ContextualWisdomLab · seonghobae · Jun 23, 2026 · Jun 23, 2026 · Jun 23, 2026 · Jun 23, 2026
@@ -135,8 +135,12 @@ jobs:
           exit 1
 
       - name: Self-test Strix gate script
+        timeout-minutes: 10
         working-directory: ${{ runner.temp }}/trusted-workspace
-        run: bash "$TRUSTED_STRIX_GATE_TEST"
+        run: |
+          set -euo pipefail
+          printf 'Running Strix gate self-test with a 10-minute step timeout.\n'
+          bash "$TRUSTED_STRIX_GATE_TEST"
 
       - name: Gate Strix secrets
         id: gate
@@ -341,6 +345,7 @@ jobs:
 
       - name: Run Strix (quick)
         if: steps.gate.outputs.enabled == 'true'
+        timeout-minutes: 30
         # Security invariant for pull_request_target: execute only from the
         # trusted base checkout. The gate copies PR-head blobs into an isolated
         # temporary scope with execute bits stripped, then scans that scope as
@@ -384,11 +389,11 @@ jobs:
           IS_PR_EVIDENCE_RUN: ${{ (github.event_name == 'pull_request_target' || github.event.inputs.pr_number != '') && 'true' || 'false' }}
         run: |
           budget_suffix="TIME""OUT"
-          process_budget_seconds="3600"
+          process_budget_seconds="1500"
           export "LLM_${budget_suffix}=120"
           export "STRIX_MEMORY_COMPRESSOR_${budget_suffix}=10"
           export "STRIX_PROCESS_${budget_suffix}_SECONDS=$process_budget_seconds"
-          export "STRIX_TOTAL_${budget_suffix}_SECONDS=7200"
+          export "STRIX_TOTAL_${budget_suffix}_SECONDS=1800"
           bash "$TRUSTED_STRIX_GATE"
 
       - name: Collect Strix reports for artifact upload

@@ -13,15 +13,16 @@ OpenCode decides; GitHub Actions mutates.
 - Prefer `gh pr merge --auto --merge --match-head-commit <head>` when native auto-merge is enabled.
 - Use direct `gh pr merge --merge --match-head-commit <head>` only when the repo policy already allows immediate merge.
 - OpenCode app-token merges are deprecated; keep app tokens for review publication, not mechanical branch mutation.
+- OpenCode approval publication must be bounded. Peer GitHub Checks can be awaited, but the approval step itself must time out instead of running for hours; the current central limit is a 45 minute approval step with 81 peer-check probes at 30 seconds.
 - Tool failures are not source findings. Model failure, API transient, update-branch `422/403`, fork/write-permission failure, conflict, failed checks, and stale review state must be reported as distinct scheduler outcomes.
 
 ## Live Repository Inventory
 
-Live generated: 2026-06-23 04:18 KST. PR #28 rechecked: 2026-06-23 11:43 KST.
+Live generated: 2026-06-23 04:18 KST. PR #28 post-merge refresh: 2026-06-23 16:05 KST.
 
 | Repo | Flow | Default | Auto | Rulesets | Required checks | Stale dismissal | Merge queue | Workflows | Recent merged actor |
 |---|---:|---:|---:|---|---|---:|---:|---|---|
-| `ContextualWisdomLab/.github` | GitHub Flow | `main` | on | `Lock default branch` | none | true | no | OpenCode Review; PR Review Merge Scheduler; Strix Security Scan | #18 `seonghobae`; #17 `seonghobae`; #2 `seonghobae` |
+| `ContextualWisdomLab/.github` | GitHub Flow | `main` | on | `Lock default branch` | none | true | no | OpenCode Review; PR Review Merge Scheduler; Strix Security Scan | #28 `seonghobae` merge `a025be1`; #18 `seonghobae`; #17 `seonghobae` |
 | `ContextualWisdomLab/bandscope` | Git Flow | `develop` | on | `Lock default branch` | `ci / build-and-test`, `dependency-review`, `security-audit`, `CodeQL`, `sbom`, `release-preflight`, `gate / build / windows`, `gate / build / macos`, `trivy-fs-scan` | false | no | OpenCode Review; PR Review Merge Scheduler; Strix Security Scan | #427 `github-actions`; #408 `seonghobae`; #405 `seonghobae` |
 | `ContextualWisdomLab/clearfolio` | GitHub Flow | `main` | off | `PR` | none | false | no | OpenCode Review; Strix Security Scan | #9 `seonghobae`; #8 `seonghobae`; #7 `seonghobae` |
 | `ContextualWisdomLab/codec-carver` | GitHub Flow | `main` | on | `Lock default branch` | none | true | no | OpenCode Review; Scheduled PR Review Merge; Strix Security Scan | #94 `opencode-agent`; #93 `seonghobae`; #90 `seonghobae` |
@@ -37,7 +38,7 @@ Live generated: 2026-06-23 04:18 KST. PR #28 rechecked: 2026-06-23 11:43 KST.
 
