Compare commits

...

13 Commits

Author SHA1 Message Date
devops-engineer 7ccd59e630 Merge branch 'main' into fix/plugin-uninstall-exec-errors
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 4s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 5s
CI / Python Lint & Test (pull_request) Successful in 7s
E2E API Smoke Test / detect-changes (pull_request) Successful in 7s
Harness Replays / detect-changes (pull_request) Successful in 5s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 5s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 4s
Harness Replays / Harness Replays (pull_request) Successful in 2s
E2E Chat / detect-changes (pull_request) Successful in 13s
CI / Detect changes (pull_request) Successful in 14s
sop-checklist / review-refire (pull_request_target) Has been skipped
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 14s
sop-checklist / all-items-acked (pull_request) [info tier:low] acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4
sop-checklist / na-declarations (pull_request) N/A: (none)
qa-review / approved (pull_request_target) Failing after 7s
E2E Chat / E2E Chat (pull_request) Successful in 3s
gate-check-v3 / gate-check (pull_request_target) Successful in 10s
sop-checklist / all-items-acked (pull_request_target) Successful in 7s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 12s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 1s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 15s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 2s
CI / Canvas (Next.js) (pull_request) Successful in 6s
security-review / approved (pull_request_target) Failing after 11s
CI / Canvas Deploy Status (pull_request) Has been skipped
sop-tier-check / tier-check (pull_request_target) Failing after 16s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 58s
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1m7s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m15s
CI / Platform (Go) (pull_request) Successful in 6m36s
CI / all-required (pull_request) Successful in 3s
qa-review / approved (pull_request_review) Has been skipped
security-review / approved (pull_request_review) Has been skipped
sop-tier-check / tier-check (pull_request_review) Successful in 7s
audit-force-merge / audit (pull_request_target) Successful in 5s
2026-06-06 16:15:21 +00:00
devops-engineer d768d8667b Merge PR #2364 via Gitea merge queue
ci-arm64-advisory / fast-checks (push) Waiting to run
CI / Detect changes (push) Successful in 7s
Block internal-flavored paths / Block forbidden paths (push) Successful in 9s
CI / Python Lint & Test (push) Successful in 7s
E2E API Smoke Test / detect-changes (push) Successful in 7s
E2E Chat / detect-changes (push) Successful in 7s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (push) Successful in 5s
E2E Staging Canvas (Playwright) / detect-changes (push) Successful in 7s
Handlers Postgres Integration / detect-changes (push) Successful in 8s
CI / Platform (Go) (push) Successful in 6s
Secret scan / Scan diff for credential-shaped strings (push) Successful in 4s
CI / Shellcheck (E2E scripts) (push) Successful in 6s
CI / Canvas (Next.js) (push) Successful in 7s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (push) Successful in 8s
E2E API Smoke Test / E2E API Smoke Test (push) Successful in 2s
review-check-tests / review-check.sh regression tests (push) Successful in 7s
E2E Chat / E2E Chat (push) Successful in 5s
CI / Canvas Deploy Status (push) Successful in 3s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (push) Successful in 4s
CI / all-required (push) Successful in 3s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (push) Successful in 15s
Ops Scripts Tests / Ops scripts (unittest) (push) Successful in 55s
Handlers Postgres Integration / Handlers Postgres Integration (push) Successful in 1m5s
publish-workspace-server-image / build-and-push (push) Successful in 4m5s
publish-workspace-server-image / Production auto-deploy (push) Successful in 2m32s
Serialized merge by gitea-merge-queue after current-main, genuine approvals, and required CI checks were green.
2026-06-06 15:35:31 +00:00
devops-engineer dcf9d3cdc1 Merge branch 'main' into fix/plugin-uninstall-exec-errors
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 5s
CI / Python Lint & Test (pull_request) Successful in 3s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 5s
E2E API Smoke Test / detect-changes (pull_request) Successful in 7s
CI / Detect changes (pull_request) Successful in 11s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 5s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 7s
CI / Canvas (Next.js) (pull_request) Successful in 2s
E2E Chat / detect-changes (pull_request) Successful in 16s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 1s
Harness Replays / detect-changes (pull_request) Successful in 14s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 17s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 12s
qa-review / approved (pull_request_target) Successful in 5s
sop-checklist / review-refire (pull_request_target) Has been skipped
gate-check-v3 / gate-check (pull_request_target) Successful in 7s
CI / Canvas Deploy Status (pull_request) Has been skipped
security-review / approved (pull_request_target) Failing after 4s
Harness Replays / Harness Replays (pull_request) Successful in 1s
E2E Chat / E2E Chat (pull_request) Successful in 2s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 15s
sop-checklist / all-items-acked (pull_request) [info tier:low] acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4
sop-checklist / na-declarations (pull_request) N/A: (none)
sop-checklist / all-items-acked (pull_request_target) Successful in 19s
sop-tier-check / tier-check (pull_request_target) Failing after 18s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 17s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m9s
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1m8s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 1m30s
CI / Platform (Go) (pull_request) Successful in 3m52s
CI / all-required (pull_request) Successful in 2s
2026-06-06 13:30:27 +00:00
Molecule AI Dev Engineer A (Kimi) b1475b1f71 fix(ci): enforce official=true + current-head binding unconditionally in review-check.sh
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 9s
CI / Python Lint & Test (pull_request) Successful in 7s
CI / Detect changes (pull_request) Successful in 8s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 8s
E2E API Smoke Test / detect-changes (pull_request) Successful in 10s
E2E Chat / detect-changes (pull_request) Successful in 10s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 12s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 2s
CI / Canvas (Next.js) (pull_request) Successful in 3s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 9s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 15s
CI / Platform (Go) (pull_request) Successful in 7s
E2E Chat / E2E Chat (pull_request) Successful in 4s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 8s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 6s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 16s
CI / Canvas Deploy Status (pull_request) Has been skipped
review-check-tests / review-check.sh regression tests (pull_request) Successful in 10s
CI / all-required (pull_request) Successful in 4s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 6s
sop-checklist / review-refire (pull_request_target) Has been skipped
gate-check-v3 / gate-check (pull_request_target) Failing after 13s
sop-checklist / all-items-acked (pull_request) acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4 — body-unfilled: comprehensive-testing, local-postgres-e2
sop-checklist / na-declarations (pull_request) N/A: (none)
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 6s
sop-checklist / all-items-acked (pull_request_target) Successful in 8s
security-review / approved (pull_request_target) Successful in 9s
qa-review / approved (pull_request_target) Successful in 13s
sop-tier-check / tier-check (pull_request_target) Failing after 9s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m9s
Ops Scripts Tests / Ops scripts (unittest) (pull_request) Successful in 1m4s
qa-review / approved (pull_request_review) Has been skipped
security-review / approved (pull_request_review) Has been skipped
sop-tier-check / tier-check (pull_request_review) Failing after 11s
audit-force-merge / audit (pull_request_target) Successful in 9s
Addresses CR2 REQUEST_CHANGES on PR #2364:

1. Change select(.official != false) → select(.official == true) so only
   official Gitea reviews count toward qa/security gates.

2. Remove the REVIEW_CHECK_STRICT conditional and always require
   select(.commit_id == ) so stale approvals on old commits are
   rejected.

3. Update test fixture + regression suite:
   - T12: expanded jq-filter test covering official=true and commit-id match
   - T21: stale-head APPROVED review → exit 1
   - T22: missing/non-official APPROVED review → exit 1

All 44 tests pass locally.
2026-06-06 13:21:19 +00:00
Molecule AI Dev Engineer A (Kimi) b2d5f88f98 fix(ci): remove all comment-based approval bypasses from review-check.sh
Issue comments (both generic keywords APPROVED/LGTM/ACCEPTED and agent-
prefix comments [core-qa-agent]/[core-security-agent]) previously
satisfied the qa-review/security-review gate without an official Gitea
review. Both paths are bypasses:

1. Generic keywords: any team member could type 'APPROVED' in a comment.
2. Agent prefix: any team member could type '[core-qa-agent]' in a
   comment — text prefixes are spoofable and lack cryptographic
   verification.

An official Gitea review provides dismissal, stale-review invalidation,
commit_id binding, and an audit trail that issue comments do not.

Changes:
- Removed the entire issue-comments fallback section. Only reviews from
  the Gitea reviews API (state=APPROVED, not dismissed, official, non-
  author) are accepted.
- Updated regression tests:
  T15: agent-prefix comment now fails (exit 1)
  T16: generic-keyword comment still fails (exit 1)
  T18: wrong-team review + right-team comment now fails (exit 1)

Tests: 38 pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-06 13:21:19 +00:00
devops-engineer 31283a292a Merge PR #2362 via Gitea merge queue
ci-arm64-advisory / fast-checks (push) Waiting to run
Block internal-flavored paths / Block forbidden paths (push) Successful in 4s
CI / Python Lint & Test (push) Successful in 3s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (push) Has started running
publish-workspace-server-image / build-and-push (push) Has started running
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (push) Successful in 6s
Handlers Postgres Integration / detect-changes (push) Successful in 7s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (push) Successful in 5s
publish-workspace-server-image / Production auto-deploy (push) Blocked by required conditions
E2E Chat / detect-changes (push) Successful in 10s
E2E Staging Canvas (Playwright) / detect-changes (push) Successful in 11s
E2E API Smoke Test / detect-changes (push) Successful in 16s
CI / Detect changes (push) Successful in 18s
E2E Chat / E2E Chat (push) Successful in 3s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (push) Successful in 3s
Secret scan / Scan diff for credential-shaped strings (push) Successful in 6s
CI / Canvas (Next.js) (push) Successful in 2s
CI / Shellcheck (E2E scripts) (push) Successful in 3s
CI / Platform (Go) (push) Successful in 7s
E2E API Smoke Test / E2E API Smoke Test (push) Successful in 7s
CI / Canvas Deploy Status (push) Successful in 3s
CI / all-required (push) Successful in 2s
Ops Scripts Tests / Ops scripts (unittest) (push) Successful in 1m3s
Handlers Postgres Integration / Handlers Postgres Integration (push) Successful in 1m7s
Serialized merge by gitea-merge-queue after current-main, genuine approvals, and required CI checks were green.
2026-06-06 12:50:47 +00:00
Molecule AI Dev Engineer A (Kimi) bc7c45f3d6 fix(security): remove SOP_FAIL_OPEN bypass branches from sop-tier-check.sh (HIGH)
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
CI / Detect changes (pull_request) Successful in 6s
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 12s
CI / Python Lint & Test (pull_request) Successful in 11s
E2E API Smoke Test / detect-changes (pull_request) Successful in 5s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 13s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 5s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 6s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 18s
E2E Chat / detect-changes (pull_request) Successful in 22s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 18s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 7s
qa-review / approved (pull_request_target) Failing after 7s
gate-check-v3 / gate-check (pull_request_target) Successful in 16s
security-review / approved (pull_request_target) Failing after 7s
sop-checklist / review-refire (pull_request_target) Has been skipped
CI / Platform (Go) (pull_request) Successful in 3s
sop-checklist / all-items-acked (pull_request) acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4 — body-unfilled: comprehensive-testing, local-postgres-e2
CI / Canvas (Next.js) (pull_request) Successful in 4s
sop-checklist / na-declarations (pull_request) N/A: (none)
CI / Shellcheck (E2E scripts) (pull_request) Successful in 2s
sop-tier-check / tier-check (pull_request_target) Failing after 8s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 2s
E2E Chat / E2E Chat (pull_request) Successful in 2s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 1s
CI / Canvas Deploy Status (pull_request) Has been skipped
sop-checklist / all-items-acked (pull_request_target) Successful in 18s
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1s
CI / all-required (pull_request) Successful in 3s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m2s
Ops Scripts Tests / Ops scripts (unittest) (pull_request) Successful in 1m37s
qa-review / approved (pull_request_review) Has been skipped
security-review / approved (pull_request_review) Has been skipped
sop-tier-check / tier-check (pull_request_review) Failing after 11s
audit-force-merge / audit (pull_request_target) Successful in 6s
The workflow already removed SOP_FAIL_OPEN env (fix/core-ci-fail-closed,
2026-06-05), but the script still carried executable bypass branches that
would exit 0 when the env was set. Remove all of them:

- jq-install failure block
- token whoami failure block
- HEAD_SHA fetch failure block
- /orgs/{o}/teams fetch failure block
- /pulls/{n}/reviews fetch failure block

