What problem does it solve? Single-reviewer code reviews miss blind spots and inflate nitpicks. This Skill runs adversarial multi-model review so independent LLM reviewers challenge a diff from different angles, then a lead judgment step filters noise into an actionable verdict. ## Core Features & Use Cases - Multi-Model Adversarial Review: Spawns one reviewer per configured model (Claude, GPT, Grok, etc.) with the same prompt, rubric, and code-quality lens so agreement across models signals high-confidence findings. - Structured Synthesis: Deduplicates findings, maps consensus and disagreement, and categorizes every item as Act On, Consider, Noted, or Dismissed with rationale. - Lead Judgment Filtering: Applies a pragmatic senior-engineer framework to reject hypothetical, nitpicky, or context-missing findings before presenting the verdict. - Use Case: Before merging a feature branch, run an adversarial review on git diff main...HEAD to catch correctness, security, and structural issues that a single reviewer would miss. ## Quick Start Ask the agent to interrogate the current branch changes and produce a synthesized adversarial review verdict.
