What problem does it solve? Single-reviewer code reviews miss blind spots and inherit one model's biases. This Skill runs adversarial multi-model review so independent LLM reviewers challenge a diff from different angles, then synthesizes their findings into one categorized verdict. ## Core Features & Use Cases - Multi-Model Review: Spawns one read-only reviewer per configured model, each applying the same prompt, rubric, and code-quality lens to the same diff. - Lead Judgment Synthesis: Deduplicates findings, maps cross-model agreement, and buckets every item as act on, consider, noted, or dismissed with rationale. - Guardrails: Holds when scope, intent, or reviewer list is missing, never auto-applies patches, and falls back to valid model slugs when a configured slug fails. - Use Case: Before merging a feature branch, run an adversarial review over git diff main...HEAD to surface correctness, security, and maintainability issues that a single reviewer would miss. ## Quick Start Ask the agent to interrogate the current branch diff against main and return a synthesized verdict with act-on, consider, noted, and dismissed findings.
