Refine serious-scout overlap and cosmetic intent handling

[VrtxOmega]

Summary

• treat any returned matching open PR as conclusive overlap evidence and stop later identifier searches for that issue
• keep zero-result incomplete or contradictory overlap responses fail-closed
• distinguish incidental cosmetic vocabulary in bug bodies from explicit cosmetic intent in titles or labels
• stop treating operational version cleanup failures as generic maintenance chores
• document the behavior and lock the live residue into focused regressions

Live residue

The isolated 20260711-1542 scout completed all six broad issue searches with zero collection errors after the rate-limit repair, then correctly returned NO_ACTION because two required overlap searches were marked incomplete/truncated. The same run also demoted three strong reports because issue templates or operational language contained documentation, docs, or cleanup vocabulary.

This change preserves the fail-closed authority boundary while correcting those false negatives. A targeted local scoring replay of the three public issue payloads now produces three candidate rows, but that fixture result is not worker authorization: live overlap, policy, ownership, and current-upstream reproduction gates remain required.

Validation

• node --test test/serious-scout.test.mjs test/github-client.test.mjs (43/43)
• npm run ci:gates (257/257 tests, 77/77 benchmark, 29/29 adversarial, maintainer demo PASS)
• npm run mcp:smoke
• npm pack --dry-run --json (75 files)
• git diff --check
• independent review reproduced 42/42 focused tests before the final mixed-subquery regression and returned ACCEPT

Non-claims

• no target repository was cloned or modified
• no issue or pull request was commented on, labeled, or contacted
• no candidate is authorized for coding by the targeted fixture replay
• remote CI on this exact commit is still required before merge

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.