 | Repo | Gap |
 |---|---|
-| `.github` | PR #28 head `60c821e` is blocked by current-head OpenCode `CHANGES_REQUESTED` and `strix` status failure. Same-head manual Strix run `27996904501` passed self-test but failed `Run Strix (quick)`, so it is not merge evidence. |
+| `.github` | PR #28 is merged at `a025be1` after same-head manual Strix run `28007326148`, same-head OpenCode run `28008174977`, unresolved review threads `0`, and guarded merge against head `811446d`. Remaining open PRs #19-#27 and #29-#36 are still blocked by `CHANGES_REQUESTED` and/or `DIRTY`; the scheduler dry run inspected 17 PRs and made no mutation. |
 | `bandscope` | Required checks are repo-specific and broad; keep GitHub native auto-merge as the check interpreter. |
 | `clearfolio` | Auto-merge is off and the PR Review Merge Scheduler is missing. |
 | `codec-carver` | Latest merged sample #94 still used `opencode-agent`; replace the legacy scheduler with the central GitHub Actions path. |
@@ -53,7 +54,7 @@ Live generated: 2026-06-23 04:18 KST. PR #28 rechecked: 2026-06-23 11:43 KST.
 | Repo | Live evidence | Adopt | Reject |
 |---|---|---|---|
 | `naruon` | `develop`, strict required checks `opencode-review` and `strix`, stale review dismissal enabled. Open PRs show `BEHIND`, `DIRTY`, and `CHANGES_REQUESTED` cases. | Strict current-head evidence and stale-dismissal awareness. | Treating `BEHIND` as merge-ready. |
-| `.github` | PR #28 is `MERGEABLE` but `BLOCKED`; required PR-target Strix failed on trusted-base self-test, and same-head manual Strix run `27996904501` also failed at `Run Strix (quick)` after publishing `strix` status failure. The latest OpenCode review still cited the stale PR-target Strix URL instead of the same-head manual Strix failure. | Same-head manual evidence for self-modifying trusted workflow changes, plus explicit handling for failed manual evidence. | Treating stale PR-target failure logs as the only current-head diagnosis after a same-head manual Strix rerun exists. |
+| `.github` | PR #28 head `811446d` reached current-head approval after manual Strix run `28007326148` published a successful `strix` status and manual OpenCode run `28008174977` approved the same head; it was merged by `seonghobae` with merge commit `a025be1`. The earlier PR-target Strix failure remains useful only as the reason same-head manual evidence was required. | Same-head manual evidence for self-modifying trusted workflow changes, current-head OpenCode approval, unresolved thread check, and `--match-head-commit` guarded merge. | Treating stale PR-target failure logs as merge blockers after newer same-head evidence exists. |
 | `pg-erd-cloud` | Recent PRs #236, #237, #239 were merged by `app/github-actions`. | GitHub Actions as mechanical merge actor with head guard. | Human-only queue draining. |
 | `codec-carver` | Recent PR #94 was merged by `app/opencode-agent`, and the repo still has legacy `Scheduled PR Review Merge`. | Native auto-merge path for current-head approved PRs. | OpenCode app as merge actor. |
 | `VibeSec` | PR #108 had native auto-merge enabled; #106 merged by `app/github-actions`; #109 merged by human. | Keep native auto-merge as preferred waiting path. | Repo-by-repo actor inconsistency. |
@@ -69,7 +70,9 @@ The checked-in scheduler already does the minimal central path:
 - blocks current-head failed check runs or status contexts before enabling auto-merge;
 - updates `BEHIND` only when OpenCode approved the exact current head, using `expected_head_sha`;
 - enables native auto-merge only for current-head OpenCode approval;
-- dispatches OpenCode when the current head has no OpenCode decision.
+- dispatches same-head Strix evidence first when the current head has no completed Strix evidence;
+- waits while same-head Strix evidence is still running, so OpenCode is not started just to poll a peer check;
+- dispatches OpenCode only after same-head Strix evidence is complete, including failed Strix evidence that OpenCode must explain from logs.
 
