next).toMatchObject({ done: true })
   })
+
+  it('deletes the registration, keeps its session and folder, and streams one removal', async () => {
+    const { api, ctx } = await harness()
+    const workspace = expectOk(await api.workspace.create(request({ name: 'delete-me' }))).workspace
+    const sessionId = SessionId('session-kept-after-workspace-delete')
+    expectOk(await api.sessions.create(request({ workspaceId: workspace.workspaceId, sessionId })))
+
+    const abort = new AbortController()
+    const stream: AsyncIterator<RpcRequest<HostFrame>> =
+      api.events.host(request({}), abort.signal)[Symbol.asyncIterator]()
+    const removed = nextHostFrame(stream)
+    expectOk(await api.workspace.delete(request({ workspaceId: workspace.workspaceId })))
+    expect(await removed).toMatchObject({
+      payload: { type: 'host/workspace-removed', workspaceId: workspace.workspaceId },
+    })
+    expect(expectOk(await api.workspace.list(request({}))).items).toEqual([])
+    expect(expectOk(await api.sessions.list(request({}))).items.map(item => item.sessionId)).toContain(sessionId)
+    expect(ctx.agents.get(sessionId)).toBeDefined()
+    expect(existsSync(workspace.path)).toBe(true)
+
+    const missing = await api.workspace.delete(request({ workspaceId: workspace.workspaceId }))
+    expect(missing.result).toMatchObject({
+      ok: false,
+      error: { code: 'workspace-not-found', details: { workspaceId: workspace.workspaceId } },
+    })
+    abort.abort()
+  })
 })
diff --git a/packages/host/apiproxy/tests/client-handler.spec.ts b/packages/host/apiproxy/tests/client-handler.spec.ts
index a9a5eac9ba..38ad7a52c9 100644
--- a/packages/host/apiproxy/tests/client-handler.spec.ts
+++ b/packages/host/apiproxy/tests/client-handler.spec.ts
@@ -40,6 +40,7 @@ function scriptedApi(overrides: {
       list: r => ok(r, { items: [] }),
       create: r => ok(r, { workspace: { workspaceId: 'w1' as never, path: '/t', title: 't', sessionIds: [], createdAt: '0', updatedAt: '0' }, created: true }),
       rename: r => ok(r, { workspace: { workspaceId: 'w1' as never, path: '/t', title: 't', sessionIds: [], createdAt: '0', updatedAt: '0' } }),
+      delete: r => ok(r, { deleted: true as const }),
       insertSessionBefore: r => ok(r, { workspace: { workspaceId: 'w1' as never, path: '/t', title: 't', sessionIds: [], createdAt: '0', updatedAt: '0' } }),
     },
     commands: {
@@ -76,13 +77,15 @@ describe('unary round trip', () => {
     expect(response.result).toEqual({ ok: true, value: { items: [{ sessionId: 's1', updatedAt: 7, running: false, blank: false }] } })
   })
 
-  it('routes workspace rename and insertSessionBefore through the wire', async () => {
+  it('routes workspace rename, delete, and insertSessionBefore through the wire', async () => {
     const api = scriptedApi()
     const c = client(api)
     const renamed = await c.workspace.rename({ workspaceId: 'w1' as never, title: 'next' })
     expect(renamed.result.ok).toBe(true)
     const blankTitle = await c.workspace.rename({ workspaceId: 'w1' as never, title: '   ' })
     expect(blankTitle.result).toMatchObject({ ok: false, error: { code: 'bad-request' } })
+    const deleted = await c.workspace.delete({ workspaceId: 'w1' as never })
+    expect(deleted.result).toEqual({ ok: true, value: { deleted: true } })
     const anchored = await c.workspace.insertSessionBefore({ workspaceId: 'w1' as never, sessionId: sid('s1'), beforeSessionId: sid('s2') })
     expect(anchored.result.ok).toBe(true)
     const appended = await c.workspace.insertSessionBefore({ workspaceId: 'w1' as never, sessionId: sid('s1') })
diff --git a/packages/host/apiproxy/tests/fetch-carrier.spec.ts b/packages/host/apiproxy/tests/fetch-carrier.spec.ts
index e8d65d2a62..0a92b9f5e5 100644
--- a/packages/host/apiproxy/tests/fetch-carrier.spec.ts
+++ b/packages/host/apiproxy/tests/fetch-carrier.spec.ts
@@ -58,6 +58,9 @@ function fakeApi(overrides: Partial<{ muxFrames: MuxFrame[]; hostFrames: HostFra
           result: { ok: true, value: { workspace: { workspaceId: 'w1' as never, path: '/w', title: 'w', sessionIds: [], createdAt: 't', updatedAt: 't' } } },
         }
       },
+      async delete(request) {
+        return { rpcId: request.rpcId, result: { ok: true, value: { deleted: true as const } } }
+      },
       async insertSessionBefore(request) {
         return {
           rpcId: request.rpcId,
diff --git a/packages/host/apiproxy/tests/rpc-schemas.spec.ts b/packages/host/apiproxy/tests/rpc-schemas.spec.ts
index 02ca8dec22..c1af02b7b9 100644
--- a/packages/host/apiproxy/tests/rpc-schemas.spec.ts
+++ b/packages/host/apiproxy/tests/rpc-schemas.spec.ts
@@ -14,6 +14,7 @@ import {
 import { hostDescribeRequestSchema, hostDescribeValueSchema } from '../src/api/host.schema.ts'
 import {
   workspaceCreateRequestSchema, workspaceCreateValueSchema, workspaceIdSchema,
+  workspaceDeleteRequestSchema, workspaceDeleteValueSchema,
   workspaceInsertSessionBeforeRequestSchema, workspaceInsertSessionBeforeValueSchema,
   workspaceListRequestSchema, workspaceListValueSchema,
   workspaceRenameRequestSchema, workspaceRenameValueSchema, workspaceViewSchema,
@@ -177,6 +178,13 @@ describe('workspace domain schemas', () => {
     expect(workspaceRenameValueSchema.parse({ workspace: view }).workspace.workspaceId).toBe('w1')
   })
 
+  it('validates workspace deletion payload and receipt', () => {
+    expect(workspaceDeleteRequestSchema.parse({ workspaceId: 'w1' }).workspaceId).toBe('w1')
+    expect(() => workspaceDeleteRequestSchema.parse({})).toThrow()
+    expect(workspaceDeleteValueSchema.parse({ deleted: true })).toEqual({ deleted: true })
+    expect(() => workspaceDeleteValueSchema.parse({ deleted: false })).toThrow()
+  })
+
   it('insertSessionBefore accepts an anchored and an anchorless move', () => {
     expect(workspaceInsertSessionBeforeRequestSchema.parse({ workspaceId: 'w1', sessionId: 's1', beforeSessionId: 's2' }).beforeSessionId).toBe('s2')
     expect(workspaceInsertSessionBeforeRequestSchema.parse({ workspaceId: 'w1', sessionId: 's1' }).beforeSessionId).toBeUndefined()
@@ -273,6 +281,11 @@ describe('events frame schemas', () => {
       { type: 'host/session-removed', sessionId: 's' },
       { type: 'host/session-status', sessionId: 's', running: true },
       { type: 'host/agent-error', sessionId: 's', message: 'boom' },
+      { type: 'host/workspace-changed', workspace: {
+        workspaceId: 'w', path: '/w', title: 'w', sessionIds: [],
+        createdAt: '0', updatedAt: '0',
+      } },
+      { type: 'host/workspace-removed', workspaceId: 'w' },
       { type: 'host/commands-changed' },
       { type: 'stream/error', error: { code: 'internal', message: 'm', details: {} } },
     ]
diff --git a/packages/workspace/README.i18n.yaml b/packages/workspace/README.i18n.yaml
index a5400bc218..6d62c08c0b 100644
--- a/packages/workspace/README.i18n.yaml
+++ b/packages/workspace/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 0d5ebabfbbb2922a369adb3a5d67ea4aafbe700f
-README.zh.md: b82e8e6138f3e97c3c047cf1812cee8e558ea29b
+#   pnpm run verify-translation-pairing --write packages/workspace/README.md
+README.md: ba92e95d3cde0a95eaaeae5a9b4384c3b8c9c4b8
+README.zh.md: 8c8146bba5fa6d81c0ce5d2ed29add77b4083270
diff --git a/packages/workspace/README.md b/packages/workspace/README.md
index 0d5ebabfbb..ba92e95d3c 100644
--- a/packages/workspace/README.md
+++ b/packages/workspace/README.md
@@ -2,10 +2,10 @@
 
 English | [中文](README.zh.md)
 
-The workspace family owns the persistent workspace concept: a directory the user works in, with a title and the ordered list of sessions that belong to it. Design record: [domain KV storage Agent Note](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md).
+The workspace family owns the persistent workspace concept: a directory the user works in, with a title and the ordered list of sessions that belong to it. Design record: [domain KV storage Agent Note](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md).
 
 | Package | Role | ctx key |
 |---|---|---|
 | `workspace/` | `WorkspaceRegistry` service over the storage domain form: realpath-unique paths, session-ownership accounting, entity cache | `ctx.workspace` |
 
-Ownership truth lives in the workspace record's `sessionIds` (ordered), never derived from session cwd; `attachSession` verifies the session header's cwd resolves to the workspace path, so one session structurally belongs to at most one workspace. Deletion (workspace and session cascade) is deliberately absent this phase and ships with the session-side primitives.
+Ownership truth lives in the workspace record's `sessionIds` (ordered), never derived from session cwd; `attachSession` verifies the session header's cwd resolves to the workspace path, so one session structurally belongs to at most one workspace. Deleting a Workspace removes only this registry record and account: directories, user files, and session logs remain, and the Sessions become Ungrouped ([decision](../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)).
diff --git a/packages/workspace/README.zh.md b/packages/workspace/README.zh.md
index b82e8e6138..8c8146bba5 100644
--- a/packages/workspace/README.zh.md
+++ b/packages/workspace/README.zh.md
@@ -2,10 +2,10 @@
 
 [English](README.md) | 中文
 
-Workspace 系列拥有持久 workspace 概念：用户工作所在的目录，包含标题以及属于它的有序会话列表。设计记录：[领域 KV 存储 Agent Note](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.zh.md)。
+Workspace 系列拥有持久 workspace 概念：用户工作所在的目录，包含标题以及属于它的有序会话列表。设计记录：[领域 KV 存储 Agent Note](../../.agents/notes/proposed/architecture/2026-07-24-domain-kv-storage-and-workspace.md)。
 
 | 包 | 职责 | ctx 键 |
 |---|---|---|
 | `workspace/` | 位于存储领域形式之上的 `WorkspaceRegistry` 服务：按 realpath 唯一的路径、会话所有权计数、实体缓存 | `ctx.workspace` |
 
-所有权真相存在 workspace 记录的 `sessionIds`（有序）中，绝不从会话 cwd 派生；`attachSession` 会验证会话头的 cwd 解析到 workspace 路径，因此一个会话在结构上最多属于一个 workspace。本阶段有意不提供删除（workspace 与会话级联）；该功能将与会话侧原语一起交付。
+所有权真相存在 workspace 记录的 `sessionIds`（有序）中，绝不从会话 cwd 派生；`attachSession` 会验证会话头的 cwd 解析到 workspace 路径，因此一个会话在结构上最多属于一个 workspace。删除 Workspace 只会移除该注册表记录及账本：目录、用户文件和会话日志都会保留，相关会话则进入 Ungrouped（参见[决策记录](../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)）。
diff --git a/packages/workspace/workspace/README.i18n.yaml b/packages/workspace/workspace/README.i18n.yaml
index b3e0df9280..0904711ad3 100644
--- a/packages/workspace/workspace/README.i18n.yaml
+++ b/packages/workspace/workspace/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 0d395ecc58fc5e3362cb5f3c565a0539bb09c4dd
-README.zh.md: 017e1e4d3aae9f8708ead3565f8b5b59d9b249ca
+#   pnpm run verify-translation-pairing --write packages/workspace/workspace/README.md
+README.md: 52d03b33b3482dcb6a2f5feddbc15ac9fefee0a8
+README.zh.md: f899abdc3dd2a551179cd710c6dda84f804a8e80
diff --git a/packages/workspace/workspace/README.md b/packages/workspace/workspace/README.md
index 0d395ecc58..52d03b33b3 100644
--- a/packages/workspace/workspace/README.md
+++ b/packages/workspace/workspace/README.md
@@ -10,6 +10,7 @@ The entity/storage rationale lives in the [domain Agent Note](../../../.agents/n
 
 - `ctx.workspace.create(path, title?)` — canonicalizes `path` via `fs.realpath`, rejects a nonexistent or non-directory path, creates at most one record per canonical path, and prepends a new record to durable workspace order. Repeated calls for that path return the existing workspace without changing its title; a different path cannot create a duplicate title.
 - `ctx.workspace.get(id)` / `list()` / `resolveByPath(path)` — cache-served lookups. `list()` is synchronous and follows durable registry order; `resolveByPath` is async because it applies the same `realpath` canon and rejects a missing path rather than creating it.
+- `ctx.workspace.delete(id)` — removes only the Workspace registration, its durable order entry, and its session account. Unknown ids return `false`; a removed record returns `true`. The directory, user files, live Sessions, and persisted session logs are never touched, so those Sessions become Ungrouped. A table-write failure restores the prior order and published entity.
 - `Workspace.attachSession(id)` — validates a live or persisted session header cwd against the workspace path and prepends a new id. Unknown sessions, absent/unresolvable/non-directory cwd values, and mismatches reject without writing. `detachSession` removes only the candidate index entry.
 - `ctx.workspace.touchSession(id)` — moves only that validated, accounted session to the front. Ungrouped or filtered sessions are no-ops, and workspace order never changes.
 - `Workspace.sessionIds` — synchronous id-plus-canonical-cwd membership projection in durable candidate order. Missing headers, invalid cwd values, and mismatches are filtered; the next workspace mutation prunes them. A medium indexing one session under two workspaces, claiming one path from two records, or diverging from durable workspace order rejects at startup.
@@ -35,5 +36,5 @@ Independent of live requests: the package never touches a request prefix, so it
 
 ## Known Limitations and Deferred Work
 
-- No delete entry point in this phase — workspace deletion ships as one complete semantic together with the session-delete primitive and cascade orchestration (future-work section of the Agent Note); a half "drop the record, keep the sessions" operation is deliberately not exposed.
+- Session deletion and destructive folder removal are separate, absent capabilities; Workspace registration deletion never substitutes for either ([decision](../../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)).
 - The header index refreshes at startup and when attach must resolve an uncached persisted id; deletion or cwd damage performed by another process is observed after the next refresh or restart.
diff --git a/packages/workspace/workspace/README.zh.md b/packages/workspace/workspace/README.zh.md
index 017e1e4d3a..f899abdc3d 100644
--- a/packages/workspace/workspace/README.zh.md
+++ b/packages/workspace/workspace/README.zh.md
@@ -10,6 +10,7 @@ DeepSeek Harness 的 Workspace 实体注册表（`ctx.workspace`）：通过领
 
 - `ctx.workspace.create(path, title?)`：规范化 `path` 时使用 `fs.realpath`，拒绝不存在或非目录的路径，每个规范路径最多创建一条记录，并将新记录前置到持久 workspace 顺序。对同一路径重复调用会返回现有 workspace，且不改变其标题；不同路径不能创建重复标题。
 - `ctx.workspace.get(id)`/`list()`/`resolveByPath(path)`：由缓存提供的查找。`list()` 为同步操作，并遵循持久注册表顺序；`resolveByPath` 为异步操作，因为它应用同一 `realpath` 规范，并会拒绝缺失路径，而不是创建路径。
+- `ctx.workspace.delete(id)`：只移除 Workspace 注册记录、对应的持久顺序条目及会话账本。未知 id 返回 `false`，成功移除记录则返回 `true`。目录、用户文件、实时会话和持久化会话日志绝不受影响，因此相关会话会进入 Ungrouped。表写入失败时会恢复原顺序和此前发布的实体。
 - `Workspace.attachSession(id)`：对照 workspace 路径验证实时或已持久化的会话头 cwd，并将新 id 前置。未知会话、缺失／无法解析／非目录的 cwd 值和不匹配情况都会在不写入的前提下被拒绝。`detachSession` 只移除候选索引条目。
 - `ctx.workspace.touchSession(id)`：仅将已验证、已记账的会话移到最前。未分组或被过滤的会话为空操作，workspace 顺序绝不改变。
 - `Workspace.sessionIds`：按持久候选顺序提供同步 id 加规范 cwd 成员投影。缺失头部、无效 cwd 值和不匹配情况都被过滤；下一次 workspace 变更会剪除它们。如果同一存储介质将一个会话索引到两个 workspace 下、从两条记录声明同一路径，或偏离持久 workspace 顺序，启动会被拒绝。
@@ -35,5 +36,5 @@ DeepSeek Harness 的 Workspace 实体注册表（`ctx.workspace`）：通过领
 
 ## 已知限制与延后工作
 
-- 本阶段没有删除入口：workspace 删除将与会话删除原语和级联编排一起作为完整语义交付（参见 Agent Note 的未来工作一节）；系统有意不公开「删除记录、保留会话」的半成品操作。
+- 会话删除与破坏性的文件夹移除是彼此独立且尚未提供的功能；删除 Workspace 注册记录绝不能替代二者（参见[决策记录](../../../.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md)）。
 - 头部索引会在启动时刷新，也会在 attach 必须解析未缓存持久 id 时刷新；另一进程执行的删除或 cwd 破坏会在下次刷新或重启后被观测。
diff --git a/packages/workspace/workspace/src/index.ts b/packages/workspace/workspace/src/index.ts
index 0f849e7374..2699365608 100644
--- a/packages/workspace/workspace/src/index.ts
+++ b/packages/workspace/workspace/src/index.ts
@@ -168,6 +168,18 @@ export class WorkspaceRegistry extends Service {
     })
   }
 
+  /**
+   * Delete one workspace registration while retaining its directory and every
+   * session log. The durable order is updated before the table deletion; a
+   * failed table write restores the prior order and keeps the entity
+   * published. Unknown ids are an idempotent no-op for domain callers.
+   * @param id - Workspace registration to remove.
+   * @returns `true` when a record was deleted, `false` when it was unknown.
+   */
+  delete(id: WorkspaceId): Promise<boolean> {
+    return this.enqueueOperation(() => this.deleteKnown(id))
+  }
+
   /**
    * Resolve by canonical directory path without creating or mutating a
    * workspace. A missing path rejects during `realpath`; an existing unowned
@@ -231,6 +243,33 @@ export class WorkspaceRegistry extends Service {
     return entity
   }
 
+  private async deleteKnown(id: WorkspaceId): Promise<boolean> {
+    const entity = this.entities.get(id)
+    if (entity === undefined) return false
+    const state = this.requireState()
+    const nextState = {
+      initialized: true,
+      workspaceIds: state.workspaceIds.filter(workspaceId => workspaceId !== id),
+    }
+    await this.setState(nextState)
+    this.entities.delete(id)
+    try {
+      await this.requireTable().delete(id)
+    } catch (error) {
+      this.entities.set(id, entity)
+      try {
+        await this.setState(state)
+      } catch (rollbackError) {
+        throw new AggregateError(
+          [error, rollbackError],
+          `workspace '${id}' record deletion and registry-order rollback both failed`,
+        )
+      }
+      throw error
+    }
+    return true
+  }
+
   private async bootstrap(headers: readonly SessionHeader[]): Promise<void> {
     const table = this.requireTable()
     const state = this.requireState()
diff --git a/packages/workspace/workspace/src/invariant.ts b/packages/workspace/workspace/src/invariant.ts
index 1764ce2fe3..808ce1dedf 100644
--- a/packages/workspace/workspace/src/invariant.ts
+++ b/packages/workspace/workspace/src/invariant.ts
@@ -20,8 +20,9 @@ export const inject = ['invariants']
  * domain's durable table. Every `domain/changed` for the `workspaces` table
  * must name a record the cache already holds an entity for (the registry
  * caches before the durable put and mutates only through cached entities).
- * A delete is valid only for create rollback, after the provisional cache
- * entry has been removed; deleting a published entity proves a bypass.
+ * A delete is valid only after the registry has removed the entity from its
+ * cache, whether for create rollback or an explicit registration deletion;
+ * deleting while the cache still publishes the entity proves a bypass.
  */
 const install: InvariantInstaller = Object.assign(
   (ctx: Context, fail: (message: string) => never) => {
diff --git a/packages/workspace/workspace/tests/invariant.spec.ts b/packages/workspace/workspace/tests/invariant.spec.ts
index ea1fbaa64c..0d0556a44a 100644
--- a/packages/workspace/workspace/tests/invariant.spec.ts
+++ b/packages/workspace/workspace/tests/invariant.spec.ts
@@ -49,7 +49,7 @@ describe('workspace cache/table invariant', () => {
       .toThrow(/cache still publishes/)
   })
 