Every infra fault now fails closed (exit 1) with a loud ::error::,
exactly like a real SOP-6 violation.
2026-06-06 11:05:42 +00:00
devops-engineer bf0db08c7c Merge branch 'main' into fix/plugin-uninstall-exec-errors
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 6s
CI / Detect changes (pull_request) Successful in 8s
CI / Python Lint & Test (pull_request) Successful in 6s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 6s
Harness Replays / detect-changes (pull_request) Successful in 6s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 2s
E2E API Smoke Test / detect-changes (pull_request) Successful in 14s
CI / Canvas (Next.js) (pull_request) Successful in 10s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 17s
E2E Chat / detect-changes (pull_request) Successful in 18s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 4s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 6s
Harness Replays / Harness Replays (pull_request) Successful in 2s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 14s
qa-review / approved (pull_request_target) Successful in 7s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 16s
gate-check-v3 / gate-check (pull_request_target) Successful in 16s
sop-checklist / review-refire (pull_request_target) Has been skipped
CI / Canvas Deploy Status (pull_request) Has been skipped
security-review / approved (pull_request_target) Failing after 7s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 4s
sop-checklist / all-items-acked (pull_request) [info tier:low] acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4
sop-checklist / na-declarations (pull_request) N/A: (none)
E2E Chat / E2E Chat (pull_request) Successful in 3s
sop-checklist / all-items-acked (pull_request_target) Successful in 9s
sop-tier-check / tier-check (pull_request_target) Failing after 5s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 1m2s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m17s
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1m52s
CI / Platform (Go) (pull_request) Successful in 4m3s
CI / all-required (pull_request) Successful in 1s
2026-06-06 10:50:18 +00:00
devops-engineer e441def8b3 Merge PR #2356 via Gitea merge queue
ci-arm64-advisory / fast-checks (push) Waiting to run
Block internal-flavored paths / Block forbidden paths (push) Successful in 9s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (push) Successful in 14s
E2E API Smoke Test / detect-changes (push) Successful in 6s
E2E Chat / detect-changes (push) Successful in 6s
E2E Staging Canvas (Playwright) / detect-changes (push) Successful in 9s
Handlers Postgres Integration / detect-changes (push) Successful in 4s
Lint curl status-code capture / Scan workflows for curl status-capture pollution (push) Successful in 4s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (push) Successful in 3s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (push) Successful in 7s
lint-required-workflows-docker-host-pinned / Lint docker-host pin on docker-touching workflows (push) Successful in 10s
CI / Python Lint & Test (push) Successful in 1m8s
CI / Detect changes (push) Successful in 1m12s
E2E API Smoke Test / E2E API Smoke Test (push) Successful in 3s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (push) Successful in 4s
Secret scan / Scan diff for credential-shaped strings (push) Successful in 17s
E2E Chat / E2E Chat (push) Successful in 13s
CI / Platform (Go) (push) Successful in 2s
CI / Shellcheck (E2E scripts) (push) Successful in 1s
CI / Canvas (Next.js) (push) Successful in 17s
lint-continue-on-error-tracking / lint-continue-on-error-tracking (push) Successful in 1m15s
CI / all-required (push) Successful in 6s
Ops Scripts Tests / Ops scripts (unittest) (push) Successful in 1m3s
Lint workflow YAML (Gitea-1.22.6-hostile shapes) / Lint workflow YAML for Gitea-1.22.6-hostile shapes (push) Successful in 1m16s
CI / Canvas Deploy Status (push) Successful in 1s
Handlers Postgres Integration / Handlers Postgres Integration (push) Successful in 3m31s
publish-workspace-server-image / build-and-push (push) Successful in 8m38s
publish-workspace-server-image / Production auto-deploy (push) Successful in 3m11s
Serialized merge by gitea-merge-queue after current-main, genuine approvals, and required CI checks were green.
2026-06-06 10:13:50 +00:00
devops-engineer 51f83260df merge-queue: scan past non-ready candidates (HOL fix) + draft opt-out label
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 16s
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 4s
CI / Detect changes (pull_request) Successful in 7s
CI / Python Lint & Test (pull_request) Successful in 12s
E2E API Smoke Test / detect-changes (pull_request) Successful in 10s
E2E Chat / detect-changes (pull_request) Successful in 10s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 11s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 11s
Lint curl status-code capture / Scan workflows for curl status-capture pollution (pull_request) Successful in 5s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 4s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 12s
lint-required-workflows-docker-host-pinned / Lint docker-host pin on docker-touching workflows (pull_request) Successful in 5s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 6s
gate-check-v3 / gate-check (pull_request_target) Successful in 4s
qa-review / approved (pull_request_target) Failing after 5s
security-review / approved (pull_request_target) Failing after 5s
sop-checklist / all-items-acked (pull_request) acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4 — body-unfilled: comprehensive-testing, local-postgres-e2
sop-checklist / na-declarations (pull_request) N/A: (none)
sop-checklist / all-items-acked (pull_request_target) Successful in 5s
sop-checklist / review-refire (pull_request_target) Has been skipped
sop-tier-check / tier-check (pull_request_target) Failing after 4s
CI / Platform (Go) (pull_request) Successful in 1s
CI / Canvas (Next.js) (pull_request) Successful in 2s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 1s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 3s
Lint pre-flip continue-on-error / Verify continue-on-error flips have run-log proof (pull_request) Successful in 57s
E2E Chat / E2E Chat (pull_request) Successful in 3s
CI / all-required (pull_request) Successful in 7s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 2s
CI / Canvas Deploy Status (pull_request) Has been skipped
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m3s
Ops Scripts Tests / Ops scripts (unittest) (pull_request) Successful in 54s
lint-continue-on-error-tracking / lint-continue-on-error-tracking (pull_request) Successful in 1m24s
lint-required-context-exists-in-bp / lint-required-context-exists-in-bp (pull_request) Successful in 1m19s
Lint workflow YAML (Gitea-1.22.6-hostile shapes) / Lint workflow YAML for Gitea-1.22.6-hostile shapes (pull_request) Successful in 1m37s
qa-review / approved (pull_request_review) Has been skipped
security-review / approved (pull_request_review) Has been skipped
sop-tier-check / tier-check (pull_request_review) Failing after 4s
audit-force-merge / audit (pull_request_target) Successful in 15s
Researcher REQUEST_CHANGES (review 9085, head 0c311bbc) caught a real
head-of-line defect in the new auto-discovery: choose_next_candidate_issue()
selected only the OLDEST non-opted-out PR and process_once() evaluated just
that one per tick. A false candidate (e.g. #1519: open + unlabeled but
mergeable=false, current-head official REQUEST_CHANGES, <2 genuine approvals)
returns decision=wait and is re-selected every tick, HOL-blocking all newer
ready PRs forever.

Fix:
- Add choose_candidate_issues() returning the FULL FIFO-ordered eligible list;
  process_once() now SCANS THROUGH it, skipping any `wait` candidate
  (REQUEST_CHANGES / mergeable!=True / insufficient genuine approvals / red
  required CI) and acting on the first ACTIONABLE one (an `update` that advances
  a stale branch, or a fully-ready `merge`). A non-ready PR no longer blocks
  newer ready PRs. The merge bar is UNCHANGED and fail-closed: a skipped PR is
  never merged. Per-PR evaluation factored into _evaluate_candidate(); the
  permanent-permission HOLD path now `continue`s the scan instead of returning.
- Add literal `draft` to the default OPT_OUT_LABELS (Gitea draft STATE was
  already skipped; the label is an additional explicit human opt-out).

Tests (§SOP-22): non-ready oldest is skipped and a newer ready PR merges in the
same tick (no HOL); #1519-style false candidate is never merged and never
blocks; red-required-CI candidate skipped for the ready PR; all-unready merges
nothing; draft-label opt-out; choose_candidate_issues full-list ordering.
41 existing tests stay green (47 total).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 03:07:06 -07:00
devops-engineer 2fa68b1f23 merge-queue: auto-discovery (opt-OUT, label-optional) for self-sustaining autonomy
The external Gitea merge queue only considered PRs that already carried the
`merge-queue` label. Agent Gitea tokens lack `write:issue` (labels are
issue-scoped), so agents could never self-label a ready PR — the queue stalled
waiting on a human to add the label, blocking core-PR autonomy (#2355).

Fix: merge-on-criteria, label-optional. The cron now AUTO-DISCOVERS every open
same-repo PR and considers any that meets the unchanged merge bar. The
`merge-queue` label is now optional metadata, not a gate — this fully removes
the write:issue dependency (the cron itself never needs to add a label).

SAFETY is preserved as opt-OUT: a PR carrying any opt-out label
(`merge-queue-hold`, `do-not-auto-merge`, or `wip`) or marked draft is skipped
(never auto-considered, never merged). A human keeps a PR out of autonomous
merging by adding one of those labels. `AUTO_DISCOVER=0` restores legacy opt-IN.

The merge bar is UNCHANGED: still 2 genuine official approvals on the CURRENT
head from {agent-reviewer, agent-researcher, agent-reviewer-cr2}, all
branch-protection-required contexts green, mergeable=True (fail-closed on
None/False per #2349/#2352), and no open REQUEST_CHANGES. Auto-discovery only
changes WHICH PRs are considered, not whether they may merge.

- new `do-not-auto-merge` (id 78) + `wip` (id 79) repo labels
- `choose_next_candidate_issue` / `list_candidate_issues` for the opt-OUT,
  draft-skipping selection; legacy `choose_next_queued_issue` retained
- defensive opt-out/draft re-check on the live pull payload (stale-listing race)
- 15 new §SOP-22 regression tests; existing 26 kept green (41 total)
- workflow + runbook updated (AUTO_DISCOVER / OPT_OUT_LABELS documented)

Verified live (dry-run): auto-discovery selects unlabeled PR #1519 (the old
code never touched it); AUTO_DISCOVER=0 still selects only labeled #2346.

Helps #2355 (autonomy expansion).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-06 03:02:58 -07:00
Molecule AI Dev Engineer A (Kimi) 48b6011e17 fix(2047): pass workspaceID to stripPluginMarkersFromMemory
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
CI / Python Lint & Test (pull_request) Successful in 6s
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 8s
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 2s
CI / Detect changes (pull_request) Successful in 8s
E2E API Smoke Test / detect-changes (pull_request) Successful in 9s
Harness Replays / detect-changes (pull_request) Successful in 6s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 5s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 11s
gate-check-v3 / gate-check (pull_request_target) Successful in 5s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 9s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 13s
security-review / approved (pull_request_target) Failing after 5s
qa-review / approved (pull_request_target) Failing after 6s
E2E Chat / detect-changes (pull_request) Successful in 16s
CI / Canvas (Next.js) (pull_request) Successful in 1s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 1s
CI / Canvas Deploy Status (pull_request) Has been skipped
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 2s
E2E Chat / E2E Chat (pull_request) Successful in 2s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 13s
Harness Replays / Harness Replays (pull_request) Successful in 6s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Successful in 54s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 1m32s
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Successful in 1m49s
CI / Platform (Go) (pull_request) Successful in 3m54s
CI / all-required (pull_request) Successful in 7s
qa-review / approved (pull_request_review) Has been skipped
security-review / approved (pull_request_review) Has been skipped
sop-tier-check / tier-check (pull_request_review) Successful in 5s
sop-checklist / review-refire (pull_request_target) Has been skipped
sop-checklist / all-items-acked (pull_request) [info tier:low] acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4
sop-checklist / na-declarations (pull_request) N/A: (none)
sop-tier-check / tier-check (pull_request_target) Successful in 5s
sop-checklist / all-items-acked (pull_request_target) Successful in 8s
2026-06-05 04:09:20 +00:00
Molecule AI Dev Engineer A (Kimi) cc99d3fff4 fix(plugins): log silently ignored execAsRoot errors during uninstall
ci-arm64-advisory / fast-checks (pull_request) Waiting to run
Lint shellcheck (arm64 pilot) / shellcheck-arm64 (pilot) (pull_request) Successful in 1s
Block internal-flavored paths / Block forbidden paths (pull_request) Successful in 3s
CI / Python Lint & Test (pull_request) Successful in 3s
CI / Detect changes (pull_request) Successful in 6s
E2E API Smoke Test / detect-changes (pull_request) Successful in 5s
E2E Chat / detect-changes (pull_request) Successful in 5s
E2E Staging Canvas (Playwright) / detect-changes (pull_request) Successful in 5s
Handlers Postgres Integration / detect-changes (pull_request) Successful in 6s
Lint forbidden tenant-env keys / Scan workspace_secrets writers for forbidden env keys (pull_request) Successful in 3s
Lint forbidden tenant-env keys / Scan for repo-host token write into tenant workspace surface (pull_request) Successful in 2s
Secret scan / Scan diff for credential-shaped strings (pull_request) Successful in 4s
gate-check-v3 / gate-check (pull_request_target) Successful in 3s
Harness Replays / detect-changes (pull_request) Successful in 17s
qa-review / approved (pull_request_target) Failing after 4s
sop-checklist / review-refire (pull_request_target) Has been skipped
sop-checklist / all-items-acked (pull_request) acked: 0/7 — missing: comprehensive-testing, local-postgres-e2e, staging-smoke, +4
sop-checklist / na-declarations (pull_request) N/A: (none)
sop-checklist / all-items-acked (pull_request_target) Successful in 3s
security-review / approved (pull_request_target) Failing after 8s
sop-tier-check / tier-check (pull_request_target) Successful in 4s
CI / Canvas (Next.js) (pull_request) Successful in 1s
CI / Shellcheck (E2E scripts) (pull_request) Successful in 2s
E2E Chat / E2E Chat (pull_request) Successful in 2s
E2E Staging Canvas (Playwright) / Canvas tabs E2E (pull_request) Successful in 2s
lint-required-no-paths / lint-required-no-paths (pull_request) Successful in 58s
E2E API Smoke Test / E2E API Smoke Test (pull_request) Failing after 25s
CI / Platform (Go) (pull_request) Failing after 36s
CI / all-required (pull_request) Has been skipped
Harness Replays / Harness Replays (pull_request) Successful in 2s
CI / Canvas Deploy Status (pull_request) Has been skipped
Handlers Postgres Integration / Handlers Postgres Integration (pull_request) Failing after 48s
Plugin uninstall had two sites where execAsRoot errors were discarded:
- Skill directory removal (plugins_install.go:125) — orphaned skill dirs
  if rm -rf failed silently
- CLAUDE.md marker stripping (plugins_install_pipeline.go:326) — stale
  plugin content left in CLAUDE.md if awk script failed