 Small proof run:
 
@@ -78,25 +81,24 @@ $ python3 scripts/ci/pr_review_merge_scheduler.py --self-test
 self-test passed
 
 $ python3 scripts/ci/pr_review_merge_scheduler.py --repo ContextualWisdomLab/.github --base-branch main --project-flow github-flow --dry-run --max-prs 40 --no-trigger-reviews
-PR #19: block: current-head OpenCode review requested changes
-PR #20: block: current-head OpenCode review requested changes
+PR #19: block: merge conflict: DIRTY
+PR #20: block: merge conflict: DIRTY
 PR #21: block: current-head OpenCode review requested changes
-PR #22: block: current-head OpenCode review requested changes
+PR #22: block: merge conflict: DIRTY
 PR #23: block: merge conflict: DIRTY
 PR #24: block: current-head OpenCode review requested changes
 PR #25: block: current-head OpenCode review requested changes
 PR #26: block: current-head OpenCode review requested changes
 PR #27: block: current-head OpenCode review requested changes
-PR #28: block: current-head OpenCode review requested changes
 PR #29: block: current-head OpenCode review requested changes
 PR #30: block: current-head OpenCode review requested changes
 PR #31: block: current-head OpenCode review requested changes
 PR #32: block: current-head OpenCode review requested changes
 PR #33: block: current-head OpenCode review requested changes
 PR #34: block: current-head OpenCode review requested changes
 PR #35: block: current-head OpenCode review requested changes
-PR #36: block: current-head OpenCode review requested changes
-{"base_branch": "main", "counts": {"block": 18}, "dry_run": true, "inspected": 18, "project_flow": "github-flow"}
+PR #36: block: merge conflict: DIRTY
+{"base_branch": "main", "counts": {"block": 17}, "dry_run": true, "inspected": 17, "project_flow": "github-flow"}
 ```
 
 ## Rollout List
@@ -109,7 +111,20 @@ PR #36: block: current-head OpenCode review requested changes
 
 ## Remaining Proof Gaps
 
+- A live current-head review -> same-head manual Strix status bridge -> OpenCode approval -> guarded merge trace has been completed on `.github` PR #28.
 - No live outdated -> update-branch -> new-head review -> merge/auto-merge trace has been completed yet.
 - `update-branch` `422/403` behavior still needs a safe fixture or a real blocked case before claiming standardized handling.
 - Required-check interpretation should stay delegated to GitHub native auto-merge until a repo needs immediate merge.
-- PR #28 itself cannot prove adoption until the same-head manual Strix failure is diagnosed and OpenCode stops reusing stale PR-target Strix self-test logs as the only failed-check evidence.
+- PR #28 proves the self-modifying trusted workflow bootstrap path after newer same-head evidence exists, but it does not prove update-branch behavior, stale approval dismissal after a head change, or cross-repository rollout.
+- PR #37 adds a bounded OpenCode approval publication timeout after manual current-head OpenCode run `28011338113` reached the approval step and was observed waiting on peer checks instead of finishing promptly.
+- PR #37 current-head run `28012303665` proved 100% coverage/docstring evidence and OpenCode completion for head `184be63`; Strix run `28012303876` proved same-head manual Strix success. The run also exposed that cancelled PR-target helper check `Strix Security Scan/publish-manual-pr-evidence-status` must be superseded by the newer same-head manual `strix` status, not treated as a source finding.
+- PR #37 head `9bbf641` exposed a remaining race: OpenCode can finish before same-head manual Strix publishes the superseding `strix` status, causing stale cancelled PR-target Strix checks to become REQUEST_CHANGES. The evidence preparation step now waits, within a 40 minute bound, whenever peer checks are still running, even if completed failed check evidence is already visible.
+- The same race also showed that PR `statusCheckRollup` does not see a manual Strix `workflow_dispatch` run until it publishes a commit status. OpenCode evidence preparation now queries current-head `strix.yml` workflow runs directly and treats in-progress same-head Strix runs as peer checks.