-  it('allows deletion only after a provisional create cache entry was removed for rollback', async () => {
+  it('allows deletion after the registry removed the cache entry for rollback or explicit deletion', async () => {
     const ctx = await setup([])
     expect(() => { ctx.emit('domain/changed', deleted()) }).not.toThrow()
   })
diff --git a/packages/workspace/workspace/tests/workspace.spec.ts b/packages/workspace/workspace/tests/workspace.spec.ts
index 8ce70cc7d5..ed08b5ba50 100644
--- a/packages/workspace/workspace/tests/workspace.spec.ts
+++ b/packages/workspace/workspace/tests/workspace.spec.ts
@@ -424,6 +424,40 @@ describe('WorkspaceRegistry create and lookup', () => {
     expect(pool.media.get('workspace')!.tables.get('workspaces')!.size).toBe(1)
   })
 
+  it('deletes only the registration and leaves its directory and session headers untouched', async () => {
+    const dir = await makeDir('delete-registration')
+    const result = await harness({ sessions: [header('kept-session', dir)] })
+    const workspace = await result.registry.create(dir)
+    await workspace.attachSession(SessionId('kept-session'))
+
+    await expect(result.registry.delete(workspace.id)).resolves.toBe(true)
+    await expect(result.registry.delete(workspace.id)).resolves.toBe(false)
+    expect(result.registry.get(workspace.id)).toBeUndefined()
+    expect(result.registry.list()).toEqual([])
+    expect(storedState(result.pool)).toEqual({ initialized: true, workspaceIds: [] })
+    expect(result.pool.media.get('workspace')!.tables.get('workspaces')!.has(workspace.id)).toBe(false)
+    await expect(realpath(dir)).resolves.toBe(dir)
+    expect(result.list).toHaveBeenCalledTimes(1)
+    expect(result.load).not.toHaveBeenCalled()
+    expect(result.inspect).not.toHaveBeenCalled()
+  })
+
+  it('rolls registry order and cache back when record deletion fails', async () => {
+    const dir = await makeDir('delete-rollback')
+    const pool = new MemoryMediaPool()
+    const result = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { deleteAt: 1 }),
+    })
+    const workspace = await result.registry.create(dir)
+
+    await expect(result.registry.delete(workspace.id)).rejects.toThrow(/selected rollback delete failure/)
+    expect(result.registry.get(workspace.id)).toBe(workspace)
+    expect(result.registry.list()).toEqual([workspace])
+    expect(storedState(pool).workspaceIds).toEqual([workspace.id])
+    expect(storedRecord(pool, workspace.id)).toMatchObject({ path: dir })
+  })
+
   it('rejects table access before the registry has started', async () => {
     const dir = await makeDir('unstarted')
     const registry = new WorkspaceRegistry(new Context())

From 5fa74343aab77f543c3dfdac2ee7d387e769132c Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 12:54:35 +0800
Subject: [PATCH 13/27] docs(ci): six always-on instances, no pre-registered
 spares

The spare tier is retired. Steady-state pool load is one serial standby
job per master push, so six always-on instances already are the
failover capacity; pre-registered offline runners are a silently
expiring guarantee (GitHub garbage-collects them after 30 days
offline). Incident-time extra capacity is a one-minute org-token
registration, now documented in the runbook.
---
 ...-07-22-evidence-based-larger-hosted-runners.i18n.yaml | 6 +++---
 .../2026-07-22-evidence-based-larger-hosted-runners.md   | 2 +-
 ...2026-07-22-evidence-based-larger-hosted-runners.zh.md | 2 +-
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml     | 6 +++---
 .../process/2026-07-26-ci-failover-runbook.md            | 9 +++------
 .../process/2026-07-26-ci-failover-runbook.zh.md         | 9 +++------
 6 files changed, 14 insertions(+), 20 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 84a10e5ab9..5ebd95248c 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-22-evidence-based-larger-hosted-runners.md: 21e602b2b5850176df981dcf448f4f827b756719
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: ba49ff18ac304f4078d4c8ebfd00bb1a85ada0b3
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+2026-07-22-evidence-based-larger-hosted-runners.md: 5b399be5571ddaf1f775ba43a2233198b8e09b18
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: 40970ec33c1a16af85ea47be3fc932209efdd654
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 21e602b2b5..5b399be557 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -52,7 +52,7 @@ The process-bound coverage project contains exactly five suite files. Thirty-two
 
 Complete serial Linux, macOS, and Windows references run only when `master` moves. Pull requests use the enterprise required path plus standard-hosted compatibility jobs, while other larger-runner sizes run only by manual dispatch.
 
-An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with four always-on systemd-managed runner instances plus four registered spares) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the admin-only `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
+An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with six always-on systemd-managed runner instances) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the admin-only `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index ba49ff18ac..40970ec33c 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -52,7 +52,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
-另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 4 个常驻的 systemd 管理运行器实例，另有 4 个已注册备用位）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.zh.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
 
 ## 曾考虑的替代方案
 
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index a2725da1b2..efb5fdd1cc 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-26-ci-failover-runbook.md: db8e0676ecc6eeaea16438e7868ccf9ac43887cc
-2026-07-26-ci-failover-runbook.zh.md: b3b4149f460784e88ce03458fc556f402c38fa2f
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+2026-07-26-ci-failover-runbook.md: 0bce83e0f9c842fa3dd73ae9c0a3eefc0975cdae
+2026-07-26-ci-failover-runbook.zh.md: 4bc6c67bab754ad3f0127557b0d5e04f7934c8a2
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index db8e0676ec..0bce83e0f9 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -14,7 +14,7 @@ Each of the three required Linux jobs resolves its runner pool through the `DSH_
 
 ### What the in-house pool is
 
-`vm-backup`: one 64-core VM, four always-on systemd-managed runner instances, four registered spares. Check the latest `serial / linux (self-hosted standby)` run before switching: a green standby is verified-yesterday capacity.
+`vm-backup`: one 64-core VM, six always-on systemd-managed runner instances. Check the latest `serial / linux (self-hosted standby)` run before switching: a green standby is verified-yesterday capacity.
 
 ### Switch (repo admin, ~1 minute, no merge)
 
@@ -24,15 +24,12 @@ Each of the three required Linux jobs resolves its runner pool through the `DSH_
 
 ### Capacity during failover
 
-Four always-on instances absorb normal PR traffic. If queues build, bring the four registered spares online on the VM (no token needed — they are already registered):
+Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner) — cloning an existing runner directory and running `config.sh` takes about a minute per instance.
 
-```bash
-for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
-```
 
 ### Switch back
 
-Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhosted`). New runs resolve back to the hosted enterprise pools. Stop the spare instances if they were started.
+Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhosted`). New runs resolve back to the hosted enterprise pools. Remove any extra instances that were registered during the incident.
 
 ### Trust boundary
 
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index b3b4149f46..4bc6c67bab 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -14,7 +14,7 @@ Status: implemented
 
 ### 自有池是什么
 
-`vm-backup`：一台 64 核虚拟机，4 个常驻 systemd 管理的运行器实例，另有 4 个已注册备用位。切换前先看 `serial / linux (self-hosted standby)` 最近一次运行：绿色 = 这套环境昨天刚被全量验证过。
+`vm-backup`：一台 64 核虚拟机，6 个常驻 systemd 管理的运行器实例。切换前先看 `serial / linux (self-hosted standby)` 最近一次运行：绿色 = 这套环境昨天刚被全量验证过。
 
 ### 切换步骤（仓库管理员，约 1 分钟，无需合并）
 
@@ -24,15 +24,12 @@ Status: implemented
 
 ### 切换期间的容量
 
-4 个常驻实例可承接正常 PR 流量。若出现排队，在虚拟机上把 4 个已注册的备用位拉起（无需 token——它们已注册）：
+6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例——复制现有 runner 目录再跑 `config.sh`，每个约一分钟。
 
-```bash
-for i in 7 8 9 10; do cd /data_local/actions-runner-$i && sudo ./svc.sh install ubuntu && sudo ./svc.sh start; done
-```
 
 ### 切回
 
-删除 `DSH_CI_FAILOVER` 变量（或改为 `selfhosted` 以外的任何值），新的运行即解析回托管企业池。若启动过备用实例，将其停止。
+删除 `DSH_CI_FAILOVER` 变量（或改为 `selfhosted` 以外的任何值），新的运行即解析回托管企业池。若故障期间追加注册过实例，将其移除。
 
 ### 信任边界
 

From 1a8225ee6cc62438a0c7c54e19ee76ba76b61b33 Mon Sep 17 00:00:00 2001
From: Ubuntu <ubuntu@localhost.localdomain>
Date: Mon, 27 Jul 2026 14:35:05 +0800
Subject: [PATCH 14/27] ci: retrigger after failover switch


From 7bd96af5eb8d889b0652f1d4972a4e3e4c9649e2 Mon Sep 17 00:00:00 2001
From: NI0317 <stniii317@gmail.com>
Date: Mon, 27 Jul 2026 14:37:04 +0800
Subject: [PATCH 15/27] fix(workspace): make deletion recoverable

---
 ...-workspace-registration-deletion.i18n.yaml |   4 +-
 ...6-07-27-workspace-registration-deletion.md |   8 +-
 ...7-27-workspace-registration-deletion.zh.md |   8 +-
 apps/web/tests/workspace-management.e2e.ts    |  29 ++++
 .../runtime/src/client/workspaces/manager.ts  |  10 +-
 .../ui-workspace/src/client/rows/Rows.tsx     |   4 +
 .../tests/api-proxy-workspace.spec.ts         |   6 +
 packages/workspace/workspace/README.i18n.yaml |   4 +-
 packages/workspace/workspace/README.md        |   2 +
 packages/workspace/workspace/README.zh.md     |   2 +
 packages/workspace/workspace/src/index.ts     |  75 ++++++++-
 packages/workspace/workspace/src/spec.ts      |  11 ++
 .../workspace/tests/workspace.spec.ts         | 151 +++++++++++++++++-
 13 files changed, 294 insertions(+), 20 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
index 847b040457..93c78373c6 100644
--- a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
-2026-07-27-workspace-registration-deletion.md: cae01d529bc6fd97da6fb61839bd5ec8e21557e2
-2026-07-27-workspace-registration-deletion.zh.md: 76377ebc5e93101e1e3efce1d29c3c654df032c2
+2026-07-27-workspace-registration-deletion.md: 58ae5c4bef2cf1cb0a0158eda5eb37daf2e9703d
+2026-07-27-workspace-registration-deletion.zh.md: 7a79a1ccc53a0d4fd7e5ab453239ade955313c6e
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
index cae01d529b..58ae5c4bef 100644
--- a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
@@ -22,6 +22,8 @@ Registry operations serialize create and delete. Deletion first writes the Works
 
 The Host stream keeps its committed-id set through the preceding global-order write and removes the id only on the table deletion. Create rollback therefore emits no false removal, while every connected tab receives exactly the id needed to delete its projection.
 
+Create and delete write a durable `pendingMutation` before their record/order pair can diverge. Startup completes only the named create or delete and clears the marker; it never infers crash provenance from an orphan row alone. Unmarked order/table divergence therefore retains the registry's fail-loud corruption behavior. A deletion whose table write committed but marker cleanup failed still reports success—the requested state and removal frame are already committed—and the next startup clears that marker idempotently.
+
 ## Client convergence
 
 `WorkspaceManager` treats both `host/workspace-changed` and `host/workspace-removed` as ordered deltas replayed over an in-flight `workspace.list` response. A successful unary delete removes the row immediately instead of waiting for its own stream echo. Removal is idempotent, and a process-local tombstone rejects late changed frames or stale baseline rows for the never-reused Workspace id. A reconnect still refreshes from `workspace.list`; Session state is never pruned by a Workspace delta.
@@ -40,14 +42,16 @@ The menu, Modal, and buttons retain their existing structure and design tokens.
 
 **Delete the table row and repair order later.** Rejected because a crash or write failure would leave an initialized registry whose order and table disagree. The registry updates both under one serialized operation and restores the prior order on table failure.
 
+**Delete every unreferenced row at startup.** Rejected because the same shape can come from unexplained order corruption; silently discarding it could lose Workspace metadata and Session accounting. Recovery requires the explicit pending marker written by the owning mutation.
+
 **Refetch both lists after success.** Rejected because the committed removal frame plus immediate unary echo is sufficient, preserves the current Session object, and avoids turning a local mutation into two list requests. Reconnect baselines remain the repair path.
 
 ## Verification
 
-Workspace package tests pin successful metadata-only deletion, unknown-id idempotence, table-failure rollback, and cache/table invariant behavior. Apiproxy and carrier tests pin the schema, handler, `workspace-not-found`, retained Session/folder, and committed `host/workspace-removed` frame. Client tests pin unary direct echo, duplicate removal, late changed frames, and deletion racing an in-flight baseline. Component tests pin confirmation, pending-state duplicate suppression, success, failure, Cancel, Escape, and Close.
+Workspace package tests pin successful metadata-only deletion, same-path re-registration, unknown-id idempotence, table-failure rollback, explicit-marker restart recovery, unexplained-corruption rejection, and cache/table invariant behavior. Apiproxy and carrier tests pin the schema, handler, `workspace-not-found`, retained Session/folder, fresh-id re-registration, and committed `host/workspace-removed` frame. Client tests pin unary direct echo, duplicate removal, late changed frames, and deletion racing an in-flight baseline. Component tests pin confirmation, pending-state duplicate suppression, success, failure, Cancel, Escape, and Close.
 
 The assembled keyless Web scenario registers an existing temporary project directory, accounts a persisted Session, makes that Session current, confirms deletion in Chromium, and verifies the Workspace group disappears while Ungrouped retains the current Session. It checks the user file and JSONL log before and after deletion and repeats the UI, directory, and log assertions after reload.
 
 ## Consequences
 
-Deleting a Workspace is intentionally reversible by registering the same directory again, although its prior manual Session order is gone; re-registration does not automatically re-adopt existing Sessions after bootstrap. The operation gives up a one-click cleanup of Session histories or source directories in exchange for a deletion boundary that matches what the record actually owns.
+Deleting a Workspace is intentionally reversible by registering the same directory again with a fresh id, although its prior manual Session order is gone; re-registration does not automatically re-adopt existing Sessions after bootstrap. The operation gives up a one-click cleanup of Session histories or source directories in exchange for a deletion boundary that matches what the record actually owns.
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
index 76377ebc5e..7a79a1ccc5 100644
--- a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
@@ -22,6 +22,8 @@ Workspace 注册已有代码目录，使 GUI 能够为目录命名，并对其
 
 Host 流在前一笔全局顺序写入期间继续保留其已提交 id 集合，只在删除表行时移除该 id。因此，创建回滚不会发出错误的移除帧，而每个已连接标签页都能收到从自身投影中删除该记录所需的准确 id。
 
+Create 与 delete 会在记录／顺序对可能分叉之前写入持久 `pendingMutation`。启动时只补全其中明确命名的 create 或 delete，并清除该标记；系统绝不会仅凭孤立表行的形状推断崩溃来源。因此，没有标记的顺序／表分叉仍会保持注册表原有的损坏直接失败语义。如果删除的表写入已经提交、但标记清理失败，操作仍会报告成功——请求状态和移除帧都已经提交——下一次启动会以幂等方式清除该标记。
+
 ## 客户端收敛
 
 `WorkspaceManager` 将 `host/workspace-changed` 与 `host/workspace-removed` 都视为有序增量，并在进行中的 `workspace.list` 响应之上回放。成功的一元删除会立即移除行，无需等待本次操作自己的流回显。移除操作具有幂等性；由于 Workspace id 永不复用，进程本地删除标记会拒绝延迟到达的 changed 帧或陈旧基线行。重连仍从 `workspace.list` 刷新；Workspace 增量绝不会剪除会话状态。
@@ -40,14 +42,16 @@ Host 流在前一笔全局顺序写入期间继续保留其已提交 id 集合
 
 **先删除表行，之后再修复顺序。** 不予采纳，因为崩溃或写入失败会使已初始化注册表的顺序与表不一致。注册表会在同一串行操作内更新二者，并在表操作失败时恢复此前顺序。
 
+**启动时删除所有未引用表行。** 不予采纳，因为来源不明的顺序损坏也会呈现相同形状；静默丢弃可能损失 Workspace 元数据和 Session 账本。恢复必须依赖拥有该变更的操作预先写入的明确待处理标记。
+
 **成功后重新拉取两个列表。** 不予采纳，因为已提交的移除帧与即时一元回显已足够，既能保留当前会话对象，也避免将局部变更扩大为两次列表请求。重连基线仍是修复路径。
 
 ## Verification
 
-Workspace 包测试固定了仅删除元数据的成功路径、未知 id 的幂等行为、表操作失败回滚，以及缓存／表不变量行为。Apiproxy 与载体测试固定了 schema、处理器、`workspace-not-found`、保留会话／文件夹，以及已提交的 `host/workspace-removed` 帧。客户端测试固定了一元直接回显、重复移除、延迟到达的 changed 帧，以及删除与进行中基线并发的行为。组件测试固定了确认交互、待处理状态下抑制重复提交、成功、失败、Cancel、Escape 与 Close。
+Workspace 包测试固定了仅删除元数据的成功路径、同路径重新注册、未知 id 的幂等行为、表操作失败回滚、明确标记的重启恢复、来源不明损坏的拒绝，以及缓存／表不变量行为。Apiproxy 与载体测试固定了 schema、处理器、`workspace-not-found`、保留会话／文件夹、使用新 id 重新注册，以及已提交的 `host/workspace-removed` 帧。客户端测试固定了一元直接回显、重复移除、延迟到达的 changed 帧，以及删除与进行中基线并发的行为。组件测试固定了确认交互、待处理状态下抑制重复提交、成功、失败、Cancel、Escape 与 Close。
 
 组装后的无密钥 Web 场景会注册一个已有临时项目目录，将持久化会话计入账本，把该会话设为当前会话，在 Chromium 中确认删除，并验证 Workspace 分组消失，而 Ungrouped 保留当前会话。该场景在删除前后检查用户文件和 JSONL 日志，并在刷新后重复验证 UI、目录与日志。
 
 ## Consequences
 
-删除 Workspace 后仍可重新注册同一目录，因此该操作有意设计为可逆；但此前的手动会话顺序会丢失，重新注册后，系统也不会在 bootstrap 结束后自动重新收编现有会话。该操作放弃一键清理会话历史或源码目录，以换取与记录实际所有权一致的删除边界。
+删除 Workspace 后仍可使用新 id 重新注册同一目录，因此该操作有意设计为可逆；但此前的手动会话顺序会丢失，重新注册后，系统也不会在 bootstrap 结束后自动重新收编现有会话。该操作放弃一键清理会话历史或源码目录，以换取与记录实际所有权一致的删除边界。
diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts
index 4dbcca36ae..3239dcfc12 100644
--- a/apps/web/tests/workspace-management.e2e.ts
+++ b/apps/web/tests/workspace-management.e2e.ts
@@ -169,6 +169,34 @@ describe('web e2e: workspace management (create / rename / flat view / hover car
     await stat(logLocation.path)
     expect((await scaffold.ctx.sessionPersistence.inspect(SessionId(SEED_ID))).events.length).toBeGreaterThan(0)
 
