F6c: run every T9 suite in CI and add the live probe workflow

.github/workflows/agent-isolation.yml

@@ -1,44 +1,9 @@
 name: Agent isolation
+# Runs on every pull request and on pushes to the same branches as CI, so `real-docker` can be a required check.
 on:
   push:
-    branches: [main]
-    paths:
-      - 'agents/**'
-      - 'git/clone.ts'
-      - 'test/agent-*.test.ts'
-      - 'runner/question-*.ts'
-      - 'runner/questions.ts'
-      - 'runner/planning*.ts'
-      - 'runner/workspace.ts'
-      - 'runner/runner-repository.ts'
-      - 'runner/execution.ts'
-      - 'core/run-audit.ts'
-      - 'runner/review.ts'
-      - 'runner/store.ts'
-      - 'git/history.ts'
-      - 'test/runner-workspace.test.ts'
-      - 'test/vitest-config.test.ts'
-      - 'vitest.config.ts'
-      - '.github/workflows/agent-isolation.yml'
+    branches: [main, 'codex/**']
   pull_request:
-    paths:
-      - 'agents/**'
-      - 'git/clone.ts'
-      - 'test/agent-*.test.ts'
-      - 'runner/question-*.ts'
-      - 'runner/questions.ts'
-      - 'runner/planning*.ts'
-      - 'runner/workspace.ts'
-      - 'runner/runner-repository.ts'
-      - 'runner/execution.ts'
-      - 'core/run-audit.ts'
-      - 'runner/review.ts'
-      - 'runner/store.ts'
-      - 'git/history.ts'
-      - 'test/runner-workspace.test.ts'
-      - 'test/vitest-config.test.ts'
-      - 'vitest.config.ts'
-      - '.github/workflows/agent-isolation.yml'
 permissions:
   contents: read
 jobs:

.github/workflows/ci.yml

@@ -8,6 +8,9 @@ permissions:
 jobs:
   test:
     runs-on: ubuntu-latest
+    # The longest recent run took 20 min, about half of this limit. A hung browser suite would otherwise hold the
+    # runner for GitHub's default of 360 min.
+    timeout-minutes: 45
     steps:
       - uses: actions/checkout@v4
       - uses: actions/setup-node@v4
@@ -16,8 +19,9 @@ jobs:
           cache: npm
       - run: npm ci --ignore-scripts
       - run: npm run typecheck
-      # The Docker agent suites run one file at a time in the Agent isolation workflow; running them here
-      # would put them in parallel against the same image tag and daemon.
-      - run: npm test -- --exclude test/agent-container.test.ts --exclude test/agent-network.test.ts --exclude test/agent-adapter.test.ts --exclude test/agent-supervisor.test.ts --exclude test/agent-gate.test.ts --exclude test/agent-question.test.ts --exclude test/agent-planning.test.ts --exclude test/runner-workspace.test.ts
+      # The Docker-backed files run one at a time in the Agent isolation workflow; running them here would put them in
+      # parallel against the same image tag and daemon. Exclude exactly the dockerTests list in vitest.config.ts;
+      # test/vitest-config.test.ts fails if the two differ.
+      - run: npm test -- --exclude test/agent-container.test.ts --exclude test/agent-gate.test.ts --exclude test/agent-network.test.ts --exclude test/agent-planning.test.ts --exclude test/agent-question.test.ts --exclude test/agent-supervisor.test.ts --exclude test/runner-workspace.test.ts
       - run: npx playwright install --with-deps chromium
       - run: npm run test:browser

