What problem does it solve? After a code review returns a rework verdict, developers often overreach: while fixing one flagged issue they silently refactor unrelated code, mask missing evidence with better mock tests, or forget to rerun the full regression suite. This Skill enforces disciplined, point-by-point repair of only what the reviewer flagged. ## Core Features & Use Cases - Scoped repair: Reads the full code-review report first and converts every critical and important finding into an explicit code or test change, touching nothing else. - Evidence integrity: Requires real verification instead of disguised mock tests when a finding questions test coverage, and mandates rerunning the full regression after fixes. - Report hygiene: Updates the existing coverage table and changed-files list in test_report in place, records accepted non-critical debt, and writes a separate repair report to .process/<task-id>/implementation_repair.md. - Use Case: A code-reviewer returns a rework verdict with three critical findings on a game feature slice; this Skill fixes exactly those findings, reruns regression, commits the changes, and returns a structured status for the repeated review gate. ## Quick Start Use the implementation-repair skill to fix the findings from the latest code-review report and update the test report.