Both now log the error without failing the overall uninstall (best-effort
 cleanup), giving operators visibility into incomplete uninstalls.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-05 03:52:07 +00:00
10 changed files with 1067 additions and 227 deletions
+325 -90
View File
@@ -4,7 +4,11 @@
Gitea 1.22.6+ has auto-merge (`pull_auto_merge`) but no GitHub-style merge
queue. This script provides the missing serialized policy in user space:
1. Pick the oldest open PR carrying QUEUE_LABEL (skipping HOLD_LABEL).
1. Scan open same-repo PRs that are NOT opted out (auto-discovery, see below),
oldest-first, skipping drafts, until an ACTIONABLE one is found. A non-ready
candidate (REQUEST_CHANGES, mergeable!=True, insufficient genuine approvals,
or red required CI) is SKIPPED so it cannot head-of-line block newer ready
PRs; the scan continues to the next candidate.
2. Refuse to act unless main's BP-required contexts are green.
3. Refuse fork PRs; the queue may only mutate same-repo branches.
4. If the PR branch does not contain current main, call Gitea's
@@ -29,13 +33,42 @@ Authoritative gates (fail-closed):
approvals present). It is NEVER used to bypass a failing REQUIRED context
or missing approvals.
Head-of-line (HOL) safety: a permanent permission/4xx merge error
(403/404/405) HOLDS the PR (applies HOLD_LABEL) so the queue advances to the
next PR instead of re-selecting the same wedged PR every tick. Likewise, a
persistent branch-update conflict (the /update endpoint returns HTTP 409
because the PR branch cannot be merged with main without manual rebase) HOLDS
the PR — a conflict will not self-resolve, so retrying it every tick would
HOL-block every ready PR behind it (issue #2352).
Auto-discovery (opt-OUT, label-optional):
The queue is SELF-SUSTAINING — a ready PR does NOT need a human (or an agent)
to add the `merge-queue` label first. When AUTO_DISCOVER is on (default), the
queue enumerates ALL open same-repo PRs and considers any that meets the full
merge bar (genuine approvals on current head + BP-required green + mergeable +
no open REQUEST_CHANGES). The merge bar above is UNCHANGED; auto-discovery only
changes WHICH PRs are considered, not whether they are mergeable.
This deliberately removes the historical dependency on an agent adding the
`merge-queue` label — agent Gitea tokens lack `write:issue` (labels are
issue-scoped), so they could never self-label and the queue stalled. The label
is now OPTIONAL metadata, not a gate.
SAFETY is preserved as opt-OUT: any PR carrying an opt-out label
(OPT_OUT_LABELS — `merge-queue-hold`, `do-not-auto-merge`, `wip`, `draft` by
default) is skipped (never auto-considered, never merged). Draft PRs
(draft=true STATE) are also skipped; the literal `draft` LABEL is an
additional explicit opt-out a human can apply without converting to a draft.
A human who wants to keep a PR out of autonomous merging just adds one of
those labels. Setting AUTO_DISCOVER=0 restores the legacy opt-IN behaviour
(only PRs already carrying QUEUE_LABEL are considered).
Head-of-line (HOL) safety has two complementary layers:
(a) The queue SCANS THROUGH the FIFO candidate list and skips any non-ready
PR (REQUEST_CHANGES, mergeable!=True, insufficient genuine approvals, or
red required CI) instead of locking on the oldest and waiting, so a PR
that can never become ready without human action does not block newer
ready PRs.
(b) For the candidate the scan acts on, two permanent failure modes HOLD the
PR (apply HOLD_LABEL) and let the scan CONTINUE to the next candidate
rather than re-selecting the same wedged PR every tick:
- a permanent permission/4xx merge error (403/404/405), and
- a persistent branch-update conflict (the /update endpoint returns
HTTP 409 because the PR branch cannot be merged with main without a
manual rebase). A conflict will not self-resolve, so retrying it
every tick would HOL-block every ready PR behind it (issue #2352).
Status-fetch is fail-closed: if the combined status for a sha cannot be
fetched, the PR is skipped this tick (never treated as green).
@@ -68,6 +101,33 @@ WATCH_BRANCH = _env("WATCH_BRANCH", default="main")
QUEUE_LABEL = _env("QUEUE_LABEL", default="merge-queue")
HOLD_LABEL = _env("HOLD_LABEL", default="merge-queue-hold")
UPDATE_STYLE = _env("UPDATE_STYLE", default="merge")
# Auto-discovery (opt-OUT). When truthy (default), the queue considers ALL open
# same-repo PRs that meet the merge bar, not only PRs already carrying
# QUEUE_LABEL — so the queue is self-sustaining without any human/agent labeling
# (agent tokens lack write:issue and cannot self-label). Set AUTO_DISCOVER=0 to
# restore the legacy opt-IN behaviour (QUEUE_LABEL required to be considered).
AUTO_DISCOVER = _env("AUTO_DISCOVER", default="1").strip().lower() not in {
"0",
"false",
"no",
"off",
"",
}
# Opt-OUT labels. A PR carrying ANY of these is skipped (never auto-considered,
# never merged) — the human escape hatch from autonomous merging. HOLD_LABEL is
# always included so the existing hold semantics keep working. `do-not-auto-merge`
# and `wip` let a human keep a PR out of the auto-merge path without removing it.
# `draft` is included as a literal label too: Gitea draft STATE (draft=true) is
# already skipped via _issue_is_draft, but a "draft" LABEL is an additional,
# explicit opt-out signal a human can apply without converting the PR to a draft.
OPT_OUT_LABELS = {
name.strip()
for name in _env(
"OPT_OUT_LABELS",
default="do-not-auto-merge,wip,draft",
).split(",")
if name.strip()
} | ({HOLD_LABEL} if HOLD_LABEL else set())
REQUIRED_CONTEXTS_RAW = _env(
"REQUIRED_CONTEXTS",
default=(
@@ -410,6 +470,85 @@ def choose_next_queued_issue(
return candidates[0] if candidates else None
def _issue_is_draft(issue: dict) -> bool:
"""True if the issue/PR is a draft.
The /issues listing exposes draft state under the `pull_request` sub-object
(`{"draft": true}`); some Gitea versions also surface a top-level `draft`.
Either is honoured. Drafts are never auto-considered for merging.
"""
pr = issue.get("pull_request")
if isinstance(pr, dict) and pr.get("draft") is True:
return True
return issue.get("draft") is True
def choose_candidate_issues(
issues: list[dict],
*,
queue_label: str,
opt_out_labels: set[str],
auto_discover: bool,
) -> list[dict]:
"""All open PRs eligible for a merge attempt this tick, oldest-first.
This is the auto-discovery selector. It does NOT change the merge bar — it
only changes WHICH PRs are considered:
- auto_discover=True (default): every open same-repo PR is a candidate,
EXCEPT those carrying an opt-out label or marked draft. The QUEUE_LABEL
is optional metadata, not a gate, so a ready PR reaches the queue with no
human/agent labeling (the write:issue gap is removed).
- auto_discover=False: legacy opt-IN — only PRs carrying queue_label are
candidates (still skipping opt-out labels and drafts).
Opt-out is the safety escape hatch: any opt_out_labels member present skips
the PR entirely (never considered, never merged). Ordering is oldest-first
(created_at, then number) to preserve the serialized FIFO ordering.
Returns the FULL ordered list (not just the head) so process_once can SCAN
THROUGH non-ready candidates instead of locking on the oldest. A non-ready
auto-discovered PR (e.g. one with REQUEST_CHANGES or mergeable=false, which
can never become ready without human action) must NOT head-of-line block the
newer ready PRs behind it — the readiness check happens per-candidate in
process_once, and a `wait` candidate is skipped to the next one.
"""
candidates = []
for issue in issues:
if "pull_request" not in issue:
continue
labels = label_names(issue)
if opt_out_labels & labels:
continue # opt-out: human kept this PR out of autonomous merging
if _issue_is_draft(issue):
continue # drafts are never auto-merged
if not auto_discover and queue_label not in labels:
continue # legacy opt-IN: require the queue label
candidates.append(issue)
candidates.sort(key=lambda issue: (issue.get("created_at") or "", int(issue["number"])))
return candidates
def choose_next_candidate_issue(
issues: list[dict],
*,
queue_label: str,
opt_out_labels: set[str],
auto_discover: bool,
) -> dict | None:
"""The oldest eligible candidate, or None. Thin head-of-list wrapper around
choose_candidate_issues; retained for callers/tests that only want the head.
process_once uses the full list (choose_candidate_issues) so it can scan past
non-ready PRs rather than HOL-block on the oldest."""
candidates = choose_candidate_issues(
issues,
queue_label=queue_label,
opt_out_labels=opt_out_labels,
auto_discover=auto_discover,
)
return candidates[0] if candidates else None
def pr_contains_base_sha(commits: list[dict], base_sha: str) -> bool:
for commit in commits:
sha = commit.get("sha") or commit.get("id")
@@ -577,6 +716,31 @@ def list_queued_issues() -> list[dict]:
return body
def list_candidate_issues(*, auto_discover: bool) -> list[dict]:
"""Open PR issues eligible for consideration this tick.
With auto_discover=True (default) this enumerates ALL open PRs (no label
filter) so the queue is self-sustaining — a ready PR is considered without
any human/agent first adding QUEUE_LABEL. With auto_discover=False it falls
back to the legacy label-filtered listing (opt-IN). Opt-out filtering and
draft-skipping happen in choose_next_candidate_issue, not here.
"""
if not auto_discover:
return list_queued_issues()
_, body = api(
"GET",
f"/repos/{OWNER}/{NAME}/issues",
query={
"state": "open",
"type": "pulls",
"limit": "50",
},
)
if not isinstance(body, list):
raise ApiError("candidate issues response not list")
return body
def get_pull(pr_number: int) -> dict:
_, body = api("GET", f"/repos/{OWNER}/{NAME}/pulls/{pr_number}")
if not isinstance(body, dict):
@@ -731,45 +895,181 @@ def process_once(*, dry_run: bool = False) -> int:
print(f"::notice::queue paused: {WATCH_BRANCH}@{main_sha[:8]} required contexts not green: {', '.join(main_bad)}")
return 0
issue = choose_next_queued_issue(
list_queued_issues(),
candidates = choose_candidate_issues(
list_candidate_issues(auto_discover=AUTO_DISCOVER),
queue_label=QUEUE_LABEL,
hold_label=HOLD_LABEL,
opt_out_labels=OPT_OUT_LABELS,
auto_discover=AUTO_DISCOVER,
)
if not issue:
print("::notice::merge queue empty")
if not candidates:
print(
"::notice::no merge candidates "
f"(auto_discover={'on' if AUTO_DISCOVER else 'off'})"
)
return 0
# HOL fix: SCAN THROUGH the FIFO candidate list until a PR we can ACT on is
# found, instead of locking on the oldest and waiting. A non-ready candidate
# (decision.action == "wait": REQUEST_CHANGES, mergeable!=True, insufficient
# genuine approvals, or red required CI) is SKIPPED — it must NOT head-of-line
# block the newer ready PRs behind it. The merge bar is unchanged: a skipped
# PR is never merged, and the first ACTIONABLE candidate (an "update" that
# advances a stale branch, or a fully-ready "merge") terminates the scan.
#
# `update` is treated as actionable, not skippable: a PR whose head merely
# lacks current main is in a legitimate in-progress state (updating it +
# rerunning CI moves it toward ready), unlike a PR that can never become
# ready without a human (RC / conflict), which is a `wait` and gets skipped.
for issue in candidates:
decision, ctx = _evaluate_candidate(
issue,
main_sha=main_sha,
main_status=main_status,
required_contexts=contexts,
required_approvals=required_approvals,
dry_run=dry_run,
)
if decision is None:
continue # not merge-eligible (not-open / opted-out / fork / wrong base)
pr_number = ctx["pr_number"]
print(f"::notice::PR #{pr_number} decision={decision.action}: {decision.reason}")
if decision.action == "wait":
# Non-ready: skip to the next candidate (no HOL block, no merge).
continue
if decision.action == "update":
try:
update_pull(pr_number, dry_run=dry_run)
except BranchUpdateConflictError as exc:
# The branch cannot be updated with main because of a real
# conflict (HTTP 409 from /update). This is the #2352 HOL guard:
# a conflict will not self-resolve without a human/agent rebase,
# so re-attempting the update every tick would head-of-line block
# every ready PR behind it. HOLD this PR (apply HOLD_LABEL, which
# is an opt-out label so later ticks skip it) and CONTINUE the
# scan so a newer ready PR can still merge this tick. Fail-closed:
# a held PR is skipped, never merged.
sys.stderr.write(
