What problem does it solve? Standard code reviews stop when the reviewer stops finding issues, which means a later re-run often surfaces problems that were present all along. This Skill replaces that subjective stopping point with a verifiable finish line: every unit of the changed surface must carry an explicit verdict from every review lane before the review is considered done. ## Core Features & Use Cases - Frozen review surface: Enumerates changed hunks, whole touched files, and every caller of changed symbols into a ledger stored at .git/exhaustive-review/<branch>.md before any reviewing begins. - Six parallel review lanes: Dispatches one subagent per lane (correctness, concurrency, security, contracts/blast radius, tests, reuse/consistency), each covering the entire unit list with per-unit verdicts. - Gap closure and re-review: Re-dispatches lanes with empty cells, deduplicates findings by file:line, and rebuilds the surface after fixes so re-reviews cover moved code. - Use Case: Before merging a large branch touching authentication and database migrations, run this Skill to confirm every hunk, file, and caller was checked by all six lanes, with pre-existing issues reported separately from introduced ones. ## Quick Start Ask the AI to run an exhaustive code review of the current branch against its merge base and confirm the six-lane fan-out when prompted.
