What problem does it solve? Code review reports list bugs, but a Story marked done does not prove the bugs were actually fixed. This Skill closes that gap by reading the real source code for each finding, judging whether it is FIXED, PARTIAL, OPEN, NOT_APPLICABLE, or FALSE_POSITIVE, and persisting the verdict into the review_findings and tech_debt_items database tables so nothing falls through before the Production Gate. ## Core Features & Use Cases - Code-level fix verification: Reads the affected file at the reported line number and compares it against the bug's root cause before marking anything fixed, with file:line evidence recorded. - Bug query and search modes: Full-text search (FTS5), per-file bug lookup, per-module overview, and severity/status statistics across the review_findings table. - Status synchronization: Writes FIXED results to review_findings and syncs DEFERRED bugs into tech_debt_items with severity mapping (P0→critical, P1→high, etc.). - Use Case: Before a code-review Production Gate, run verification for a Story, confirm all P0/P1 bugs are fixed with code evidence, defer the rest into tech debt, and block the gate if any critical bug remains open. ## Quick Start Ask the AI to verify whether all bugs assigned to story X are actually fixed in the code and update the review findings database with the results.
