What problem does it solve? Code review feedback often contains false positives, stale findings, or unverifiable claims, and acting on them blindly wastes effort and introduces regressions. This Skill enforces a disciplined verdict process so only confirmed findings lead to code changes. ## Core Features & Use Cases - Evidence-Based Verdicts: Every finding is judged as CONFIRMED or REJECTED only after reading the current content at the cited location. - Structured Rejection Reasons: Rejections carry explicit reasons such as stale, unverifiable, or yagni, preventing vague dismissals. - Minimal Targeted Fixes: Only confirmed findings are fixed, one at a time, at the cited location, followed by re-running the relevant verification. - Use Case: After an automated or peer review emits a list of findings on a pull request, use this Skill to verify each finding against the actual code, reject the false positives with reasons, and fix only what is genuinely broken. ## Quick Start Ask the assistant to evaluate the review findings emitted by the requesting-code-review step and fix only the confirmed ones.
