What problem does it solve? When a model generates code and then reviews its own output in the same session, the review is biased by the generation context and misses real defects. This Skill replaces self-review with independent reviewers in clean sessions and structures large PR reviews into two passes, while rejecting majority-vote quorum filtering that discards rare legitimate findings. ## Core Features & Use Cases - Independent Reviewer Sessions: Each file is reviewed in a fresh session that never saw the code-generation reasoning chain, eliminating self-review bias. - Two-Pass Review Structure: Pass A performs a per-file deep dive with typed findings; Pass B performs cross-file integration over only the Pass A summaries to detect broken contracts and duplicate findings. - Minority Finding Preservation: Rejects N-of-M quorum consensus so a bug reported by a single reviewer is preserved for evidence-based validation instead of being voted away. - Use Case: A team generates a 14-file payments module with an AI model. Instead of asking the same session to review its own work and consensus-filtering 2-of-3 reviews, this Skill runs independent per-file reviews plus a cross-file integration pass and keeps every finding. ## Quick Start Review this multi-file pull request using independent clean-session reviewers with a per-file pass followed by a cross-file integration pass, and do not discard any minority findings by majority vote.