f"::error::branch-update conflict for PR #{pr_number}: {exc}\n"
)
hold_note = (
"merge-queue: could not update this branch with "
f"`{WATCH_BRANCH}` — the update returned a merge conflict "
f"(HTTP 409) that the queue cannot auto-resolve ({exc}). "
f"Applied `{HOLD_LABEL}` to unblock the queue (HOL guard). "
f"Fix: rebase/merge `{WATCH_BRANCH}` into this branch and "
f"resolve the conflicts, then remove `{HOLD_LABEL}` to requeue."
)
hold_pr(pr_number, hold_note, dry_run=dry_run)
continue # held — keep scanning for a mergeable candidate
post_comment(
pr_number,
(
f"merge-queue: updated this branch with `{WATCH_BRANCH}` at "
f"`{main_sha[:12]}`. Waiting for CI on the refreshed head."
),
dry_run=dry_run,
)
return 0
if decision.ready:
latest_main_sha = get_branch_head(WATCH_BRANCH)
if latest_main_sha != main_sha:
print(
f"::notice::main moved {main_sha[:8]} -> {latest_main_sha[:8]}; "
"deferring to next tick"
)
return 0
try:
merge_pull(pr_number, dry_run=dry_run, force=decision.force)
except MergePermissionError as exc:
# Permanent merge failure (HTTP 403/404/405). HOLD this PR by
# applying HOLD_LABEL (it becomes an opt-out label, so subsequent
# ticks skip it) and CONTINUE scanning so the queue still advances
# to the next ready PR this tick rather than stalling.
sys.stderr.write(f"::error::merge permission error for PR #{pr_number}: {exc}\n")
hold_note = (
"merge-queue: merge failed with a permanent permission error "
f"({exc}). No available token has Can-merge permission for this "
f"PR. Applied `{HOLD_LABEL}` to unblock the queue (HOL guard). "
f"Fix: grant Can-merge to the queue token, then remove "
f"`{HOLD_LABEL}` to requeue."
)
try:
add_label_by_name(pr_number, HOLD_LABEL, dry_run=dry_run)
except ApiError as label_exc:
# If we cannot even apply the hold label, fall back to a comment
# so the wedge is at least visible; do NOT loop on this PR.
sys.stderr.write(
f"::error::could not apply HOLD_LABEL to PR #{pr_number}: {label_exc}\n"
)
hold_note += (
f"\n\n(NOTE: could not apply the hold label automatically: "
f"{label_exc}. Please add `{HOLD_LABEL}` manually.)"
)
post_comment(pr_number, hold_note, dry_run=dry_run)
continue # held — keep scanning for a mergeable candidate
return 0
return 0
def _evaluate_candidate(
issue: dict,
*,
main_sha: str,
main_status: dict,
required_contexts: list[str],
required_approvals: int,
dry_run: bool,
) -> tuple[MergeDecision | None, dict]:
"""Evaluate a single auto-discovered candidate against the full merge bar.
Returns (decision, ctx) where ctx carries {"pr_number"}. A None decision
means the PR is not merge-eligible at all (not open / opted-out / draft /
fork / wrong base) and the caller should skip to the next candidate; for
fork / wrong-base the explanatory comment is posted here before returning.
The merge bar is UNCHANGED from the single-PR path — this only factors the
per-PR evaluation out so process_once can scan multiple candidates. A failed
status fetch still raises (fail-closed): it propagates to the caller so the
PR is never treated as green.
"""
pr_number = int(issue["number"])
ctx = {"pr_number": pr_number}
pr = get_pull(pr_number)
if pr.get("state") != "open":
print(f"::notice::PR #{pr_number} is not open; skipping")
return 0
return None, ctx
# Defensive opt-out/draft re-check on the authoritative pull payload: the
# /issues listing's label/draft view can lag, but the merge bar must respect
# the live pull state. (choose_candidate_issues already filtered on the
# listing; this guards against a stale listing racing a just-added opt-out.)
if OPT_OUT_LABELS & label_names(pr):
print(f"::notice::PR #{pr_number} carries an opt-out label; skipping")
return None, ctx
if pr.get("draft") is True:
print(f"::notice::PR #{pr_number} is a draft; skipping")
return None, ctx
if pr.get("base", {}).get("ref") != WATCH_BRANCH:
post_comment(pr_number, f"merge-queue: skipped; base branch is not `{WATCH_BRANCH}`.", dry_run=dry_run)
return 0
return None, ctx
if pr.get("head", {}).get("repo_id") != pr.get("base", {}).get("repo_id"):
post_comment(pr_number, "merge-queue: skipped; fork PRs are not supported by the serialized queue.", dry_run=dry_run)
return 0
return None, ctx
head_sha = pr.get("head", {}).get("sha")
if not isinstance(head_sha, str) or len(head_sha) < 7:
raise ApiError(f"PR #{pr_number} missing head sha")
commits = get_pull_commits(pr_number)
current_base = pr_has_current_base(pr, commits, main_sha)
# Fail-closed: a failed status fetch raises here and the tick is skipped
# (the PR is never treated as green).
# Fail-closed: a failed status fetch raises here and propagates (the PR is
# never treated as green).
pr_status = get_combined_status(head_sha)
pr_labels = label_names(pr)
# FAIL-CLOSED: Gitea returns mergeable=None (or omits the field) while it is
# still COMPUTING conflict state. Only the literal True is decisive proof the
# PR is conflict-free; None and False both mean "not (yet) mergeable". We must
# NOT autonomously merge on an unknown — treat anything but True as not-yet-
# mergeable so evaluate_merge_readiness returns a transient "wait" decision.
# This is transient: process_once returns 0 (no hold label, no dequeue) and
# the PR is re-checked next tick once Gitea has finished computing mergeability.
mergeable_field = pr.get("mergeable")
mergeable = mergeable_field is True
# mergeable so evaluate_merge_readiness returns a "wait" decision.
mergeable = pr.get("mergeable") is True
reviews = get_pull_reviews(pr_number)
approvers, request_changes = genuine_approvals(
@@ -779,7 +1079,7 @@ def process_once(*, dry_run: bool = False) -> int:
decision = evaluate_merge_readiness(
main_status=main_status,
pr_status=pr_status,
required_contexts=contexts,
required_contexts=required_contexts,
required_approvals=required_approvals,
approvers=approvers,
request_changes=request_changes,
@@ -787,72 +1087,7 @@ def process_once(*, dry_run: bool = False) -> int:
mergeable=mergeable,
pr_labels=pr_labels,
)
print(f"::notice::PR #{pr_number} decision={decision.action}: {decision.reason}")
if decision.action == "update":
try:
update_pull(pr_number, dry_run=dry_run)
except BranchUpdateConflictError as exc:
# The branch cannot be updated with main because of a real conflict
# (HTTP 409). This is the HOL fix for issue #2352: previously the
# 409 propagated to main() and the tick exited 0 with the PR still
# queued, so the NEXT tick re-selected the SAME conflicted PR and
# retried the failing update forever — head-of-line-blocking every
# ready PR behind it. A conflict will not self-resolve; it needs a
# human/agent rebase. So HOLD this PR (HOL guard) and advance to the
# next candidate. Fail-closed: a held PR is skipped, never merged.
sys.stderr.write(
f"::error::branch-update conflict for PR #{pr_number}: {exc}\n"
)
hold_note = (
"merge-queue: could not update this branch with "
f"`{WATCH_BRANCH}` — the update returned a merge conflict "
f"(HTTP 409) that the queue cannot auto-resolve ({exc}). "
f"Applied `{HOLD_LABEL}` to unblock the queue (HOL guard). "
f"Fix: rebase/merge `{WATCH_BRANCH}` into this branch and "
f"resolve the conflicts, then remove `{HOLD_LABEL}` to requeue."
)
hold_pr(pr_number, hold_note, dry_run=dry_run)
return 0
post_comment(
pr_number,
(
f"merge-queue: updated this branch with `{WATCH_BRANCH}` at "
f"`{main_sha[:12]}`. Waiting for CI on the refreshed head."
),
dry_run=dry_run,
)
return 0
if decision.ready:
latest_main_sha = get_branch_head(WATCH_BRANCH)
if latest_main_sha != main_sha:
print(
f"::notice::main moved {main_sha[:8]} -> {latest_main_sha[:8]}; "
"deferring to next tick"
)
return 0
try:
merge_pull(pr_number, dry_run=dry_run, force=decision.force)
except MergePermissionError as exc:
# Permanent merge failure (HTTP 403/404/405). This is the
# head-of-line (HOL) bug fix: previously we returned 0 with the PR
# still queued, so the next tick re-selected the SAME wedged PR
# forever and the queue never advanced. Instead, HOLD this PR by
# applying HOLD_LABEL (choose_next_queued_issue skips held PRs), so
# the queue moves on to the next candidate. A maintainer removes
# the hold once the permission issue is fixed.
sys.stderr.write(f"::error::merge permission error for PR #{pr_number}: {exc}\n")
hold_note = (
"merge-queue: merge failed with a permanent permission error "
f"({exc}). No available token has Can-merge permission for this "
f"PR. Applied `{HOLD_LABEL}` to unblock the queue (HOL guard). "
f"Fix: grant Can-merge to the queue token, then remove "
f"`{HOLD_LABEL}` to requeue."
)
hold_pr(pr_number, hold_note, dry_run=dry_run)
return 0
return 0
return 0
return decision, ctx
def main() -> int:
+14 -53
View File
@@ -197,19 +197,15 @@ if [ "$HTTP_CODE" != "200" ]; then
exit 1
fi
# Filter: state=APPROVED, not-dismissed, non-author. Optionally strict-mode
# adds commit_id==head.sha (off by default; see header).
# Filter: state=APPROVED, official=true, not-dismissed, non-author,
# commit_id matches current PR head. All conditions are mandatory.
JQ_FILTER='.[]
| select(.state == "APPROVED")
| select(.official == true)
| select(.dismissed != true)
| select(.official != false)
| select(.user.login != $author)'
if [ "${REVIEW_CHECK_STRICT:-}" = "1" ]; then
JQ_FILTER="${JQ_FILTER}
| select(.commit_id == \$head)"
fi
JQ_FILTER="${JQ_FILTER}
| .user.login"
| select(.user.login != $author)
| select(.commit_id == $head)
| .user.login'
REVIEW_CANDIDATES=$(jq -r --arg author "$PR_AUTHOR" --arg head "$PR_HEAD_SHA" "$JQ_FILTER" "$REVIEWS_JSON" | sort -u)
debug "candidate non-author approvers: $(echo "$REVIEW_CANDIDATES" | tr '\n' ' ')"
@@ -241,49 +237,14 @@ if [ -z "$REVIEW_CANDIDATES" ]; then
fi
# --- Fallback/extension (internal#348): check issue comments for agent-approval ---
# core-qa-agent and core-security-agent can approve via issue comments. Always
# include comment candidates, even if the reviews API returned approvals for a
# different team; team membership below is the authoritative filter.
COMMENT_CANDIDATES=""
AGENT_PATTERN=""
case "$TEAM" in
qa) AGENT_PATTERN="\\[core-qa-agent\\]" ;;
security) AGENT_PATTERN="\\[core-security-agent\\]" ;;
esac
HTTP_CODE=$(curl -sS -o "$COMMENTS_JSON" -w '%{http_code}' \
-K "$CURL_AUTH_FILE" "${API}/repos/${OWNER}/${NAME}/issues/${PR_NUMBER}/comments")
debug "GET /issues/${PR_NUMBER}/comments → HTTP ${HTTP_CODE}"
if [ "$HTTP_CODE" = "200" ]; then
# JQ expression: select non-author comments that match either the
# agent-prefix pattern (case-insensitive) OR a generic approval keyword.
JQ_APPROVALS='
.[] |
select(.user.login != $author) |
. as $cmt |
if ($agent_pattern | length) > 0 and ($cmt.body // "" | test($agent_pattern; "i")) then
$cmt.user.login
elif ($cmt.body // "" | test("\\b(APPROVED|LGTM|ACCEPTED)\\b"; "i")) then
$cmt.user.login
else
empty
end
'
COMMENT_CANDIDATES=$(jq -r \
--arg author "$PR_AUTHOR" \
--arg agent_pattern "$AGENT_PATTERN" \
"$JQ_APPROVALS" \
"$COMMENTS_JSON" 2>/dev/null | sort -u)
debug "comment-based approval candidates: $(echo "$COMMENT_CANDIDATES" | tr '\n' ' ')"
if [ -n "$COMMENT_CANDIDATES" ]; then
echo "::notice::${TEAM}-review: found $(echo "$COMMENT_CANDIDATES" | wc -w | xargs) comment-based approval candidate(s) — verifying team membership..."
fi
else
debug "could not fetch issue comments (HTTP ${HTTP_CODE})"
fi
CANDIDATES=$(printf '%s\n%s\n' "$REVIEW_CANDIDATES" "$COMMENT_CANDIDATES" | sed '/^$/d' | sort -u)
# --- COMMENT APPROVAL REMOVED (security hardening) ---
# Previous versions accepted issue comments containing generic approval
# keywords (APPROVED/LGTM/ACCEPTED) or agent prefixes ([core-qa-agent],
# [core-security-agent]) as satisfying the gate. Both paths are bypasses:
# a comment lacks the audit trail, dismissal, stale-review invalidation,
# and commit_id binding that an official Gitea review provides.
# Only APPROVED reviews from the Gitea reviews API count.
CANDIDATES="$REVIEW_CANDIDATES"
if [ -z "${CANDIDATES:-}" ]; then
echo "::error::${TEAM}-review awaiting non-author APPROVE from ${TEAM} team (no candidates from reviews API or issue comments)"
+2 -26
View File
@@ -48,7 +48,6 @@ set -euo pipefail
