Does this skill require external tools?

Yes, it requires the Codex CLI to be installed and authenticated on your system to perform the external reviews.

What standards does it apply during the review?

It references a local standards.md file and uses predefined JSON schemas (plan-review and review-result) to ensure all outputs meet established quality criteria.

How does it handle subsequent reviews?

It uses a session management system via a .codex-session-active file to resume reviews, ensuring that previously identified issues have been addressed.

What is the primary purpose of Review Codex?

It acts as a final gatekeeper, using the Codex CLI to independently review plans and code for security, style, and logic issues before a task is marked as complete.

Can it review both plans and implemented code?

Yes, it automatically detects the review type based on the presence of .task/plan-refined.json or .task/impl-result.json files.

Review Codex

Name: Review Codex
Author: cskwork

bycskwork

0•

安全与测试

Automates high-fidelity plan and code reviews using the Codex CLI to ensure security, feasibility, and adherence to standards.

Review Codex serves as an independent quality assurance gate for Claude Code, leveraging the Codex CLI to perform rigorous assessments of development plans and code implementations. By utilizing structured JSON schemas and domain-specific standards, it identifies potential security vulnerabilities, logic flaws, and architectural inconsistencies before final approval. This skill is particularly useful in multi-AI workflows where a secondary, specialized model is required to validate the output of a primary coding agent, ensuring production-grade reliability and compliance with project-specific guidelines.

主要功能

01Automated plan and implementation validation

02Security and feasibility auditing against custom standards

03Schema-enforced JSON review outputs for consistency

040 GitHub stars

05Independent assessment via specialized Codex CLI

06Context-aware incremental review session tracking

使用场景

01Final approval gate for complex codebase refactoring

02Automated security auditing of new feature implementations

03Verifying architectural plan feasibility before execution

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add cskwork/llm-review review-codex

For use in Claude.ai and ChatGPT