Trevin Chow ee64fa237b feat(ce-resolve-pr-feedback, ce-pr-comment-resolver): filter bot wrappers, split test responsibility ...

- Filter CodeRabbit, Codex, Gemini, Copilot, and Codecov top-level comments
  and review bodies in get-pr-comments. These bots put actionable feedback in
  inline review threads; their wrapper bodies are reliable noise and the
  agent repeatedly narrated ignoring them. Also trim unused url/createdAt
  fields from the GraphQL payload.
- Add silent-drop rule so the agent stops announcing non-actionable items.
- Split test responsibility: resolvers run only targeted tests for what they
  changed (specific file, pattern, or the test they wrote), never the full
  project suite. Orchestrator runs the project's validation once against the
  combined diff before commit. Covers pre-existing failures (note in footer),
  failures in touched files (one inline retry, then escalate), and the
  zero-changes skip path.

2026-04-20 00:25:02 -07:00

..

coding-tutor

Remove cursor target and add plugin metadata

2026-02-13 12:14:48 -05:00

compound-engineering

feat(ce-resolve-pr-feedback, ce-pr-comment-resolver): filter bot wrappers, split test responsibility

2026-04-20 00:25:02 -07:00