+    // Re-registering the exact deleted path immediately, without a reload, is
+    // a supported reversible flow. It creates a fresh Workspace id without
+    // re-adopting the retained Session.
+    await page.getByRole('button', { name: 'Create workspace' }).click()
+    await page.getByRole('menuitem', { name: 'Create workspace' }).hover()
+    await page.getByRole('menuitem', { name: 'Use an existing folder' }).click()
+    const reuseFolder = page.getByRole('dialog', { name: 'Use an existing folder' })
+    await reuseFolder.getByLabel('Existing folder path').fill(scaffold.workspaceCwd)
+    await reuseFolder.getByRole('button', { name: 'Use folder' }).click()
+    await expect.poll(() => reuseFolder.count(), { timeout: 10_000 }).toBe(0)
+    const reregistered = await scaffold.ctx.workspace.resolveByPath(scaffold.workspaceCwd)
+    expect(reregistered?.id).toBeDefined()
+    expect(reregistered?.id).not.toBe(workspace.id)
+    expect(reregistered?.sessionIds).toEqual([])
+    await expect.poll(() => page.getByText('Ungrouped', { exact: true }).count(), { timeout: 10_000 })
+      .toBeGreaterThanOrEqual(1)
+    expect(await readFile(join(scaffold.workspaceCwd, 'workspace', 'a.txt'), 'utf8')).toBe('alpha\n')
+    await stat(logLocation.path)
+
+    // Restore the deleted-registry state so reload still verifies deletion
+    // persistence independently of the successful re-registration above.
+    if (reregistered === undefined) throw new Error('same-path re-registration did not materialize')
+    await scaffold.ctx.workspace.delete(reregistered.id)
+    await expect.poll(
+      () => page.getByRole('button', { name: `Workspace actions for ${reregistered.title}` }).count(),
+      { timeout: 10_000 },
+    ).toBe(0)
+
     const warningStart = tripwire.warnings.length
     await page.reload({ waitUntil: 'load' })
     await page.waitForSelector('[class*="frame"]', { timeout: 30_000 })
@@ -183,6 +211,7 @@ describe('web e2e: workspace management (create / rename / flat view / hover car
     expect(await readFile(join(scaffold.workspaceCwd, 'workspace', 'a.txt'), 'utf8')).toBe('alpha\n')
     await stat(logLocation.path)
     expect((await scaffold.ctx.sessionPersistence.inspect(SessionId(SEED_ID))).events.length).toBeGreaterThan(0)
+
     expect(tripwire.pageErrors).toEqual([])
   }, 90_000)
 
diff --git a/packages/client/runtime/src/client/workspaces/manager.ts b/packages/client/runtime/src/client/workspaces/manager.ts
index 83275e9a2a..7179ed9eb9 100644
--- a/packages/client/runtime/src/client/workspaces/manager.ts
+++ b/packages/client/runtime/src/client/workspaces/manager.ts
@@ -33,6 +33,14 @@ export class WorkspaceManager {
   private error: RpcError | null = null
   private inflight: Promise<void> | null = null
   private refreshFrames: WorkspaceDelta[] | null = null
+  /**
+   * Ids this process has seen removed, kept for the connection's lifetime so
+   * a late changed frame or a stale baseline row cannot resurrect a deleted
+   * row. Correctness rests on Host ids never being reused (the registry mints
+   * a fresh `randomUUID` per record, including when the same directory is
+   * registered again) — a path-derived id scheme would turn these entries
+   * into permanent blindfolds and must clear them instead.
+   */
   private readonly removedIds = new Set<WorkspaceId>()
   private snapshotCache: WorkspaceListSnapshot
   private readonly notifier = new Notifier(() => {
@@ -266,7 +274,7 @@ function upsertWorkspace(items: readonly WorkspaceView[], workspace: WorkspaceVi
     : items.map((item, position) => position === index ? workspace : item)
 }
 
-
+/** Replay one ordered delta over a baseline: upsert in place, or drop the removed id. */
 function applyWorkspaceDelta(items: readonly WorkspaceView[], delta: WorkspaceDelta): WorkspaceView[] {
   return delta.type === 'upsert'
     ? upsertWorkspace(items, delta.workspace)
diff --git a/packages/client/ui-workspace/src/client/rows/Rows.tsx b/packages/client/ui-workspace/src/client/rows/Rows.tsx
index ae866f58cf..e245f8b217 100644
--- a/packages/client/ui-workspace/src/client/rows/Rows.tsx
+++ b/packages/client/ui-workspace/src/client/rows/Rows.tsx
@@ -75,6 +75,10 @@ export function ProjectRowItem({ group, onToggle, onCreate, actions }: {
             items={WORKSPACE_MENU_ITEMS}
             onSelect={(id) => {
               setMenuOpen(false)
+              // Unknown ids leave before the dispatch: a future menu row must
+              // not inherit the destructive branch as an else fallback.
+              /* v8 ignore next -- WORKSPACE_MENU_ITEMS carries exactly these two rows today. */
+              if (id !== 'rename' && id !== 'delete') return
               if (id === 'rename') actions.rename()
               else actions.delete()
             }}
diff --git a/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts b/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts
index d5ba628590..bbd57cb6dc 100644
--- a/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts
+++ b/packages/host/apiproxy/tests/api-proxy-workspace.spec.ts
@@ -269,6 +269,12 @@ describe('Host Workspace increments', () => {
       ok: false,
       error: { code: 'workspace-not-found', details: { workspaceId: workspace.workspaceId } },
     })
+
+    const reregistered = expectOk(await api.workspace.create(request({ path: workspace.path }))).workspace
+    expect(reregistered.workspaceId).not.toBe(workspace.workspaceId)
+    expect(reregistered.path).toBe(workspace.path)
+    expect(reregistered.sessionIds).toEqual([])
+    expect(expectOk(await api.sessions.list(request({}))).items.map(item => item.sessionId)).toContain(sessionId)
     abort.abort()
   })
 })
diff --git a/packages/workspace/workspace/README.i18n.yaml b/packages/workspace/workspace/README.i18n.yaml
index 0904711ad3..b5eaefa98c 100644
--- a/packages/workspace/workspace/README.i18n.yaml
+++ b/packages/workspace/workspace/README.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write packages/workspace/workspace/README.md
-README.md: 52d03b33b3482dcb6a2f5feddbc15ac9fefee0a8
-README.zh.md: f899abdc3dd2a551179cd710c6dda84f804a8e80
+README.md: bee3e4fcb5dded273f30942ee2e42ee93b839e62
+README.zh.md: 7960a2d13df4f881687fd88cdb07e237b3abb7c8
diff --git a/packages/workspace/workspace/README.md b/packages/workspace/workspace/README.md
index 52d03b33b3..bee3e4fcb5 100644
--- a/packages/workspace/workspace/README.md
+++ b/packages/workspace/workspace/README.md
@@ -18,6 +18,8 @@ The entity/storage rationale lives in the [domain Agent Note](../../../.agents/n
 
 `storageDomain` and `sessionPersistence` are required startup dependencies. An unavailable peer leaves the plugin pending and cannot commit an empty initialized marker. On the first successful start, the registry calls `SessionPersistence.list()` and uses only header `id`, `cwd`, and `createdAt` to group valid historical directories and persist initial order; it never reads event bodies. The initialized marker is written last, so partial bootstrap writes are reused safely after restart. Later cwd-only sessions remain Ungrouped.
 
+Create and delete persist an explicit pending-mutation marker before their record and order can diverge. Startup completes only the marked mutation, then clears the marker; an unmarked order/table mismatch remains unexplained corruption and fails loud. Deleting and re-registering the same path creates a fresh Workspace id and does not automatically re-adopt the retained Sessions.
+
 ## Model Experience
 
 ### Workspace records and session accounts
diff --git a/packages/workspace/workspace/README.zh.md b/packages/workspace/workspace/README.zh.md
index f899abdc3d..7960a2d13d 100644
--- a/packages/workspace/workspace/README.zh.md
+++ b/packages/workspace/workspace/README.zh.md
@@ -18,6 +18,8 @@ DeepSeek Harness 的 Workspace 实体注册表（`ctx.workspace`）：通过领
 
 `storageDomain` 和 `sessionPersistence` 是启动必需依赖。对等服务不可用时，插件保持待处理，且不能提交空的已初始化标记。首次成功启动时，注册表调用 `SessionPersistence.list()`，仅使用头部 `id`、`cwd` 和 `createdAt` 对有效历史目录分组并持久化初始顺序；它绝不读取事件正文。已初始化标记最后写入，因此重启后可安全复用部分启动写入。后续仅有 cwd 的会话仍属于 Ungrouped。
 
+Create 与 delete 会在记录和顺序可能分叉之前，先持久化明确的待处理变更标记。启动时只补全被该标记证明的变更，随后清除标记；没有标记的顺序／表不一致仍属于来源不明的损坏，并会直接失败。删除后重新注册同一路径会生成新的 Workspace id，且不会自动重新接纳保留下来的 Session。
+
 ## 模型体验
 
 ### Workspace 记录与会话记账
diff --git a/packages/workspace/workspace/src/index.ts b/packages/workspace/workspace/src/index.ts
index 2699365608..5172c63805 100644
--- a/packages/workspace/workspace/src/index.ts
+++ b/packages/workspace/workspace/src/index.ts
@@ -109,6 +109,7 @@ export class WorkspaceRegistry extends Service {
     this.global = domain.global
     this.state = domain.global.get()
 
+    await this.recoverPendingMutation()
     this.validateStoredState(this.state)
     if (!this.state.initialized) {
       const headers = await this.ctx.sessionPersistence.list()
@@ -218,10 +219,28 @@ export class WorkspaceRegistry extends Service {
     }
     const entity = new WorkspaceEntity(this.host, id, record)
     this.entities.set(id, entity)
+    const pendingState: WorkspaceDomainState = {
+      ...state,
+      pendingMutation: { operation: 'create', workspaceId: id },
+    }
+    try {
+      await this.setState(pendingState)
+    } catch (error) {
+      this.entities.delete(id)
+      throw error
+    }
     try {
       await table.put(id, record)
     } catch (error) {
       this.entities.delete(id)
+      try {
+        await this.setState(state)
+      } catch (rollbackError) {
+        throw new AggregateError(
+          [error, rollbackError],
+          `workspace '${id}' record write and pending-marker rollback both failed`,
+        )
+      }
       throw error
     }
 
@@ -232,10 +251,17 @@ export class WorkspaceRegistry extends Service {
       try {
         await table.delete(id)
       } catch (rollbackError) {
-        this.entities.set(id, entity)
         throw new AggregateError(
           [error, rollbackError],
-          `workspace '${id}' was stored but its registry order and rollback both failed`,
+          `workspace '${id}' order write and record rollback both failed; the pending marker remains recoverable`,
+        )
+      }
+      try {
+        await this.setState(state)
+      } catch (rollbackError) {
+        throw new AggregateError(
+          [error, rollbackError],
+          `workspace '${id}' order write and pending-marker rollback both failed`,
         )
       }
       throw error
@@ -251,7 +277,10 @@ export class WorkspaceRegistry extends Service {
       initialized: true,
       workspaceIds: state.workspaceIds.filter(workspaceId => workspaceId !== id),
     }
-    await this.setState(nextState)
+    await this.setState({
+      ...nextState,
+      pendingMutation: { operation: 'delete', workspaceId: id },
+    })
     this.entities.delete(id)
     try {
       await this.requireTable().delete(id)
@@ -260,6 +289,10 @@ export class WorkspaceRegistry extends Service {
       try {
         await this.setState(state)
       } catch (rollbackError) {
+        // The durable marker still says to finish deletion, so the cache must
+        // agree with that recoverable direction rather than republish a row
+        // absent from the persisted order.
+        this.entities.delete(id)
         throw new AggregateError(
           [error, rollbackError],
           `workspace '${id}' record deletion and registry-order rollback both failed`,
@@ -267,9 +300,38 @@ export class WorkspaceRegistry extends Service {
       }
       throw error
     }
+    try {
+      await this.setState(nextState)
+    } catch (error) {
+      // The deletion committed at the table write and was already published
+      // to Host streams. Keep the durable marker for startup recovery rather
+      // than reporting failure after the requested state became true.
+      this.ctx.logger.warn(
+        `workspace '${id}' was deleted but its pending marker could not be cleared: ${String(error)}`,
+      )
+    }
     return true
   }
 
+  /**
+   * Complete the one mutation explicitly named by durable state. Unexplained
+   * order/table divergence still reaches {@link validateStoredState} and
+   * fails loud; this path never infers provenance from shape alone.
+   */
+  private async recoverPendingMutation(): Promise<void> {
+    const state = this.requireState()
+    const pending = state.pendingMutation
+    if (pending === undefined) return
+    if (state.workspaceIds.includes(pending.workspaceId)) {
+      throw new Error(
+        `workspace domain is inconsistent: pending ${pending.operation} workspace `
+        + `'${pending.workspaceId}' is still present in registry order`,
+      )
+    }
+    await this.requireTable().delete(pending.workspaceId)
+    await this.setState({ initialized: state.initialized, workspaceIds: state.workspaceIds })
+  }
+
   private async bootstrap(headers: readonly SessionHeader[]): Promise<void> {
     const table = this.requireTable()
     const state = this.requireState()
@@ -493,7 +555,12 @@ export class WorkspaceRegistry extends Service {
   }
 
   private enqueueOperation<T>(operation: () => Promise<T>): Promise<T> {
-    const result = this.operationTail.then(operation)
+    const result = this.operationTail.then(async () => {
+      // A committed delete may leave only its marker cleanup pending. Retry
+      // recovery before another create/delete can overwrite that provenance.
+      await this.recoverPendingMutation()
+      return await operation()
+    })
     this.operationTail = result.then(() => {}, () => {})
     return result
   }
diff --git a/packages/workspace/workspace/src/spec.ts b/packages/workspace/workspace/src/spec.ts
index 8df908949a..7b1a6a41d0 100644
--- a/packages/workspace/workspace/src/spec.ts
+++ b/packages/workspace/workspace/src/spec.ts
@@ -29,6 +29,16 @@ export const workspaceRecord = z.object({
 /** One stored workspace record, inferred from {@link workspaceRecord}. */
 export type WorkspaceRecord = z.infer<typeof workspaceRecord>
 
+/**
+ * Recoverable two-write mutation marker. The marker is persisted before the
+ * record/order pair can diverge, so startup can distinguish an interrupted
+ * registry operation from unexplained medium corruption.
+ */
+const workspacePendingMutation = z.discriminatedUnion('operation', [
+  z.object({ operation: z.literal('create'), workspaceId }),
+  z.object({ operation: z.literal('delete'), workspaceId }),
+])
+
 /**
  * Durable registry state. `initialized` distinguishes a valid empty registry
  * from one that still needs the header-only history bootstrap;
@@ -37,6 +47,7 @@ export type WorkspaceRecord = z.infer<typeof workspaceRecord>
 export const workspaceDomainState = z.object({
   initialized: z.boolean(),
   workspaceIds: z.array(workspaceId),
+  pendingMutation: workspacePendingMutation.optional(),
 })
 
 /** Durable registry state inferred from {@link workspaceDomainState}. */
diff --git a/packages/workspace/workspace/tests/workspace.spec.ts b/packages/workspace/workspace/tests/workspace.spec.ts
index ed08b5ba50..4576155f3b 100644
--- a/packages/workspace/workspace/tests/workspace.spec.ts
+++ b/packages/workspace/workspace/tests/workspace.spec.ts
@@ -89,7 +89,7 @@ async function storageContext(pool: MemoryMediaPool, backend: StorageBackend = n
 /** Backend wrapper that injects one selected bootstrap write failure. */
 function selectiveFailureBackend(
   pool: MemoryMediaPool,
-  failure: { putAt?: number; deleteAt?: number; globalAt?: number },
+  failure: { putAt?: number; deleteAt?: number; globalAt?: number | readonly number[] },
 ): StorageBackend {
   const inner = new MemoryStorageBackend(pool)
   let puts = 0
@@ -113,7 +113,8 @@ function selectiveFailureBackend(
           },
           setGlobal: async (value) => {
             globals += 1
-            if (globals === failure.globalAt) throw new Error('selected bootstrap marker failure')
+            const failAt = Array.isArray(failure.globalAt) ? failure.globalAt : [failure.globalAt]
+            if (failAt.includes(globals)) throw new Error('selected bootstrap marker failure')
             await unit.setGlobal(value)
           },
           close: () => unit.close(),
@@ -394,19 +395,34 @@ describe('WorkspaceRegistry create and lookup', () => {
 
   it('rolls back the provisional cache when the record write fails', async () => {
     const dir = await makeDir('write-failure')
-    const result = await harness()
-    result.pool.failNextWrites = 1
-    await expect(result.registry.create(dir)).rejects.toThrow(/injected/)
+    const pool = new MemoryMediaPool()
+    const result = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { putAt: 1 }),
+    })
+    await expect(result.registry.create(dir)).rejects.toThrow(/selected bootstrap put failure/)
     expect(result.registry.list()).toEqual([])
     expect(await result.registry.create(dir)).toBeDefined()
   })
 
+  it('does not publish a Workspace when its pending marker cannot be written', async () => {
+    const dir = await makeDir('pending-marker-write-failure')
+    const pool = new MemoryMediaPool()
+    const result = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { globalAt: 2 }),
+    })
+    await expect(result.registry.create(dir)).rejects.toThrow(/selected bootstrap marker failure/)
+    expect(result.registry.list()).toEqual([])
+    expect(pool.media.get('workspace')!.tables.get('workspaces')?.size ?? 0).toBe(0)
+  })
+
   it('rolls back a record when registry-order persistence fails', async () => {
     const dir = await makeDir('order-write-failure')
     const pool = new MemoryMediaPool()
     const result = await harness({
       pool,
-      backend: selectiveFailureBackend(pool, { globalAt: 2 }),
+      backend: selectiveFailureBackend(pool, { globalAt: 3 }),
     })
     await expect(result.registry.create(dir)).rejects.toThrow(/marker failure/)
     expect(result.registry.list()).toEqual([])
@@ -418,12 +434,38 @@ describe('WorkspaceRegistry create and lookup', () => {
     const pool = new MemoryMediaPool()
     const result = await harness({
       pool,
-      backend: selectiveFailureBackend(pool, { globalAt: 2, deleteAt: 1 }),
+      backend: selectiveFailureBackend(pool, { globalAt: 3, deleteAt: 1 }),
     })
     await expect(result.registry.create(dir)).rejects.toBeInstanceOf(AggregateError)
     expect(pool.media.get('workspace')!.tables.get('workspaces')!.size).toBe(1)
   })
 
