multi-agent-review-and-fix

Orchestrate multi-agent code review and automated fixes across Codex and Claude.

Updated Aug 23, 2026

Search Tags:#automation #validation #codex #claude #multi-agent #orchestration #code-review

Ready to run

npx skills add https://github.com/yosuke7040/dotfiles --skill multi-agent-review-and-fix

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: multi-agent-review-and-fix
Source: https://github.com/yosuke7040/dotfiles/tree/main/claude/skills/multi-agent-review-and-fix
Command: npx skills add https://github.com/yosuke7040/dotfiles --skill multi-agent-review-and-fix

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Harnesses a four-role iterative workflow to orchestrate autonomous review and semi-automatic fixes across Codex and Claude, reducing manual debugging cycles and improving code quality.

Core Features & Use Cases

Orchestrated review loop: Codex performs the review, Claude triages and fixes, then validates changes within a six-iteration cap.
Directive-driven corrections: Fixed directives guide subsequent iterations to converge on high-quality changes.
Quality assurance: Maintains history and diffs to verify improvements and guard against regressions.

Quick Start

Initiate the multi-agent review loop on the target repository to start iterative code quality improvement.

Frequently Asked Questions about multi-agent-review-and-fix

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

FAQPage Schema

How do I automate code review and fixes across multiple agents?▼

Multi-agent code review automation works by orchestrating Codex to perform the review and Claude to triage and fix issues, running iteratively to progressively enhance code quality while reducing manual debugging cycles.

What is a multi-agent orchestration workflow for code review?▼

A multi-agent orchestration workflow for code review is a four-role iterative process where Codex performs the review, Claude triages and fixes issues, and changes are validated before commit to ensure structured code quality enhancement.

Can I use Codex and Claude together to validate code changes before committing?▼

Yes, Codex and Claude can be used together to validate code changes before commit. The workflow maintains history and diffs to verify improvements and guard against regressions across up to six iterations.

How many iterations does the multi-agent review loop run before stopping?▼

The multi-agent review loop runs across up to six iterations. It uses directive-driven corrections to guide subsequent iterations, converging on high-quality changes while maintaining history to verify improvements.

What is the best way to prevent code regressions during automated fixes?▼

The best way to prevent regressions during automated fixes is to maintain history and diffs for change validation before commit. Directive-driven corrections guide subsequent iterations to converge on high-quality changes safely.

Do I need a specific repository setup to start the iterative code review loop?▼

You need a target repository to initiate the multi-agent review loop. The workflow applies directive governance and change validation, requiring a structured codebase to triage, correct, and validate improvements effectively.