What problem does it solve? Single-perspective code reviews miss blind spots, and raw multi-reviewer output is noisy. This Skill runs adversarial reviews of a diff across multiple independent LLM models, then applies lead-reviewer judgment to filter findings into an actionable verdict. ## Core Features & Use Cases - Multi-Model Adversarial Review: Launches one reviewer per configured model (e.g., opus and sonnet) with the same prompt and rubric, so agreement across models is a high-confidence signal. - Structured Rubric and Quality Lens: Reviewers apply a shared rubric covering correctness, root causes, structural integrity, verification, complexity, and security, plus a strict code-quality lens targeting structural simplification. - Lead Judgment Synthesis: Findings are deduplicated, consensus is identified, and every item is categorized as Act On, Consider, Noted, or Dismissed with rationale. - Use Case: Before merging a feature branch, ask for an adversarial review of git diff main...HEAD and receive a synthesized verdict listing only the issues that would block a real PR. ## Quick Start Ask the AI to interrogate the changes on your current feature branch and produce a synthesized adversarial review verdict.