+  it('reports a record write and pending-marker rollback failure together', async () => {
+    const dir = await makeDir('record-marker-rollback-failure')
+    const pool = new MemoryMediaPool()
+    const result = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { putAt: 1, globalAt: 3 }),
+    })
+    await expect(result.registry.create(dir)).rejects.toBeInstanceOf(AggregateError)
+    expect(storedState(pool)).toMatchObject({
+      pendingMutation: { operation: 'create' },
+    })
+  })
+
+  it('reports an order write and pending-marker rollback failure together', async () => {
+    const dir = await makeDir('order-marker-rollback-failure')
+    const pool = new MemoryMediaPool()
+    const result = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { globalAt: [3, 4] }),
+    })
+    await expect(result.registry.create(dir)).rejects.toBeInstanceOf(AggregateError)
+    expect(storedState(pool)).toMatchObject({
+      pendingMutation: { operation: 'create' },
+    })
+  })
+
   it('deletes only the registration and leaves its directory and session headers untouched', async () => {
     const dir = await makeDir('delete-registration')
     const result = await harness({ sessions: [header('kept-session', dir)] })
@@ -440,6 +482,11 @@ describe('WorkspaceRegistry create and lookup', () => {
     expect(result.list).toHaveBeenCalledTimes(1)
     expect(result.load).not.toHaveBeenCalled()
     expect(result.inspect).not.toHaveBeenCalled()
+
+    const reregistered = await result.registry.create(dir)
+    expect(reregistered.id).not.toBe(workspace.id)
+    expect(reregistered.path).toBe(dir)
+    expect(reregistered.sessionIds).toEqual([])
   })
 
   it('rolls registry order and cache back when record deletion fails', async () => {
@@ -458,11 +505,58 @@ describe('WorkspaceRegistry create and lookup', () => {
     expect(storedRecord(pool, workspace.id)).toMatchObject({ path: dir })
   })
 
+  it('commits deletion and leaves a recoverable marker when marker cleanup fails', async () => {
+    const dir = await makeDir('delete-marker-cleanup')
+    const pool = new MemoryMediaPool()
+    const first = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { globalAt: 5 }),
+    })
+    const workspace = await first.registry.create(dir)
+
+    await expect(first.registry.delete(workspace.id)).resolves.toBe(true)
+    expect(first.registry.list()).toEqual([])
+    expect(storedState(pool)).toEqual({
+      initialized: true,
+      workspaceIds: [],
+      pendingMutation: { operation: 'delete', workspaceId: workspace.id },
+    })
+    const reregistered = await first.registry.create(dir)
+    expect(reregistered.id).not.toBe(workspace.id)
+    expect(storedState(pool)).toEqual({
+      initialized: true,
+      workspaceIds: [reregistered.id],
+    })
+    await first.fiber.dispose()
+
+    const restarted = await harness({ pool })
+    expect(restarted.registry.list().map(item => item.id)).toEqual([reregistered.id])
+  })
+
+  it('keeps the failed deletion unpublished when record and order rollback both fail', async () => {
+    const dir = await makeDir('delete-double-failure')
+    const pool = new MemoryMediaPool()
+    const result = await harness({
+      pool,
+      backend: selectiveFailureBackend(pool, { deleteAt: 1, globalAt: 5 }),
+    })
+    const workspace = await result.registry.create(dir)
+
+    await expect(result.registry.delete(workspace.id)).rejects.toBeInstanceOf(AggregateError)
+    expect(result.registry.get(workspace.id)).toBeUndefined()
+    expect(storedState(pool)).toMatchObject({
+      workspaceIds: [],
+      pendingMutation: { operation: 'delete', workspaceId: workspace.id },
+    })
+  })
+
   it('rejects table access before the registry has started', async () => {
     const dir = await makeDir('unstarted')
     const registry = new WorkspaceRegistry(new Context())
     await expect(registry.create(dir)).rejects.toThrow(/not started/)
     expect(() => registry.list()).toThrow(/not started/)
+    const internals = registry as unknown as { requireTable(): unknown }
+    expect(() => internals.requireTable()).toThrow(/not started/)
   })
 })
 
@@ -650,6 +744,49 @@ describe('header-validated membership projection', () => {
     internals.entities.delete(workspace.id)
     expect(() => result.registry.list()).toThrow(/references missing workspace/)
   })
+
+  it('recovers only an explicitly marked interrupted create or delete', async () => {
+    const createDir = await makeDir('pending-create')
+    const deleteDir = await makeDir('pending-delete')
+    const createId = WorkspaceId('00000000-0000-4000-8000-000000000004')
+    const deleteId = WorkspaceId('00000000-0000-4000-8000-000000000005')
+
+    const interruptedCreate = storedPool(
+      [[createId, record(createDir, [])]],
+      {
+        initialized: true,
+        workspaceIds: [],
+        pendingMutation: { operation: 'create', workspaceId: createId },
+      },
+    )
+    const createRecovery = await harness({ pool: interruptedCreate })
+    expect(createRecovery.registry.list()).toEqual([])
+    expect(interruptedCreate.media.get('workspace')!.tables.get('workspaces')!.has(createId)).toBe(false)
+    expect(storedState(interruptedCreate)).toEqual({ initialized: true, workspaceIds: [] })
+
+    const interruptedDelete = storedPool(
+      [[deleteId, record(deleteDir, [])]],
+      {
+        initialized: true,
+        workspaceIds: [],
+        pendingMutation: { operation: 'delete', workspaceId: deleteId },
+      },
+    )
+    const deleteRecovery = await harness({ pool: interruptedDelete })
+    expect(deleteRecovery.registry.list()).toEqual([])
+    expect(interruptedDelete.media.get('workspace')!.tables.get('workspaces')!.has(deleteId)).toBe(false)
+    expect(storedState(interruptedDelete)).toEqual({ initialized: true, workspaceIds: [] })
+
+    const corruptPending = storedPool(
+      [[deleteId, record(deleteDir, [])]],
+      {
+        initialized: true,
+        workspaceIds: [deleteId],
+        pendingMutation: { operation: 'delete', workspaceId: deleteId },
+      },
+    )
+    await expect(harness({ pool: corruptPending })).rejects.toThrow(/still present in registry order/)
+  })
 })
 
 describe('workspace mutation and status', () => {

From be80eb04ad4876dd3c60e000d9b7e1836bed3a1f Mon Sep 17 00:00:00 2001
From: Ubuntu <ubuntu@localhost.localdomain>
Date: Mon, 27 Jul 2026 14:45:54 +0800
Subject: [PATCH 16/27] ci: retrigger after runner-group policy fix


From fe246e4a0a14a4ce154e05e52188bac098dea80c Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 15:17:48 +0800
Subject: [PATCH 17/27] =?UTF-8?q?ci:=20failover=20round=20=E2=80=94=20aggr?=
 =?UTF-8?q?egate=20follows=20the=20selector,=20tighter=20shared-VM=20bound?=
 =?UTF-8?q?s?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- all-checks-passed now resolves its pool through the same
  DSH_CI_FAILOVER expression as the worker jobs it aggregates.
  Pinned to the hosted pool it would leave the branch-protection
  verdict queued on the failed pool after every failover job passed —
  observed live during the 2026-07-27 outage as a required check
  looping against dead capacity.
- Coverage worker bound under failover drops 12 → 8 and snapshot
  concurrency 16 → 12: the pool now runs six always-on instances (the
  spare tier was retired), so worst case is 6 × 8 = 48 coverage
  workers on the shared 64-core VM.
---
 .github/workflows/ci.yml | 22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 5afa5d7f62..df3d386e39 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -102,11 +102,12 @@ jobs:
           || 'dsh-enterprise-ubuntu-24-04-32core-test' }}
     name: node 24 / coverage
     env:
-      # Failover halves the worker bound: the hosted 32-core runner is
+      # Failover shrinks the worker bound: the hosted 32-core runner is
       # exclusive to one job, but the failover pool shares one 64-core VM
-      # across four runner instances, and the timing-sensitive process
-      # suites have documented aggregate-contention failures.
-      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '12' || '24' }}
+      # across six always-on runner instances, and the timing-sensitive
+      # process suites have documented aggregate-contention failures.
+      # 8 × 6 instances = 48 workers worst case on 64 cores.
+      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '8' || '24' }}
       DSH_GATE_CONCURRENCY: '8'
     steps:
       - uses: actions/checkout@v6
@@ -160,7 +161,7 @@ jobs:
       DSH_NODE_COMPAT_SKIP_TYPECHECK: '1'
       DSH_PUBLINT_CONCURRENCY: '8'
       # Failover halves snapshot concurrency for the shared 64-core VM.
-      DSH_SNAPSHOT_MAX_CONCURRENCY: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '16' || '32' }}
+      DSH_SNAPSHOT_MAX_CONCURRENCY: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '12' || '32' }}
     steps:
       - uses: actions/checkout@v6
         with:
@@ -765,8 +766,15 @@ jobs:
   # 'cancelled' and 'skipped'.
   all-checks-passed:
     name: all checks passed
-    # The required verdict must not add a separate standard-hosted billing dependency.
-    runs-on: dsh-enterprise-ubuntu-latest-32core-test
+    # The required verdict must not add a separate standard-hosted billing
+    # dependency — and it must follow the failover selector like the worker
+    # jobs it aggregates: if it stayed pinned to the hosted pool, every
+    # failover-passed run would still leave the branch-protection verdict
+    # queued forever on the failed pool.
+    runs-on: >-
+      ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
+          || 'dsh-enterprise-ubuntu-latest-32core-test' }}
     needs: [node-24, node-24-coverage, node-24-consumers, node-compat, python-sdk, windows]
     if: always() && github.event_name == 'pull_request'
     steps:

From aedf7fbf349df99a89faa72ddc17ec95ea2aff53 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 15:27:40 +0800
Subject: [PATCH 18/27] docs(i18n): keep the runbook link target identical
 across the pair

The pairing gate requires link target #9 to be byte-identical between
the language sides; my earlier 'fix' pointed the zh side at the zh
runbook and broke the contract. Reverted to the shared target and
re-recorded the pairing hash.
---
 .../2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml   | 2 +-
 .../2026-07-22-evidence-based-larger-hosted-runners.zh.md       | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 5ebd95248c..99cabc76bb 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
 2026-07-22-evidence-based-larger-hosted-runners.md: 5b399be5571ddaf1f775ba43a2233198b8e09b18
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: 40970ec33c1a16af85ea47be3fc932209efdd654
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: f77516e2375bfc0557679d05fd275bd9cee7d8eb
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index 40970ec33c..f77516e237 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -52,7 +52,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
-另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.zh.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
 
 ## 曾考虑的替代方案
 

From caabf8f671d194194b4d8b876566b847d2f73ddf Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 15:36:11 +0800
Subject: [PATCH 19/27] ci: dependabot stays hosted under failover; runbook
 matches shipped bounds
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- All four failover selectors (three workers + the verdict job) and the
  paired env/cache expressions now exclude dependabot[bot]: under
  failover, dependency-supplied code keeps queueing for the hosted pool
  instead of executing on the persistent VM. A delayed Dependabot PR
  during an outage is an acceptable cost; dependency code on the
  privileged host is not.
- Runbook (both languages): records the shipped failover bounds
  (coverage 8, snapshots 12, sized for six instances) and documents
  that the verdict job follows the selector too — operators previously
  had no explanation for a verdict queued after all workers passed.
- Local static gate green: 32 passed, 0 failed (translation pairing
  519 pairs consistent).
---
 .../2026-07-26-ci-failover-runbook.i18n.yaml       |  4 ++--
 .../process/2026-07-26-ci-failover-runbook.md      |  6 +++---
 .../process/2026-07-26-ci-failover-runbook.zh.md   |  6 +++---
 .github/workflows/ci.yml                           | 14 +++++++++-----
 4 files changed, 17 insertions(+), 13 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index efb5fdd1cc..26f7f23f85 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: 0bce83e0f9c842fa3dd73ae9c0a3eefc0975cdae
-2026-07-26-ci-failover-runbook.zh.md: 4bc6c67bab754ad3f0127557b0d5e04f7934c8a2
+2026-07-26-ci-failover-runbook.md: ab1a727caa045d2074a9c577416f96f45efcd0aa
+2026-07-26-ci-failover-runbook.zh.md: 5dfaca0c1c0f443307bea28bb6544385ebb68bb7
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index 0bce83e0f9..ab1a727caa 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -6,11 +6,11 @@ English | [中文](2026-07-26-ci-failover-runbook.zh.md)
 
 ## Problem
 
-The three required Linux jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) run on the hosted enterprise 32-core pools. When those pools degrade — jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. An outage therefore needs a switch a repository admin can throw without merging anything.
+The three required Linux worker jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) and the required verdict job that aggregates them (`all checks passed`) run on the hosted enterprise 32-core pools. When those pools degrade — jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. An outage therefore needs a switch a repository admin can throw without merging anything.
 
 ## Decision
 
-Each of the three required Linux jobs resolves its runner pool through the `DSH_CI_FAILOVER` repository variable. Unset (normal), they run on the hosted enterprise pools. Set to `selfhosted` by a repository admin, all three retarget onto the in-house self-hosted `vm-backup` pool, coverage and snapshot concurrency drop to shared-VM bounds, and the hosted-path pnpm cache restores are skipped. The switch is admin-only repository state, not a merge, so it works while every check is red. The in-house pool's readiness is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push.
+Each of the three required Linux worker jobs — and the `all checks passed` verdict job, which would otherwise stay queued on the failed pool even after every worker passed — resolves its runner pool through the `DSH_CI_FAILOVER` repository variable. Unset (normal), they run on the hosted enterprise pools. Set to `selfhosted` by a repository admin, all four retarget onto the in-house self-hosted `vm-backup` pool, coverage and snapshot concurrency drop to shared-VM bounds, and the hosted-path pnpm cache restores are skipped. The switch is admin-only repository state, not a merge, so it works while every check is red. The in-house pool's readiness is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push.
 
 ### What the in-house pool is
 
@@ -20,7 +20,7 @@ Each of the three required Linux jobs resolves its runner pool through the `DSH_
 
 1. Repository **Settings → Secrets and variables → Actions → Variables → New repository variable**: name `DSH_CI_FAILOVER`, value `selfhosted`.
 2. Retrigger the required jobs so they re-resolve their pool. Jobs already **queued** for the hosted labels do not retarget and cannot be re-run in place, so for the documented indefinite-queue outage, cancel the stuck run and re-run all jobs, or push a new commit; "Re-run failed jobs" only helps once a job has actually failed rather than queued.
-3. That is the entire switch. Under failover the workflow also, automatically: halves `DSH_COVERAGE_MAX_WORKERS` to 12 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 16 (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
+3. That is the entire switch. Under failover the workflow also, automatically: drops `DSH_COVERAGE_MAX_WORKERS` to 8 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 12 (sized for six always-on instances: worst case 6 × 8 = 48 coverage workers on the 64-core VM) (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
 
 ### Capacity during failover
 
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index 4bc6c67bab..5dfaca0c1c 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -6,11 +6,11 @@ Status: implemented
 
 ## 问题
 
-[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）运行在托管的企业级 32 核池上。当这些托管池发生故障——作业无限排队、企业标签消失或 GitHub 侧容量故障——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。因此故障需要一个仓库管理员无需合并任何代码即可触发的开关。
+[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 工作作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）以及聚合它们的必需判定作业（`all checks passed`）运行在托管的企业级 32 核池上。当这些托管池发生故障——作业无限排队、企业标签消失或 GitHub 侧容量故障——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。因此故障需要一个仓库管理员无需合并任何代码即可触发的开关。
 
 ## 决策
 
-三个必需的 Linux 作业各自通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。变量不存在（正常）时它们运行在托管企业池上；由仓库管理员设为 `selfhosted` 时，三者全部切换到公司自有的自托管 `vm-backup` 池，coverage 与 snapshot 的并发降到共享虚拟机上限，并跳过托管路径的 pnpm 缓存恢复。这个开关是仅限管理员的仓库状态而非一次合并，因此在所有检查都是红色时仍然有效。自有池的就绪状态由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。
+三个必需的 Linux 工作作业——以及 `all checks passed` 判定作业（若不随切换，即使全部工作作业通过，它仍会滞留在故障池的队列中）——各自通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。变量不存在（正常）时它们运行在托管企业池上；由仓库管理员设为 `selfhosted` 时，四者全部切换到公司自有的自托管 `vm-backup` 池，coverage 与 snapshot 的并发降到共享虚拟机上限，并跳过托管路径的 pnpm 缓存恢复。这个开关是仅限管理员的仓库状态而非一次合并，因此在所有检查都是红色时仍然有效。自有池的就绪状态由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。
 
 ### 自有池是什么
 
@@ -20,7 +20,7 @@ Status: implemented
 
 1. 仓库 **Settings → Secrets and variables → Actions → Variables → New repository variable**：名称 `DSH_CI_FAILOVER`，值 `selfhosted`。
 2. 重新触发必需作业，使其重新解析运行器池。已经为托管标签**排队**的作业不会重定向，也无法原地 re-run，因此对于本手册所述的无限排队故障，应取消卡住的运行并 re-run all jobs，或推送一个新提交；“Re-run failed jobs”只有在作业真正失败（而非仍在排队）时才有用。
-3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 12、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 16（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
+3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 8、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 12（按 6 个常驻实例定容：最坏 6 × 8 = 48 个覆盖率工作进程对 64 核）（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
 
 ### 切换期间的容量
 
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index df3d386e39..8306b7e034 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -44,6 +44,7 @@ jobs:
     if: github.event_name == 'pull_request'
     runs-on: >-
       ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && github.event.pull_request.user.login != 'dependabot[bot]'
           && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
           || 'dsh-enterprise-ubuntu-latest-32core-test' }}
     name: node 24 / static
@@ -60,7 +61,7 @@ jobs:
       # compression and upload on the paid latency-critical path. Skipped
       # under failover — see the coverage lane's identical rationale.
       - uses: actions/cache/restore@v4
-        if: vars.DSH_CI_FAILOVER != 'selfhosted'
+        if: vars.DSH_CI_FAILOVER != 'selfhosted' || github.event.pull_request.user.login == 'dependabot[bot]'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
@@ -98,6 +99,7 @@ jobs:
     if: github.event_name == 'pull_request'
     runs-on: >-
       ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && github.event.pull_request.user.login != 'dependabot[bot]'
           && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
           || 'dsh-enterprise-ubuntu-24-04-32core-test' }}
     name: node 24 / coverage
@@ -107,7 +109,7 @@ jobs:
       # across six always-on runner instances, and the timing-sensitive
       # process suites have documented aggregate-contention failures.
       # 8 × 6 instances = 48 workers worst case on 64 cores.
-      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '8' || '24' }}
+      DSH_COVERAGE_MAX_WORKERS: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && github.event.pull_request.user.login != 'dependabot[bot]' && '8' || '24' }}
       DSH_GATE_CONCURRENCY: '8'
     steps:
       - uses: actions/checkout@v6
@@ -118,7 +120,7 @@ jobs:
       # serves warm installs directly, and this hosted-path restore would
       # spend ~52 s pulling ~180 MB into a path pnpm never reads there.
       - uses: actions/cache/restore@v4
-        if: vars.DSH_CI_FAILOVER != 'selfhosted'
+        if: vars.DSH_CI_FAILOVER != 'selfhosted' || github.event.pull_request.user.login == 'dependabot[bot]'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
@@ -151,6 +153,7 @@ jobs:
     if: github.event_name == 'pull_request'
     runs-on: >-
       ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && github.event.pull_request.user.login != 'dependabot[bot]'
           && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
           || 'dsh-enterprise-ubuntu-latest-32core-test' }}
     name: node 24 / snapshots and artifacts
@@ -161,7 +164,7 @@ jobs:
       DSH_NODE_COMPAT_SKIP_TYPECHECK: '1'
       DSH_PUBLINT_CONCURRENCY: '8'
       # Failover halves snapshot concurrency for the shared 64-core VM.
-      DSH_SNAPSHOT_MAX_CONCURRENCY: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && '12' || '32' }}
+      DSH_SNAPSHOT_MAX_CONCURRENCY: ${{ vars.DSH_CI_FAILOVER == 'selfhosted' && github.event.pull_request.user.login != 'dependabot[bot]' && '12' || '32' }}
     steps:
       - uses: actions/checkout@v6
         with:
@@ -177,7 +180,7 @@ jobs:
 