# workflow-level jq install can fail on runners with network restrictions
# (GitHub releases not reachable from some runner networks — infra#241
# follow-up). This fallback is idempotent — no-op when jq is already on PATH.
# SOP_FAIL_OPEN=1 makes this always exit 0 so CI never blocks on jq absence.
if ! command -v jq >/dev/null 2>&1; then
echo "::notice::jq not found on PATH — attempting install..."
_jq_installed="no"
@@ -67,12 +66,6 @@ if ! command -v jq >/dev/null 2>&1; then
if ! command -v jq >/dev/null 2>&1; then
echo "::error::jq installation failed — apt-get and GitHub binary both failed."
echo "::error::sop-tier-check requires jq for all JSON API parsing."
# SOP_FAIL_OPEN=1 is set in the workflow step's env — makes script always
# exit 0 so CI never blocks. The SOP-6 tier review gate remains enforced.
if [ "${SOP_FAIL_OPEN:-}" = "1" ]; then
echo "::warning::SOP_FAIL_OPEN=1 — exiting 0 so CI does not block."
exit 0
fi
exit 1
fi
fi
@@ -101,15 +94,10 @@ echo "::notice::tier-check start: repo=$OWNER/$NAME pr=$PR_NUMBER author=$PR_AUT
# cause the script to exit prematurely when the token is empty/invalid — the
# if check below handles that case gracefully. Without || true, a 401 from an
# empty/invalid token causes jq to exit 1, triggering set -e and exiting the
# entire script before SOP_FAIL_OPEN can be evaluated (the check is in the jq-
# install block; if jq is already on PATH, that block is skipped entirely).
# entire script before the error can be logged.
WHOAMI=$(curl -sS -H "$AUTH" "${API}/user" | jq -r '.login // ""') || true
if [ -z "$WHOAMI" ]; then
echo "::error::GITEA_TOKEN cannot resolve a user via /api/v1/user — check the token scope and that the secret is wired correctly."
if [ "${SOP_FAIL_OPEN:-}" = "1" ]; then
echo "::warning::SOP_FAIL_OPEN=1 — exiting 0 so CI does not block."
exit 0
fi
exit 1
fi
echo "::notice::token resolves to user: $WHOAMI"
@@ -119,10 +107,6 @@ echo "::notice::token resolves to user: $WHOAMI"
HEAD_SHA=$(curl -sS -H "$AUTH" "${API}/repos/${OWNER}/${NAME}/pulls/${PR_NUMBER}" | jq -r '.head.sha // ""') || true
if [ -z "$HEAD_SHA" ]; then
echo "::error::Failed to fetch PR head SHA — token may be invalid."
if [ "${SOP_FAIL_OPEN:-}" = "1" ]; then
echo "::warning::SOP_FAIL_OPEN=1 — exiting 0 so CI does not block."
exit 0
fi
exit 1
fi
debug "pr-head-sha=$HEAD_SHA"
@@ -215,10 +199,6 @@ if [ "${SOP_DEBUG:-}" = "1" ]; then
fi
if [ "$_HTTP_EXIT" -ne 0 ] || [ "$HTTP_CODE" != "200" ]; then
echo "::error::GET /orgs/${OWNER}/teams failed (curl exit=$_HTTP_EXIT HTTP=$HTTP_CODE) — token may lack read:org scope or be invalid."
if [ "${SOP_FAIL_OPEN:-}" = "1" ]; then
echo "::warning::SOP_FAIL_OPEN=1 — exiting 0 so CI does not block."
exit 0
fi
exit 1
fi
@@ -265,17 +245,13 @@ done
# 5. Read approving reviewers. set +e disables set -e temporarily so that curl
# failures (e.g. empty/invalid token → HTTP 401) do not abort the script before
# SOP_FAIL_OPEN is evaluated. set -e is restored immediately after.
# set -e is restored immediately after.
set +e
REVIEWS=$(curl -sS -H "$AUTH" "${API}/repos/${OWNER}/${NAME}/pulls/${PR_NUMBER}/reviews")
_REVIEWS_EXIT=$?
set -e
if [ $_REVIEWS_EXIT -ne 0 ] || [ -z "$REVIEWS" ]; then
echo "::error::Failed to fetch reviews (curl exit=$_REVIEWS_EXIT) — token may be invalid or unreachable."
if [ "${SOP_FAIL_OPEN:-}" = "1" ]; then
echo "::warning::SOP_FAIL_OPEN=1 — exiting 0 so CI does not block."
exit 0
fi
exit 1
fi
APPROVERS=$(echo "$REVIEWS" | jq -r --arg head_sha "$HEAD_SHA" '[.[] | select(.state=="APPROVED" and .commit_id == $head_sha) | .user.login] | unique | .[]') || true
+17 -6
View File
@@ -109,23 +109,34 @@ class Handler(http.server.BaseHTTPRequestHandler):
return self._json(200, [{
"state": "APPROVED",
"dismissed": True,
"official": True,
"user": {"login": "core-devops"},
"commit_id": "abc1234",
"commit_id": "deadbeef0000111122223333444455556666",
}])
if sc == "T3_reviews_approved_non_author":
return self._json(200, [
{"state": "CHANGES_REQUESTED", "dismissed": False, "user": {"login": "bob"}, "commit_id": "abc1234"},
{"state": "APPROVED", "dismissed": False, "user": {"login": "core-devops"}, "commit_id": "abc1234"},
{"state": "CHANGES_REQUESTED", "dismissed": False, "official": True, "user": {"login": "bob"}, "commit_id": "deadbeef0000111122223333444455556666"},
{"state": "APPROVED", "dismissed": False, "official": True, "user": {"login": "core-devops"}, "commit_id": "deadbeef0000111122223333444455556666"},
])
if sc == "T19_ai_sop_ack_approved":
# ai-sop-ack member submitted APPROVED review — must NOT count
# toward qa-review (team_id=20) or security-review (team_id=21).
return self._json(200, [
{"state": "APPROVED", "dismissed": False, "user": {"login": "ai-reviewer"}, "commit_id": "abc1234"},
{"state": "APPROVED", "dismissed": False, "official": True, "user": {"login": "ai-reviewer"}, "commit_id": "deadbeef0000111122223333444455556666"},
])
# Default: one non-author APPROVED
if sc == "T21_stale_head_approved":
# APPROVED review but on an old commit (stale head) → must be rejected
return self._json(200, [
{"state": "APPROVED", "dismissed": False, "official": True, "user": {"login": "core-devops"}, "commit_id": "oldsha0000000000000000000000000000"},
])
if sc == "T22_missing_official":
# APPROVED review with no official field → must be rejected
return self._json(200, [
{"state": "APPROVED", "dismissed": False, "user": {"login": "core-devops"}, "commit_id": "deadbeef0000111122223333444455556666"},
])
# Default: one non-author APPROVED (current head, official)
return self._json(200, [
{"state": "APPROVED", "dismissed": False, "user": {"login": "core-devops"}, "commit_id": "abc1234"},
{"state": "APPROVED", "dismissed": False, "official": True, "user": {"login": "core-devops"}, "commit_id": "deadbeef0000111122223333444455556666"},
])
# GET /repos/{owner}/{name}/issues/{pr_number}/comments
+598 -11
View File
@@ -308,6 +308,8 @@ def test_process_once_holds_pr_on_permanent_merge_error(monkeypatch):
monkeypatch.setattr(mq, "WATCH_BRANCH", "main")
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
monkeypatch.setattr(mq, "HOLD_LABEL", "merge-queue-hold")
monkeypatch.setattr(mq, "AUTO_DISCOVER", True)
monkeypatch.setattr(mq, "OPT_OUT_LABELS", {"merge-queue-hold", "do-not-auto-merge", "wip"})
monkeypatch.setattr(mq, "REVIEWER_SET", REVIEWERS)
monkeypatch.setattr(mq, "get_branch_protection", lambda branch: mq.BranchProtection(
@@ -324,7 +326,7 @@ def test_process_once_holds_pr_on_permanent_merge_error(monkeypatch):
return {"state": "success", "statuses": [{"context": ctx, "status": "success"}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
monkeypatch.setattr(mq, "list_queued_issues", lambda: [
monkeypatch.setattr(mq, "list_candidate_issues", lambda *, auto_discover: [
{"number": 100, "pull_request": {}, "labels": [{"name": "merge-queue"}],
"created_at": "2026-06-01T00:00:00Z"},
])
@@ -374,6 +376,8 @@ def _fully_ready_process_once_monkeypatch(monkeypatch, mergeable, calls):
monkeypatch.setattr(mq, "WATCH_BRANCH", "main")
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
monkeypatch.setattr(mq, "HOLD_LABEL", "merge-queue-hold")
monkeypatch.setattr(mq, "AUTO_DISCOVER", True)
monkeypatch.setattr(mq, "OPT_OUT_LABELS", {"merge-queue-hold", "do-not-auto-merge", "wip"})
monkeypatch.setattr(mq, "REVIEWER_SET", REVIEWERS)
monkeypatch.setattr(mq, "get_branch_protection", lambda branch: mq.BranchProtection(
required_contexts=["CI / all-required (pull_request)"],
@@ -389,7 +393,7 @@ def _fully_ready_process_once_monkeypatch(monkeypatch, mergeable, calls):
return {"state": "success", "statuses": [{"context": ctx, "status": "success"}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
monkeypatch.setattr(mq, "list_queued_issues", lambda: [
monkeypatch.setattr(mq, "list_candidate_issues", lambda *, auto_discover: [
{"number": 102, "pull_request": {}, "labels": [{"name": "merge-queue"}],
"created_at": "2026-06-01T00:00:00Z"},
])
@@ -484,6 +488,8 @@ def test_status_fetch_failure_is_fail_closed(monkeypatch):
monkeypatch.setattr(mq, "WATCH_BRANCH", "main")
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
monkeypatch.setattr(mq, "HOLD_LABEL", "merge-queue-hold")
monkeypatch.setattr(mq, "AUTO_DISCOVER", True)
monkeypatch.setattr(mq, "OPT_OUT_LABELS", {"merge-queue-hold", "do-not-auto-merge", "wip"})
monkeypatch.setattr(mq, "REVIEWER_SET", REVIEWERS)
monkeypatch.setattr(mq, "get_branch_protection", lambda branch: mq.BranchProtection(
required_contexts=["CI / all-required (pull_request)"],
@@ -501,7 +507,7 @@ def test_status_fetch_failure_is_fail_closed(monkeypatch):
raise mq.ApiError("GET /commits/HEAD/status -> HTTP 502: bad gateway")
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
monkeypatch.setattr(mq, "list_queued_issues", lambda: [
monkeypatch.setattr(mq, "list_candidate_issues", lambda *, auto_discover: [
{"number": 101, "pull_request": {}, "labels": [{"name": "merge-queue"}],
"created_at": "2026-06-01T00:00:00Z"},
])
@@ -595,6 +601,8 @@ def _stale_pr_update_409_monkeypatch(monkeypatch, queued_issues, calls):
monkeypatch.setattr(mq, "WATCH_BRANCH", "main")
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
monkeypatch.setattr(mq, "HOLD_LABEL", "merge-queue-hold")
monkeypatch.setattr(mq, "AUTO_DISCOVER", True)
monkeypatch.setattr(mq, "OPT_OUT_LABELS", {"merge-queue-hold", "do-not-auto-merge", "wip"})
monkeypatch.setattr(mq, "REVIEWER_SET", REVIEWERS)
monkeypatch.setattr(mq, "get_branch_protection", lambda branch: mq.BranchProtection(
required_contexts=["CI / all-required (pull_request)"],
@@ -610,7 +618,8 @@ def _stale_pr_update_409_monkeypatch(monkeypatch, queued_issues, calls):
return {"state": "success", "statuses": [{"context": ctx, "status": "success"}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
monkeypatch.setattr(mq, "list_queued_issues", lambda: queued_issues)
# Scan-loop process_once enumerates candidates via list_candidate_issues.
monkeypatch.setattr(mq, "list_candidate_issues", lambda *, auto_discover: queued_issues)
monkeypatch.setattr(mq, "get_pull", lambda n: {
"state": "open", "number": n, "mergeable": True,
"base": {"ref": "main", "repo_id": 1},
@@ -640,6 +649,7 @@ def _stale_pr_update_409_monkeypatch(monkeypatch, queued_issues, calls):
def fake_add_label(pr_number, label_name, *, dry_run):
calls["hold_label"] = (pr_number, label_name)
calls.setdefault("holds", []).append((pr_number, label_name))
monkeypatch.setattr(mq, "add_label_by_name", fake_add_label)
monkeypatch.setattr(mq, "post_comment", lambda *a, **k: None)
@@ -669,9 +679,12 @@ def test_process_once_holds_pr_on_409_conflict_on_update(monkeypatch):
def test_queue_advances_past_held_conflicted_pr(monkeypatch):
"""End-to-end HOL proof for #2352: PR #1409 (oldest) hits a 409-on-update
and is held; on the NEXT tick choose_next_queued_issue must SKIP the held
PR and select the next ready PR (#1500) instead of stalling on #1409."""
"""End-to-end HOL proof for #2352 under the scan-loop architecture: PR #1409
(oldest) hits a 409-on-update and is HELD (HOLD_LABEL applied); once held it
carries an opt-out label so it is excluded from candidate selection and can
never re-block the queue. The 409-conflict hold (#2354) and the
scan-through-skip (#2356) coexist: a held conflicted PR is both held AND no
longer a candidate, so newer ready PRs behind it are unblocked."""
calls = {"update_attempts": 0, "merge_attempts": 0, "hold_label": None}
conflicted = {"number": 1409, "pull_request": {},
"labels": [{"name": "merge-queue"}],
@@ -685,16 +698,30 @@ def test_queue_advances_past_held_conflicted_pr(monkeypatch):
calls=calls,
)
# Tick 1: oldest (#1409) is selected, 409-on-update → held.
# Tick 1: oldest (#1409) is selected, 409-on-update → held, then the scan
# CONTINUES to #1500 (which also 409s in this fixture and is likewise held).
# The key #2352 property: the conflicted oldest PR is held and does NOT stop
# the scan from advancing past it.
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["hold_label"] == (1409, "merge-queue-hold")
assert (1409, "merge-queue-hold") in calls["holds"]
assert calls["merge_attempts"] == 0 # held, not merged — fail-closed
# Simulate the label now present on #1409 (as the real hold would persist).
conflicted["labels"] = [{"name": "merge-queue"}, {"name": "merge-queue-hold"}]
