---

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

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

Copy# 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.

Quick Start

1. Create Request File

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

Copy# 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.

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

3. Run Review (Custom Provider/rounds)

Copy# 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).

Copycode_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).

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

find

Find past review requests.

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

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.