       # Skipped under failover — see the coverage lane's identical rationale.
       - uses: actions/cache/restore@v4
-        if: vars.DSH_CI_FAILOVER != 'selfhosted'
+        if: vars.DSH_CI_FAILOVER != 'selfhosted' || github.event.pull_request.user.login == 'dependabot[bot]'
         with:
           path: /home/runner/.local/share/pnpm/store/v11
           key: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-pnpm-${{ hashFiles('pnpm-lock.yaml') }}
@@ -773,6 +776,7 @@ jobs:
     # queued forever on the failed pool.
     runs-on: >-
       ${{ vars.DSH_CI_FAILOVER == 'selfhosted'
+          && github.event.pull_request.user.login != 'dependabot[bot]'
           && fromJSON('["self-hosted", "linux", "x64", "vm-backup"]')
           || 'dsh-enterprise-ubuntu-latest-32core-test' }}
     needs: [node-24, node-24-coverage, node-24-consumers, node-compat, python-sdk, windows]

From d606ab6877a4516a422ab45e4160873843aa9a55 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 15:47:19 +0800
Subject: [PATCH 20/27] =?UTF-8?q?ci:=20standby=20archive=20base=20ref;=20r?=
 =?UTF-8?q?unbook=20=E2=80=94=20dependabot=20exception,=20clean=20bootstra?=
 =?UTF-8?q?p,=20variable=20trust=20model?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- serial-linux-selfhosted now fetches depth 2 and passes
  DSH_ARCHIVE_BASE_REF=github.event.before, running the same
  frozen-archive comparison as serial-linux instead of diffing the
  new manifest against itself.
- Runbook (both languages): documents the deliberate dependabot
  exception (queued-on-hosted during failover is expected, not a
  failed switch); corrects the emergency-capacity bootstrap to
  exclude .runner/.credentials when cloning a runner directory; and
  replaces the 'admin-only' variable claim with the accurate
  trust-model statement — repository variables are writer-manageable,
  which in this private fork-disabled repo with an all-workflows
  runner group is routing among members, not an escalation.
Static gate green locally: 32 passed, 0 failed.
---
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml      | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.md | 8 ++++++--
 .../process/2026-07-26-ci-failover-runbook.zh.md          | 8 ++++++--
 .github/workflows/ci.yml                                  | 6 ++++++
 4 files changed, 20 insertions(+), 6 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index 26f7f23f85..7b8d08befe 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: ab1a727caa045d2074a9c577416f96f45efcd0aa
-2026-07-26-ci-failover-runbook.zh.md: 5dfaca0c1c0f443307bea28bb6544385ebb68bb7
+2026-07-26-ci-failover-runbook.md: 55c1350593562d62463e751451d50a79cf45a1d6
+2026-07-26-ci-failover-runbook.zh.md: 13977b78244440a23722d089849ea7ff6b751aea
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index ab1a727caa..55c1350593 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -22,9 +22,13 @@ Each of the three required Linux worker jobs — and the `all checks passed` ver
 2. Retrigger the required jobs so they re-resolve their pool. Jobs already **queued** for the hosted labels do not retarget and cannot be re-run in place, so for the documented indefinite-queue outage, cancel the stuck run and re-run all jobs, or push a new commit; "Re-run failed jobs" only helps once a job has actually failed rather than queued.
 3. That is the entire switch. Under failover the workflow also, automatically: drops `DSH_COVERAGE_MAX_WORKERS` to 8 and `DSH_SNAPSHOT_MAX_CONCURRENCY` to 12 (sized for six always-on instances: worst case 6 × 8 = 48 coverage workers on the 64-core VM) (shared-VM contention bounds), and skips the hosted-path pnpm cache restores (the VM's persistent store serves warm installs).
 
-### Capacity during failover
+#**Dependabot exception.** All four selectors deliberately exclude `dependabot[bot]`: under failover, Dependabot PRs stay queued for the hosted pool rather than executing dependency-supplied code on the persistent VM. A Dependabot PR that remains queued during an outage is expected behavior, not a failed switch; it completes when the hosted pool recovers.
 
-Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner) — cloning an existing runner directory and running `config.sh` takes about a minute per instance.
+**Who can flip the variable.** GitHub's API lets any collaborator with write access manage repository variables, so the switch is writer-level, not strictly admin-only. In this repository's trust model that is not an escalation: the runner group admits all workflows of this private, fork-disabled repository (a deliberate trade to make PR-ref failover possible at all), so any writer could already reach the VM by pushing a branch workflow. The boundary against untrusted code is repository membership; the variable only routes work for members.
+
+## Capacity during failover
+
+Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` — then run `config.sh`; copying `.runner`/`.credentials` verbatim makes `config.sh` refuse with "already configured". About a minute per instance.
 
 
 ### Switch back
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index 5dfaca0c1c..13977b7824 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -22,9 +22,13 @@ Status: implemented
 2. 重新触发必需作业，使其重新解析运行器池。已经为托管标签**排队**的作业不会重定向，也无法原地 re-run，因此对于本手册所述的无限排队故障，应取消卡住的运行并 re-run all jobs，或推送一个新提交；“Re-run failed jobs”只有在作业真正失败（而非仍在排队）时才有用。
 3. 切换到此完成。故障切换状态下工作流还会自动：把 `DSH_COVERAGE_MAX_WORKERS` 降为 8、`DSH_SNAPSHOT_MAX_CONCURRENCY` 降为 12（按 6 个常驻实例定容：最坏 6 × 8 = 48 个覆盖率工作进程对 64 核）（共享虚拟机的争抢上限），并跳过托管路径的 pnpm 缓存恢复（虚拟机的持久 store 直接提供热安装）。
 
-### 切换期间的容量
+#**Dependabot 例外。**四个选择器都刻意排除了 `dependabot[bot]`：故障切换期间，Dependabot 拉取请求继续在托管池排队，而不是把依赖方提供的代码放到持久化虚拟机上执行。故障期间 Dependabot PR 持续排队是预期行为而非切换失败；托管池恢复后它会自行完成。
 
-6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例——复制现有 runner 目录再跑 `config.sh`，每个约一分钟。
+**谁能扳动这个变量。**GitHub 的 API 允许任何具有写权限的协作者管理仓库变量，因此该开关实际是写者级而非严格的管理员级。在本仓库的信任模型下这并不构成越权：runner group 接纳本私有、禁 fork 仓库的全部工作流（这是让 PR 引用的故障切换得以成立的刻意取舍），因此任何写者本就可以通过推送分支工作流触达这台虚拟机。抵御不可信代码的边界是仓库成员资格；变量只是为成员路由工作。
+
+## 切换期间的容量
+
+6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`——再跑 `config.sh`；原样拷贝 `.runner`/`.credentials` 会使 `config.sh` 以 "already configured" 拒绝。每个约一分钟。
 
 
 ### 切回
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 8306b7e034..8c3b854ae6 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -423,7 +423,12 @@ jobs:
     name: serial / linux (self-hosted standby)
     runs-on: [self-hosted, linux, x64, vm-backup]
     steps:
+      # fetch-depth 2 + DSH_ARCHIVE_BASE_REF below: same frozen-archive
+      # comparison as serial-linux — without the prior commit the archive
+      # verifier defaults to HEAD and compares the new manifest with itself.
       - uses: actions/checkout@v6
+        with:
+          fetch-depth: 2
 
       - uses: actions/setup-node@v6
         with:
@@ -440,6 +445,7 @@ jobs:
 
       - name: Run complete unsharded primary Node CI serially
         env:
+          DSH_ARCHIVE_BASE_REF: ${{ github.event.before }}
           DSH_COVERAGE_MAX_WORKERS: '1'
           DSH_E2E_MAX_WORKERS: '1'
           DSH_ESLINT_CACHE: '1'

From 4701373fc2cf87e14fb53885f9fa434b54ca0e20 Mon Sep 17 00:00:00 2001
From: NI0317 <stniii317@gmail.com>
Date: Mon, 27 Jul 2026 15:52:35 +0800
Subject: [PATCH 21/27] fix(workspace): remove transient duplicate warning

---
 ...-workspace-registration-deletion.i18n.yaml |  4 +-
 ...6-07-27-workspace-registration-deletion.md |  4 +-
 ...7-27-workspace-registration-deletion.zh.md |  4 +-
 apps/web/tests/workspace-management.e2e.ts    | 88 +++++++++++++++++++
 .../runtime/src/client/workspaces/manager.ts  | 15 +++-
 .../src/client/WorkspaceBrowser.tsx           | 15 +++-
 .../src/client/WorkspacePicker.tsx            |  2 +-
 .../tests/workspace-browser.spec.tsx          |  7 +-
 .../tests/workspace-picker.spec.tsx           | 33 +++++--
 9 files changed, 155 insertions(+), 17 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
index 93c78373c6..d576fb10e5 100644
--- a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
-2026-07-27-workspace-registration-deletion.md: 58ae5c4bef2cf1cb0a0158eda5eb37daf2e9703d
-2026-07-27-workspace-registration-deletion.zh.md: 7a79a1ccc53a0d4fd7e5ab453239ade955313c6e
+2026-07-27-workspace-registration-deletion.md: 8168b0832ca39e6023f6981815ffe758b5695361
+2026-07-27-workspace-registration-deletion.zh.md: b0df6982ac81426a5b0ce2f0e2b0e744212e3f5b
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
index 58ae5c4bef..8168b0832c 100644
--- a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.md
@@ -28,6 +28,8 @@ Create and delete write a durable `pendingMutation` before their record/order pa
 
 `WorkspaceManager` treats both `host/workspace-changed` and `host/workspace-removed` as ordered deltas replayed over an in-flight `workspace.list` response. A successful unary delete removes the row immediately instead of waiting for its own stream echo. Removal is idempotent, and a process-local tombstone rejects late changed frames or stale baseline rows for the never-reused Workspace id. A reconnect still refreshes from `workspace.list`; Session state is never pruned by a Workspace delta.
 
+The delete confirmation remains pending until the React Workspace projection has committed the removed id, so the next create gesture cannot observe one stale list frame. During create, duplicate-name validation is suppressed while the request is pending because the committed `host/workspace-changed` frame may publish the newly created Workspace before its unary response; after failure returns the form to editing, validation uses the latest list again.
+
 ## Confirmation interaction
 
 The existing Workspace row menu opens a shared `Modal` before deletion. The text states all three consequences: the Workspace leaves the list, the folder and session logs remain, and its Sessions appear under Ungrouped. While the request is pending, the confirm and Cancel controls are disabled, duplicate confirmation is ignored, and Escape or Close cannot dismiss the operation. Failure keeps the Modal open with the error; Cancel, Escape, and Close before submission never delete.
@@ -48,7 +50,7 @@ The menu, Modal, and buttons retain their existing structure and design tokens.
 
 ## Verification
 
-Workspace package tests pin successful metadata-only deletion, same-path re-registration, unknown-id idempotence, table-failure rollback, explicit-marker restart recovery, unexplained-corruption rejection, and cache/table invariant behavior. Apiproxy and carrier tests pin the schema, handler, `workspace-not-found`, retained Session/folder, fresh-id re-registration, and committed `host/workspace-removed` frame. Client tests pin unary direct echo, duplicate removal, late changed frames, and deletion racing an in-flight baseline. Component tests pin confirmation, pending-state duplicate suppression, success, failure, Cancel, Escape, and Close.
+Workspace package tests pin successful metadata-only deletion, same-path re-registration, unknown-id idempotence, table-failure rollback, explicit-marker restart recovery, unexplained-corruption rejection, and cache/table invariant behavior. Apiproxy and carrier tests pin the schema, handler, `workspace-not-found`, retained Session/folder, fresh-id re-registration, and committed `host/workspace-removed` frame. Client tests pin unary direct echo, duplicate removal, late changed frames, and deletion racing an in-flight baseline. Component tests pin confirmation, projection-settled closing, pending-state duplicate suppression, success-frame-before-unary ordering, failure, Cancel, Escape, and Close. The browser scenario observes every transient alert, slot error, console error, and page error while reusing a deleted title for a different directory.
 
 The assembled keyless Web scenario registers an existing temporary project directory, accounts a persisted Session, makes that Session current, confirms deletion in Chromium, and verifies the Workspace group disappears while Ungrouped retains the current Session. It checks the user file and JSONL log before and after deletion and repeats the UI, directory, and log assertions after reload.
 
diff --git a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
index 7a79a1ccc5..b0df6982ac 100644
--- a/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-27-workspace-registration-deletion.zh.md
@@ -28,6 +28,8 @@ Create 与 delete 会在记录／顺序对可能分叉之前写入持久 `pendin
 
 `WorkspaceManager` 将 `host/workspace-changed` 与 `host/workspace-removed` 都视为有序增量，并在进行中的 `workspace.list` 响应之上回放。成功的一元删除会立即移除行，无需等待本次操作自己的流回显。移除操作具有幂等性；由于 Workspace id 永不复用，进程本地删除标记会拒绝延迟到达的 changed 帧或陈旧基线行。重连仍从 `workspace.list` 刷新；Workspace 增量绝不会剪除会话状态。
 
+删除确认框会保持待处理，直到 React Workspace 投影已经提交目标 id 的移除，因此下一次创建操作不会读到一帧陈旧列表。创建请求进行中会暂停重复名称校验，因为已提交的 `host/workspace-changed` 帧可能先于一元响应发布刚创建的 Workspace；如果请求失败并让表单回到可编辑状态，系统会重新使用最新列表执行校验。
+
 ## 确认交互
 
 现有 Workspace 行菜单会在删除前打开共享 `Modal`。文案明确说明三项后果：Workspace 会从列表中移除，文件夹和会话日志会保留，相关会话会出现在 Ungrouped 下。请求待处理期间，确认与 Cancel 控件均被禁用，重复确认会被忽略，Escape 或 Close 也无法关闭此次操作。失败时 `Modal` 保持打开并显示错误；提交前使用 Cancel、Escape 或 Close 绝不会触发删除。
@@ -48,7 +50,7 @@ Create 与 delete 会在记录／顺序对可能分叉之前写入持久 `pendin
 
 ## Verification
 
-Workspace 包测试固定了仅删除元数据的成功路径、同路径重新注册、未知 id 的幂等行为、表操作失败回滚、明确标记的重启恢复、来源不明损坏的拒绝，以及缓存／表不变量行为。Apiproxy 与载体测试固定了 schema、处理器、`workspace-not-found`、保留会话／文件夹、使用新 id 重新注册，以及已提交的 `host/workspace-removed` 帧。客户端测试固定了一元直接回显、重复移除、延迟到达的 changed 帧，以及删除与进行中基线并发的行为。组件测试固定了确认交互、待处理状态下抑制重复提交、成功、失败、Cancel、Escape 与 Close。
+Workspace 包测试固定了仅删除元数据的成功路径、同路径重新注册、未知 id 的幂等行为、表操作失败回滚、明确标记的重启恢复、来源不明损坏的拒绝，以及缓存／表不变量行为。Apiproxy 与载体测试固定了 schema、处理器、`workspace-not-found`、保留会话／文件夹、使用新 id 重新注册，以及已提交的 `host/workspace-removed` 帧。客户端测试固定了一元直接回显、重复移除、延迟到达的 changed 帧，以及删除与进行中基线并发的行为。组件测试固定了确认交互、投影稳定后关闭、待处理状态下抑制重复提交、成功帧先于一元响应、失败、Cancel、Escape 与 Close。浏览器场景会在为不同目录复用已删除名称时，观测每一次瞬时 alert、slot error、console error 与 page error。
 
 组装后的无密钥 Web 场景会注册一个已有临时项目目录，将持久化会话计入账本，把该会话设为当前会话，在 Chromium 中确认删除，并验证 Workspace 分组消失，而 Ungrouped 保留当前会话。该场景在删除前后检查用户文件和 JSONL 日志，并在刷新后重复验证 UI、目录与日志。
 
diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts
index 3239dcfc12..98a2338064 100644
--- a/apps/web/tests/workspace-management.e2e.ts
+++ b/apps/web/tests/workspace-management.e2e.ts
@@ -108,6 +108,31 @@ describe('web e2e: workspace management (create / rename / flat view / hover car
 
   it('deletes only the Workspace registration and keeps its current Session, folder, and log', async () => {
     onTestFailed(() => saveFailureShot(page, 'web-e2e-ws-delete'))
+    const slotConsoleErrors: string[] = []
+    const transientSlotErrors: string[] = []
+    page.on('console', (message) => {
+      if (message.type() === 'error' && /slot entry crashed/i.test(message.text())) {
+        slotConsoleErrors.push(message.text())
+      }
+    })
+    await page.exposeFunction('recordDshSlotError', (key: string) => {
+      if (!transientSlotErrors.includes(key)) transientSlotErrors.push(key)
+    })
+    await page.evaluate(() => {
+      const target = window as unknown as { recordDshSlotError(key: string): Promise<void> }
+      const seen = new Set<string>()
+      const collect = (): void => {
+        for (const node of document.querySelectorAll<HTMLElement>('[data-slot-error]')) {
+          const key = node.dataset.slotError ?? ''
+          if (!seen.has(key)) {
+            seen.add(key)
+            void target.recordDshSlotError(key)
+          }
+        }
+      }
+      new MutationObserver(collect).observe(document.documentElement, { childList: true, subtree: true })
+      collect()
+    })
     // Register the scaffold's existing project directory through the real UI.
     await page.getByRole('button', { name: 'Create workspace' }).click()
     await page.getByRole('menuitem', { name: 'Create workspace' }).hover()
@@ -212,6 +237,69 @@ describe('web e2e: workspace management (create / rename / flat view / hover car
     await stat(logLocation.path)
     expect((await scaffold.ctx.sessionPersistence.inspect(SessionId(SEED_ID))).events.length).toBeGreaterThan(0)
 
+    expect(transientSlotErrors).toEqual([])
+    expect(slotConsoleErrors).toEqual([])
+    expect(tripwire.pageErrors).toEqual([])
+  }, 90_000)
+
+  it('reuses a deleted title for a different new directory without any transient error surface', async () => {
+    onTestFailed(() => saveFailureShot(page, 'web-e2e-ws-reuse-title'))
+    const title = 'same-name'
+    const oldPath = join(scaffold.workspaceCwd, 'adopted', title)
+    await mkdir(oldPath, { recursive: true })
+    const transientErrors: string[] = []
+    const consoleErrors: string[] = []
+    page.on('console', (message) => {
+      if (message.type() === 'error') consoleErrors.push(message.text())
+    })
+    await page.exposeFunction('recordDshTransientWorkspaceError', (message: string) => {
+      if (!transientErrors.includes(message)) transientErrors.push(message)
+    })
+    await page.evaluate(() => {
+      const target = window as unknown as {
+        recordDshTransientWorkspaceError(message: string): Promise<void>
+      }
+      const collect = (): void => {
+        for (const node of document.querySelectorAll<HTMLElement>('[data-slot-error], [role="alert"]')) {
+          const message = node.dataset.slotError ?? node.textContent?.trim() ?? ''
+          if (message !== '') void target.recordDshTransientWorkspaceError(message)
+        }
+      }
+      new MutationObserver(collect).observe(document.documentElement, { childList: true, subtree: true })
+      collect()
+    })
+
+    await page.getByRole('button', { name: 'Create workspace' }).click()
+    await page.getByRole('menuitem', { name: 'Create workspace' }).hover()
+    await page.getByRole('menuitem', { name: 'Use an existing folder' }).click()
+    const adopt = page.getByRole('dialog', { name: 'Use an existing folder' })
+    await adopt.getByLabel('Existing folder path').fill(oldPath)
+    await adopt.getByRole('button', { name: 'Use folder' }).click()
+    await expect.poll(() => adopt.count(), { timeout: 10_000 }).toBe(0)
+    const oldWorkspace = await scaffold.ctx.workspace.resolveByPath(oldPath)
+    if (oldWorkspace === undefined) throw new Error('old same-name Workspace was not registered')
+
+    const oldRow = page.locator('[role="treeitem"]').filter({ hasText: title }).first()
+    await oldRow.hover()
+    await page.getByRole('button', { name: `Workspace actions for ${title}` }).click()
+    await page.getByRole('menuitem', { name: 'Delete workspace' }).click()
+    await page.getByRole('dialog', { name: 'Delete workspace' })
+      .getByRole('button', { name: 'Delete workspace' }).click()
+    await expect.poll(() => scaffold.ctx.workspace.get(oldWorkspace.id), { timeout: 10_000 }).toBeUndefined()
+
+    await page.getByRole('button', { name: 'Create workspace' }).click()
+    await page.getByRole('menuitem', { name: 'Create workspace' }).hover()
+    await page.getByRole('menuitem', { name: 'Create a new workspace' }).click()
+    const create = page.getByRole('dialog', { name: 'Create a new workspace' })
+    await create.getByLabel('New workspace name').fill(title)
+    await create.getByRole('button', { name: 'Create workspace' }).click()
+    await expect.poll(() => create.count(), { timeout: 10_000 }).toBe(0)
+    const fresh = scaffold.ctx.workspace.list().find(workspace => workspace.title === title)
+    expect(fresh?.id).toBeDefined()
+    expect(fresh?.id).not.toBe(oldWorkspace.id)
+    expect(fresh?.path).toBe(join(scaffold.workspaceCwd, title))
+    expect(transientErrors).toEqual([])
+    expect(consoleErrors).toEqual([])
     expect(tripwire.pageErrors).toEqual([])
   }, 90_000)
 
diff --git a/packages/client/runtime/src/client/workspaces/manager.ts b/packages/client/runtime/src/client/workspaces/manager.ts
index 7179ed9eb9..ce4198cd01 100644
--- a/packages/client/runtime/src/client/workspaces/manager.ts
+++ b/packages/client/runtime/src/client/workspaces/manager.ts
@@ -133,7 +133,7 @@ export class WorkspaceManager {
    */
   async delete(workspaceId: WorkspaceId): Promise<RpcResult<{ deleted: true }>> {
     const { result } = await this.api.workspace.delete({ workspaceId })
-    if (result.ok) this.remove(workspaceId)
+    if (result.ok) this.remove(workspaceId, true)
     return result
   }
 
