name: code-review
description: >
Submit code review requests to multiple AI providers (GitHub Copilot, Anthropic Claude,
OpenAI Codex, Google Gemini) and get patches back. Use when user says "code review",
"review this code", "get a patch for", or needs AI-generated unified diffs for code fixes.
allowed-tools: Bash, Read
triggers:
- code review
- review this code
- review my changes
- review these changes
- get a patch
- generate a patch
- generate diff
- copilot review
- codex review
- claude review
- review request
- full review
- code review loop
- run a code review
- request code review
- use codex to review
- use claude to review
- opus vs codex
- coder reviewer loop
- 3 round review
- multi-round review
- assess and review
- review based on changes
- review with gpt-5
- review with codex
metadata:
short-description: Multi-provider AI code review CLI

Code Review Skill

Submit structured code review requests to multiple AI providers and get unified diffs back.

Supported Providers & Models

⚠️ COST WARNING: Only use github provider to avoid API charges. The anthropic, openai, and google providers make direct API calls that cost money.

✅ RECOMMENDED: Use --provider github --model claude-sonnet-4.5 for Claude models at no additional cost beyond your GitHub Copilot subscription.

Context Bridging: For providers that don't support session persistence (OpenAI, Gemini), the skill automatically injects previous round outputs into the next prompt to enable multi-round iteration.

Prerequisites

# Check provider availability
python .pi/skills/code-review/code_review.py check

Agent Actions (How to use)

Use the table below to map user requests to the correct command.

💡 COST-SAVING TIP: Always use --provider github for Claude models to avoid API charges. The github provider includes Claude models at no additional cost beyond your GitHub Copilot subscription.

Quick Start

1. Create Request File

First, creating a request file is recommended to define the scope.

# Auto-generate request context from git status
python .pi/skills/code-review/code_review.py build -A -t "Fix crash in Auth" -o request.md

2. Run Review (Standard)

Run the full 3-step pipeline (Generate -> Judge -> Finalize).
Default: Uses GitHub Copilot (gpt-5) with 2 rounds.

python .pi/skills/code-review/code_review.py review-full --file request.md

3. Run Review (Custom Provider/rounds)

# Example: 4 rounds using OpenAI Codex
python .pi/skills/code-review/code_review.py review-full \
  --file request.md \
  --provider openai \
  --model gpt-5.2-codex \
  --rounds 4

Commands

review-full (Recommended)

Run the iterative review pipeline.

• Supports session continuity for all providers (native or bridged).
• Generates a final unified diff.

loop (Coder vs Reviewer)

Advanced: Run a feedback loop between two different agents (e.g., Anthropic Coder vs OpenAI Reviewer).

code_review.py loop \
  --coder-provider anthropic --coder-model opus-4.5 \
  --reviewer-provider openai --reviewer-model gpt-5.2-codex \
  --rounds 5 --file request.md

bundle

Bundle request for copy/paste into GitHub Copilot web (if CLI is unavailable).

code_review.py bundle --file request.md --clipboard

find

Find past review requests.

code_review.py find --dir . --pattern "*.md"

Cost Comparison

Best Practice: Always use --provider github to access Claude models (like claude-sonnet-4.5) at no additional cost.

Project Agent Workflow

1. Interpret User Request: e.g., "Fix the bug in auth"
2. Build Request: code_review.py build -A -t "Fix Auth Bug" -o request.md
3. Execute Review: code_review.py review-full --file request.md
4. Apply Patch: Parse output and apply valid diffs.