fix(predict-gate): resolve contributor confirmation for gate prediction

[galuis116]

Summary

The pre-submission gate prediction — the gittensory_predict_gate MCP tool (#693/#712) and the /v1/local/branch-analysis self-gate (#688) — never passed confirmedContributor to buildPredictedGateVerdict, so it stayed undefined.

The prediction promises strict parity with the maintainer gate (see predicted-gate.ts header: "the same verdict the gate would compute post-submission" and "Only confirmed Gittensor contributors are ever hard-blocked"). The real gate enforces that in evaluateGateCheck:

if (policy.confirmedContributor === false && blockers.length > 0) {
  return { conclusion: "neutral", ... }; // non-confirmed authors are never hard-blocked
}

The maintainer pipeline resolves a concrete boolean (processors.ts): const confirmedContributor = official?.status === "confirmed";. The prediction callers didn't — and undefined === false is false, so the neutral branch was skipped and any fired blocker produced conclusion: "failure".

Impact: under the default gittensor pack, every non-confirmed contributor (i.e. the first-time / agent users this self-check exists for) whose synthetic PR trips an opted-in blocker (e.g. duplicates: "block", the default) was told "failure — your PR will be blocked", when the real post-submission gate returns neutral / non-blocking for them. The prediction was the opposite of the real outcome for exactly its target audience.

Fix

Resolve the caller's own confirmed status the same way the pipeline does (official Gittensor API → confirmed) and pass it through:

• API route (/v1/local/branch-analysis): reuse the already-fetched context.gittensorSnapshot — it's non-null only when fetchGittensorContributorSnapshot confirms the login, matching official?.status === "confirmed". Zero new fetch.
• MCP predictGate: resolve fetchGittensorContributorSnapshot(login) !== null, gated behind the pack so the oss-anti-slop pack (which drops the contributor gate entirely) stays account-free and fetch-free.

No change to buildPredictedGateVerdict itself — it already accepted confirmedContributor and handled the oss-anti-slop override; the callers simply never supplied it.

Tests

The existing predicted-gate.test.ts locked the function-level parity by explicitly passing confirmedContributor: false — a path no real caller exercised, which masked the bug. Added caller-level regression tests in mcp-predict-gate.test.ts (stubbing the Gittensor /miners endpoint) under the gittensor pack:

• non-confirmed contributor + blocker → neutral (confirmedContributor: false),
• confirmed contributor + same blocker → failure (confirmedContributor: true).

Full coverage suite green locally (only the known CRLF-local gittensory-focus-manifest and the local mcp-cli timeouts fail locally; both pass in CI).

[ghost]

Note

Gittensory Gate skipped

PR closed before full evaluation. No late first comment was created.

Signal	Result	Evidence	Action
Gate result	⚠️ Skipped	#715 is no longer open.	No action.

---

💰 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.

[superagent-security]

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

[ghost]

gittensory · advisory review

Reviewed 3 changed file(s) — two independent AI reviewers.

Changed files: package.json, src/mcp/server.ts, test/unit/mcp-predict-gate.test.ts

Suggested action: ✅ Safe to merge — both reviewers found no blocking issues.

Reviewer A · gpt-oss-120b — recommends ✅ merge
The PR adds contributor confirmation handling to gate prediction, updates dependencies, and introduces comprehensive tests verifying the new logic under various scenarios.

Suggestions

• Remove the newly added ws, tar, and js-yaml dependencies from package.json if they are not used, to keep the bundle size minimal.
• Confirm that fetchGittensorContributorSnapshot safely catches network errors and returns null so the added await does not cause unhandled rejections.
• Verify that buildPredictedGateVerdict’s TypeScript signature includes the optional confirmedContributor field to avoid compile‑time errors.

Worth double-checking

• If fetchGittensorContributorSnapshot does not handle fetch failures internally, the prediction endpoint could throw on network errors.
• Unused dependencies could increase build time and surface attack surface.

Reviewer B · mistral-small-3.1-24b-instruct — recommends ✅ merge
This PR fixes a parity issue in the contributor confirmation logic for gate prediction in the Gittensory MCP server. It adds necessary checks to ensure that only confirmed Gittensor contributors are hard-blocked, and it includes comprehensive tests to cover the new behavior. The changes are well-documented and follow the project's conventions.

Suggestions

• Consider adding a comment explaining why ws, tar, and js-yaml are being added to package.json.
• Ensure that the fetchGittensorContributorSnapshot function is thoroughly tested, especially for edge cases like network failures.

Worth double-checking

• Verify that the new logic does not introduce any performance bottlenecks, especially with the additional network call to the Gittensor API.
• Double-check that the tests cover all possible scenarios, including cases where the Gittensor API might return unexpected results.

[JSONbored]

PR #715 — fix(predict-gate): resolve contributor confirmation for gate prediction
Author: galuis116 | CI: ✅ pass | Status: Concerns
ACTION: REQUEST CHANGES

Highlights:

• Correct gate-parity fix: passes confirmedContributor so non-confirmed contributors get neutral not failure — mirrors evaluateGateCheck. Moves in the SAFE direction (less likely to block outsiders), does not weaken auth.
• oss-anti-slop pack stays account/fetch-free; gittensor pack resolves via fetchGittensorContributorSnapshot(login) !== null, matching the confirmed-status pipeline.
• Caller-level regression tests added (confirmed→failure, non-confirmed→neutral).

Change requests / reasons:

• New fetchGittensorContributorSnapshot is now on the prediction path: add SSRF-guard confirmation + a network-failure/error-path test.
• failing CI

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.68%. Comparing base (8bdd2bd) to head (be92de6).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #715   +/-   ##
=======================================
  Coverage   96.68%   96.68%           
=======================================
  Files          93       93           
  Lines       13769    13771    +2     
  Branches     5015     5017    +2     
=======================================
+ Hits        13312    13314    +2     
  Misses         88       88           
  Partials      369      369           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[JSONbored]

PR #715 — fix(predict-gate): resolve contributor confirmation for gate prediction
Verdict: MERGE
Resolves issue: N/A (bugfix referencing #693/#712/#688, no direct "Fixes")
CI: All green (validate, test x2, lint, workers, ui, security, Superagent Security Scan = success, codecov). gittensory/Context neutral.

Highlights:

• Correctly diagnoses the real bug: callers never passed confirmedContributor, leaving it undefined; since undefined === false is false, the "non-confirmed authors are never hard-blocked" neutral branch was skipped and fired blockers produced a false failure — opposite of the real gate outcome for the first-time/agent users this self-check targets.
• Resolves confirmed status the same way the maintainer pipeline does: API route reuses the already-fetched gittensorSnapshot (zero new fetch); MCP predictGate resolves fetchGittensorContributorSnapshot(login) !== null, gated behind the pack so oss-anti-slop stays account/fetch-free. buildPredictedGateVerdict itself unchanged (right surface).
• Meaningful caller-level regression tests: non-confirmed+blocker -> neutral, confirmed+blocker -> failure, API-failure -> fail-safe neutral.

Concerns / required changes:

• package.json adds ws/tar/js-yaml under the lovable overrides block — unrelated to a predict-gate fix, most likely a rebase/transitive-pin artifact. Confirm it's an intentional security pin rather than stray scope; not a blocker.

Completeness vs issue: Fully closes the parity gap in both callers with the correct neutral/failure/fail-safe regression matrix. Correct and complete as a standalone bugfix.