# Tick 2: the queue must ADVANCE — choose_next_queued_issue skips the held
# #1409 and selects the next ready candidate #1500, NOT re-select #1409.
# Next selection: the scan-loop candidate selector must SKIP the now-held
# #1409 (HOLD_LABEL is in OPT_OUT_LABELS) and surface the next ready
# candidate #1500 — the held PR no longer head-of-line blocks. The legacy
# opt-IN selector (choose_next_queued_issue) honours the same hold.
opt_out = {"merge-queue-hold", "do-not-auto-merge", "wip"}
remaining = mq.choose_candidate_issues(
[conflicted, next_ready],
queue_label="merge-queue",
opt_out_labels=opt_out,
auto_discover=True,
)
assert [i["number"] for i in remaining] == [1500]
selected = mq.choose_next_queued_issue(
[conflicted, next_ready],
queue_label="merge-queue",
@@ -702,3 +729,563 @@ def test_queue_advances_past_held_conflicted_pr(monkeypatch):
)
assert selected is not None
assert selected["number"] == 1500
# --------------------------------------------------------------------------
# §SOP-22: AUTO-DISCOVERY (opt-OUT, label-optional). The queue must be
# self-sustaining — a ready PR is considered/merged with NO `merge-queue`
# label, while opt-out labels (merge-queue-hold / do-not-auto-merge / wip) and
# drafts are skipped. The merge bar (approvals/required-green/mergeable) is
# unchanged; only candidate selection changes.
# --------------------------------------------------------------------------
OPT_OUT = {"merge-queue-hold", "do-not-auto-merge", "wip"}
def _issue(number, labels, *, created="2026-06-01T00:00:00Z", draft=False, is_pr=True):
pr = {"draft": draft} if is_pr else None
out = {
"number": number,
"labels": [{"name": n} for n in labels],
"created_at": created,
}
if pr is not None:
out["pull_request"] = pr
return out
def test_auto_discover_selects_unlabeled_ready_pr():
"""A ready PR with NO merge-queue label is auto-considered (the autonomy fix:
agents cannot self-label because their token lacks write:issue)."""
issues = [_issue(50, labels=[])] # no merge-queue label at all
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert selected is not None
assert selected["number"] == 50
def test_auto_discover_skips_opt_out_labels():
"""Each opt-out label keeps a PR OUT of autonomous merging (the human escape
hatch). A PR carrying any of them is never selected even though it is open."""
for optout in OPT_OUT:
issues = [_issue(60, labels=[optout])]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert selected is None, f"{optout!r} should opt the PR out"
def test_auto_discover_skips_opt_out_even_when_queue_labeled():
"""An opt-out label beats the merge-queue label: a held/wip PR that also
carries merge-queue is still skipped."""
issues = [_issue(61, labels=["merge-queue", "wip"])]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert selected is None
def test_auto_discover_skips_drafts():
issues = [_issue(62, labels=[], draft=True)]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert selected is None
def test_auto_discover_skips_non_pull_issues():
"""A plain issue (no pull_request key) is never a merge candidate."""
issues = [_issue(63, labels=[], is_pr=False)]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert selected is None
def test_auto_discover_oldest_first_skipping_opt_out():
"""Selection is FIFO (oldest created_at first), and the opt-out PR is passed
over for the next-oldest eligible PR."""
issues = [
_issue(70, labels=["do-not-auto-merge"], created="2026-06-01T01:00:00Z"),
_issue(71, labels=[], created="2026-06-01T02:00:00Z"),
_issue(72, labels=["merge-queue"], created="2026-06-01T03:00:00Z"),
]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert selected["number"] == 71 # 70 opted out, 71 is next-oldest eligible
def test_opt_in_mode_requires_queue_label():
"""AUTO_DISCOVER off restores legacy opt-IN: only merge-queue-labeled PRs are
candidates; an unlabeled ready PR is NOT selected."""
issues = [
_issue(80, labels=[], created="2026-06-01T01:00:00Z"),
_issue(81, labels=["merge-queue"], created="2026-06-01T02:00:00Z"),
]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=False
)
assert selected["number"] == 81
def test_opt_in_mode_still_honours_opt_out():
"""Even in opt-IN mode, an opt-out label on a queue-labeled PR skips it."""
issues = [_issue(82, labels=["merge-queue", "merge-queue-hold"])]
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=False
)
assert selected is None
def test_list_candidate_issues_omits_label_filter_when_auto_discover(monkeypatch):
"""The auto-discovery listing must NOT pass a `labels` filter (so unlabeled
PRs are enumerated); the opt-IN listing must keep filtering by QUEUE_LABEL."""
captured = {}
def fake_api(method, path, *, query=None, **kw):
captured["query"] = dict(query or {})
return 200, []
monkeypatch.setattr(mq, "api", fake_api)
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
mq.list_candidate_issues(auto_discover=True)
assert "labels" not in captured["query"]
assert captured["query"].get("type") == "pulls"
mq.list_candidate_issues(auto_discover=False)
assert captured["query"].get("labels") == "merge-queue"
def _wire_ready_process_once(monkeypatch, *, issues, pr_payload, calls):
"""Wire process_once fully green EXCEPT candidate selection / pull payload,
which the caller supplies to exercise auto-discovery end-to-end."""
monkeypatch.setattr(mq, "OWNER", "molecule-ai")
monkeypatch.setattr(mq, "NAME", "molecule-core")
monkeypatch.setattr(mq, "WATCH_BRANCH", "main")
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
monkeypatch.setattr(mq, "HOLD_LABEL", "merge-queue-hold")
monkeypatch.setattr(mq, "AUTO_DISCOVER", True)
monkeypatch.setattr(mq, "OPT_OUT_LABELS", OPT_OUT)
monkeypatch.setattr(mq, "REVIEWER_SET", REVIEWERS)
monkeypatch.setattr(mq, "get_branch_protection", lambda branch: mq.BranchProtection(
required_contexts=["CI / all-required (pull_request)"],
required_approvals=2, block_on_rejected_reviews=True,
))
main_sha = "b" * 40
head_sha = "a" * 40
monkeypatch.setattr(mq, "get_branch_head", lambda branch: main_sha)
def fake_combined(sha):
ctx = "CI / all-required (push)" if sha == main_sha else "CI / all-required (pull_request)"
return {"state": "success", "statuses": [{"context": ctx, "status": "success"}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
monkeypatch.setattr(mq, "list_candidate_issues", lambda *, auto_discover: issues)
monkeypatch.setattr(mq, "get_pull", lambda n: dict(pr_payload, number=n))
monkeypatch.setattr(mq, "get_pull_commits", lambda n: [{"sha": main_sha}, {"sha": head_sha}])
monkeypatch.setattr(mq, "get_pull_reviews", lambda n: [
{"state": "APPROVED", "user": {"login": "agent-researcher"},
"official": True, "stale": False, "dismissed": False, "commit_id": head_sha},
{"state": "APPROVED", "user": {"login": "agent-reviewer-cr2"},
"official": True, "stale": False, "dismissed": False, "commit_id": head_sha},
])
def fake_merge(pr_number, *, dry_run, force=False):
calls["merged"] = pr_number
monkeypatch.setattr(mq, "merge_pull", fake_merge)
monkeypatch.setattr(mq, "update_pull", lambda *a, **k: calls.__setitem__("updated", True))
monkeypatch.setattr(mq, "post_comment", lambda *a, **k: None)
monkeypatch.setattr(mq, "add_label_by_name", lambda *a, **k: None)
return main_sha, head_sha
def test_process_once_auto_merges_unlabeled_ready_pr(monkeypatch):
"""End-to-end: a fully-ready PR with NO merge-queue label is auto-merged.
This is the core autonomy fix — no human/agent labeling required."""
calls = {"merged": None, "updated": False}
head_sha = "a" * 40
_wire_ready_process_once(
monkeypatch,
issues=[_issue(90, labels=[])], # NO merge-queue label
pr_payload={
"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": head_sha, "repo_id": 1},
"labels": [],
},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["merged"] == 90 # merged despite no merge-queue label
def test_process_once_skips_opt_out_labeled_pr(monkeypatch):
"""A fully-ready PR carrying an opt-out label is NOT merged (skipped)."""
for optout in OPT_OUT:
calls = {"merged": None, "updated": False}
head_sha = "a" * 40
_wire_ready_process_once(
monkeypatch,
issues=[_issue(91, labels=[optout])],
pr_payload={
"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": head_sha, "repo_id": 1},
"labels": [{"name": optout}],
},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["merged"] is None, f"{optout!r} PR must not be merged"
def test_process_once_does_not_merge_unapproved_pr(monkeypatch):
"""A not-ready PR (only one genuine approval) is auto-considered but NOT
merged — auto-discovery does not lower the merge bar."""
calls = {"merged": None, "updated": False}
head_sha = "a" * 40
main_sha, _ = _wire_ready_process_once(
monkeypatch,
issues=[_issue(92, labels=[])],
pr_payload={
"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": head_sha, "repo_id": 1},
"labels": [],
},
calls=calls,
)
# Only ONE genuine approval → below the required 2.
monkeypatch.setattr(mq, "get_pull_reviews", lambda n: [
{"state": "APPROVED", "user": {"login": "agent-researcher"},
"official": True, "stale": False, "dismissed": False, "commit_id": head_sha},
])
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["merged"] is None
def test_process_once_does_not_merge_red_required_pr(monkeypatch):
"""A not-ready PR (required context red) is auto-considered but NOT merged."""
calls = {"merged": None, "updated": False}
head_sha = "a" * 40
main_sha = "b" * 40
_wire_ready_process_once(
monkeypatch,
issues=[_issue(93, labels=[])],
pr_payload={
"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": head_sha, "repo_id": 1},
"labels": [],
},
calls=calls,
)
# Required PR context is FAILURE; main stays green.
def fake_combined(sha):
if sha == main_sha:
return {"state": "success",
"statuses": [{"context": "CI / all-required (push)", "status": "success"}]}
return {"state": "failure",
"statuses": [{"context": "CI / all-required (pull_request)", "status": "failure"}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["merged"] is None
def test_process_once_does_not_merge_unmergeable_pr(monkeypatch):
"""A not-ready PR (mergeable False = conflicts) is auto-considered but NOT
merged."""
calls = {"merged": None, "updated": False}
head_sha = "a" * 40
_wire_ready_process_once(
monkeypatch,
issues=[_issue(94, labels=[])],
pr_payload={
"state": "open", "mergeable": False, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": head_sha, "repo_id": 1},
"labels": [],
},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["merged"] is None
# --------------------------------------------------------------------------
# §SOP-22 (cont.): HEAD-OF-LINE (HOL) — a non-ready auto-discovered candidate
# must NOT block the newer ready PRs behind it. The queue SCANS THROUGH the
# FIFO candidate list, skipping `wait` candidates (REQUEST_CHANGES, mergeable
# != True, insufficient genuine approvals, or red required CI), and merges the
# first ready PR in the SAME tick. (Regression for the #1519-style false
# candidate the reviewer caught: open + unlabeled + mergeable=false + current-
# head official REQUEST_CHANGES + <2 genuine approvals.)
# --------------------------------------------------------------------------
MAIN_SHA = "b" * 40
def _wire_multi_candidate_process_once(monkeypatch, *, issues, pulls, reviews, calls):
"""Wire process_once for MULTIPLE candidates, dispatching get_pull /
get_pull_reviews / head-status BY PR NUMBER so each candidate can have a
different readiness. `pulls` maps number -> pull payload; `reviews` maps
number -> reviews list. Main is green; each PR head status is green."""
monkeypatch.setattr(mq, "OWNER", "molecule-ai")
monkeypatch.setattr(mq, "NAME", "molecule-core")
monkeypatch.setattr(mq, "WATCH_BRANCH", "main")
monkeypatch.setattr(mq, "QUEUE_LABEL", "merge-queue")
monkeypatch.setattr(mq, "HOLD_LABEL", "merge-queue-hold")
monkeypatch.setattr(mq, "AUTO_DISCOVER", True)
monkeypatch.setattr(mq, "OPT_OUT_LABELS", OPT_OUT)
monkeypatch.setattr(mq, "REVIEWER_SET", REVIEWERS)
monkeypatch.setattr(mq, "get_branch_protection", lambda branch: mq.BranchProtection(
required_contexts=["CI / all-required (pull_request)"],
required_approvals=2, block_on_rejected_reviews=True,
))
monkeypatch.setattr(mq, "get_branch_head", lambda branch: MAIN_SHA)
def fake_combined(sha):
ctx = "CI / all-required (push)" if sha == MAIN_SHA else "CI / all-required (pull_request)"