@@ -224,14 +224,21 @@ export class WorkspaceManager {
   }
 
   /** Remove one id idempotently and retain a tombstone against late echoes. */
-  private remove(workspaceId: WorkspaceId): void {
+  private remove(workspaceId: WorkspaceId, direct = false): void {
     this.refreshFrames?.push({ type: 'remove', workspaceId })
     this.removedIds.add(workspaceId)
     const items = this.items.filter(item =>
       item.getSnapshot().view?.workspaceId !== workspaceId)
-    if (items.length === this.items.length) return
+    if (items.length === this.items.length) {
+      // The Host frame may have removed the row first but left its batched
+      // notification pending. A successful unary echo still flushes that
+      // committed state before the user action resolves.
+      if (direct) this.notifier.notifyNow()
+      return
+    }
     this.items = items
-    this.notifier.markDirty()
+    if (direct) this.notifier.notifyNow()
+    else this.notifier.markDirty()
   }
 
   private installViews(views: readonly WorkspaceView[]): void {
diff --git a/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx b/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
index 0090164928..c56de93c56 100644
--- a/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
+++ b/packages/client/ui-workspace/src/client/WorkspaceBrowser.tsx
@@ -307,7 +307,15 @@ export function WorkspaceBrowser({
   // unmount that row without tearing down the in-flight confirmation state.
   const [deleteTarget, setDeleteTarget] = useState<{ workspaceId: WorkspaceId; title: string } | null>(null)
   const [deleting, setDeleting] = useState(false)
+  const [deleteCommittedId, setDeleteCommittedId] = useState<WorkspaceId | null>(null)
   const [deleteError, setDeleteError] = useState<string | null>(null)
+  useEffect(() => {
+    if (deleteCommittedId === null
+      || workspaces.some(workspace => workspace.workspaceId === deleteCommittedId)) return
+    setDeleting(false)
+    setDeleteCommittedId(null)
+    setDeleteTarget(null)
+  }, [deleteCommittedId, workspaces])
   const closeDelete = () => {
     if (deleting) return
     setDeleteTarget(null)
@@ -317,10 +325,13 @@ export function WorkspaceBrowser({
     /* v8 ignore next -- the Modal is absent without a target and its button is disabled while deleting. */
     if (deleting || deleteTarget === null) return
     setDeleting(true)
+    setDeleteCommittedId(null)
     setDeleteError(null)
     deleteWorkspace(deleteTarget.workspaceId).then(() => {
-      setDeleting(false)
-      setDeleteTarget(null)
+      // Keep the confirmation pending until this component has rendered the
+      // committed list projection without the deleted id. Closing earlier
+      // exposes one stale React frame to the next Create Workspace gesture.
+      setDeleteCommittedId(deleteTarget.workspaceId)
     }).catch((reason: unknown) => {
       setDeleting(false)
       setDeleteError(reason instanceof Error ? reason.message : String(reason))
diff --git a/packages/client/ui-workspace/src/client/WorkspacePicker.tsx b/packages/client/ui-workspace/src/client/WorkspacePicker.tsx
index 2be39875bc..99ed2831a6 100644
--- a/packages/client/ui-workspace/src/client/WorkspacePicker.tsx
+++ b/packages/client/ui-workspace/src/client/WorkspacePicker.tsx
@@ -60,7 +60,7 @@ export function WorkspaceCreateFlow({
   const [creating, setCreating] = useState(false)
   const [modalError, setModalError] = useState<string | null>(null)
   const normalizedWorkspaceName = workspaceName.trim()
-  const duplicateWorkspaceName = normalizedWorkspaceName !== ''
+  const duplicateWorkspaceName = !creating && normalizedWorkspaceName !== ''
     && workspaces.some(workspace => workspace.title === normalizedWorkspaceName)
 
   const items: MenuEntry[] = [
diff --git a/packages/client/ui-workspace/tests/workspace-browser.spec.tsx b/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
index 1dbe895b74..33abdc1231 100644
--- a/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
+++ b/packages/client/ui-workspace/tests/workspace-browser.spec.tsx
@@ -461,7 +461,7 @@ describe('WorkspaceBrowser', () => {
   it('confirms Workspace deletion, explains retention, and blocks duplicate submission', async () => {
     let resolveDelete!: () => void
     const deleteWorkspace = vi.fn(() => new Promise<void>((resolve) => { resolveDelete = resolve }))
-    mount({
+    const browser = mount({
       useWorkspaces: hook(workspaceState([workspace('alpha', ['session'], 'Alpha')])),
       deleteWorkspace,
     })
@@ -484,6 +484,11 @@ describe('WorkspaceBrowser', () => {
     fireEvent.click(screen.getByRole('button', { name: 'Close' }))
     expect(screen.getByRole('dialog', { name: 'Delete workspace' })).toBeTruthy()
     await act(async () => { resolveDelete() })
+    // RPC success alone does not close: the component waits until its
+    // useWorkspaces projection has committed the removal, preventing a stale
+    // duplicate-name frame from leaking into the next create gesture.
+    expect(screen.getByRole('dialog', { name: 'Delete workspace' })).toBeTruthy()
+    rerender(browser, { useWorkspaces: hook(workspaceState([])) })
     expect(screen.queryByRole('dialog', { name: 'Delete workspace' })).toBeNull()
   })
 
diff --git a/packages/client/ui-workspace/tests/workspace-picker.spec.tsx b/packages/client/ui-workspace/tests/workspace-picker.spec.tsx
index d487fae5f8..a5510178e7 100644
--- a/packages/client/ui-workspace/tests/workspace-picker.spec.tsx
+++ b/packages/client/ui-workspace/tests/workspace-picker.spec.tsx
@@ -35,18 +35,25 @@ function anchor(): { current: HTMLElement } {
 function mount(items: readonly WorkspaceView[] = [workspace('alpha', 'Alpha')], createWorkspace = vi.fn()) {
   const onPick = vi.fn()
   const onClose = vi.fn()
-  const view = render(
+  const anchorRef = anchor()
+  const renderPicker = (nextItems: readonly WorkspaceView[]) => (
     <WorkspacePicker
       open
-      anchorRef={anchor()}
+      anchorRef={anchorRef}
       useSessions={hook(sessions)}
-      useWorkspaces={hook(workspaceState(items))}
+      useWorkspaces={hook(workspaceState(nextItems))}
       onPick={onPick}
       onClose={onClose}
       createWorkspace={createWorkspace}
-    />,
+    />
   )
-  return { view, onPick, onClose, createWorkspace }
+  const view = render(
+    renderPicker(items),
+  )
+  return {
+    view, onPick, onClose, createWorkspace,
+    rerenderItems: (nextItems: readonly WorkspaceView[]) => { view.rerender(renderPicker(nextItems)) },
+  }
 }
 
 function chooseCreateItem(name: 'Use an existing folder' | 'Create a new workspace'): void {
@@ -106,6 +113,22 @@ describe('WorkspacePicker', () => {
     expect(b.createWorkspace).not.toHaveBeenCalled()
   })
 
+  it('does not flash a duplicate alert when the successful create frame arrives before its unary response', async () => {
+    let resolve!: (workspace: WorkspaceView) => void
+    const pending = new Promise<WorkspaceView>((settle) => { resolve = settle })
+    const created = workspace('fresh', 'same-name')
+    const b = mount([], vi.fn(() => pending))
+    chooseCreateItem('Create a new workspace')
+    fireEvent.change(screen.getByLabelText('New workspace name'), { target: { value: 'same-name' } })
+    fireEvent.click(screen.getByRole('button', { name: 'Create workspace' }))
+
+    b.rerenderItems([created])
+    expect(screen.getByRole('status').textContent).toBe('Creating workspace…')
+    expect(screen.queryByRole('alert')).toBeNull()
+    await act(async () => { resolve(created); await pending })
+    expect(b.onPick).toHaveBeenCalledWith(created.workspaceId)
+  })
+
   it('exposes creation phase and error text while retaining the modal for retry', async () => {
     let reject!: (reason: unknown) => void
     const pending = new Promise<WorkspaceView>((_resolve, rejectPromise) => { reject = rejectPromise })

From e2eca69e9c0ba88f591afe84727b4634d596c3b3 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 15:54:59 +0800
Subject: [PATCH 22/27] docs(ci): writer-level trust boundary stated
 everywhere; serial note counts four references
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Sweep every remaining 'admin-only' claim (workflow comments, runbook
  lines 13/40, topology note, all zh pairs): the variable is
  writer-manageable, and the boundary against untrusted code is
  repository membership (private, forking disabled, Dependabot
  excluded) — stated identically at every site instead of only in the
  'who can flip' paragraph.
- Serial cross-platform reference note (both languages): master now
  runs four references — the three hosted OS legs plus the self-hosted
  standby drill, linked to the failover runbook.
Static gate green locally: 32 passed, 0 failed.
---
 ...2026-07-21-serial-cross-platform-ci-reference.i18n.yaml | 6 +++---
 .../2026-07-21-serial-cross-platform-ci-reference.md       | 6 +++---
 .../2026-07-21-serial-cross-platform-ci-reference.zh.md    | 6 +++---
 ...26-07-22-evidence-based-larger-hosted-runners.i18n.yaml | 4 ++--
 .../2026-07-22-evidence-based-larger-hosted-runners.md     | 2 +-
 .../2026-07-22-evidence-based-larger-hosted-runners.zh.md  | 2 +-
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml       | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.md  | 4 ++--
 .../process/2026-07-26-ci-failover-runbook.zh.md           | 4 ++--
 .github/workflows/ci.yml                                   | 7 ++++---
 10 files changed, 23 insertions(+), 22 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.i18n.yaml b/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.i18n.yaml
index 17edb300cc..50ac9c830b 100644
--- a/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-21-serial-cross-platform-ci-reference.md: 5433d2c51831ce61d06a16ee0b0ed982911f9218
-2026-07-21-serial-cross-platform-ci-reference.zh.md: 041d53d13e14354c995e4b65defce94a97646b0a
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.md
+2026-07-21-serial-cross-platform-ci-reference.md: 3e3d3ed06a16baf81b940b50c3d3deb75b7d8894
+2026-07-21-serial-cross-platform-ci-reference.zh.md: e05f92c05ab66d5a444f29186c605b4609d36110
diff --git a/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.md b/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.md
index 5433d2c518..3e3d3ed06a 100644
--- a/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.md
+++ b/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.md
@@ -14,15 +14,15 @@ Reviewers also need a direct answer to a simpler question: what happens when the
 
 ## Decision
 
-[CI](../../../../.github/workflows/ci.yml) gives pull-request and master-push events complementary responsibilities. Pull requests run consolidated Linux and Windows jobs plus the Node compatibility and Python contracts on standard GitHub-hosted capacity. A push to `master` skips those jobs and runs three explicit references named `serial / linux`, `serial / macos`, and `serial / windows`. They intentionally duplicate their short checkout, runtime setup, and immutable install sequences instead of hiding the operating systems behind a matrix or reusable workflow. `workflow_dispatch` is reserved for runner benchmarks.
+[CI](../../../../.github/workflows/ci.yml) gives pull-request and master-push events complementary responsibilities. Pull requests run consolidated Linux and Windows jobs plus the Node compatibility and Python contracts on standard GitHub-hosted capacity. A push to `master` skips those jobs and runs four explicit references: `serial / linux`, `serial / macos`, and `serial / windows` on standard hosted runners, plus `serial / linux (self-hosted standby)` on the in-house `vm-backup` pool — the hot-standby drill that continuously re-proves the failover target described in the [failover runbook](2026-07-26-ci-failover-runbook.md). They intentionally duplicate their short checkout, runtime setup, and immutable install sequences instead of hiding the operating systems behind a matrix or reusable workflow. `workflow_dispatch` is reserved for runner benchmarks.
 
-Each reference job runs `pnpm run check:ci` without any shard selector. `DSH_GATE_CONCURRENCY=1` makes the top-level aggregate execute one ready gate at a time; coverage, snapshot replay, built-bin smoke, and publication validation also receive worker counts of one. The three operating-system jobs may run beside one another, but each host's repository gates are serial and complete. Linux installs bubblewrap before replaying snapshots, and Windows enables Developer Mode before installing the symlinked workspace.
+Each reference job runs `pnpm run check:ci` without any shard selector. `DSH_GATE_CONCURRENCY=1` makes the top-level aggregate execute one ready gate at a time; coverage, snapshot replay, built-bin smoke, and publication validation also receive worker counts of one. The reference jobs may run beside one another, but each host's repository gates are serial and complete. Linux installs bubblewrap before replaying snapshots, and Windows enables Developer Mode before installing the symlinked workspace.
 
 Platform ownership remains explicit inside that complete aggregate. `pty-local` supports Linux and macOS and therefore owns its unit and per-file coverage contract on POSIX rather than loading a backend that rejects `win32`; the Windows run still executes every portable package. Portable fixtures derive native paths through `node:path`, compare canonical identities with the same native realpath implementation as production, and use filenames legal on every host. ACP snapshot runs also pass both JavaScript and native realpath spellings of their generated cwd to the normalizer, which replaces aliases longest-first so Windows short and long paths cannot churn shared fixtures.
 
 The macOS reference runs the ordinary Vitest project in forked processes. Node 24 on macOS arm64 has aborted in its CJS lexer from a worker thread; the process boundary contains that external runtime failure without removing any test from the aggregate, while Linux and Windows retain the lower-overhead thread pool. Repository-owned races are fixed at their observation boundaries: dev bundle polling stages each candidate table, graph, and watch-baseline map before publishing a rescan, and a missing bundle remains dirty until a successful content hash. PTY readiness retains a prompt candidate while polling checks foreground ownership; the ordinary silence bound covers inherited markers from interactive children. Real PTY fixtures assemble synchronization tokens at runtime so the interactive shell's input echo cannot satisfy a child-readiness wait. The live-link package-manager e2e preserves the workflow-prepared Corepack home and pnpm metadata/store caches while isolating the other managers' mutable caches, so it does not discard reusable package-manager state before the install.
 
-Master reference jobs are diagnostic and do not participate in the pull request's required `all checks passed` result. A pull request runs only its required jobs; a master push runs only the three serial references. Performance is evaluated from completed hosted-job timestamps and reported as a measurement; it is not encoded as a `timeout-minutes` value.
+Master reference jobs are diagnostic and do not participate in the pull request's required `all checks passed` result. A pull request runs only its required jobs; a master push runs only the serial references. Performance is evaluated from completed hosted-job timestamps and reported as a measurement; it is not encoded as a `timeout-minutes` value.
 
 The portable reference uses GitHub's standard `ubuntu-latest`, `macos-latest`, and `windows-2025` labels. Required pull-request jobs use the same portable Linux and Windows capacity under the [required-CI decision](2026-07-23-portable-required-pull-request-ci.md). Higher-core hosted runners remain manual benchmarks because a correctness path must remain runnable without repository-external runner configuration.
 
diff --git a/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.zh.md b/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.zh.md
index 041d53d13e..e05f92c05a 100644
--- a/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.zh.md
+++ b/.agents/notes/implemented/process/2026-07-21-serial-cross-platform-ci-reference.zh.md
@@ -14,15 +14,15 @@ Status: implemented
 
 ## 决策
 
-[CI](../../../../.github/workflows/ci.yml) 为拉取请求事件与 master 推送事件赋予互补的职责。拉取请求在 GitHub 标准托管容量上运行合并后的 Linux 和 Windows 作业，以及 Node 兼容性与 Python 契约。向 `master` 推送时会跳过这些作业，改为运行三个显式参考作业，名称分别为 `serial / linux`、`serial / macos` 和 `serial / windows`。这些作业有意分别重复简短的代码检出、运行时设置和依赖锁定的安装步骤，不用矩阵或可复用工作流把操作系统差异隐藏起来。`workflow_dispatch` 仅用于运行器基准测试。
+[CI](../../../../.github/workflows/ci.yml) 为拉取请求事件与 master 推送事件赋予互补的职责。拉取请求在 GitHub 标准托管容量上运行合并后的 Linux 和 Windows 作业，以及 Node 兼容性与 Python 契约。向 `master` 推送时会跳过这些作业，改为运行四个显式参考作业：在标准托管运行器上的 `serial / linux`、`serial / macos` 和 `serial / windows`，以及在公司自有 `vm-backup` 池上的 `serial / linux (self-hosted standby)`——后者是热备演练，持续验证[故障切换手册](2026-07-26-ci-failover-runbook.md)所描述的切换目标。这些作业有意分别重复简短的代码检出、运行时设置和依赖锁定的安装步骤，不用矩阵或可复用工作流把操作系统差异隐藏起来。`workflow_dispatch` 仅用于运行器基准测试。
 
-每个参考作业均在不设置任何分片选择器的情况下运行 `pnpm run check:ci`。`DSH_GATE_CONCURRENCY=1` 使顶层聚合每次只执行一个已经就绪的门禁；覆盖率、快照回放、built-bin 冒烟测试和发布验证的并发数也设为 1。三种操作系统的作业可以彼此并行，但每台主机上的仓库门禁都串行运行且完整执行。Linux 在回放快照前安装 bubblewrap，Windows 则在安装采用符号链接的工作区前启用开发人员模式。
+每个参考作业均在不设置任何分片选择器的情况下运行 `pnpm run check:ci`。`DSH_GATE_CONCURRENCY=1` 使顶层聚合每次只执行一个已经就绪的门禁；覆盖率、快照回放、built-bin 冒烟测试和发布验证的并发数也设为 1。各参考作业可以彼此并行，但每台主机上的仓库门禁都串行运行且完整执行。Linux 在回放快照前安装 bubblewrap，Windows 则在安装采用符号链接的工作区前启用开发人员模式。
 
