Multi Agent Review — AI skill for Claude Code
Claude Code reviews PRs made by Codex and other AI agents — like CodeRabbit, but agent-to-agent.
How to install Multi Agent Review
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open TheCraigHewitt/multi-agent-review and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Multi Agent Review does
Claude Code reviews PRs made by Codex and other AI agents — like CodeRabbit, but agent-to-agent.
Alternatives in AI
- Flow Deliver — Multi-AI validation, scoring, and review using Codex and Gemini CLIs (Double Diamond Deliver phase) 2.8k ★
- Flow Next — Plan-first AI workflow plugin for Claude Code, OpenAI Codex, and Factory Droid 569 ★
- Claudexor — Multi-harness control plane for Claude Code, Codex, Cursor, and OpenCode: quota-aware rotation across multiple 425 ★
README
multi-agent-review
Automated PR reviews by Claude Code for commits made by Codex and other AI agents. Like CodeRabbit, but for agent-to-agent code review.
**Flow:** Codex writes code → pushes a PR → Claude Code reviews it → posts a structured GitHub PR review with inline comments.
Quick Start
npm install -g multi-agent-review
cd your-project
mar init
# Reviews happen automatically when Codex finishes a turn.
# Or run manually:
mar review codex/my-feature
Prerequisites
- Node.js >= 20
- Claude Code CLI (
claudecommand) - GitHub CLI (
ghcommand, authenticated) - Codex CLI (optional, for automatic reviews)
How It Works
**`mar init`** sets up your project:
- Creates
.mar/config.jsonwith detected repo and branch defaults - Adds a
notifyhook to~/.codex/config.tomlso Codex triggers reviews automatically - Adds
.mar/to.gitignore
- Creates
**When Codex finishes a turn**, the notify handler checks:
- Is the branch a
codex/*branch? (configurable pattern) - Are there new commits since the last review?
- Is there an open PR for this branch?
- If all yes: spawns a background review
- Is the branch a
**The review** runs `claude -p` with a structured review prompt:
- Reads the cumulative diff against the base branch
- Analyzes for security issues, bugs, performance problems, convention violations
- Returns structured JSON with findings
**Posts to GitHub** as a proper PR review:
- Summary with verdict and collapsible file walkthrough
- Inline comments on specific diff lines with severity badges
- Uses
COMMENTevent (advisory only, never blocks merges) - Updates existing review instead of posting duplicates
GitHub Action
Add Claude reviews to any repo without installing the CLI:
# .github/workflows/review.yml
name: Claude Review
on: pull_request
jobs:
review:
runs-on:
Related Skills
Cerber
AI code-review cockpit: Claude reviews PRs into local artifacts; nothing reaches GitHub until you say so
Pi Plugin Cc
Drive the Pi coding agent from Claude Code or Codex: multi-model review panels, model racing, incremental & sh
Lad MCP Server
Lad MCP Server: Autonomous code & system design review for AI coding agents (Claude Code, Cursor, Codex, etc.)
Multipov All
Run a comprehensive multi-agent codebase review on $ARGUMENTS (default: the entire repo; may also be a subdir
Review Pro
Repository-aware AI code review for coding agents — triage → 13 specialist reviewers → synthesis. Reviews what
Iago
🦜 Greptile-style Mermaid diagrams for AI code reviews — drawn by your own agent. A Claude Code / Codex / Copi
Related Agents
Review Fixer
Applies what CodeRabbit and Sourcery found on a pull request. Use after fetch_review has written .claude/forge
PR Review Validator
Harvest one PR's bot reviews (Copilot, CodeRabbit, Greptile, code scanning), report where each auto-running bo
History Investigator
Answers "why is this code like this?" and "when did this change?" from git history and linked PRs. Use before