codex-review

Automate Codex-based code diff reviews with structured JSON summaries.

13|Updated May 19, 2024
One-click install
npx skills add https://github.com/david-crespo/dotfiles --skill codex-review-david-crespo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex-review
Source: https://github.com/david-crespo/dotfiles/tree/main/claude/skills/codex-review
Command: npx skills add https://github.com/david-crespo/dotfiles --skill codex-review-david-crespo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a Codex-based second-opinion workflow for evaluating code changes by sending diffs to Codex, interpreting its feedback, and synthesizing actionable results for the user.

Core Features & Use Cases

  • Codex diff review: automatically routes your diffs to Codex for a thorough review, highlighting potential issues and suggested improvements.
  • Multi-round dialogue: supports back-and-forth discussions with Codex to clarify feedback and refine recommendations.
  • Synthesis & handoff: presents a concise, actionable synthesis that you can act on in your pull requests or code reviews.

Quick Start

Provide the current code changes to Claude Codex for review using the codex-call.ts script to receive a structured feedback synthesis

Frequently Asked Questions about codex-review

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

FAQPage Schema
How do I automate a Codex code review for my git diffs?

Automated Codex code review routes your generated diffs to Codex for evaluation, interprets the feedback, and synthesizes actionable results for iterative improvements.

Can I have a multi-round dialogue with Codex to clarify code review feedback?

Yes, multi-round dialogue supports session-based back-and-forth with Codex to clarify feedback and refine recommendations across deterministic workflow rounds.

What is a cost-aware JSON summary in an AI-assisted code review?

A cost-aware JSON summary is the structured output format generated after obtaining a Codex second opinion, synthesizing contextual feedback into actionable results.

Do I need a read-only Codex integration to review code changes?

Yes, the workflow requires a read-only Codex integration wrapper to evaluate diffs, interpret feedback, and support resumable session-based iterations.

How do I resume a code review session across multiple rounds of feedback?

You resume a code review session using the deterministic workflow that supports back-and-forth dialogue, allowing iterative improvements to be continued across rounds.

What's the best way to get a second opinion on code changes before a pull request?

The best way to get a second opinion is routing diffs to Codex for review, synthesizing its feedback into a concise, actionable handoff for your pull requests.