 该完整聚合流程仍明确划分平台归属。`pty-local` 支持 Linux 与 macOS，因此其单元测试和逐文件覆盖率契约由 POSIX 平台负责，而不会在 Windows 上加载一个明确拒绝 `win32` 的后端；Windows 仍会执行所有可移植包（package）。可移植 fixture（测试前置数据）通过 `node:path` 派生原生路径，使用与生产代码相同的原生 realpath 实现比较规范化后的路径标识，并采用所有宿主机均允许的文件名。ACP（Agent Client Protocol）快照运行还会把生成的 cwd 分别通过 realpath 的 JavaScript 实现与原生实现得到的两种表示一并传给规范化器；规范化器按长度从长到短替换这些别名，避免 Windows 的短路径与长路径表示差异导致共享 fixture 反复变化。
 
 macOS 参考流程使用 fork 进程运行常规 Vitest 项目。macOS arm64 上的 Node 24 曾在工作线程中执行 CJS 词法分析器时异常终止；进程边界能够隔离这一外部运行时故障，且无需从聚合流程中删除任何测试，而 Linux 与 Windows 仍使用开销更低的线程池。仓库自身引入的竞态均在相应的观测边界修复：开发构建产物的轮询逻辑每次发布重新扫描结果前，都会先暂存候选表、候选图和候选监视基线映射；构建产物缺失后会一直保持脏状态，直到成功计算内容哈希。PTY 就绪检测会在轮询检查前台进程组归属期间保留提示符候选项；常规静默时限也适用于交互式子进程继承提示符标记的情况。真实 PTY fixture 会在运行时拼接同步标记，使就绪等待逻辑不会把交互式 shell 的输入回显误判为子进程已就绪。实时链接场景下的包管理器 e2e 会保留由工作流预先准备的 Corepack 主目录、pnpm 元数据缓存和 store 缓存，同时隔离其他包管理器的可变缓存，因此不会在安装前丢弃可复用的包管理器状态。
 
-master 分支的参考作业仅用于诊断，不参与拉取请求所要求的 `all checks passed` 结果。拉取请求只运行其必需作业；向 master 推送时只运行三个串行参考作业。系统根据已完成托管作业的时间戳评估性能，并将其报告为测量结果，而不是写成 `timeout-minutes` 值。
+master 分支的参考作业仅用于诊断，不参与拉取请求所要求的 `all checks passed` 结果。拉取请求只运行其必需作业；向 master 推送时只运行串行参考作业。系统根据已完成托管作业的时间戳评估性能，并将其报告为测量结果，而不是写成 `timeout-minutes` 值。
 
 可移植的参考流程使用 GitHub 标准的 `ubuntu-latest`、`macos-latest` 和 `windows-2025` 标签。依据[必需 CI 决策](2026-07-23-portable-required-pull-request-ci.md)，拉取请求必需作业使用相同的可移植 Linux 和 Windows 容量。更高核心数的托管运行器仍仅用于手动基准测试，因为正确性路径必须无需仓库外部的运行器配置即可运行。
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index 99cabc76bb..f65ebb1ba4 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
-2026-07-22-evidence-based-larger-hosted-runners.md: 5b399be5571ddaf1f775ba43a2233198b8e09b18
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: f77516e2375bfc0557679d05fd275bd9cee7d8eb
+2026-07-22-evidence-based-larger-hosted-runners.md: 180cc03ad091b2e9e96a86311515250f92065c6b
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: b81f67805fd543e81ede02ceeac2dda1831f4cef
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 5b399be557..180cc03ad0 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -52,7 +52,7 @@ The process-bound coverage project contains exactly five suite files. Thirty-two
 
 Complete serial Linux, macOS, and Windows references run only when `master` moves. Pull requests use the enterprise required path plus standard-hosted compatibility jobs, while other larger-runner sizes run only by manual dispatch.
 
-An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with six always-on systemd-managed runner instances) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the admin-only `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
+An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with six always-on systemd-managed runner instances) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the writer-manageable `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index f77516e237..b81f67805f 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -52,7 +52,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
-另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过仅限管理员的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过写者可管理的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
 
 ## 曾考虑的替代方案
 
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index 7b8d08befe..de9ba10469 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: 55c1350593562d62463e751451d50a79cf45a1d6
-2026-07-26-ci-failover-runbook.zh.md: 13977b78244440a23722d089849ea7ff6b751aea
+2026-07-26-ci-failover-runbook.md: 05014454fa3e38045b89a857c346db0f897ab5a6
+2026-07-26-ci-failover-runbook.zh.md: e106a0de40799ca1c218217ea66c24068697dc53
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index 55c1350593..05014454fa 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -10,7 +10,7 @@ The three required Linux worker jobs in [CI](../../../../.github/workflows/ci.ym
 
 ## Decision
 
-Each of the three required Linux worker jobs — and the `all checks passed` verdict job, which would otherwise stay queued on the failed pool even after every worker passed — resolves its runner pool through the `DSH_CI_FAILOVER` repository variable. Unset (normal), they run on the hosted enterprise pools. Set to `selfhosted` by a repository admin, all four retarget onto the in-house self-hosted `vm-backup` pool, coverage and snapshot concurrency drop to shared-VM bounds, and the hosted-path pnpm cache restores are skipped. The switch is admin-only repository state, not a merge, so it works while every check is red. The in-house pool's readiness is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push.
+Each of the three required Linux worker jobs — and the `all checks passed` verdict job, which would otherwise stay queued on the failed pool even after every worker passed — resolves its runner pool through the `DSH_CI_FAILOVER` repository variable. Unset (normal), they run on the hosted enterprise pools. Set to `selfhosted` by any repository writer, all four retarget onto the in-house self-hosted `vm-backup` pool, coverage and snapshot concurrency drop to shared-VM bounds, and the hosted-path pnpm cache restores are skipped. The switch is writer-manageable repository state, not a merge, so it works while every check is red. The in-house pool's readiness is continuously re-proven by the `serial / linux (self-hosted standby)` lane, which runs the complete unsharded aggregate on every master push.
 
 ### What the in-house pool is
 
@@ -37,7 +37,7 @@ Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhos
 
 ### Trust boundary
 
-The variable is repository-admin-only state: a pull request can neither set it nor read a different value into effect, and the expressions live in the base branch's workflow definition. This failover path therefore adds no PR-editable route to the self-hosted pool. Runner-side enforcement — an org-level runner group restricting these runners to the master-ref workflow — is tracked separately and composes with this mechanism.
+The variable is writer-manageable repository state; a pull request event itself can neither set it nor read a different value into effect, and the selector expressions live in workflow definitions. Note that under failover, `pull_request` runs execute the PR merge ref's own workflow definition — the boundary against untrusted code is repository membership (private, forking disabled, Dependabot excluded by the selectors), not the variable. Runner-side enforcement — an org-level runner group restricting these runners to the master-ref workflow — is tracked separately and composes with this mechanism.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index 13977b7824..e106a0de40 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -10,7 +10,7 @@ Status: implemented
 
 ## 决策
 
-三个必需的 Linux 工作作业——以及 `all checks passed` 判定作业（若不随切换，即使全部工作作业通过，它仍会滞留在故障池的队列中）——各自通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。变量不存在（正常）时它们运行在托管企业池上；由仓库管理员设为 `selfhosted` 时，四者全部切换到公司自有的自托管 `vm-backup` 池，coverage 与 snapshot 的并发降到共享虚拟机上限，并跳过托管路径的 pnpm 缓存恢复。这个开关是仅限管理员的仓库状态而非一次合并，因此在所有检查都是红色时仍然有效。自有池的就绪状态由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。
+三个必需的 Linux 工作作业——以及 `all checks passed` 判定作业（若不随切换，即使全部工作作业通过，它仍会滞留在故障池的队列中）——各自通过仓库变量 `DSH_CI_FAILOVER` 解析运行器池。变量不存在（正常）时它们运行在托管企业池上；由任何具备写权限的协作者设为 `selfhosted` 时，四者全部切换到公司自有的自托管 `vm-backup` 池，coverage 与 snapshot 的并发降到共享虚拟机上限，并跳过托管路径的 pnpm 缓存恢复。这个开关是写者可管理的仓库状态而非一次合并，因此在所有检查都是红色时仍然有效。自有池的就绪状态由 `serial / linux (self-hosted standby)` 通道持续验证——每次 master 推送都在其上运行完整的未分片聚合流程。
 
 ### 自有池是什么
 
@@ -37,7 +37,7 @@ Status: implemented
 
 ### 信任边界
 
-该变量是仅限仓库管理员的状态：拉取请求既不能设置它，也不能让不同的值生效，且表达式存在于基线分支的工作流定义中。因此这条故障切换路径没有增加任何可由 PR 编辑的自托管池访问途径。运行器侧的强制约束——通过组织级 runner group 把这批运行器限定到 master 引用的工作流——另行跟踪，与本机制互补。
+该变量是写者可管理的仓库状态；`pull_request` 事件本身既不能设置它，也不能让不同的值生效，选择器表达式存在于工作流定义中。需要注意：故障切换期间，`pull_request` 运行执行的是 PR merge 引用自带的工作流定义——抵御不可信代码的边界是仓库成员资格（私有、禁 fork、选择器排除 Dependabot），而非该变量。（运行器侧的组织级 runner group 约束另行跟踪，与本机制互补。）
 
 ## 曾考虑的替代方案
 
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 8c3b854ae6..9fe393b93c 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -34,8 +34,9 @@ jobs:
   # FAILOVER: each Linux enterprise job resolves its pool through the
   # DSH_CI_FAILOVER repository variable. Unset (normal), the expressions
   # pick the hosted enterprise pools below. Setting the variable to
-  # 'selfhosted' (repo Settings → Actions → Variables; admin-only, not
-  # PR-editable, no merge required) retargets all three onto the in-house
+  # 'selfhosted' (repo Settings → Actions → Variables; writer-manageable
+  # repository state — not PR-editable, no merge required) retargets all
+  # three onto the in-house
   # vm-backup pool and re-running the failed jobs is the entire switch —
   # see .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md. The
   # in-house pool's readiness is re-proven on every master push by the
@@ -411,7 +412,7 @@ jobs:
   # Hot-standby drill for the in-house self-hosted pool: every master move
   # re-runs the complete unsharded aggregate on the persistent 64-core VM,
   # continuously proving that environment can take over a required lane if
-  # the hosted pools degrade (the switch is then setting the admin-only
+  # the hosted pools degrade (the switch is then setting the writer-manageable
   # DSH_CI_FAILOVER variable — see the failover runbook, no merge required).
   # Push-triggered, so it always executes the base branch's own workflow
   # definition — no PR-editable path selects these runners. Non-blocking for

From 24d7211f09d28055c0795eac4c91f924d3593e3c Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 16:02:19 +0800
Subject: [PATCH 23/27] docs(ci): stop claiming no PR-editable path reaches the
 standby pool
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The standby lane itself is push-only, but under failover pull_request
jobs do reach these runners with the PR merge ref's workflow. The
workflow comment and the larger-runner note (both languages) now state
that plainly and name the actual boundary — repository membership
(private, forking disabled, Dependabot excluded) — matching the
runbook. Static gate green locally: 32 passed, 0 failed.
---
 ...26-07-22-evidence-based-larger-hosted-runners.i18n.yaml | 4 ++--
 .../2026-07-22-evidence-based-larger-hosted-runners.md     | 2 +-
 .../2026-07-22-evidence-based-larger-hosted-runners.zh.md  | 2 +-
 .github/workflows/ci.yml                                   | 7 +++++--
 4 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
index f65ebb1ba4..68b4098d4f 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
-2026-07-22-evidence-based-larger-hosted-runners.md: 180cc03ad091b2e9e96a86311515250f92065c6b
-2026-07-22-evidence-based-larger-hosted-runners.zh.md: b81f67805fd543e81ede02ceeac2dda1831f4cef
+2026-07-22-evidence-based-larger-hosted-runners.md: 67fc7ded5cffc6a219665f135a4c9e1cc4752691
+2026-07-22-evidence-based-larger-hosted-runners.zh.md: 71c5c067361b57fab5aae9e9ffa3850a30609db3
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
index 180cc03ad0..67fc7ded5c 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.md
@@ -52,7 +52,7 @@ The process-bound coverage project contains exactly five suite files. Thirty-two
 
 Complete serial Linux, macOS, and Windows references run only when `master` moves. Pull requests use the enterprise required path plus standard-hosted compatibility jobs, while other larger-runner sizes run only by manual dispatch.
 
-An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with six always-on systemd-managed runner instances) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the writer-manageable `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). Because the standby lane is push-triggered, it always executes the base branch's workflow definition — no pull-request-editable path can route code to these runners, and the repository additionally keeps forking disabled.
+An additional serial Linux reference runs on the in-house self-hosted pool (`vm-backup` label: a 64-core VM with six always-on systemd-managed runner instances) on every `master` push. It is a hot-standby drill, not a required check: each run re-proves that the persistent VM can execute the complete unsharded aggregate. The actual switch is pre-wired: the three required Linux jobs resolve their pool through the writer-manageable `DSH_CI_FAILOVER` repository variable, so an outage response is setting one variable and re-running — no merge, which would be deadlocked behind the failing checks themselves ([runbook](2026-07-26-ci-failover-runbook.md)). The standby lane is push-triggered, so it always executes the base branch's workflow definition. Under failover, however, `pull_request` jobs do reach these runners with the PR merge ref's own workflow definition — the trust boundary is repository membership (the repository is private with forking disabled, and the selectors exclude Dependabot), as the [failover runbook](2026-07-26-ci-failover-runbook.md) records.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
index b81f67805f..71c5c06736 100644
--- a/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
+++ b/.agents/notes/implemented/process/2026-07-22-evidence-based-larger-hosted-runners.zh.md
@@ -52,7 +52,7 @@ Windows 仓库工作在超过 16 核后收益很小，但 32 核池可以让完
 
 只有在 `master` 移动时，才运行完整的 Linux、macOS 和 Windows 串行参考。拉取请求使用企业级运行器必需路径和标准托管兼容性作业，其他大型运行器规格仅通过手动触发运行。
 
-另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过写者可管理的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义——不存在任何可由拉取请求编辑的路径能把代码路由到这些运行器上；此外仓库继续保持禁用 fork。
+另有一条串行 Linux 参考在每次 `master` 推送时运行于公司自有的自托管池（`vm-backup` 标签：一台 64 核虚拟机，运行 6 个常驻的 systemd 管理运行器实例）。它是热备演练而非必需检查：每次运行都重新证明这台持久化虚拟机能够执行完整的未分片聚合流程。实际切换机制已预先布线：三个必需 Linux 作业通过写者可管理的仓库变量 `DSH_CI_FAILOVER` 解析运行器池，因此故障响应就是设置一个变量并重跑——无需合并（合并本身会被正在失败的检查死锁）（[切换手册](2026-07-26-ci-failover-runbook.md)）。该热备通道由 push 触发，执行的始终是基线分支自身的工作流定义。但需要注意：故障切换期间，`pull_request` 作业确实会带着 PR merge 引用自带的工作流定义到达这些运行器——信任边界是仓库成员资格（仓库为私有且禁用 fork，选择器排除 Dependabot），详见[故障切换手册](2026-07-26-ci-failover-runbook.md)的记录。
 
 ## 曾考虑的替代方案
 
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 9fe393b93c..140ae00446 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -414,8 +414,11 @@ jobs:
   # continuously proving that environment can take over a required lane if
   # the hosted pools degrade (the switch is then setting the writer-manageable
   # DSH_CI_FAILOVER variable — see the failover runbook, no merge required).
-  # Push-triggered, so it always executes the base branch's own workflow
-  # definition — no PR-editable path selects these runners. Non-blocking for
+  # Push-triggered, so this lane always executes the base branch's own
+  # workflow definition. (Under failover, pull_request jobs do reach these
+  # runners with the PR merge ref's workflow — the boundary there is
+  # repository membership: private, forking disabled, Dependabot excluded.)
+  # Non-blocking for
   # pull requests; no cache steps because the VM's persistent pnpm store and
   # tool caches make them redundant (and saving here would poison the hosted
   # cache namespace with self-hosted paths).

From ce3b13bb0816d3a23b68b916087b3beef29fcc83 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 16:13:30 +0800
Subject: [PATCH 24/27] =?UTF-8?q?ci:=20standby=20fetches=20full=20history;?=
 =?UTF-8?q?=20runbook=20=E2=80=94=20writer=20wording=20throughout,=20maste?=
 =?UTF-8?q?r-ref=20pinning=20incompatibility?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- serial-linux-selfhosted checks out fetch-depth 0: depth 2 misses
  github.event.before on multi-commit or force pushes, failing the
  archive verifier on a valid tree. Full fetch is cheap against the
  VM's local mirror.
- Runbook (both languages): every remaining admin phrasing (problem
  statement, switch heading, alternatives, consequences) now says
  writer; and the 'composes with this mechanism' claim about a
  master-ref-pinned runner group is replaced with the truth observed
  live on 2026-07-27 — master-ref pinning blocks PR failover, and the
  shipped posture is repository-scoped all-workflow group access.
Static gate green locally: 32 passed, 0 failed.
---
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml   |  4 ++--
 .../process/2026-07-26-ci-failover-runbook.md          | 10 +++++-----
 .../process/2026-07-26-ci-failover-runbook.zh.md       | 10 +++++-----
 .github/workflows/ci.yml                               |  9 +++++----
 4 files changed, 17 insertions(+), 16 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index de9ba10469..658a85ce34 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: 05014454fa3e38045b89a857c346db0f897ab5a6
-2026-07-26-ci-failover-runbook.zh.md: e106a0de40799ca1c218217ea66c24068697dc53
+2026-07-26-ci-failover-runbook.md: ca4349661d03ff4e28d7c3c2b6e910106ff4aa30
+2026-07-26-ci-failover-runbook.zh.md: 1d59bd537879f531c9075e833c9e1dbfbd4bb0a2
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index 05014454fa..ca4349661d 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -6,7 +6,7 @@ English | [中文](2026-07-26-ci-failover-runbook.zh.md)
 
 ## Problem
 
-The three required Linux worker jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) and the required verdict job that aggregates them (`all checks passed`) run on the hosted enterprise 32-core pools. When those pools degrade — jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. An outage therefore needs a switch a repository admin can throw without merging anything.
+The three required Linux worker jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) and the required verdict job that aggregates them (`all checks passed`) run on the hosted enterprise 32-core pools. When those pools degrade — jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. An outage therefore needs a switch any responder with repository write access can throw without merging anything.
 
 ## Decision
 
@@ -16,7 +16,7 @@ Each of the three required Linux worker jobs — and the `all checks passed` ver
 
 `vm-backup`: one 64-core VM, six always-on systemd-managed runner instances. Check the latest `serial / linux (self-hosted standby)` run before switching: a green standby is verified-yesterday capacity.
 
-### Switch (repo admin, ~1 minute, no merge)
+### Switch (any repository writer, ~1 minute, no merge)
 
 1. Repository **Settings → Secrets and variables → Actions → Variables → New repository variable**: name `DSH_CI_FAILOVER`, value `selfhosted`.
 2. Retrigger the required jobs so they re-resolve their pool. Jobs already **queued** for the hosted labels do not retarget and cannot be re-run in place, so for the documented indefinite-queue outage, cancel the stuck run and re-run all jobs, or push a new commit; "Re-run failed jobs" only helps once a job has actually failed rather than queued.
@@ -37,14 +37,14 @@ Delete the `DSH_CI_FAILOVER` variable (or set it to anything other than `selfhos
 
 ### Trust boundary
 
-The variable is writer-manageable repository state; a pull request event itself can neither set it nor read a different value into effect, and the selector expressions live in workflow definitions. Note that under failover, `pull_request` runs execute the PR merge ref's own workflow definition — the boundary against untrusted code is repository membership (private, forking disabled, Dependabot excluded by the selectors), not the variable. Runner-side enforcement — an org-level runner group restricting these runners to the master-ref workflow — is tracked separately and composes with this mechanism.
+The variable is writer-manageable repository state; a pull request event itself can neither set it nor read a different value into effect, and the selector expressions live in workflow definitions. Note that under failover, `pull_request` runs execute the PR merge ref's own workflow definition — the boundary against untrusted code is repository membership (private, forking disabled, Dependabot excluded by the selectors), not the variable. Note on runner-group policy: pinning the runner group to the master-ref workflow is **incompatible** with this failover — the four failover jobs are `pull_request` runs evaluated from PR merge refs, and a master-pinned group leaves them queued (observed live on 2026-07-27; the group was widened to all workflows of this repository to unblock the switch). A stricter runner-side policy therefore costs PR failover; the shipped posture accepts repository-scoped, all-workflow group access.
 
