cross-review-pr

What problem does it solve?

This Skill reduces missed issues and one-model bias by running a pull request through two independent reviewers, then checking whether Reviewer B's findings hold up under Reviewer A's validation. It is designed for change reviews where confidence matters more than speed, especially when the scope spans multiple files or risks regressions.

Core Features & Use Cases

• Independent dual review: Runs two separate LLM reviews over the same PR, branch, commit, or codebase scope without sharing findings during the first pass.
• One-way validation: Has Reviewer A check Reviewer B's findings only, so the final report distinguishes overlap, independent-only issues, and challenged items.
• Deep review mode: Breaks large or critical changes into focused areas and reviews each area with symmetric reviewer coverage.
• Typical use cases: Use it for high-stakes pull requests, architecture-sensitive refactors, security-sensitive changes, or any review where a second model perspective is valuable.

Quick Start

Ask the agent to run a cross-model review of PR 47 using Claude and Codex.