What problem does it solve?
This Skill eliminates the risk of merging code with critical safety, security, and correctness issues that manual PR reviews often miss, preventing production incidents, data leaks, and costly rollbacks.
Core Features & Use Cases
- Two-Pass Critical Review: First pass flags high-severity issues like SQL injection, race conditions, LLM output trust violations, and shell injection; second pass catches informational issues like async/sync mixing, column name mismatches, and dead code.
- Specialist Subagent Checks: Runs parallel specialist reviews for API contracts, data migrations, maintainability, performance, security, and red team adversarial analysis for diffs over 200 lines.
- Use Case: For example, if you are merging a PR that adds a new user-facing feature with database queries and LLM-generated content, this Skill will flag unparameterized SQL queries, missing N+1 eager loading, and unvalidated LLM output saved to the database.
Quick Start
Ask the AI to run the review skill on your open PR diff to catch critical bugs and safety issues before you merge.