 ## Alternatives considered
 
-**Merge a workflow change to switch pools.** Rejected because the outage that motivates the switch is exactly the state in which no PR can merge: the required checks are the ones failing. A repository variable is admin-controlled state that takes effect on re-run without a merge.
+**Merge a workflow change to switch pools.** Rejected because the outage that motivates the switch is exactly the state in which no PR can merge: the required checks are the ones failing. A repository variable is writer-manageable state that takes effect on re-run without a merge.
 
 **Keep the self-hosted pool always in the required path.** Rejected because it trades hosted-pool availability for the in-house VM's, moving a single point of failure rather than adding a fallback. The variable keeps the hosted pools primary and the self-hosted pool a proven, one-action standby.
 
 ## Consequences
 
-Recovering from a hosted-pool outage is a single admin variable plus a re-run, with no merge on the critical path. The cost is a second runner topology to keep working: the standby lane exercises it on every master push so the failover target never goes stale, and the concurrency and cache-restore branches in `ci.yml` carry a `selfhosted` leg that must stay in step with the hosted leg.
+Recovering from a hosted-pool outage is a single variable (any writer) plus a re-run, with no merge on the critical path. The cost is a second runner topology to keep working: the standby lane exercises it on every master push so the failover target never goes stale, and the concurrency and cache-restore branches in `ci.yml` carry a `selfhosted` leg that must stay in step with the hosted leg.
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index e106a0de40..1d59bd5378 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -6,7 +6,7 @@ Status: implemented
 
 ## 问题
 
-[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 工作作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）以及聚合它们的必需判定作业（`all checks passed`）运行在托管的企业级 32 核池上。当这些托管池发生故障——作业无限排队、企业标签消失或 GitHub 侧容量故障——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。因此故障需要一个仓库管理员无需合并任何代码即可触发的开关。
+[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 工作作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）以及聚合它们的必需判定作业（`all checks passed`）运行在托管的企业级 32 核池上。当这些托管池发生故障——作业无限排队、企业标签消失或 GitHub 侧容量故障——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。因此故障需要一个任何具备仓库写权限的响应者都能在不合并任何代码的情况下触发的开关。
 
 ## 决策
 
@@ -16,7 +16,7 @@ Status: implemented
 
 `vm-backup`：一台 64 核虚拟机，6 个常驻 systemd 管理的运行器实例。切换前先看 `serial / linux (self-hosted standby)` 最近一次运行：绿色 = 这套环境昨天刚被全量验证过。
 
-### 切换步骤（仓库管理员，约 1 分钟，无需合并）
+### 切换步骤（任何具备写权限的协作者，约 1 分钟，无需合并）
 
 1. 仓库 **Settings → Secrets and variables → Actions → Variables → New repository variable**：名称 `DSH_CI_FAILOVER`，值 `selfhosted`。
 2. 重新触发必需作业，使其重新解析运行器池。已经为托管标签**排队**的作业不会重定向，也无法原地 re-run，因此对于本手册所述的无限排队故障，应取消卡住的运行并 re-run all jobs，或推送一个新提交；“Re-run failed jobs”只有在作业真正失败（而非仍在排队）时才有用。
@@ -37,14 +37,14 @@ Status: implemented
 
 ### 信任边界
 
-该变量是写者可管理的仓库状态；`pull_request` 事件本身既不能设置它，也不能让不同的值生效，选择器表达式存在于工作流定义中。需要注意：故障切换期间，`pull_request` 运行执行的是 PR merge 引用自带的工作流定义——抵御不可信代码的边界是仓库成员资格（私有、禁 fork、选择器排除 Dependabot），而非该变量。（运行器侧的组织级 runner group 约束另行跟踪，与本机制互补。）
+该变量是写者可管理的仓库状态；`pull_request` 事件本身既不能设置它，也不能让不同的值生效，选择器表达式存在于工作流定义中。需要注意：故障切换期间，`pull_request` 运行执行的是 PR merge 引用自带的工作流定义——抵御不可信代码的边界是仓库成员资格（私有、禁 fork、选择器排除 Dependabot），而非该变量。关于 runner group 策略的说明：把 runner group 绑定到 master 引用的工作流与本故障切换机制**不兼容**——四个故障切换作业是从 PR merge 引用求值的 `pull_request` 运行，master 绑定的组会让它们持续排队（2026-07-27 实际故障中亲历；当时将组放宽为本仓库全部工作流才疏通了切换）。更严格的运行器侧策略以牺牲 PR 故障切换为代价；当前采用的形态是仓库范围、全工作流的组访问。
 
 ## 曾考虑的替代方案
 
-**通过合并一次工作流改动来切换池。** 否决，因为触发切换的故障状态恰恰是任何 PR 都无法合并的状态：必需检查正是失败的那些。仓库变量是管理员控制的状态，重跑即生效，无需合并。
+**通过合并一次工作流改动来切换池。** 否决，因为触发切换的故障状态恰恰是任何 PR 都无法合并的状态：必需检查正是失败的那些。仓库变量是写者可管理的状态，重跑即生效，无需合并。
 
 **让自托管池长期处于必需路径中。** 否决，因为这是拿托管池的可用性去换自有虚拟机的可用性，只是搬移了单点故障而非增加回退。该变量让托管池保持主路径，自托管池作为一个经过验证、一步即可启用的热备。
 
 ## 后果
 
-从托管池故障中恢复只需一个管理员变量加一次重跑，关键路径上没有合并。代价是要维护第二套运行器拓扑：热备通道在每次 master 推送时都运行它，使故障切换目标永不失效；而 `ci.yml` 中的并发与缓存恢复分支带有一条 `selfhosted` 支路，必须与托管支路保持同步。
+从托管池故障中恢复只需一个变量（任何写者可设）加一次重跑，关键路径上没有合并。代价是要维护第二套运行器拓扑：热备通道在每次 master 推送时都运行它，使故障切换目标永不失效；而 `ci.yml` 中的并发与缓存恢复分支带有一条 `selfhosted` 支路，必须与托管支路保持同步。
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 140ae00446..3c952a7bb3 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -427,12 +427,13 @@ jobs:
     name: serial / linux (self-hosted standby)
     runs-on: [self-hosted, linux, x64, vm-backup]
     steps:
-      # fetch-depth 2 + DSH_ARCHIVE_BASE_REF below: same frozen-archive
-      # comparison as serial-linux — without the prior commit the archive
-      # verifier defaults to HEAD and compares the new manifest with itself.
+      # Full history + DSH_ARCHIVE_BASE_REF below: same frozen-archive
+      # comparison as serial-linux. Depth 2 would miss github.event.before
+      # on multi-commit or force pushes; full fetch is cheap here because
+      # checkout resolves against the VM's local mirror.
       - uses: actions/checkout@v6
         with:
-          fetch-depth: 2
+          fetch-depth: 0
 
       - uses: actions/setup-node@v6
         with:

From 3cf2853b3f04fdceb3ff99109dea8f764d28fbc3 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 16:22:30 +0800
Subject: [PATCH 25/27] docs(ci): bootstrap procedure starts the listener
 service

config.sh only registers; the runner stays offline until svc.sh
install/start. Both language sides updated so emergency capacity
actually comes online.
---
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml          | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.md     | 2 +-
 .../implemented/process/2026-07-26-ci-failover-runbook.zh.md  | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index 658a85ce34..0e3b0820c0 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: ca4349661d03ff4e28d7c3c2b6e910106ff4aa30
-2026-07-26-ci-failover-runbook.zh.md: 1d59bd537879f531c9075e833c9e1dbfbd4bb0a2
+2026-07-26-ci-failover-runbook.md: 80dd7c4291e3de11c2f13b3247af56762396c720
+2026-07-26-ci-failover-runbook.zh.md: 7933a857f1559c540fccc2cd89352c4fe351dd7e
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index ca4349661d..80dd7c4291 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -28,7 +28,7 @@ Each of the three required Linux worker jobs — and the `all checks passed` ver
 
 ## Capacity during failover
 
-Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` — then run `config.sh`; copying `.runner`/`.credentials` verbatim makes `config.sh` refuse with "already configured". About a minute per instance.
+Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` — then run `config.sh` (copying `.runner`/`.credentials` verbatim makes it refuse with "already configured"), and **start the listener**: `sudo ./svc.sh install ubuntu && sudo ./svc.sh start`. Registration alone leaves the runner offline; only a started service adds capacity. About a minute per instance.
 
 
 ### Switch back
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index 1d59bd5378..7933a857f1 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -28,7 +28,7 @@ Status: implemented
 
 ## 切换期间的容量
 
-6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`——再跑 `config.sh`；原样拷贝 `.runner`/`.credentials` 会使 `config.sh` 以 "already configured" 拒绝。每个约一分钟。
+6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`——再跑 `config.sh`（原样拷贝 `.runner`/`.credentials` 会使其以 "already configured" 拒绝），然后**启动监听器**：`sudo ./svc.sh install ubuntu && sudo ./svc.sh start`。仅注册不会上线；只有启动了服务的 runner 才会增加容量。每个约一分钟。
 
 
 ### 切回

From ebb5bf4c1941cb5bddf7395a3ba6a7e718f9f9eb Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 16:25:45 +0800
Subject: [PATCH 26/27] docs(ci): starting a new failover runner needs svc.sh,
 not just config.sh

config.sh registers the instance without starting a listener, so the
procedure as written left the new runner offline and added no capacity.
---
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml          | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.md     | 2 +-
 .../implemented/process/2026-07-26-ci-failover-runbook.zh.md  | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index de9ba10469..db702da620 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: 05014454fa3e38045b89a857c346db0f897ab5a6
-2026-07-26-ci-failover-runbook.zh.md: e106a0de40799ca1c218217ea66c24068697dc53
+2026-07-26-ci-failover-runbook.md: b93c86d73f319f40706a4f9b31f448804e7b5ba8
+2026-07-26-ci-failover-runbook.zh.md: 25b83981e70070800c5a4037807d26811e55124d
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index 05014454fa..b93c86d73f 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -28,7 +28,7 @@ Each of the three required Linux worker jobs — and the `all checks passed` ver
 
 ## Capacity during failover
 
-Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` — then run `config.sh`; copying `.runner`/`.credentials` verbatim makes `config.sh` refuse with "already configured". About a minute per instance.
+Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` — then run `config.sh`; copying `.runner`/`.credentials` verbatim makes `config.sh` refuse with "already configured". `config.sh` only registers the instance — it starts no listener, so a runner that stops there is registered and offline, adding no capacity. Install and start its service too: `sudo ./svc.sh install <user> && sudo ./svc.sh start` (this pool is systemd-managed; a foreground `./run.sh` also works but dies with the shell). Confirm the instance reports Idle in org Settings → Actions → Runners before counting it. About a minute per instance.
 
 
 ### Switch back
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index e106a0de40..25b83981e7 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -28,7 +28,7 @@ Status: implemented
 
 ## 切换期间的容量
 
-6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`——再跑 `config.sh`；原样拷贝 `.runner`/`.credentials` 会使 `config.sh` 以 "already configured" 拒绝。每个约一分钟。
+6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`——再跑 `config.sh`；原样拷贝 `.runner`/`.credentials` 会使 `config.sh` 以 "already configured" 拒绝。`config.sh` 只完成注册，不启动监听进程，因此停在这一步的 runner 处于已注册但离线状态，不增加任何容量。还须安装并启动其服务：`sudo ./svc.sh install <user> && sudo ./svc.sh start`（本池由 systemd 管理；前台运行 `./run.sh` 亦可，但会随 shell 退出而终止）。确认该实例在组织 Settings → Actions → Runners 中显示 Idle 后再计入容量。每个约一分钟。
 
 
 ### 切回

From 7b49f4a4eeabc2970c4231ba26f5b2f9fbe56440 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 16:56:02 +0800
Subject: [PATCH 27/27] =?UTF-8?q?docs(ci):=20runbook=20=E2=80=94=20glob=20?=
 =?UTF-8?q?migrated=20identity=20files;=20scope=20the=20promise=20to=20ent?=
 =?UTF-8?q?erprise-pool=20outages?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- The clone exclusion is now '.runner*' (catches .runner_migrated /
  .credentials_migrated, which GitHub writes on migrated runners and
  which equally trip the already-configured refusal — hit live during
  the 2026-07-27 org-group migration).
- The problem statement now states the verdict runs on ubuntu-latest,
  scopes the switch to enterprise Linux-pool outages, and names the
  residual dependency: node-compat/python-sdk/windows stay standard-
  hosted, so a broader hosted-capacity failure still blocks the
  verdict, with Windows the only leg lacking any in-house substitute.
Static gate green locally: 32 passed, 0 failed.
---
 .../process/2026-07-26-ci-failover-runbook.i18n.yaml          | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.md     | 4 ++--
 .../implemented/process/2026-07-26-ci-failover-runbook.zh.md  | 4 ++--
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
index 0e3b0820c0..a7b9e827fa 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
-2026-07-26-ci-failover-runbook.md: 80dd7c4291e3de11c2f13b3247af56762396c720
-2026-07-26-ci-failover-runbook.zh.md: 7933a857f1559c540fccc2cd89352c4fe351dd7e
+2026-07-26-ci-failover-runbook.md: 4e4f8ea7fc60cf76fd8308147bbf7cc0bac74798
+2026-07-26-ci-failover-runbook.zh.md: bb7e43fe55c9cced51f042de6503978ec9349d6b
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
index 80dd7c4291..4e4f8ea7fc 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.md
@@ -6,7 +6,7 @@ English | [中文](2026-07-26-ci-failover-runbook.zh.md)
 
 ## Problem
 
-The three required Linux worker jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) and the required verdict job that aggregates them (`all checks passed`) run on the hosted enterprise 32-core pools. When those pools degrade — jobs queue indefinitely, the enterprise labels vanish, or GitHub-side capacity fails — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. An outage therefore needs a switch any responder with repository write access can throw without merging anything.
+The three required Linux worker jobs in [CI](../../../../.github/workflows/ci.yml) (`node 24 / static`, `node 24 / coverage`, `node 24 / snapshots and artifacts`) run on the hosted enterprise 32-core pools; the required verdict job that aggregates them (`all checks passed`) runs on standard `ubuntu-latest`. When the enterprise pools degrade — jobs queue indefinitely or the enterprise labels vanish — every open pull request becomes unmergeable, and the ordinary recovery of merging a fix is itself deadlocked behind the very required checks that cannot run. **Scope: this switch recovers an enterprise Linux-pool outage.** The verdict's other required dependencies (`node-compat`, `python-sdk`, `windows`) stay on standard hosted runners by design (the portable boundary); in a broader GitHub-hosted capacity failure that also takes out the standard pools, those dependencies still block `all checks passed`, and only the Windows leg has no in-house substitute at all — during the 2026-07-27 outage the standard pools recovered first, which is the ordering this design bets on. An outage therefore needs a switch any responder with repository write access can throw without merging anything.
 
 ## Decision
 
@@ -28,7 +28,7 @@ Each of the three required Linux worker jobs — and the `all checks passed` ver
 
 ## Capacity during failover
 
-Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` — then run `config.sh` (copying `.runner`/`.credentials` verbatim makes it refuse with "already configured"), and **start the listener**: `sudo ./svc.sh install ubuntu && sudo ./svc.sh start`. Registration alone leaves the runner offline; only a started service adds capacity. About a minute per instance.
+Six always-on instances absorb normal PR traffic (the pool's steady-state load is one serial standby job per master push, so failover capacity is effectively the full pool). If queues still build, register additional instances with an org registration token (org Settings → Actions → Runners → New runner). Clone an existing runner directory **excluding its identity files** — `rsync -a --exclude '.runner*' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/` (the globs also catch `.runner_migrated`/`.credentials_migrated`, which GitHub writes on migrated runners and which equally trigger the already-configured refusal) — then run `config.sh` (copying `.runner`/`.credentials` verbatim makes it refuse with "already configured"), and **start the listener**: `sudo ./svc.sh install ubuntu && sudo ./svc.sh start`. Registration alone leaves the runner offline; only a started service adds capacity. About a minute per instance.
 
 
 ### Switch back
diff --git a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
index 7933a857f1..bb7e43fe55 100644
--- a/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
+++ b/.agents/notes/implemented/process/2026-07-26-ci-failover-runbook.zh.md
@@ -6,7 +6,7 @@ Status: implemented
 
 ## 问题
 
-[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 工作作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）以及聚合它们的必需判定作业（`all checks passed`）运行在托管的企业级 32 核池上。当这些托管池发生故障——作业无限排队、企业标签消失或 GitHub 侧容量故障——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。因此故障需要一个任何具备仓库写权限的响应者都能在不合并任何代码的情况下触发的开关。
+[CI](../../../../.github/workflows/ci.yml) 中三个必需的 Linux 工作作业（`node 24 / static`、`node 24 / coverage`、`node 24 / snapshots and artifacts`）运行在托管的企业级 32 核池上；聚合它们的必需判定作业（`all checks passed`）运行在标准 `ubuntu-latest` 上。当企业池发生故障——作业无限排队或企业标签消失——所有开启的拉取请求都无法合并，而"合并一个修复"这一常规恢复手段本身正被那些无法运行的必需检查死锁。**适用范围：本切换恢复的是企业级 Linux 池故障。**判定作业的其余必需依赖（`node-compat`、`python-sdk`、`windows`）按设计留在标准托管运行器上（可移植边界）；若更大范围的 GitHub 托管容量故障连标准池一并击倒，这些依赖仍会阻塞 `all checks passed`，且只有 Windows 这条腿完全没有自有替代——2026-07-27 的故障中标准池率先恢复，本设计押注的正是这一顺序。因此故障需要一个任何具备仓库写权限的响应者都能在不合并任何代码的情况下触发的开关。
 
 ## 决策
 
@@ -28,7 +28,7 @@ Status: implemented
 
 ## 切换期间的容量
 
-6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`——再跑 `config.sh`（原样拷贝 `.runner`/`.credentials` 会使其以 "already configured" 拒绝），然后**启动监听器**：`sudo ./svc.sh install ubuntu && sudo ./svc.sh start`。仅注册不会上线；只有启动了服务的 runner 才会增加容量。每个约一分钟。
+6 个常驻实例可承接正常 PR 流量（该池平时唯一的稳态负载是每次 master 推送一个串行热备作业，故障切换时几乎全池可用）。若仍出现排队，用组织级注册 token（组织 Settings → Actions → Runners → New runner）追加注册实例。复制现有 runner 目录时**必须排除身份文件**——`rsync -a --exclude '.runner*' --exclude '.credentials*' --exclude '_diag' --exclude '_work' <src>/ <dst>/`（通配同时排除 `.runner_migrated`/`.credentials_migrated`——GitHub 会在迁移过的运行器上写入这些文件，它们同样会触发 already-configured 拒绝）——再跑 `config.sh`（原样拷贝 `.runner`/`.credentials` 会使其以 "already configured" 拒绝），然后**启动监听器**：`sudo ./svc.sh install ubuntu && sudo ./svc.sh start`。仅注册不会上线；只有启动了服务的 runner 才会增加容量。每个约一分钟。
 
 
 ### 切回