pr-review

Automate dual-AI PR reviews with Codex and Claude Sonnet.

11|Updated Dec 18, 2025
One-click install
npx skills add https://github.com/brolag/neural-claude-code --skill pr-review-brolag
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pr-review
Source: https://github.com/brolag/neural-claude-code/tree/main/skills/pr-review
Command: npx skills add https://github.com/brolag/neural-claude-code --skill pr-review-brolag

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates dual-AI PR reviews by combining Codex (logic/edge cases) and Claude Sonnet (security/architecture/slop) to provide a comprehensive early feedback before human approval.

Core Features & Use Cases

  • Dual-AI PR review: Codex analyzes logic and edge cases while Claude Sonnet assesses security, architecture, and AI slop, with a consolidated PR comment.
  • Auto-detection and targeted review: automatically detects the repository from the current git context or accepts an explicit repo, enabling use across multiple projects.
  • Structured, actionable output: posts a formatted comment on the PR with dedicated sections for Codex, Claude Sonnet, and AI Slop, plus error guidance.

Quick Start

Run /pr-review <PR-number> from any directory containing the project to perform a dual-AI PR review and post the results as a PR comment.

Frequently Asked Questions about pr-review

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

FAQPage Schema
How do I automate dual-AI PR reviews for GitHub projects?

Automate dual-AI PR reviews by combining Codex and Claude Sonnet to surface logic, security, and architecture issues before human approval. The Skill posts a structured, consolidated comment on the GitHub PR with dedicated sections for each AI's feedback.

How does combining Codex and Claude Sonnet improve code review quality?

Combining Codex and Claude Sonnet improves code review by splitting responsibilities: Codex analyzes logic and edge cases while Claude Sonnet assesses security, architecture, and AI slop. This dual-AI mechanism provides comprehensive, targeted feedback before merge.

Can I run automated GitHub PR reviews locally with repository autodetection?

Yes, you can run automated GitHub PR reviews locally. The Skill auto-detects the repository from your current git context, or accepts an explicit repo parameter, enabling use across multiple local projects without manual configuration.

What is the best way to check security and architecture concerns before a PR merge?

The best way to check security and architecture concerns before a PR merge is using a dual-AI review tool. Claude Sonnet specifically targets security, architecture, and AI slop detection, while Codex validates logic, delivering structured feedback directly to the PR.

How do I start a PR review from a local project directory?

To start a PR review, run the /pr-review command with your PR number from any directory containing the project. The Skill automatically detects the repository, executes the dual-AI review, and posts the results as a formatted PR comment.

What does the structured output look like for an automated PR comment?

The structured output for an automated PR comment features separate sections for Codex, Claude Sonnet, and AI Slop. It presents consolidated, actionable feedback on logic, security, and design concerns, along with error guidance directly on the GitHub PR.