.github/workflows/live-probes.yml

@@ -0,0 +1,48 @@
+name: Live probes
+# Runs the live Claude startup probes against the real Claude service. It is not a pull request check: it runs
+# nightly on the default branch and when started by hand. See docs/implementation/ci.md.
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: '23 6 * * *'
+permissions:
+  contents: read
+concurrency:
+  group: live-probes
+  cancel-in-progress: false
+jobs:
+  live-probes:
+    runs-on: ubuntu-latest
+    # The probes' own time limits total 33 min (6, 8, 8 and 11). agent-container and agent-supervisor may each spend
+    # up to 10 min building the image before their probes; agent-question builds inside its probe's own 11 min. The
+    # teardown hooks have Vitest limits that total 9 min: agent-container 2 after each test and 2 at the end,
+    # agent-supervisor 3 and agent-question 2. That is 62 min, plus setup. These are only Vitest's limits: the afterAll
+    # hooks in agent-container and agent-question call docker through spawnSync with no timeout of their own, and
+    # Vitest's hook timer cannot fire while such a call blocks. The 75 min job limit caps the run.
+    timeout-minutes: 75
+    steps:
+      # Fail at once when the secret is missing, so a missing secret never looks like a passing run.
+      - name: Require the Claude token
+        env:
+          CLAUDE_CODE_OAUTH_TOKEN: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+        run: |
+          if [ -z "$CLAUDE_CODE_OAUTH_TOKEN" ]; then
+            echo "::error::The repository secret CLAUDE_CODE_OAUTH_TOKEN is not set. Add it to run the live probes."
+            exit 1
+          fi
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: '26.7.0'
+          cache: npm
+      - run: npm ci --ignore-scripts
+      # The docker project runs these files one at a time, as in the Agent isolation workflow. The filter keeps only
+      # the tests whose titles end in "(live Claude)".
+      - name: Run the live Claude probes
+        env:
+          CODEBOOST_RUN_AUTH_PROBES: '1'
+          CLAUDE_CODE_OAUTH_TOKEN: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+        run: npx vitest run --project docker -t '\(live Claude\)$' --reporter=default --reporter=json --outputFile=live-probes.json test/agent-container.test.ts test/agent-question.test.ts test/agent-supervisor.test.ts
+      # Vitest exits 0 when it skips every probe, so check that each probe passed and nothing else ran.
+      - name: Check that every probe passed
+        run: node scripts/live-probes.ts check live-probes.json

