co-review

Compare Claude and Codex findings on the same code diff and merge them.

Updated Jun 13, 2026
One-click install
npx skills add https://github.com/taloncjones/dotfiles --skill co-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: co-review
Source: https://github.com/taloncjones/dotfiles/tree/main/claude/skills/co-review
Command: npx skills add https://github.com/taloncjones/dotfiles --skill co-review

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires git, github-cli, codex, and includes scripts (resource) components.

What problem does it solve?

This Skill enhances code reviews by concurrently running code reviews using Claude and Codex, capturing findings from both models to produce a more thorough report.

Core Features & Use Cases

  • Parallel Code Reviews: Reviews the same code using Claude and Codex, ensuring no finding is missed.
  • Unified Reporting: Consolidates and deduplicates findings from both Claude and Codex into one comprehensive report.
  • PR Integration: Works directly with GitHub Pull Requests for a seamless integration with the existing code review workflow.
  • Use Case: Utilize this Skill on a Pull Request in your GitHub repository to automatically conduct a review, ensuring no detail is overlooked before merge.

Quick Start

Review the pull request at https://github.com/example/repo/issue/123.

Frequently Asked Questions about co-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a parallel code review using Claude and Codex on a GitHub PR?

To run a parallel code review, trigger the Skill on a GitHub Pull Request to concurrently capture findings from Claude and Codex. It merges the outcomes with deduplication and source tagging into one comprehensive report.

Do I need to install GitHub CLI and Codex to automate code reviews?

Yes, you need GitHub CLI and Codex installed to automate code reviews. The Skill requires concurrent access to both Claude and Codex, along with git, to fetch diffs and compare findings directly within your repository context.

What is the best way to compare Claude and Codex code review findings without duplicates?

Comparing Claude and Codex code review findings is best handled by consolidating and deduplicating their outputs. This Skill merges findings from both models with source tagging, ensuring no detail is missed while eliminating redundant alerts.

Can I use parallel AI code reviews for any GitHub repository diff?

Yes, you can use parallel AI code reviews for any GitHub repository diff. The Skill works directly with GitHub Pull Requests for versioned comparisons, requiring user-specific code access to ensure accurate, context-aware reviews before merge.

How do I consolidate AI code review results from multiple models?

Consolidating AI code review results involves merging findings from Claude and Codex into a single unified report. This Skill deduplicates overlapping issues and tags the source model, producing a thorough, unified review summary.

What are the limitations of using multiple AI models for GitHub PR reviews?

Limitations of using multiple AI models for GitHub PR reviews include the necessity of concurrent access to both Claude and Codex. The Skill also depends on a valid GitHub Pull Request context to perform accurate versioned comparisons.