ReviewMyWork - AI Code Review Agent

An intelligent code review agent that orchestrates LLM tool calls to provide comprehensive analysis of your git changes.

Installation

# Install globally with pip
pip install reviewmywork

# Install globally with uv
uv tool install reviewmywork

Key Features

• LLM Agent Architecture: Orchestrates multiple tool calls for thorough analysis
• Rich Terminal UI: Structured output with confidence scores, detailed issues, and actionable suggestions
• Multi-Provider Support: Works with any aisuite-compatible LLM (Anthropic, OpenAI, Azure, Ollama, etc.)
• Minimal & Fast: Focused on core functionality

How It Works

ReviewMyWork parses your git diff, builds some context, and then allows the LLM to intelligently use tools like read_file, search_content, and git_history through multi-turn conversations to understand context and provide detailed reviews with confidence scoring.

Configuration

Set your LLM provider credentials:

# Example for Anthropic
export ANTHROPIC_API_KEY=your-key-here

# Optional: Configure timeouts
export REVIEWMYWORK_TOOL_TIMEOUT=120
export REVIEWMYWORK_MAX_TURNS=10

Or use a .env file in your project root.

Requirements

• Python 3.10+
• ripgrep (rg command)
• git
• LLM API key for your chosen provider

Usage

# Review changes against main branch
reviewmywork review . --base main --model your-model

# Review specific repository
reviewmywork review --base develop --model openai:gpt-4.1 /path/to/repo 

# Get help
reviewmywork --help

Short alias: Use rmw instead of reviewmywork for faster typing.

Development

Development Setup

# Install dependencies and package
uv sync
uv pip install -e .

# Run a review
reviewmywork review . --base main --model anthropic:claude-3-5-sonnet

License

This project is licensed under the MIT License - see the LICENSE file for details.