adversarial-multi-model-review

What problem does it solve?

This Skill enables a structured adversarial review by running parallel, cross-checking perspectives across multiple AI models to surface issues a single reviewer might miss.

Core Features & Use Cases

• Four-phase adversarial review process: runs with model switching between phases for maximum adversarial value.
• Independent, cross-perspective analysis: each model evaluates engineering, security, architecture, operations, production, and devil's advocate dimensions.
• Artifact-driven synthesis: stores intermediate findings outside the repository and produces a final consolidated improvement list for broad use in code, plans, designs, or architecture reviews.

Quick Start

Run the skill with the artifact you want reviewed (file path, code block, plan, or design) to initiate Phase 1.