What problem does it solve? Code reviews often miss regressions hiding in unchanged callers, skip entire risk categories, or rubber-stamp changes without evidence. This Skill enforces a rigorous, four-phase review process that produces auditable, evidence-backed verdicts instead of superficial approvals. ## Core Features & Use Cases - Four-phase review workflow: Names the semantic review unit, builds context from tests and consumers, sweeps critical and informational categories, then self-critiques to catch missed issues. - Evidence-backed findings: Every finding carries a file:line anchor with reasoning and a concrete fix, and every checked category produces a clearance line for auditability. - Structured verdict report: Outputs BLOCKERS, CONCERNS, NITPICKS, and coverage notes with a clear APPROVE, REQUEST CHANGES, or COMMENT ONLY verdict. - Use Case: A reviewer receives a large PR mixing a new API route with refactoring. The Skill detects scope drift, checks all consumers of changed contracts outside the diff, flags a missing enum consumer as a BLOCKER, and recommends splitting the unreviewable parts. ## Quick Start Ask the AI to review the current pull request diff against the stated ticket intent and produce a verdict report with blockers and coverage.
