fix(review): meter tie-break judge budget

[JSONbored]

Closes #3679

Summary

• The order-swapped dual-AI tie-break path (fix(review): add order-swapped stability check to the dual-AI tie-break pass #2997) performs additional judge calls — with retries and
  per-model fallback — that were never counted against the free daily AI neuron budget, letting a
  contributor spend extra provider usage beyond the configured limit by triggering repeated consensus
  disagreements.
• Reserves a worst-case tieBreakAiCalls budget (2 judge calls x up to 3 attempts x a per-model
  fallback multiplier) and adds it to estimatedNeurons in runGittensoryAiReview
  (src/services/ai-review.ts), gated on mode === "block" && dual && combine === "consensus" — the
  only path that can actually reach the tie-break judge.
• Two regression tests: the default two-distinct-models fallback multiplier (x2), and a self-host
  single-model configuration where a reviewer has no distinct fallback (primaryFallback === primary.model), exercising the x1 multiplier branch.

Scope

• The PR title follows type(scope): short summary Conventional Commit format.
• This PR is focused — one source file, one test file.
• Follows CONTRIBUTING.md; no site//CNAME/VitePress changes.
• Linked issue: Closes #3679.

Validation

• git diff --check
• npm run actionlint — not run; no workflow files touched.
• npm run typecheck (clean)
• npm run test:coverage (full/unsharded) — not run locally; ran the affected test file
  (test/unit/ai-review.test.ts, 153 tests, all green) plus the full queue.test.ts integration
  suite (732 tests). Cross-referenced the exact new-diff line/branch ranges against a fresh
  coverage-final.json (--coverage.include='src/services/ai-review.ts'): 100% of the new
  statements and branches are covered (the original patch was missing the x1-fallback-multiplier
  branch; closed with a second new test). GitHub CI runs the full suite/gate on push.
• npm run test:workers / npm run build:mcp / npm run test:mcp-pack — not run; nothing in
  those surfaces touched.
• npm run ui:openapi:check / ui:lint / ui:typecheck / ui:build — not run; no API/schema or
  apps/gittensory-ui/** changes.
• npm audit --audit-level=moderate — not run; no dependency changes.
• New/changed behavior has regression tests for both branches of the new fallback-multiplier
  ternary, plus the existing quota-exceeded assertion pattern already used elsewhere in this file.

If any required check was skipped, explain why:

• Local validation here was typecheck + the specific affected test files rather than a full local
  test:ci/test:coverage/npm audit pass, since GitHub CI runs the complete gate on push and
  re-running the whole suite by hand for every change is redundant.

Safety

• No secrets, wallets, hotkeys, coldkeys, trust scores, or private scoring values are touched or
  exposed.
• Not a public-facing text change (internal budget/neuron-estimation logic only).
• Not an auth/CORS/session change; no negative-path tests needed for that reason.
• Not an API/OpenAPI change.
• Not a UI change (N/A) — no UI Evidence section included.
• No changelog edit.

[superagent-security]

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

[loopover-orb]

Tip

🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩🟩

✅ Gittensory review result - approve/merge recommended

_{Review updated: 2026-07-06 04:11:47 UTC}

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

✅ Suggested Action - Approve/Merge

• safe to merge

Review summary
This PR adds a `tieBreakAiCalls` reservation (worst-case 2 judges × 3 attempts × fallback multiplier) to `estimatedNeurons` in `runGittensoryAiReview`, gated on `mode === "block" && dual && combine === "consensus"`, closing the gap where retried/fallback tie-break judge calls could spend AI budget uncounted. The multiplier logic (x2 for distinct fallback model, x1 when `primaryFallback === primary.model`) is tested with two focused regression tests hitting `quota_exceeded`, and the change is additive/small in scope. The main gap is that only the non-zero `tieBreakAiCalls` path is exercised — no test covers the `tieBreakAiCalls === 0` branch (e.g. non-consensus combine mode or non-block mode), which lines up with the reported codecov/patch shortfall.

Nits — 6 non-blocking

• Neither new test exercises the `tieBreakAiCalls === 0` branch (e.g. `combine !== "consensus"` or `mode !== "block"`), which is likely why codecov/patch reports only 75% diff coverage on this file.
• The literals `3` (max attempts) and `512` (judge token estimate) in src/services/ai-review.ts:1736 and :1746 are unexplained magic numbers — consider naming constants (e.g. `TIE_BREAK_MAX_ATTEMPTS`, `TIE_BREAK_TOKEN_ESTIMATE`) so the worst-case math is self-documenting.
• The comment says '2 judge calls x up to 3 attempts x per-model fallback multiplier' but doesn't reference where the retry count of 3 or the 2-judge count come from in the tie-break implementation — a pointer/comment link to that code would help future readers verify the worst-case math stays in sync.
• Add a test where `combine !== "consensus"` (or `mode !== "block"`) with a dual reviewer config to confirm `tieBreakAiCalls` stays 0 and the reservation doesn't over-count budget for paths that can't reach the tie-break judge.
• Extract the `3` and `512` literals into named constants near the tie-break judge implementation so the reservation formula and the actual judge-call site can't silently drift out of sync.
• PR author also opened the linked issue — Link an issue that was opened by a different contributor, or provide a rationale for why this self-authored issue represents genuine discovery work.

Signal	Result	Evidence
Code review	✅ No blockers	1 reviewer
Linked issue	✅ Linked	#3679
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: 55 registered-repo PR(s), 47 merged, 456 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 55 PR(s), 456 issue(s).
Gate result	✅ Passing	No configured 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: 55 PR(s), 456 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Treat this as maintainer-lane context rather than normal contributor-lane activity.
• No action.

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

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.34%. Comparing base (ddf68c3) to head (b84aec0).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3671   +/-   ##
=======================================
  Coverage   93.34%   93.34%           
=======================================
  Files         315      315           
  Lines       32107    32111    +4     
  Branches    11769    11772    +3     
=======================================
+ Hits        29969    29973    +4     
  Misses       1507     1507           
  Partials      631      631           

Files with missing lines	Coverage Δ
src/services/ai-review.ts	96.99% <100.00%> (+0.02%)	⬆️

🚀 New features to boost your workflow:

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