codex-review-loop

Automate adversarial code review of pull requests with Codex CLI.

Updated Mar 8, 2026
One-click install
npx skills add https://github.com/deicer/codex-lb --skill codex-review-loop
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex-review-loop
Source: https://github.com/deicer/codex-lb/tree/main/.agents/skills/codex-review-loop
Command: npx skills add https://github.com/deicer/codex-lb --skill codex-review-loop

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill automates the process of code review for pull requests, identifying potential issues, and even suggesting and applying fixes, significantly speeding up the development cycle.

Core Features & Use Cases

  • Adversarial Code Review: Leverages AI (Codex CLI) to perform in-depth code analysis.
  • Structured Findings: Parses AI output into actionable, categorized findings.
  • Automated Fix Loop: Attempts to automatically fix identified issues and verifies them.
  • Iterative Re-review: Re-runs the review process to catch regressions introduced by fixes.
  • Use Case: Integrate this into your CI/CD pipeline to automatically review code changes, flag critical issues, and even auto-correct common problems before human review, ensuring higher code quality and faster merges.

Quick Start

Use the codex-review-loop skill to review the current pull request.

Frequently Asked Questions about codex-review-loop

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

FAQPage Schema
How do I automate AI code review for pull requests?

You can automate AI code review for pull requests by using an adversarial loop that parses AI findings into structured data and executes an atomic fix-verify-commit cycle. This automatically identifies issues and applies verified corrections.

How does an automated code review fix loop handle regressions?

An automated code review fix loop handles regressions by running iterative re-reviews after applying fixes. It verifies the changes and repeats the review process to ensure no new issues were introduced during the automated refactoring.

Can I use AI to review uncommitted changes and specific branches?

Yes, you can use AI to review uncommitted changes and specific branches. The review loop includes built-in scope resolution, allowing it to analyze targeted branches, uncommitted local changes, or entire pull requests.

What is the best way to auto-fix code issues found during a PR review?

The best way to auto-fix code issues found during a PR review is an atomic fix-verify-commit loop. It takes categorized findings, attempts automated fixes, verifies them, and commits the changes atomically to ensure codebase stability.

Does environment configuration affect automated AI code review models?

Yes, environment configuration affects automated AI code review models. The review loop supports environment overrides, allowing developers to specify which review models to use and adjust the reasoning effort applied during code analysis.

What are the limitations of automating code review with AI?

Limitations of automating code review with AI include potential misinterpretation of complex logic and the need for iterative re-reviews. The automated fix loop requires built-in error handling to manage edge cases and prevent incorrect refactoring.