feat(review): record a confirmed override when the owner closes an AI-judgment-held PR

[cleanjunc]

Summary

Closes #8123.

Implements #8106's recorded decision: the repo owner closing (not merging) a PR that was held for a low-confidence AI judgment is the explicit "the automated call was right" signal — the confirmed-side mirror of recordReversalSignals' existing reversal hooks, and the first non-inferred positive confirmation in this system.

• A new owner-close branch in recordReversalSignals (src/review/outcomes-wire.ts), using the exact ownerLogin === senderLogin check the reversal side already uses. A contributor's own close records nothing.
• "Held via aiReviewLowConfidenceHold" is detected the same way calibration: wire linked_issue_scope_mismatch into the shared signal-tracking module (fired + reversal) #8101 detects its rule: a recorded rule_fired event for either AI_JUDGMENT_BLOCKER_CODES code (ai_consensus_defect / ai_review_split, imported from src/rules/advisory.ts) against this target within the 30-day lookback calibration: wire recordRuleFired/recordHumanOverride for the remaining isConfiguredGateBlocker codes (excludes linked_issue_scope_mismatch, covered by #8101) #8104 established. Each matching code gets a "confirmed" HumanOverrideEvent via the existing createSignalStore adapter.
• Scoped to aiReviewLowConfidenceHold only — the other two hold kinds carry no ruleId-equivalent to key on (the issue's explicit boundary). resolveAiReviewLowConfidenceHold and the holds' production logic are untouched.
• Best-effort end to end: the caller attaches .catch(() => undefined) and the helper mirrors calibration: wire linked_issue_scope_mismatch into the shared signal-tracking module (fired + reversal) #8101's shape exactly — a SignalStore failure can never affect the PR-close handling (pinned by test). An owner merge-close never enters this branch (the merged path remains the reversal side's business — also pinned by test).

Scope

• The PR title follows type(scope): short summary Conventional Commit format, for example fix(api): restore profile access checks.
• This PR is focused and does not mix unrelated backend, UI, MCP, docs, dependency, and deploy changes.
• This follows CONTRIBUTING.md and does not reintroduce GitHub Pages, VitePress, site/, or CNAME.
• I linked a currently open issue this PR resolves (e.g. Closes #123) — a linked open issue is required for every contributor PR.

Validation

• git diff --check
• npm run actionlint
• npm run typecheck
• npm run test:coverage locally; codecov/patch requires ≥99% coverage of the lines AND branches you changed (aim for 100% on your diff so CI variance does not fail near the threshold). Global coverage is a non-blocking trend with a loose 90% backstop, not the gate.
• npm run test:workers
• npm run build:mcp
• npm run test:mcp-pack
• npm run ui:openapi:check
• npm run ui:lint
• npm run ui:typecheck
• npm run ui:build
• npm audit --audit-level=moderate
• New or changed behavior has unit/integration tests for new branches, fallback paths, and sanitizer boundaries

If any required check was skipped, explain why:

• Ran the full affected suite instead of the 30+-min unsharded chain: test/unit/outcomes-wire.test.ts — 79/79 green, including 7 new calibration: record a 'confirmed' HumanOverrideEvent when an owner closes an aiReviewLowConfidenceHold PR #8123 cases covering every branch the issue's own requirements enumerate: owner-vs-contributor, held-vs-not-held (a non-AI rule firing does not make it an AI hold), matching-vs-different-target, both AI codes confirmed in one close, the merged-close exclusion, and the fail-open .catch path (a rejecting SignalStore never throws out of the handler). The diff is one focused branch + one helper mirroring the adjacent, already-merged calibration: wire linked_issue_scope_mismatch into the shared signal-tracking module (fired + reversal) #8101 hook, so the changed lines are exercised end-to-end by those cases; real CI re-runs typecheck and the sharded suite here.
• workers/mcp/ui/audit/actionlint: untouched surfaces — one source file + one test file; zero dependency changes.

Safety

• No secrets, wallet details, hotkeys, coldkeys, user PATs, private keys, raw trust scores, private rankings, or private maintainer evidence are exposed.
• Public GitHub text stays sanitized, low-noise, and does not imply compensation guarantees or optimization tactics.
• Auth, cookie, CORS, GitHub App, Cloudflare, or session changes include negative-path tests.
• API/OpenAPI/MCP behavior is updated and tested where needed.
• UI changes use live API data or real empty/error/loading states, not production mock/demo fallbacks.
• Visible UI changes include a UI Evidence section below with JPG/JPEG or PNG screenshots arranged as organized, captioned, clickable thumbnails. SVG screenshots are not used as review evidence. Review-only screenshots or recordings are not committed to the repository.
• Public docs/changelogs are updated where needed; changelogs are only edited for release-prep PRs.

The three unchecked Safety boxes are N/A: no auth/session/UI change — a webhook-side recording hook plus tests.

UI Evidence

N/A — no UI change.

Notes

• With this, ai_consensus_defect/ai_review_split become the first rules whose calibration: pure confusion-matrix scorer for a candidate rule classifier against the backtest corpus #8085 backtest precision reflects genuine positive confirmations rather than a reversal-only floor — exactly the issue's Expected Outcome.

