fix: harden manifest public-safe filter

[JSONbored]

Motivation

• A repo-controlled .gittensory.yml review override could bypass the stricter public comment sanitizer and cause sensitive terms (e.g. seed phrase, private rankings) to be published under the GitHub App identity, so the manifest-level public-safe check needed to be aligned with the site-wide sanitizer.

Description

• Tighten the shared manifest public-safety guard by expanding isFocusManifestPublicSafe to block additional forbidden public-output terms and common plural/variant forms (covers seed phrase, private key, plural rankings, score/scoreability/estimate variants, and related patterns) in src/signals/focus-manifest.ts.
• Add a regression test ensuring review footer/note overrides containing terms covered by the public comment sanitizer are dropped with warnings in test/unit/focus-manifest.test.ts.
• Extend the sanitizer coverage assertions in test/unit/policy-sanitizer.test.ts so the shared guard is exercised for the newly-covered terms.
• Preserves existing behavior for safe inputs; only manifest-parsed public strings are more strictly rejected before they can reach public comment sinks.

Testing

• Ran npm exec -- vitest run test/unit/focus-manifest.test.ts test/unit/policy-sanitizer.test.ts and the tests passed.
• Ran npm run typecheck (tsc --noEmit) which completed successfully.

---

Codex Task

[ghost]

gittensory · advisory review

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

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

Reviewer A · gpt-oss-120b — recommends ✅ merge
The PR expands the public‑safe filter regex to block additional forbidden terms, adds tests to verify that review overrides containing these terms are stripped, and updates the policy sanitizer test suite accordingly. The changes are straightforward and maintain existing functionality.

Suggestions

• Consider adding a test case for a plain "score" term to confirm whether it should be filtered, given the original regex covered \bscore\w*.
• Run the full test suite to verify that the expanded regex does not introduce performance regressions on large manifest texts.

Worth double-checking

• The removal of the generic \bscore\w* pattern could allow unintended safe‑passing of simple "score" mentions; verify this aligns with policy.
• Complex regex may have edge‑case false positives; ensure no legitimate public content is inadvertently blocked.

Reviewer B · mistral-small-3.1-24b-instruct — recommends ✅ merge
This PR strengthens the public-safe filter in the focus manifest by expanding the list of forbidden terms. It includes new tests to ensure the filter works as intended. The change is clear and well-targeted.

Suggestions

• Consider adding a comment explaining the rationale behind the new terms added to the regex pattern.
• Ensure that the new terms are documented in the relevant parts of the codebase or documentation to maintain clarity.

Worth double-checking

• Verify that the expanded regex pattern does not inadvertently block legitimate terms.
• Check that the new tests cover all edge cases and that the existing tests remain valid.

[superagent-security]

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

[ghost]

Note

Gittensory Gate skipped

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

Signal	Result	Evidence	Action
Gate result	⚠️ Skipped	#659 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.

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	gittensory-ui	6beaf3c	Commit Preview URL Branch Preview URL	Jun 14 2026, 02:15 PM