code-review

Run a multi-model consensus code review using Claude sub-agents

This skill should be used when the user asks to "review my code", "do a code review", "review this project", "find bugs in my code", "security review", "review my changes", "code audit", "check my code for issues", "analyze code quality", mentions multi-model review, consensus review, or wants a thorough code review using multiple AI models. Performs a three-phase review: light models for reconnaissance, powerful models for deep analysis with consensus scoring, and verification agents that test each finding with code snippets to eliminate false positives.