fix(agent-actions): recheck non-ci staged closes

[JSONbored]

Motivation

• Fix a safety gap where fresh non-CI heuristic closes persisted with closeRequiresCiState: "not_required" could be accepted and executed without any accept-time live revalidation, allowing a stale justification (conflict/duplicate/gate) to be acted on.
• No linked issue: this is a small, self-evident safety-predicate fix in code the AI reviewer's own gate verdict on this PR pointed at directly (src/services/agent-approval-queue.ts:205), not a pre-planned feature.

Description

• Reintroduced an accept-time live-disposition recheck in decidePendingAgentAction (in src/services/agent-approval-queue.ts) that runs the same fetchLiveCiAggregate/fetchLivePullRequestMergeState/fetchLivePullRequestReviewDecision checks used for merges when the pending row is a staged merge or a non-CI heuristic close with closeRequiresCiState === "not_required".
• Bases the close-path staleness check only on mergeableState/reviewDecision, dropping the ciState === "passed" requirement: closeRequiresCiState: "not_required" already means CI was never this close's justification, so gating staleness on CI meant a cleared conflict on an otherwise-clean PR would still execute the close whenever live CI happened to be pending, failed, or unverified at accept time.
• Added regression/unit tests in test/unit/agent-approval-queue.test.ts covering: live-CI-passing + clean → superseded; live conflict remains → still executes; clean mergeability with live CI merely pending (not passed) → still superseded (the case the gate's own AI review flagged as reachable and uncovered); a reviewer requesting changes → still executes.
• Rebased onto current main (branch was ~280 commits behind).

Testing

• npm run typecheck — clean.
• git diff --check — clean.
• npx vitest run test/unit/agent-approval-queue.test.ts — 62/62 passed.
• Verified the new "clean mergeability + non-passed CI" regression test fails against the pre-fix predicate and passes against the fix (confirms it pins the fix, not a tautology).
• Scoped vitest --coverage on src/services/agent-approval-queue.ts via this test file: 100% branch/line/statement coverage.

---

Codex Task

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.06%. Comparing base (fdda351) to head (fbd20aa).
⚠️ Report is 17 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2902   +/-   ##
=======================================
  Coverage   96.06%   96.06%           
=======================================
  Files         259      259           
  Lines       28655    28660    +5     
  Branches    10428    10432    +4     
=======================================
+ Hits        27528    27533    +5     
  Misses        490      490           
  Partials      637      637           

Files with missing lines	Coverage Δ
src/services/agent-action-executor.ts	96.44% <100.00%> (+0.01%)	⬆️
src/services/agent-approval-queue.ts	100.00% <100.00%> (ø)
src/settings/agent-actions.ts	95.59% <ø> (ø)

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[gittensory-orb]

Warning

🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨

⏸️ Gittensory review result - manual review recommended

_{Review updated: 2026-07-04 08:30:01 UTC}

6 files · 1 AI reviewer · 1 blocker · readiness 100/100 · CI green · unknown

⏸️ Suggested Action - Manual Review

• AI reviewers agree on a likely critical defect: src/services/agent-approval-queue.ts:205 in decidePendingAgentAction: a conflict-justified close is still accepted after the conflict clears when live reviewDecision is CHANGES_REQUESTED, so the stale base-conflict close can execute even though the persisted close justification no longer holds. — Resolve the flagged defect, or override if the AI reviewers are mistaken, then re-run the gate.

Review summary
The change correctly introduces and round-trips a dedicated discriminator for conflict-justified heuristic closes instead of applying the live mergeability recheck to every non-CI close. However, the close accept-time predicate still treats an unrelated live review decision as sufficient to execute a stale conflict close, and it leaves already-staged rows without the new discriminator on the old unsafe path despite these rows having no expiry.

Blockers

• src/services/agent-approval-queue.ts:205 in decidePendingAgentAction: a conflict-justified close is still accepted after the conflict clears when live reviewDecision is CHANGES_REQUESTED, so the stale base-conflict close can execute even though the persisted close justification no longer holds.
• src/services/agent-approval-queue.ts:179 in decidePendingAgentAction: any already-pending conflict-justified heuristic close created before closeRequiresMergeableState existed has closeRequiresMergeableState undefined, so shouldRecheckLiveDisposition skips the live recheck and preserves the no-expiry stale-close gap this PR is meant to close.

Nits — 5 non-blocking

• nit: src/services/agent-approval-queue.ts:179 would read more clearly as an explicit boolean predicate instead of relying on pr?.headSha returning a string-or-falsey value.
• nit: src/services/agent-approval-queue.ts:186 still fetches live CI for conflict-close acceptance even though the close branch intentionally ignores CI for the stale predicate, which spends rate limit only for audit metadata.
• In src/services/agent-approval-queue.ts, make conflict-close staleness depend on the conflict signal itself: if the live mergeable-state fetch succeeds with mergeableState === "clean", reject as stale regardless of reviewDecision, and let any remaining review issues be replanned separately.
• In src/services/agent-approval-queue.ts, add an explicit legacy-row policy for heuristic closes with closeRequiresCiState === "not_required" and closeRequiresMergeableState === undefined, such as fail-closed/manual replan or a narrowly documented backfill path.
• In test/unit/agent-approval-queue.test.ts, replace the current "reviewer has since requested changes" expectation with a regression test proving a cleared conflict is superseded even when reviewDecision is CHANGES_REQUESTED.

Concerns raised — review before merging

• src/services/agent-approval-queue.ts:205 in decidePendingAgentAction: a conflict-justified close is still accepted after the conflict clears when live reviewDecision is CHANGES_REQUESTED, so the stale base-conflict close can execute even though the persisted close justification no longer holds.

Signal	Result	Evidence
Code review	❌ 1 blocker	1 reviewer
Linked issue	✅ No-issue rationale	PR body explains why no issue is linked.
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (no linked issue context).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 60 registered-repo PR(s), 51 merged, 421 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 60 PR(s), 421 issue(s).
Gate result	❌ Blocking	Repo-configured hard blocker found.

Review context

• Author: JSONbored
• Role context: owner (maintainer lane)
• Public audience mode: oss maintainer
• Lane context: Repository registration is not available in the local Gittensory cache.
• Public profile languages: Python, TypeScript, JavaScript, Ruby, Go, Kotlin, MDX, Shell
• Official Gittensor activity: 60 PR(s), 421 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Treat this as maintainer-lane context rather than normal contributor-lane activity.
• No action.
• Link the issue being solved, or explicitly explain why this is a no-issue PR.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by Gittensory, a quiet PR intelligence layer for OSS maintainers.

• Re-run Gittensory review