fix(review): scope the bare-score public-safety check per repo (metagraphed#8038)

[JSONbored]

Summary

Live, recurring issue: metagraphed#8038 (and many others) got "The AI review returned non-blocking notes for this change but did not include a separate narrative summary" instead of a real review write-up, despite the model producing one.

Root cause: sanitizePublicComment throws (by design — never redacts) whenever public-facing AI text matches a bare \bscore\w*\b, to stop loopover's own private gittensor miner trust/reward values from leaking. toPublicSafe catches that throw and returns null, and composeAdvisoryNotes silently discards the entire narrative assessment in favor of the generic fallback text. This filter is applied identically across every repo the shared engine serves — but metagraphed's own public schema legitimately has fields named totalScore, credibility, baseTotalScore, etc. Any AI review that naturally describes this code in prose ("the resolver filters results by their score...") is near-certain to trip the filter and lose its whole write-up.

Fix, scoped as narrowly as possible:

• sanitizePublicComment gains an optional { allowBareScoreTerm?: boolean }, default false (today's unchanged behavior for every existing caller that doesn't pass it). Every other forbidden phrase — wallet, hotkey, coldkey, trust score, reward, scoreability, etc. — stays enforced unconditionally, everywhere, regardless of this flag. Only the over-broad bare-word check is ever relaxable, and only where a caller has positively confirmed it's safe.
• New isPublicScoreTermSafeForRepo(env, repoFullName): a fail-closed (deny-by-default), config-as-code repo allowlist (LOOPOVER_PUBLIC_SCORE_TERMS_ALLOWED_REPOS, comma/whitespace owner/repo list). Unset ⇒ the check stays enforced for every repo, unchanged from today. Deliberately no */all wildcard (unlike the MCP allowlists) — this is a safety-relevant exemption, and a blanket opt-in-everything value would silently cover a future repo that does have private trust/reward data.
• composeAdvisoryNotes threads the flag through to its 3 relevant toPublicSafe calls (assessment, blockers, nits) only — the other 11 toPublicSafe call sites in ai-review.ts (blocker titles, tie-break judge output, consensus titles) are untouched.
• runLoopOverAiReview resolves the flag once, from isPublicScoreTermSafeForRepo(env, input.repoFullName).

This is additive and inert until an operator sets the env var for a specific repo — no behavior changes for any repo today.

Validation

• New tests in queue-intelligence.test.ts: the option passes a bare "score" through when true; still throws when false; every OTHER forbidden phrase still throws even when the flag is true; isPublicScoreTermSafeForRepo's allow/deny/no-wildcard behavior.
• New regression tests in ai-review.test.ts: a bare "score" mention drops the whole assessment by default but survives when allowlisted (with a surviving, score-free nit proving the drop doesn't silently empty unrelated findings); the explicit-phrase bans (raw trust score) stay enforced even with the flag on.
• npm run typecheck clean; npm run test:coverage (full, unsharded): all tests passing.

[github-actions]

Logic backtest

Replayed 0 historical case(s) for linked_issue_scope_mismatch through the base (c4a6426) and head (0a076df) versions of its detection logic (corpus checksum 4f53cda18c2b).

Backtest comparison: linked_issue_scope_mismatch

Verdict: unchanged — no comparable axis moved.

Advisory only — this check never blocks merge (#8105).

[superagent-security]

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

[loopover-orb]

Caution

🛑 LoopOver review result - fixes required

_{Review updated: 2026-07-24 21:22:44 UTC}

5 files · 1 AI reviewer · 2 blockers · CI failing · unstable

🛑 Suggested Action - Manual Review

Review summary
The AI review returned non-blocking notes for this change but did not include a separate narrative summary. Review the nits below before deciding this PR.

Nits — 5 non-blocking

• src/queue-intelligence.ts's new `isPublicScoreTermSafeForRepo` and `sanitizePublicComment` doc comments are quite long relative to the function bodies; could be trimmed without losing the safety rationale.
• The magic number 8038 appears only in a doc comment reference (metagraphed#8038), not in logic, so it's not a real magic-number concern despite the external brief flagging it.
• Consider whether `LOOPOVER_PUBLIC_SCORE_TERMS_ALLOWED_REPOS` documentation in env.d.ts and queue-intelligence.ts could link to each other via a shared anchor for discoverability.
• src/queue-intelligence.ts:157 — the allowlist parsing duplicates the comma/whitespace-split pattern used elsewhere (MCP allowlists); if that pattern changes again, keep both in sync manually since they're intentionally separate per the doc comment.
• test/unit/ai-review.test.ts's two new REGRESSION tests are a good template if other repo-scoped safety exemptions are added later.

Why this is blocked

• No linked issue detected: No closing reference or linked issue number was found in the PR metadata/body. — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue: This repo's maintainer focus manifest requires every PR to reference a tracked issue. — Link the relevant issue (for example Closes #123) before opening the PR.

📋 Copy for AI agents — paste into your coding agent

Fix the following blocker(s) from this PR review:

1. No linked issue detected: No closing reference or linked issue number was found in the PR metadata/body. — If this PR is intended to solve an issue, link it explicitly in the PR body.

2. Maintainer requires a linked issue: This repo's maintainer focus manifest requires every PR to reference a tracked issue. — Link the relevant issue (for example `Closes #123`) before opening the PR.

CI checks failing

• codecov/patch — 91.66% of diff hit (target 99.00%)

Decision drivers

• ❌ Code review — 2 blockers (1 reviewer)
• ❌ Gate result — Blocking (Repo-configured hard blocker found.)

Context & advisory signals — never blocks the verdict

Signal	Result	Evidence
Linked issue	⚠️ Missing	No linked issue or no-issue rationale found.
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: 13 registered-repo PR(s), 13 merged, 272 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 13 PR(s), 272 issue(s).
Improvement	✅ Minor	risk: clean · value: minor

Review context

• Author: JSONbored
• Role context: owner (maintainer lane)
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: Python, TypeScript, Ruby, Go, MDX, Shell, Solidity, JavaScript
• Official Gittensor activity: 13 PR(s), 272 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Start here: Treat this as maintainer-lane context rather than normal contributor-lane activity.
• Then work through the remaining 2 steps in the Signals table above.

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

[codecov]

Codecov Report

❌ Patch coverage is 86.36364% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.93%. Comparing base (bd273c1) to head (0a076df).
⚠️ Report is 5 commits behind head on main.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
src/queue/processors.ts	70.00%	2 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #8562      +/-   ##
==========================================
+ Coverage   89.58%   91.93%   +2.35%     
==========================================
  Files          97      795     +698     
  Lines       22706    79669   +56963     
  Branches     3872    24075   +20203     
==========================================
+ Hits        20341    73246   +52905     
- Misses       2187     5309    +3122     
- Partials      178     1114     +936     

Flag	Coverage Δ
backend	92.87% <86.36%> (?)

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

Files with missing lines	Coverage Δ
src/queue-intelligence.ts	100.00% <100.00%> (ø)
src/services/ai-review.ts	97.08% <100.00%> (ø)
src/queue/processors.ts	95.66% <70.00%> (ø)

... and 695 files with indirect coverage changes