README.md

@@ -14,7 +14,7 @@ npm run typecheck
 npm test
 ```
 
-Tests create disposable local repositories and exercise command checks in disposable credential-free containers. They do not invoke model providers or access GitHub.
+Tests create disposable local repositories and exercise command checks in disposable credential-free containers. They do not invoke model providers or access GitHub. [CI](docs/implementation/ci.md) lists the workflows, the required checks and the live Claude probes.
 
 ## Guarded GitHub merge
 

docs/designs/codeboost-plan-indexed-review.md

@@ -1922,7 +1922,7 @@ Read each row left to right: finish and validate step 1 before step 2 within tha
 | C — guarded merge gate | **C1.** Required-check and branch-rule reads (T12). **C2.** Snapshot/evidence blockers, with unavailable T6 execution evidence blocking merge. **C3.** Head-pinned, base-protected merge and refusal handling (T7). **C4.** Review UI, race regressions and final #21 / PR #23 review. | Before C1, record B0 evidence for the foundation contracts C consumes; existing work must supply that evidence before C4 completion. Continue existing work rather than restarting implemented steps. Release shared runner/UI files after C4 merges. |
 | D — agent isolation | **D1.** Invocation contract and isolated task clone (T1). **D2.** Pinned, restricted container and startup self-test (T1). **D3.** Vendor-only egress and phase/tool enforcement (T2). **D4.** Claude/Codex adapters, cancellation settlement and bounded output. **D5.** Full real-Docker and hostile-input gate for this boundary (T9). | Can run alongside C and E. F requires D5 merged; G's production invocation requires D5. Add regressions with each step; D5 integrates them rather than postponing testing. |
 | E — planning logic | **E1.** Audit existing T18 schema/parser/prompt behavior and remaining #6 gaps. **E2.** Read-only authoring-provider contract and safe prompt/response handling. **E3.** Identity/revision-bound suggestion orchestration using the existing store interface. **E4.** Import, replay, malformed-response and hostile-input acceptance fixtures (T18). | Can run alongside C and D with injected providers. G consumes E4; live invocation waits for D5. Shared schema/store fixes must go through the assigned integration owner. |
-| F — runner and pre-merge automation | **F1.** Before implementation, publish and review the lifecycle/state-holder contract: pending, running, completed, failed, cancelled, stale and closing; legal transitions; ownership and settlement for persisted records, in-memory jobs, subprocesses, admitted HTTP requests and rendered UI; guarded retry; reject-admission → drain requests → cancel/await jobs → close storage. Then implement it and the feedback-event contract under the AGENTS.md async rules. **F2.** Per-item execution, review/reject rounds, pre-PR already-fixed checks, PR opening and hostile-issue eval (build step 5; T9). **F3.** Trusted rebase and ledger mapping (remaining T3). **F4.** Foreign-commit conflict handling (T11). **F5.** Post-rebase attribution/approval refresh and head-bound command execution (T6). **F6.** Required-check refresh, already-fixed check, guarded merge handoff and #22 integration regressions. F owns common CI after C: integrate every T9 suite (Docker, adapter, hostile-input/issue, recorded-output, unit and browser) into required CI, coordinating D's dedicated workflow. T9 remains incomplete until the combined head demonstrably runs and passes every suite. | Starts after C4 and D5 merge and B0 evidence is handed off for F's consumed contracts. Recheck that evidence against merged main before F1; existing T4/T5/T10 behavior is reused rather than rebuilt. F1 owns planning persistence/API additions needed by G. F2's working reject loop supplies the learning dependency. Ask already uses D's invocation contract (`runner/question-container.ts`), so F reuses that path rather than adding a second one. |
+| F — runner and pre-merge automation | **F1.** Before implementation, publish and review the lifecycle/state-holder contract: pending, running, completed, failed, cancelled, stale and closing; legal transitions; ownership and settlement for persisted records, in-memory jobs, subprocesses, admitted HTTP requests and rendered UI; guarded retry; reject-admission → drain requests → cancel/await jobs → close storage. Then implement it and the feedback-event contract under the AGENTS.md async rules. **F2.** Per-item execution, review/reject rounds, pre-PR already-fixed checks, PR opening and hostile-issue eval (build step 5; T9). **F3.** Trusted rebase and ledger mapping (remaining T3). **F4.** Foreign-commit conflict handling (T11). **F5.** Post-rebase attribution/approval refresh and head-bound command execution (T6). **F6.** Required-check refresh, already-fixed check, guarded merge handoff and #22 integration regressions. F owns common CI after C: integrate every T9 suite (Docker, adapter, hostile-input/issue, recorded-output, unit and browser) into required CI, coordinating D's dedicated workflow. T9 remains incomplete until the combined head demonstrably runs and passes every suite. *F6c status:* the workflows are in place (`docs/implementation/ci.md`); required-check settings, a passing run of both required jobs on one head and a green `Live probes` run remain. | Starts after C4 and D5 merge and B0 evidence is handed off for F's consumed contracts. Recheck that evidence against merged main before F1; existing T4/T5/T10 behavior is reused rather than rebuilt. F1 owns planning persistence/API additions needed by G. F2's working reject loop supplies the learning dependency. Ask already uses D's invocation contract (`runner/question-container.ts`), so F reuses that path rather than adding a second one. |
 | G — planning screen | **G1.** Import and plan display UI. **G2.** Authoring and suggestion cards. **G3.** Revision-bound Apply and draft/attachment preservation. **G4.** Real provider/store integration and complete T18 browser/adapter acceptance. | G1 starts after E4 and C4 merge; G1–G3 may use fixtures. G4 waits for D5 and F1's production planning API/persistence contract. Release shared web files after G4. |