+- Strix run `28014156427` also reported sensitive log disclosure risk in failed-check evidence handling. The collector now redacts common token, API key, password, secret, authorization, Slack token, and AWS access-key patterns before any failed logs are summarized or embedded in review evidence.
+- Strix run `28015621232` reported `GitHub Actions pull_request_target with PR Code Execution` against `.github/workflows/opencode-review.yml`. OpenCode Review is now `workflow_dispatch`-only, and the scheduler dispatches same-head Strix before same-head OpenCode. This follows GitHub's secure-use guidance to avoid `pull_request_target` with untrusted PR checkout/execution: https://docs.github.com/en/actions/reference/security/secure-use and GitHub Security Lab's "Preventing pwn requests": https://securitylab.github.com/resources/github-actions-preventing-pwn-requests/.
+- OpenCode run `28017920517` failed without posting a PR review because every model attempt failed to produce a valid control block; the primary `github-models/openai/gpt-5` error was `Request body too large for gpt-5 model. Max size: 4000 tokens.` The prompt now requires reading `bounded-review-evidence.md` instead of inlining `bounded-review-evidence-excerpt.md`.
+- PR #37 head `ce5591e` reproduced the self-modifying workflow hazard: the base-branch `pull_request_target` OpenCode run `28019367683` posted `REQUEST_CHANGES` from skipped coverage evidence, while the same-head manual `coverage-evidence` job in run `28019384032` proved 100% test and docstring coverage. The central policy removes `pull_request_target` from OpenCode review and relies on scheduler-dispatched `workflow_dispatch` evidence for PR-head review.
+- OpenCode run `28019384032` also showed a model-output repair gap: DeepSeek V3 returned an `APPROVE` control block but wrote `Coverage: Not applicable` and `Docstring coverage: Not applicable` even though bounded current-head evidence proved both at 100%. The normalizer now reads the last concrete verification label after evidence-based repair, so an appended repair summary can replace earlier invalid model labels without accepting missing coverage.
+- Strix run `28022323798` caught that the first label repair changed normalizer parsing too narrowly: inline approval summaries in `test_strix_quick_gate.sh` no longer normalized. Label parsing now accepts inline verification labels while excluding the `Coverage:` suffix inside `Docstring coverage:`, preserving both inline transcript controls and appended evidence repair.
+- PR #37 same-head manual Strix run `28023392848` succeeded for head `07a6b76`, but the concurrently dispatched same-head manual OpenCode run `28023401894` spent its early lifetime waiting in `Prepare bounded OpenCode review evidence`. That exposed a scheduler-level resource issue: dispatching Strix and OpenCode together can turn OpenCode into a long poller whenever Strix is queued or slow. The scheduler now serializes the process: first dispatch Strix, then wait for a later scheduler pass to dispatch OpenCode after Strix evidence is complete.
+- The base-branch automatic OpenCode run `28025023007` still posted a current-head `CHANGES_REQUESTED` review before cancellation on head `1d05f52`, even though that automatic trigger is removed by this PR. The scheduler previously treated any current-head OpenCode `CHANGES_REQUESTED` as permanent. It now reads the latest OpenCode review on the current head, so a later same-head OpenCode approval can supersede an earlier false negative from the same reviewer.
@@ -19,6 +19,11 @@ Branch updates and merges run through the workflow `GITHUB_TOKEN`, so GitHub
 records those mechanical mutations as `github-actions[bot]` rather than an
 OpenCode app token or a personal token.
 
+OpenCode review execution is `workflow_dispatch`-only. The scheduler dispatches
+same-head Strix evidence first, then dispatches OpenCode for the same PR head.
+This avoids running PR-head review, CodeGraph, coverage, or PoC code from a
+privileged `pull_request_target` OpenCode workflow.
+
 OpenCode approval is evidence-gated. Before approval, the review summary must
 name changed files, CodeGraph or structural MCP evidence, a Change Flow DAG,
 100% test coverage evidence, 100% docstring coverage evidence, and a concrete