skills.rest
infinith4
infinith4COMMUNITY
Explore all skills by this author →
GitHub

codex-review

Automate iterative Codex and Claude Code reviews until acceptance is met.

1|Updated Jan 19, 2026
Search Tags:#automation#codex#iteration#review#claude-code#code-review#pull-request
One-click install
npx skills add https://github.com/infinith4/dev-ai-driven-development --skill codex-review-infinith4
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: codex-review
Source: https://github.com/infinith4/dev-ai-driven-development/tree/main/.claude/skills/codex-review
Command: npx skills add https://github.com/infinith4/dev-ai-driven-development --skill codex-review-infinith4

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Codex reviews code changes in read-only mode, Claude Code applies targeted fixes, and the system loops back for re-review until the acceptance criterion (ok: true) is achieved, ensuring a robust gate before merging.

Core Features & Use Cases

  • Iterative review loop: Codex reviews changes in read-only mode and Claude Code applies fixes, then rechecks until all issues are resolved.
  • Scale-aware workflow: Applies different strategies for small, medium, and large changes, coordinating multiple agents and cross-checks when needed.
  • Pre-PR validation: Used after spec/PRD/requirements updates and before merging or releasing to verify compliance with requirements and design.

Quick Start

Submit your diff to start the iterative Codex–Claude review loop until ok: true is reached.

Frequently Asked Questions about codex-review

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

FAQPage Schema
How do I automate an iterative code review loop before a pull request?▼

An iterative code review loop automates validation by having Codex review diffs in read-only mode and Claude Code apply fixes until all issues are resolved. This converges on a verified implementation before commit or release.

Can I use Claude Code to automatically fix issues found by Codex during a code review?▼

Yes, Claude Code can automatically apply targeted fixes to code changes that Codex identifies during its read-only review. The system loops back for re-review until the acceptance criterion of ok: true is achieved.

How does the iterative review workflow handle large multi-file rewrites?▼

For large multi-file rewrites, the iterative review workflow applies a scale-aware strategy that coordinates multiple agents and cross-checks. This ensures comprehensive validation for major changes like new modules or infra updates.

What is the acceptance criterion for an automated Codex and Claude Code review cycle?▼

The acceptance criterion for an automated Codex and Claude Code review cycle is a JSON-based output schema returning ok: true. This schema provides easy automation for next-review tasks and clear traceability of all recommendations.

When should I use an iterative AI code review workflow instead of a manual check?▼

You should use an iterative AI code review workflow for major changes such as new modules, public APIs, infra/config updates, or multi-file rewrites. It serves as a robust validation gate after spec updates and before merging.

Does the Codex review and Claude Code fix loop work without external dependencies?▼

Yes, the Codex review and Claude Code fix loop operates without external dependencies, using a defined JSON-based output schema for automation. It relies solely on the read-only review capabilities of Codex and the fixing capabilities of Claude Code.

Related Skills

Explore More Software Engineering Skills→
rintirinti

code-review

Evaluate uncommitted git diffs with Codex CLI review feedback.

Community
Advanced
yusaduymazyusaduymaz

gsd-ship

Push verified branches and open pull requests for merge tracking.

Community
Advanced
JubaKitiashviliJubaKitiashvili
41

quality-gate

Run parallel code-reviewer and performance-profiler checks for React Native merges.

Community
Intermediate
nnexainnexai
1

gsd-plan-review-convergence

Orchestrate iterative plan-review-replan loops with multiple AI CLI reviewers.

Community
Advanced