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.53%. Comparing base (073de61) to head (6cd2db5).
⚠️ Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #8125      +/-   ##
==========================================
- Coverage   92.02%   89.53%   -2.49%     
==========================================
  Files         760       98     -662     
  Lines       77372    22786   -54586     
  Branches    23386     3915   -19471     
==========================================
- Hits        71203    20402   -50801     
+ Misses       5061     2199    -2862     
+ Partials     1108      185     -923     

Flag	Coverage Δ
shard-1	25.82% <33.33%> (-31.25%)	⬇️
shard-2	90.61% <100.00%> (+36.29%)	⬆️
shard-3	25.82% <33.33%> (-24.50%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/review/outcomes-wire.ts	91.07% <100.00%> (+0.53%)	⬆️

... and 662 files with indirect coverage changes

[loopover-orb]

Tip

✅ LoopOver review result - approve/merge recommended

_{Review updated: 2026-07-23 00:17:00 UTC}

2 files · 1 AI reviewer · no blockers · readiness 100/100 · CI green · clean

✅ Suggested Action - Approve/Merge

• safe to merge

Review summary
Adds an owner-close-without-merge branch to recordReversalSignals that records a 'confirmed' HumanOverrideEvent for AI-judgment codes (ai_consensus_defect/ai_review_split) that fired against the target within a 30-day lookback, mirroring the existing #8101/#8104 reversal-recording pattern almost exactly (same owner-login check, same catch-and-swallow, same createSignalStore adapter). The new code path is placed correctly before the existing 'merged' branch, correctly excludes merges (checked via a dedicated test), and is well covered — degenerate sender/owner cases, multi-code confirmation, wrong-target, non-AI-code, and SignalStore-rejection are all tested against the real recordReversalSignals entry point. The lookback constant AI_JUDGMENT_CONFIRMATION_LOOKBACK_MS duplicates LINKED_ISSUE_SCOPE_MISMATCH_LOOKBACK_MS's value (30 days) rather than reusing a shared constant, a minor DRY nit but not a defect.

Nits — 5 non-blocking

• src/review/outcomes-wire.ts:513 — AI_JUDGMENT_CONFIRMATION_LOOKBACK_MS duplicates the same 30-day literal as LINKED_ISSUE_SCOPE_MISMATCH_LOOKBACK_MS (and CONFIGURED_GATE_BLOCKER_SIGNAL_LOOKBACK_MS); consider sharing one lookback constant across the three override helpers.
• src/review/outcomes-wire.ts:517-527 — recordAiJudgmentHoldConfirmations loops sequentially (await inside for) over AI_JUDGMENT_BLOCKER_CODES instead of Promise.all like recordConfiguredGateBlockerOverrides does; harmless since the list is only 2 codes, but inconsistent with the sibling helper's pattern.
• The file is now ~616 lines per the size-smell note; consider whether the three near-identical override helpers (recordConfiguredGateBlockerOverrides, recordLinkedIssueScopeMismatchOverride, recordAiJudgmentHoldConfirmations) could share a small generic helper to keep this file from growing further with each new rule category.
• Factor the three lookback-scoped 'query rule history → record override if fired' helpers into one parameterized function (ruleIds, verdict, lookbackMs) to avoid a fourth near-duplicate the next time a new confirmation/reversal source is added.
• Consider using Promise.all in recordAiJudgmentHoldConfirmations for consistency with recordConfiguredGateBlockerOverrides, even though the list is currently just 2 entries.

Decision drivers

• ✅ Code review — No blockers (1 reviewer)
• ✅ Gate result — Passing (No configured blocker found.)

Context & advisory signals — never blocks the verdict

Signal	Result	Evidence
Linked issue	✅ Linked	#8123
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 (1 linked issue).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 51 registered-repo PR(s), 16 merged, 33 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	cleanjunc; Gittensor profile; 51 PR(s), 33 issue(s).
Improvement	✅ Minor	risk: clean · value: minor · LLM: moderate

Linked issue satisfaction

Addressed
The PR adds the owner-vs-contributor closed-event branch in recordReversalSignals reusing the ownerLogin===senderLogin check, queries queryRuleHistory for both AI_JUDGMENT_BLOCKER_CODES within a 30-day lookback, and calls recordHumanOverride with verdict 'confirmed' per matching code, wrapped in .catch(() => undefined) as required. Tests cover owner vs contributor, held vs not-held, matching vs no

Review context

• Author: cleanjunc
• Role context: outside_contributor
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: not available
• Official Gittensor activity: 51 PR(s), 33 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Keep the PR focused and include validation evidence before maintainer review.

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.

🧪 Chat with LoopOver

Ask LoopOver a question about this PR directly in a comment — grounded only in the same cached, public-safe facts shown above, never a new claim.

• @loopover ask <question> answers contribution-quality Q&A with source citations and freshness.
• @loopover chat <question> answers in natural prose from cached decision-pack facts via local inference (maintainer/collaborator; read-only).
• A plain-language @loopover mention with a real question is routed to the closest matching read-only command automatically — no exact syntax required.

Full command reference: https://loopover.ai/docs/loopover-commands

🧪 Experimental — new and may change.

_{🟩 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 LoopOver, a quiet PR intelligence layer for OSS maintainers.

• Re-run LoopOver review