What problem does it solve? When you receive a code review from another AI model, a review bot, or a teammate, you cannot tell which findings are real without manually checking each one against the code. This Skill cross-checks every claimed issue against the actual codebase and classifies it, so you only act on genuine problems. ## Core Features & Use Cases - Finding-by-finding verification: Parses each discrete finding (claim, file:line location, severity) and reads the referenced code to confirm or refute it. - Structured verdicts: Classifies every finding as Valid, Invalid, Partial, Already fixed, or Needs info, with quoted code evidence and reasoning. - Triage report output: Produces a grouped summary with counts per verdict and clear next-step prompts, without modifying any code. - Use Case: A review bot flags 12 issues in your pull request. Run this Skill to discover that 4 are valid bugs, 6 are misreadings of the code, and 2 were already fixed, then fix only the real ones. ## Quick Start Validate the findings in this pasted review report against my current codebase and give me a triage of which ones are real.
