codex-review

Perform two-pass adversarial reviews of design and implementation plans using OpenAI Codex CLI.

Updated Feb 13, 2026
One-click install
npx skills add https://github.com/kroepke/codex-plan-reviewer --skill codex-review-kroepke
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex-review
Source: https://github.com/kroepke/codex-plan-reviewer/tree/main/skills/codex-review
Command: npx skills add https://github.com/kroepke/codex-plan-reviewer --skill codex-review-kroepke

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill automates a rigorous, two-pass adversarial review process for design documents and implementation plans, identifying potential flaws and inconsistencies before execution.

Core Features & Use Cases

  • Two-Pass Review: Conducts both section-by-section and holistic document reviews.
  • Automated Critique: Leverages OpenAI Codex CLI to provide detailed feedback.
  • Iterative Improvement: Supports focused, multi-round reviews of specific sections.
  • Use Case: Before committing to a complex software architecture, use this Skill to have Codex stress-test the design document for hidden flaws, contradictions, and unaddressed failure modes.

Quick Start

Use the codex-review skill to perform a two-pass review on the document located at '/path/to/my/design_doc.md'.

Frequently Asked Questions about codex-review

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

FAQPage Schema
How do I automate adversarial code review for design documents?

You can automate adversarial code review by running a two-pass critique process that analyzes design documents section-by-section and holistically for cross-cutting concerns. This identifies hidden flaws and inconsistencies before implementation begins.

What is a two-pass review process for architecture plans?

A two-pass review first analyzes architecture plans section-by-section for local issues, then evaluates the document holistically for cross-cutting concerns. This iterative critique is fed back into the document for targeted revisions.

Does the adversarial review process require the OpenAI Codex CLI?

Yes, the adversarial review process requires the OpenAI Codex CLI, Python 3.10 or higher, and authenticated OpenAI API access to automate the critique and stress-test design documents.

How do I stress-test an implementation plan for hidden flaws before execution?

To stress-test an implementation plan, run it through a two-pass adversarial review using OpenAI Codex. The process flags unaddressed failure modes and contradictions, allowing you to revise specific sections iteratively.

Can I target a specific section for iterative critique during document review?

Yes, the review process supports focused, multi-round iterative critique of specific document sections. This allows targeted refinement of individual architecture or design plan components without reviewing the entire document.