return {"state": "success", "statuses": [{"context": ctx, "status": "success"}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
monkeypatch.setattr(mq, "list_candidate_issues", lambda *, auto_discover: issues)
monkeypatch.setattr(mq, "get_pull", lambda n: dict(pulls[n], number=n))
# Each PR head contains current main (so no candidate needs an update; the
# only differentiator is readiness). head sha is the pull's own head.
monkeypatch.setattr(
mq, "get_pull_commits",
lambda n: [{"sha": MAIN_SHA}, {"sha": pulls[n]["head"]["sha"]}],
)
monkeypatch.setattr(mq, "get_pull_reviews", lambda n: reviews[n])
def fake_merge(pr_number, *, dry_run, force=False):
calls.setdefault("merged", [])
calls["merged"].append(pr_number)
monkeypatch.setattr(mq, "merge_pull", fake_merge)
monkeypatch.setattr(mq, "update_pull", lambda *a, **k: calls.__setitem__("updated", True))
monkeypatch.setattr(mq, "post_comment", lambda *a, **k: None)
monkeypatch.setattr(mq, "add_label_by_name", lambda *a, **k: None)
def _two_approvals(head_sha):
return [
{"state": "APPROVED", "user": {"login": "agent-researcher"},
"official": True, "stale": False, "dismissed": False, "commit_id": head_sha},
{"state": "APPROVED", "user": {"login": "agent-reviewer-cr2"},
"official": True, "stale": False, "dismissed": False, "commit_id": head_sha},
]
def test_hol_unready_oldest_does_not_block_newer_ready_pr(monkeypatch):
"""The OLDEST auto-discovered candidate is NOT ready (mergeable=false). The
queue must SKIP it and merge the NEWER ready PR in the SAME tick — no HOL."""
calls = {"updated": False}
old_head, new_head = "a" * 40, "c" * 40
_wire_multi_candidate_process_once(
monkeypatch,
issues=[
_issue(500, labels=[], created="2026-06-01T01:00:00Z"), # oldest, NOT ready
_issue(501, labels=[], created="2026-06-01T02:00:00Z"), # newer, READY
],
pulls={
500: {"state": "open", "mergeable": False, "draft": False, # conflict
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": old_head, "repo_id": 1}, "labels": []},
501: {"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": new_head, "repo_id": 1}, "labels": []},
},
reviews={500: _two_approvals(old_head), 501: _two_approvals(new_head)},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
# The newer ready PR merged; the non-ready oldest did not block it.
assert calls.get("merged") == [501]
def test_hol_1519_style_false_candidate_never_merged_and_never_blocks(monkeypatch):
"""Live #1519 repro: oldest, open, UNLABELED, but mergeable=false + a
current-head official REQUEST_CHANGES + only ONE genuine approval. It must
NEVER be merged and must NEVER block the newer ready PR behind it."""
calls = {"updated": False}
false_head, ready_head = "a" * 40, "c" * 40
_wire_multi_candidate_process_once(
monkeypatch,
issues=[
_issue(1519, labels=[], created="2026-05-20T00:00:00Z"), # oldest false candidate
_issue(2000, labels=[], created="2026-06-01T00:00:00Z"), # newer, READY
],
pulls={
1519: {"state": "open", "mergeable": False, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": false_head, "repo_id": 1}, "labels": []},
2000: {"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": ready_head, "repo_id": 1}, "labels": []},
},
reviews={
1519: [
# one genuine approval (below 2) ...
{"state": "APPROVED", "user": {"login": "agent-researcher"},
"official": True, "stale": False, "dismissed": False, "commit_id": false_head},
# ... plus a current-head official REQUEST_CHANGES (human action needed)
{"state": "REQUEST_CHANGES", "user": {"login": "agent-reviewer"},
"official": True, "stale": False, "dismissed": False, "commit_id": false_head},
],
2000: _two_approvals(ready_head),
},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
# #1519 is never merged; the ready PR behind it merges this same tick.
assert calls.get("merged") == [2000]
assert 1519 not in calls.get("merged", [])
def test_hol_unready_red_required_ci_is_skipped_for_ready_pr(monkeypatch):
"""A candidate whose required CI is RED is skipped (not waited-on) so the
newer ready PR merges in the same tick."""
calls = {"updated": False}
red_head, ready_head = "a" * 40, "c" * 40
_wire_multi_candidate_process_once(
monkeypatch,
issues=[
_issue(600, labels=[], created="2026-06-01T01:00:00Z"), # required CI red
_issue(601, labels=[], created="2026-06-01T02:00:00Z"), # ready
],
pulls={
600: {"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": red_head, "repo_id": 1}, "labels": []},
601: {"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": ready_head, "repo_id": 1}, "labels": []},
},
reviews={600: _two_approvals(red_head), 601: _two_approvals(ready_head)},
calls=calls,
)
# PR 600's required PR context is FAILURE; 601 (and main) stay green.
def fake_combined(sha):
if sha == MAIN_SHA:
return {"state": "success",
"statuses": [{"context": "CI / all-required (push)", "status": "success"}]}
state = "failure" if sha == red_head else "success"
return {"state": state,
"statuses": [{"context": "CI / all-required (pull_request)", "status": state}]}
monkeypatch.setattr(mq, "get_combined_status", fake_combined)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls.get("merged") == [601]
def test_hol_all_candidates_unready_merges_nothing(monkeypatch):
"""If EVERY candidate is non-ready, the queue merges nothing (fail-closed)
and does not loop — it simply finds no actionable PR this tick."""
calls = {"updated": False}
h1, h2 = "a" * 40, "c" * 40
_wire_multi_candidate_process_once(
monkeypatch,
issues=[
_issue(700, labels=[], created="2026-06-01T01:00:00Z"), # RC
_issue(701, labels=[], created="2026-06-01T02:00:00Z"), # unmergeable
],
pulls={
700: {"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": h1, "repo_id": 1}, "labels": []},
701: {"state": "open", "mergeable": False, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": h2, "repo_id": 1}, "labels": []},
},
reviews={
700: _two_approvals(h1) + [
{"state": "REQUEST_CHANGES", "user": {"login": "agent-reviewer"},
"official": True, "stale": False, "dismissed": False, "commit_id": h1},
],
701: _two_approvals(h2),
},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls.get("merged") is None # nothing merged; no HOL loop
def test_opt_out_draft_label_excludes_candidate():
"""The literal `draft` label is now an opt-out label (added to the default
OPT_OUT_LABELS), independent of Gitea draft STATE — a human can opt a PR out
by labeling it `draft` without converting it to a draft PR."""
# `draft` must be in the shipped default opt-out set.
assert "draft" in mq.OPT_OUT_LABELS
opt_out = OPT_OUT | {"draft"}
issues = [_issue(800, labels=["draft"], draft=False)] # label only, not draft STATE
selected = mq.choose_next_candidate_issue(
issues, queue_label="merge-queue", opt_out_labels=opt_out, auto_discover=True
)
assert selected is None
def test_choose_candidate_issues_returns_full_fifo_list_skipping_opt_out():
"""choose_candidate_issues returns ALL eligible candidates oldest-first (so
process_once can scan past non-ready ones), skipping opt-out/draft/non-PR."""
issues = [
_issue(72, labels=["merge-queue"], created="2026-06-01T03:00:00Z"),
_issue(70, labels=["do-not-auto-merge"], created="2026-06-01T01:00:00Z"), # opt-out
_issue(71, labels=[], created="2026-06-01T02:00:00Z"),
_issue(73, labels=[], draft=True, created="2026-06-01T00:30:00Z"), # draft
_issue(74, labels=[], is_pr=False, created="2026-06-01T00:00:00Z"), # not a PR
]
ordered = mq.choose_candidate_issues(
issues, queue_label="merge-queue", opt_out_labels=OPT_OUT, auto_discover=True
)
assert [i["number"] for i in ordered] == [71, 72] # FIFO, opt-out/draft/non-PR dropped
def test_process_once_defensive_skip_when_pull_payload_opted_out(monkeypatch):
"""If the listing missed an opt-out label but the authoritative pull payload
carries it (stale listing race), process_once must still skip the merge."""
calls = {"merged": None, "updated": False}
head_sha = "a" * 40
_wire_ready_process_once(
monkeypatch,
issues=[_issue(95, labels=[])], # listing shows no opt-out
pr_payload={
"state": "open", "mergeable": True, "draft": False,
"base": {"ref": "main", "repo_id": 1},
"head": {"sha": head_sha, "repo_id": 1},
"labels": [{"name": "do-not-auto-merge"}], # live pull is opted out
},
calls=calls,
)
rc = mq.process_once(dry_run=False)
assert rc == 0
assert calls["merged"] is None
+53 -19
View File
@@ -14,10 +14,17 @@
# T9 — team membership probe → 403 (token not in team) → script exits 1 (fail closed)
# T10 — CURL_AUTH_FILE created with mode 600 and correct header content
# T11 — bash syntax check (bash -n passes)
# T12 — jq filter: non-author APPROVED → in candidate list; dismissed → excluded
# T12 — jq filter: non-author APPROVED official current-head → in candidate list; dismissed → excluded
# T13 — missing required env GITEA_TOKEN → exits 1 with error
# T14 — non-default-base PR exits 0 without requiring review
# T18wrong-team review candidate does not block right-team comment approval
# T15comment agent-prefix approval → exit 1
# T16 — comment generic keyword approval → exit 1
# T17 — comments with no approval keywords → exit 1
# T18 — wrong-team review + right-team comment → exit 1
# T19 — ai-sop-ack APPROVED review excluded from qa-review gate
# T20 — ai-sop-ack APPROVED review excluded from security-review gate
# T21 — stale-head APPROVED review → exit 1 (commit_id mismatch)
# T22 — missing/non-official APPROVED review → exit 1 (official != true)
#
# Hostile-self-review (per feedback_assert_exact_not_substring):
# this test MUST FAIL if the script is absent. Verified by running
@@ -319,41 +326,50 @@ assert_file_contains "T10b printf header format (CURL_AUTH_FILE content)" "$T10_
assert_file_contains "T10c 'header =' curl-config syntax" "$T10_AUTHFILE" 'header = "Authorization: token '
rm -f "$T10_AUTHFILE"
# T12 — jq filter: non-author APPROVED included, dismissed excluded
# T12 — jq filter: non-author APPROVED official current-head included; dismissed/stale/missing-official excluded
echo
echo "== T12 jq filter =="
# These are tested indirectly via T3 and T6 above, but let's also test
# the jq expression directly.
JQ_FILTER='.[]
| select(.state == "APPROVED")
| select(.official == true)
| select(.dismissed != true)
| select(.user.login != "alice")
| select(.commit_id == $head)
| .user.login'
T12_INPUT='[{"state":"APPROVED","dismissed":false,"user":{"login":"core-devops"}},{"state":"CHANGES_REQUESTED","dismissed":false,"user":{"login":"bob"}},{"state":"APPROVED","dismissed":false,"user":{"login":"alice"}},{"state":"APPROVED","dismissed":true,"user":{"login":"carol"}}]'
T12_INPUT='[{"state":"APPROVED","official":true,"dismissed":false,"commit_id":"deadbeef0000111122223333444455556666","user":{"login":"core-devops"}},{"state":"CHANGES_REQUESTED","official":true,"dismissed":false,"commit_id":"deadbeef0000111122223333444455556666","user":{"login":"bob"}},{"state":"APPROVED","official":true,"dismissed":false,"commit_id":"deadbeef0000111122223333444455556666","user":{"login":"alice"}},{"state":"APPROVED","official":true,"dismissed":true,"commit_id":"deadbeef0000111122223333444455556666","user":{"login":"carol"}},{"state":"APPROVED","official":false,"dismissed":false,"commit_id":"deadbeef0000111122223333444455556666","user":{"login":"dave"}},{"state":"APPROVED","official":true,"dismissed":false,"commit_id":"oldsha0000000000000000000000000000","user":{"login":"eve"}}]'
JQ_CMD=$(command -v jq 2>/dev/null || echo /tmp/jq)
T12_CANDIDATES=$(echo "$T12_INPUT" | "$JQ_CMD" -r "$JQ_FILTER" 2>/dev/null | sort -u)
assert_contains "T12 jq: core-devops (non-author APPROVED) in candidates" "core-devops" "$T12_CANDIDATES"
T12_CANDIDATES=$(echo "$T12_INPUT" | "$JQ_CMD" -r --arg head "deadbeef0000111122223333444455556666" "$JQ_FILTER" 2>/dev/null | sort -u)
assert_contains "T12 jq: core-devops (non-author APPROVED official current-head) in candidates" "core-devops" "$T12_CANDIDATES"
assert_eq "T12 jq: alice (author) NOT in candidates" "" "$(echo "$T12_CANDIDATES" | grep '^alice$' || true)"
assert_eq "T12 jq: carol (dismissed) NOT in candidates" "" "$(echo "$T12_CANDIDATES" | grep '^carol$' || true)"
assert_eq "T12 jq: dave (official=false) NOT in candidates" "" "$(echo "$T12_CANDIDATES" | grep '^dave$' || true)"
assert_eq "T12 jq: eve (stale head) NOT in candidates" "" "$(echo "$T12_CANDIDATES" | grep '^eve$' || true)"
