feat(review): ban size and stale-base adjudication at parse time, add the explicit unknown abstention (#8833)

[JSONbored]

What

Two more objectively-checkable facts move behind the deterministic boundary, enforced exactly the way the CI-state ban (#8845) and the test-evidence ban already are — demoted at parse time on BOTH the Workers and BYOK paths, so no parse route escapes:

Fact the model can no longer flip	Deterministic owner	Enforcement
PR size ("too large", "should be split into smaller PRs")	sizeGateMode → oversized_pr	SIZE_CLAIM_PATTERN → demoteSizeClaimBlockers, unconditional
Base staleness / rebase / merge conflicts	fetchBaseAheadBy → stale_base_ref; GitHub's own mergeable_state	STALE_BASE_CLAIM_PATTERN → demoteStaleBaseClaimBlockers, unconditional

Both patterns are deliberately claim-shaped: a code-magnitude judgment ("this buffer is too large"), "this function should be split into smaller helpers", or code ABOUT rebasing/conflict-resolution never match — the ban covers assertions about the SUBMISSION and repository state, which are not even visible in a diff.

Explicit abstention (the issue's "unknown → hold" requirement): the rubric now offers "confidence": "unknown" for a model that genuinely cannot judge its blockers. parseReviewConfidence maps it to CONFIDENCE_WHEN_UNSTATED (0.5), under every sane close floor, so an honest abstain routes to the low-confidence hold disposition (a human decides) instead of a guessed verdict.

REVIEW_PROMPT_VERSION bumps to v2 — both edits shape the judge's verdict surface, which the #9477 cache fingerprint (drift = automatic cache miss) and the #8222 counterfactual replay key on.

Tests

• Pattern both-arm tests: six positive size phrasings + four code-magnitude negatives; seven positive staleness/rebase/conflict phrasings + four code-about-the-topic negatives.
• Demotion behavior: demoted claims land in nits annotated with the deterministic owner; unrelated blockers untouched; zero-demotion returns the same object (no reallocation), matching every sibling demotion.
• REGRESSION (parse-path): runWorkersOpinion demotes both claim kinds and logs ai_review_size_claim_demoted / ai_review_stale_base_claim_demoted; the BYOK provider path does the same — the LLM path can no longer flip either fact.
• Abstention: the rubric offers the exact string the parser maps, and parseReviewConfidence("unknown") lands under the 0.93 default close floor.
• Prompt-family suites (integrity, cache-input fingerprint, counterfactual replay) green against v2; the deliberate-bump cache test now spoofs v99 so it stays meaningful after this real bump.

Part of #8833 — the structured verifiable-blocker schema ({claim_kind, path, line, evidence_ref}) lands as its own PR next.

[loopover-orb]

Warning

⏸️ LoopOver review result - manual review recommended

_{Review updated: 2026-07-28 21:13:16 UTC}

4 files · 1 AI reviewer · 2 blockers · CI green · clean

⏸️ Suggested Action - Manual Review

Review summary
This PR extends the existing CI-claim/test-evidence-absence demotion pattern to two new claim shapes — whole-PR size claims and base-staleness/rebase/merge-conflict claims — using the identical unconditional demote-to-nit shape, wired into both the Workers and BYOK provider paths, plus adds an explicit "unknown" confidence abstention that already falls through the existing non-number fallback in parseReviewConfidence. The regexes are anchored on submission-noun + claim-verb shapes so code-content sentences ("this buffer is too large", "the rebase helper mutates...") don't false-positive, and both directions are covered by dedicated positive/negative test suites plus end-to-end wiring tests on both call paths. REVIEW_PROMPT_VERSION is correctly bumped to v2 and the two dependent cache/replay tests are updated in lockstep, so no stale-fingerprint gap is introduced.

Nits — 5 non-blocking

• The nit message baked into demoteSizeClaimBlockers cites `sizeGateMode`/`oversized_pr` as the deterministic owner (src/services/ai-review.ts, demoteSizeClaimBlockers) — that owner isn't visible anywhere in this diff, so it's worth double-checking the field/finding name is accurate before it's surfaced verbatim to contributors.
• The four new regex constants (SIZE_CLAIM_PATTERN, STALE_BASE_CLAIM_PATTERN) are long single-line alternations with no inline comments breaking down each alternative — consider splitting into named sub-patterns the way test cases already group them (size vs split-phrasing; behind vs rebase vs conflict) for future maintainability.
• The console.warn blocks in runWorkersOpinion/runProviderReview (src/services/ai-review.ts around the new sizeDemotion/staleBaseDemotion calls) repeat the same four-line pattern four times now (ci, evidence, test-evidence, size, stale-base) — a small helper `warnDemoted(event, count, extra)` would remove the growing duplication.
• Consider factoring the repeated `if (xDemotion && xDemotion.demoted.length > 0) console.warn(...)` blocks in runWorkersOpinion and runProviderReview into a single small helper now that there are five near-identical instances per call site.
• If sizeGateMode/oversized_pr isn't the exact live field/finding name, update the demotion nit text in demoteSizeClaimBlockers to match the actual gate-side naming so the surfaced message stays accurate to contributors.

Concerns raised — review before merging

• 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.

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: 14 registered-repo PR(s), 13 merged, 276 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 14 PR(s), 276 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 registered but has no active allocation in the current snapshot.
• Public profile languages: Python, TypeScript, Ruby, Go, MDX, Shell, Solidity, JavaScript
• Official Gittensor activity: 14 PR(s), 276 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 3 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.

Decision record

• action: hold · clause: missing_linked_issue
• config: 844a5e795a8d848738a37ec992189d0e164e822cc111877acbda51bb953e859e · pack: oss-anti-slop · ci: passed
• record: 2c1bfbc91914442e37b9be93a605bc482a096851293daa516b4301b024da0b89 (schema v5, head fb468c3)

_{🟩 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

[superagent-security]

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

[github-actions]

Logic backtest

Replayed 0 historical case(s) for linked_issue_scope_mismatch through the base (a858cbc) and head (fb468c3) 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).

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.03%. Comparing base (a858cbc) to head (fb468c3).
⚠️ Report is 2 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #9592      +/-   ##
==========================================
- Coverage   89.84%   89.03%   -0.81%     
==========================================
  Files         879      879              
  Lines      111071   111095      +24     
  Branches    26439    26449      +10     
==========================================
- Hits        99787    98914     -873     
- Misses       9992    11155    +1163     
+ Partials     1292     1026     -266     

Flag	Coverage Δ
backend	94.15% <100.00%> (-1.47%)	⬇️

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

Files with missing lines	Coverage Δ
src/services/ai-review.ts	96.88% <100.00%> (+0.09%)	⬆️

... and 3 files with indirect coverage changes