What problem does it solve? Review feedback arrives from many sources—PR comments, reviewer subagents, bots, or your human partner—and acting on it blindly leads to unverified changes, silent agreement with wrong suggestions, and partial implementations that break things. This Skill enforces a disciplined procedure: verify each item against the codebase, push back with technical reasoning where the reviewer is wrong, and implement only what survives. ## Core Features & Use Cases - Source-based trust rules: Feedback from your human partner is trusted and clarified, while feedback from bots, subagents, and other external sources must be verified against the code before acting. - Structured response procedure: Read all items first, assign severity tiers where missing, verify each item, sort into accept/pushback/unclear, and implement in severity order with per-fix testing. - Mandatory pushback rules: Push back with cited evidence when a suggestion breaks functionality, conflicts with prior decisions, violates YAGNI on unused code, or is technically incorrect for the stack. - Use Case: A reviewer subagent returns five findings on your pull request. The Skill verifies each against the code, implements the three valid ones in severity order, pushes back on one that breaks macOS 10.15 compatibility, and asks a clarifying question on the last before touching anything. ## Quick Start Address the review comments on this pull request by verifying each item, pushing back where the reviewer is wrong, and implementing the valid fixes in severity order.