 | H — issue prioritization | **H1.** Decide and record ranking policy. **H2.** Issue retrieval/normalization. **H3.** Deterministic ranking with reasons and failure/stale states. **H4.** Issue-list UI and end-to-end checks (build step 8). | H1–H3 can run alongside F/G after the issue-access contract is inspected. H4 waits for G4 to release shared web files. No existing T-ID covers this entire milestone. |
 | I — queue, schedule and recovery | **I1.** Queue admission and persisted transitions. **I2.** Run-window scheduling and cancellation. **I3.** Restart recovery, stale attempts and shutdown draining. **I4.** UI integration and controlled race acceptance (build step 7). | Starts after F6; owns shared runner/store files. UI work waits for G/H to release its exact files. No existing T-ID covers this entire milestone. |
@@ -1981,7 +1981,7 @@ F, G and H ran together within these ownership boundaries. G and H are done; for
 
 Built from this review's findings. Each task comes from a specific decision above. Run with Claude Code or Codex, and tick each one as you ship it. Effort ratios assumed: features about 30x, tests about 50x, architecture about 5x.
 
-**Status (checked against `main` on 2026-09-26, lane B0).** T1, T2, T4, T5, T10, T13 and T14 meet their Verify lines and are ticked, with evidence under each (T1, T2 and T14 re-checked 2026-09-26 after lane D merged). T9 stays open until lane F6 runs every suite in required CI. Ask calls the `agents/` boundary from `runner/question-container.ts`; it has no host adapter. The planned files `core/segments`, `core/choices` and `core/attribution` were never created. That logic lives in `core/linking.ts` (segments and ledger attribution) and `core/approvals.ts` (approvals and duplicate-segment choices). The Files lines below now name the real files.
+**Status (checked against `main` on 2026-09-26, lane B0).** T1, T2, T4, T5, T10, T13 and T14 meet their Verify lines and are ticked, with evidence under each (T1, T2 and T14 re-checked 2026-09-26 after lane D merged). T9 stays open. Lane F6 (common CI) has added the workflows in `docs/implementation/ci.md`: `test` and `real-docker` run on every pull request with no path filter, and `Live probes` runs the live Claude probes nightly and by hand. T9 can be ticked when a repository administrator requires `test` and `real-docker` on `main`, both pass on the same head, and a `Live probes` run passes with the `CLAUDE_CODE_OAUTH_TOKEN` secret. Ask calls the `agents/` boundary from `runner/question-container.ts`; it has no host adapter. The planned files `core/segments`, `core/choices` and `core/attribution` were never created. That logic lives in `core/linking.ts` (segments and ledger attribution) and `core/approvals.ts` (approvals and duplicate-segment choices). The Files lines below now name the real files.
 
 These `T` IDs are requirement identifiers, not the build-order numbers. Current merge-gate work is **build step 4, increment 1 (#21)** and spans parts of T6, T7, and T12; it is unrelated to the numbering of T4. See “Build step 4: scope and progress” for the current increment and remaining milestone criteria. An increment must not mark a broader requirement complete while any of its acceptance criteria remain deferred.
 

docs/implementation/agent-isolation.md

@@ -15,13 +15,16 @@ npm test
 `vitest.config.ts` finishes the ordinary test project first, then runs the Docker-backed
 files one at a time because they share one image tag and daemon. Use
 `npx vitest run --project docker` to rerun only that serial project. The `Agent isolation`
-workflow runs the complete gate. The main `CI` workflow skips the Docker project so that
-it can cover the ordinary project independently.
+workflow's `real-docker` job runs the complete gate on every pull request. The main `CI`
+workflow's `test` job skips the Docker-backed files so that it can cover the ordinary
+project independently. [CI](ci.md) describes both jobs.
 
-The live vendor probes need real credentials, so CI does not run them. To run them,
-set `CODEBOOST_RUN_AUTH_PROBES=1` and `CLAUDE_CODE_OAUTH_TOKEN`. Codex has no live
-probe while it is refused (#93). Do not put credentials in an issue, a pull request
-or chat.
+The live vendor probes need real credentials, so pull requests do not run them. The
+`Live probes` workflow runs them every night and when started by hand, with the
+repository secret `CLAUDE_CODE_OAUTH_TOKEN` ([CI](ci.md#live-probes)). To run them
+locally, set `CODEBOOST_RUN_AUTH_PROBES=1` and `CLAUDE_CODE_OAUTH_TOKEN`. Codex has no
+live probe while it is refused (#93). Do not put credentials in an issue, a pull
+request or chat.
 
 ## What the gate proves
 
@@ -407,10 +410,16 @@ against real Docker, including two reviews with different Ask owners on one daem
 
 ## Limits of this gate
 
-- CI does not run the live vendor probes. Run them locally with credentials before
-  a release that changes the image, the adapters or the prompts. Codex has no live
-  probe while it is refused in every phase (#93).
-- T9 as a whole is complete only when lane F runs every suite in required CI (F6).
+- Pull requests do not run the live vendor probes. The `Live probes` workflow runs them
+  every night and by hand, and fails if the `CLAUDE_CODE_OAUTH_TOKEN` secret is missing.
+  A pull request can therefore merge before a probe shows that it broke the Claude
+  startup path. Start the workflow by hand, or run the probes locally, before a release
+  that changes the image, the adapters or the prompts. Codex has no live probe while it
+  is refused in every phase (#93).
+- T9 as a whole is not complete yet. Lane F6 has added the workflows ([CI](ci.md)). What
+  remains: a repository administrator must require `test` and `real-docker` on `main`;
+  both jobs must pass on the same head after that; and a `Live probes` run must pass
+  once the secret exists.
 - Gitlink mounts and the pre-launch check apply to execute and fix only. In planning, questions and review, `/work`
   is a read-only volume, so nothing can be written beneath a gitlink there; that includes `check` attempts, which run
   as review. Running an item's `cmd` in a writable task filesystem is not built yet; when it is, each run must call