xreview

Delegate code review to Codex with a three-party verification workflow.

12|Updated Mar 10, 2026
One-click install
npx skills add https://github.com/davidleitw/xreview --skill xreview
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: xreview
Source: https://github.com/davidleitw/xreview/tree/main/.agents/skills/xreview
Command: npx skills add https://github.com/davidleitw/xreview --skill xreview

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates code review by delegating to Codex while you verify findings.

Core Features & Use Cases

  • Enables a three-party review loop (Codex reviews, you verify, user decides) to improve review quality.
  • Works across languages such as cpp and go, and coordinates with the xreview CLI for consistent workflows.
  • Produces a verifiable Fix Plan and maintains an audit trail so you can decide with confidence.

Quick Start

Ask your agent to review the code with xreview and proceed to verification before applying fixes.

Frequently Asked Questions about xreview

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

FAQPage Schema
How do I automate code review with Codex while verifying the findings?

Automated code review with Codex uses a three-party workflow where Codex provides findings, your agent verifies them, and you decide on fixes. Ask your agent to review the code with xreview and proceed to verification before applying fixes.

What is a three-party AI code review workflow?

A three-party AI code review loop assigns distinct roles to improve quality: an AI agent reviews the code, a secondary verifier validates the findings, and the user decides on fixes. This produces a verifiable Fix Plan and maintains an audit trail.

Does Codex-based code review work with C++ and Go projects?

Yes, Codex-based code review supports multiple programming languages including C++ and Go. It coordinates with the xreview CLI to maintain consistent review workflows across these different language environments.

How do I generate a verifiable Fix Plan for code review findings?

You generate a verifiable Fix Plan by running a structured three-phase review process where Codex identifies issues and your agent verifies them. This plan maintains an audit trail so you can decide on fixes with confidence.

Can I use this agent workflow without installing external dependencies?

Yes, the agent workflow operates without external dependencies. It implements a frontmatter-driven entry with structured processing and optional scripts, references, and assets as directed by SKILL.md.

What is the best way to delegate code review to an AI agent?

The best way to delegate code review is using a three-party workflow where the AI agent provides findings, a secondary process verifies them, and the user decides on fixes. This ensures review quality through an audited verification loop.