# T15 — comment-based approval via agent prefix pattern → exit 0
# T15 — comment-based approval via agent prefix pattern → exit 1
# SECURITY: agent-prefix comments are also removed. A text prefix in an
# issue comment is spoofable (any team member can type "[core-qa-agent]")
# and lacks the audit trail of an official Gitea review.
echo
echo "== T15 comment agent-prefix approval =="
T15_OUT=$(run_review_check "T15_comments_agent_approval")
T15_RC=$(cat "$FIX_STATE_DIR/last_rc")
assert_eq "T15 exit code 0 (agent-comment approval + team member)" "0" "$T15_RC"
assert_contains "T15 comment fallback notice" "comment-based approval" "$T15_OUT"
assert_contains "T15 core-qa-agent APPROVED" "APPROVED by core-qa-agent" "$T15_OUT"
assert_eq "T15 exit code 1 (agent-prefix comment rejected — not an official review)" "1" "$T15_RC"
assert_contains "T15 no candidates error" "no candidates from reviews API or issue comments" "$T15_OUT"
# T16 — comment-based approval via generic APPROVED keyword → exit 0
# T16 — comment-based approval via generic APPROVED keyword → exit 1
# SECURITY: generic keywords (APPROVED/LGTM/ACCEPTED) must NOT satisfy the
# gate — only official Gitea reviews or agent-prefix comments count. A plain
# comment from a team member is a bypass if it skips the review UI.
echo
echo "== T16 comment generic keyword approval =="
T16_OUT=$(run_review_check "T16_comments_generic_approval")
T16_RC=$(cat "$FIX_STATE_DIR/last_rc")
assert_eq "T16 exit code 0 (generic-approval comment + team member)" "0" "$T16_RC"
assert_contains "T16 comment fallback notice" "comment-based approval" "$T16_OUT"
assert_eq "T16 exit code 1 (generic-approval comment rejected — not an official review)" "1" "$T16_RC"
assert_contains "T16 no candidates error" "no candidates from reviews API or issue comments" "$T16_OUT"
# T17 — no approval keywords in comments → exit 1
echo
@@ -363,16 +379,16 @@ T17_RC=$(cat "$FIX_STATE_DIR/last_rc")
assert_eq "T17 exit code 1 (no candidates from comments)" "1" "$T17_RC"
assert_contains "T17 no candidates error" "no candidates from reviews API or issue comments" "$T17_OUT"
# T18 — a wrong-team PR review candidate must not suppress a right-team
# comment approval. This matches PR #1790, where QA had an APPROVED review
# and security approved via the agent comment convention.
# T18 — wrong-team review + right-team comment → exit 1
# SECURITY: with comment approval fully removed, a wrong-team review plus
# a right-team comment yields NO valid candidates. Only official reviews
# from the target team count.
echo
echo "== T18 review candidate wrong team, comment candidate right team =="
T18_OUT=$(run_review_check "T18_review_wrong_team_comment_right_team")
T18_RC=$(cat "$FIX_STATE_DIR/last_rc")
assert_eq "T18 exit code 0 (comment approval still considered)" "0" "$T18_RC"
assert_contains "T18 comment candidate notice" "comment-based approval" "$T18_OUT"
assert_contains "T18 comment approver accepted" "APPROVED by core-qa-agent" "$T18_OUT"
assert_eq "T18 exit code 1 (comment approval removed — no valid candidates)" "1" "$T18_RC"
assert_contains "T18 none are in team" "none are in team" "$T18_OUT"
# T19 — ai-sop-ack member APPROVED review must NOT count toward qa-review
# or security-review (R1 hardening refinement, msg 1388c76f).
@@ -393,6 +409,24 @@ assert_eq "T20 exit code 1 (ai-sop-ack not in security team)" "1" "$T20_RC"
assert_contains "T20 ai-reviewer excluded from security" "candidates: ai-reviewer" "$T20_OUT"
assert_contains "T20 none are in security team" "none are in team" "$T20_OUT"
# T21 — stale-head APPROVED review must be rejected (commit_id mismatch).
# SECURITY: an approval on an old commit does not cover the current head.
echo
echo "== T21 stale-head APPROVED review rejected =="
T21_OUT=$(run_review_check "T21_stale_head_approved")
T21_RC=$(cat "$FIX_STATE_DIR/last_rc")
assert_eq "T21 exit code 1 (stale-head approval rejected)" "1" "$T21_RC"
assert_contains "T21 no candidates error" "no candidates from reviews API or issue comments" "$T21_OUT"
# T22 — missing/non-official APPROVED review must be rejected.
# SECURITY: only official Gitea reviews count; comments and non-official reviews lack audit trail.
echo
echo "== T22 missing official flag APPROVED review rejected =="
T22_OUT=$(run_review_check "T22_missing_official")
T22_RC=$(cat "$FIX_STATE_DIR/last_rc")
assert_eq "T22 exit code 1 (missing official rejected)" "1" "$T22_RC"
assert_contains "T22 no candidates error" "no candidates from reviews API or issue comments" "$T22_OUT"
echo
echo "------"
echo "PASS=$PASS FAIL=$FAIL"
+19 -3
View File
@@ -7,10 +7,13 @@ name: gitea-merge-queue
# the user-space queue bot, one PR per tick, using the non-bypass merge actor.
#
# Queue contract:
# - add label `merge-queue` to an open same-repo PR
# - auto-discovery (default): any open same-repo PR is considered — no
# `merge-queue` label required (the label is optional metadata now)
# - bot updates stale PR heads with current main, then waits for CI
# - bot merges only when current main is green and required PR contexts pass
# - add `merge-queue-hold` to pause a queued PR without removing it
# - bot merges only when current main is green, genuine approvals are present
# on the current head, required PR contexts pass, and the PR is mergeable
# - add `merge-queue-hold`, `do-not-auto-merge`, or `wip` to keep a PR OUT of
# autonomous merging; draft PRs are also skipped
on:
# Schedule moved to operator-config:
@@ -48,6 +51,19 @@ jobs:
WATCH_BRANCH: ${{ github.event.repository.default_branch }}
QUEUE_LABEL: merge-queue
HOLD_LABEL: merge-queue-hold
# Auto-discovery (opt-OUT). When on (default), the queue considers ALL
# open same-repo PRs that meet the merge bar — it does NOT wait for a
# human/agent to add `merge-queue`. Agent Gitea tokens lack
# write:issue (labels are issue-scoped) and could never self-label,
# which stalled the queue; the label is now OPTIONAL metadata. The
# merge bar is UNCHANGED — only candidate selection widens. Set
# AUTO_DISCOVER=0 to restore legacy opt-IN (require the merge-queue
# label to be considered).
AUTO_DISCOVER: "1"
# Opt-OUT labels: any of these on a PR keeps it OUT of autonomous
# merging (the human escape hatch). HOLD_LABEL is always also honoured.
# A human who wants a PR held just adds one of these labels.
OPT_OUT_LABELS: do-not-auto-merge,wip
UPDATE_STYLE: merge
# Recognised official-reviewer set. A merge needs >= required_approvals
# DISTINCT genuine official approvals from these accounts on the
+30 -14
View File
@@ -8,26 +8,39 @@ against the latest `main`.
## Queue Contract
Add the `merge-queue` label to an open PR when it is ready to merge.
**Auto-discovery (opt-OUT, default).** You do NOT need to label a PR. The bot
auto-discovers every open same-repo PR and merges any that meets the bar. The
`merge-queue` label is now optional metadata, not a gate. This removed the
historical autonomy gap: agent Gitea tokens lack `write:issue` (labels are
issue-scoped), so agents could never self-label and ready PRs stalled.
To keep a PR OUT of autonomous merging, add an opt-OUT label:
`merge-queue-hold`, `do-not-auto-merge`, or `wip`. Draft PRs are also skipped.
The bot processes one PR per tick:
1. Confirms `main` is green.
2. Selects the oldest open PR carrying `merge-queue`.
3. Skips PRs with `merge-queue-hold`.
4. Rejects fork PRs because the queue may only update same-repo branches.
5. If the PR head does not contain current `main`, calls Gitea's
1. Confirms `main`'s branch-protection-required push contexts are green.
2. Selects the oldest open same-repo PR that is NOT opt-out-labeled and NOT a
draft (auto-discovery). With `AUTO_DISCOVER=0` it falls back to legacy
opt-IN: only PRs carrying `merge-queue` are considered.
3. Rejects fork PRs because the queue may only update same-repo branches.
4. If the PR head does not contain current `main`, calls Gitea's
`/pulls/{n}/update?style=merge` endpoint and waits for CI on the new head.
6. Merges only after the current PR head has required contexts green:
- `CI / all-required (pull_request)`
- `sop-checklist / all-items-acked (pull_request)`
5. Merges only when, on the PR's CURRENT head sha:
- `>= required_approvals` distinct genuine official `APPROVED` reviews from
the recognised reviewer set (read from branch protection; default 2),
- no open official `REQUEST_CHANGES`,
- every branch-protection-required status context is green, and
- the PR is `mergeable` (Gitea returns `True`; `None`/`False` = wait).
The workflow is serialized with `concurrency`, so two queued PRs cannot be
The merge bar is unchanged by auto-discovery — only WHICH PRs are considered
changes. The workflow is serialized with `concurrency`, so two PRs cannot be
merged against the same observed `main`.
## Operator Commands
Queue a PR:
Queue a PR (optional — auto-discovery already considers every ready PR; the
label is just visible metadata):
```bash
curl -fsS -X POST \
@@ -37,7 +50,8 @@ curl -fsS -X POST \
-d '{"labels":["merge-queue"]}'
```
Temporarily hold a queued PR:
Keep a PR OUT of autonomous merging (opt-OUT — use `merge-queue-hold`,
`do-not-auto-merge`, or `wip`):
```bash
curl -fsS -X POST \
@@ -56,9 +70,11 @@ REPO=molecule-ai/molecule-core \
WATCH_BRANCH=main \
QUEUE_LABEL=merge-queue \
HOLD_LABEL=merge-queue-hold \
AUTO_DISCOVER=1 \
OPT_OUT_LABELS=do-not-auto-merge,wip \
REVIEWER_SET=agent-reviewer,agent-researcher,agent-reviewer-cr2 \
UPDATE_STYLE=merge \
REQUIRED_CONTEXTS='CI / all-required (pull_request),sop-checklist / all-items-acked (pull_request)' \
python3 .gitea/scripts/gitea-merge-queue.py
python3 .gitea/scripts/gitea-merge-queue.py --dry-run
```
Dry run:
@@ -161,7 +161,7 @@ func (h *PluginsHandler) uninstallViaDocker(ctx context.Context, c *gin.Context,
// 1. Strip plugin's rule/fragment markers from CLAUDE.md (mirrors
// AgentskillsAdaptor.uninstall lines 184-188). Best-effort: if
// the user edited CLAUDE.md, our marker stays untouched.
h.stripPluginMarkersFromMemory(ctx, containerName, pluginName)
h.stripPluginMarkersFromMemory(ctx, workspaceID, containerName, pluginName)
// 2. Remove copied skill dirs declared in the plugin's plugin.yaml.
for _, skill := range skillNames {
@@ -171,9 +171,11 @@ func (h *PluginsHandler) uninstallViaDocker(ctx context.Context, c *gin.Context,
log.Printf("Plugin uninstall: skipping invalid skill name %q in %s: %v", skill, pluginName, err)
continue
}
_, _ = h.execAsRoot(ctx, containerName, []string{
if _, rmErr := h.execAsRoot(ctx, containerName, []string{
"rm", "-rf", "/configs/skills/" + skill,
})
}); rmErr != nil {
log.Printf("Plugin uninstall: failed to remove skill %s from %s: %v", skill, workspaceID, rmErr)
}
}
// 3. Delete the plugin directory itself (as root to handle file ownership).
@@ -393,7 +393,7 @@ func (h *PluginsHandler) readPluginSkillsFromContainer(ctx context.Context, cont
// `# Plugin: <name> /` — mirrors AgentskillsAdaptor.uninstall's stripping
// logic so install/uninstall are symmetric. Best-effort: silent on read or
// write failure, since the rest of uninstall must still succeed.
func (h *PluginsHandler) stripPluginMarkersFromMemory(ctx context.Context, containerName, pluginName string) {
func (h *PluginsHandler) stripPluginMarkersFromMemory(ctx context.Context, workspaceID, containerName, pluginName string) {
// Use sed via bash -c for atomic in-place delete: drop the marker line
// and the blank line that follows it (install adds a leading blank line
// before the marker via append_to_memory). Three sed passes mirror the
@@ -417,7 +417,9 @@ func (h *PluginsHandler) stripPluginMarkersFromMemory(ctx context.Context, conta
`awk 'BEGIN{skip=0; blanks=0} /^%s/{skip=1; blanks=0; next} skip==1 && /^[[:space:]]*$/{blanks++; if(blanks>=2){skip=0; print; next} next} /^# Plugin: /{if(skip==1)skip=0} skip==1{next} {print}' /configs/CLAUDE.md > /tmp/claude.new && mv /tmp/claude.new /configs/CLAUDE.md`,
regexpEscapeForAwk(marker),
)
_, _ = h.execAsRoot(ctx, containerName, []string{"bash", "-c", script})
if _, awkErr := h.execAsRoot(ctx, containerName, []string{"bash", "-c", script}); awkErr != nil {
log.Printf("Plugin uninstall: failed to strip markers from CLAUDE.md for %s in %s: %v", pluginName, workspaceID, awkErr)
}
}
// regexpEscapeForAwk escapes characters that have special meaning inside an