autoengineer-cli 0.1.0

Autonomous Software Engineering Multi-Agent System using AI agents to analyze, implement, test, and review code

AutoEngineer-CLI 🤖

Autonomous Software Engineering Multi-Agent System

A production-grade CLI tool that uses 5 specialized AI agents to autonomously analyze, implement, test, and review code from natural language descriptions.

✨ Features

• 5 Specialized AI Agents: Manager, Architect, Coder, QA, and Reviewer
• Autonomous Code Generation: Creates complete, working implementations from task descriptions
• Sandboxed Execution: Tests code safely in Podman containers
• Comprehensive Code Review: Security analysis, best practices, and performance recommendations
• Free-Tier LLM Support: Works with OpenRouter's free models (262k+ context)

🚀 Installation

Recommended: Using pipx (Isolated Installation)

pipx install autoengineer-cli

Alternative: Using pip

pip install autoengineer-cli

Alternative: Using uv

uv tool install autoengineer-cli

⚙️ Setup

1. Get an OpenRouter API Key (Free):

   • Visit OpenRouter.ai
   • Create an account and get your API key

2. Set the Environment Variable:

   # Linux/macOS
   export OPENROUTER_API_KEY="your-api-key-here"
   
   # Windows PowerShell
   $env:OPENROUTER_API_KEY="your-api-key-here"
   
   # Or create a .env file in your project
   echo "OPENROUTER_API_KEY=your-api-key-here" > .env
   

3. Install Podman (Optional, for sandboxed testing):

   • Podman Installation Guide

📖 Usage

Basic Usage

# Create a new project directory
mkdir my-project && cd my-project

# Run AutoEngineer with a task
autoengineer --repo . --task "Create a REST API with user authentication"

CLI Options

autoengineer --help

Options:
  -r, --repo PATH       Path to the repository to work on (required)
  -t, --task TEXT       Description of the task to complete (required)
  -o, --output PATH     Save the final report to a file
  -v, --verbose         Enable verbose output
  --max-retries INT     Maximum retry attempts (default: 3)

Examples

# Create a simple script
autoengineer -r ./my-project -t "Create a Python script that calculates Fibonacci numbers"

# Build an authentication system
autoengineer -r ./my-project -t "Add JWT-based authentication with login, register, and token refresh"

# Add a feature to existing code
autoengineer -r ./existing-project -t "Add logging to all API endpoints"

# Save the report to a file
autoengineer -r ./my-project -t "Create a CLI calculator" -o report.md

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│ User: "Add authentication API"                                  │
└─────────────────────────────────────────────────────────────────┘
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│ Manager Agent → Parses requirements, creates task breakdown     │
└─────────────────────────────────────────────────────────────────┘
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│ Architect Agent → Analyzes codebase, creates execution plan     │
└─────────────────────────────────────────────────────────────────┘
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│ Coder Agent → Writes actual code files using file tools         │
└─────────────────────────────────────────────────────────────────┘
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│ QA Agent → Tests code in Podman sandbox, reports issues         │
└─────────────────────────────────────────────────────────────────┘
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│ Reviewer Agent → Generates comprehensive code review report     │
└─────────────────────────────────────────────────────────────────┘

🔧 Tech Stack

Component	Technology
AI Orchestration	CrewAI
LLM Gateway	LiteLLM + OpenRouter
Containerization	Podman
CLI Framework	Click
Language	Python 3.10+

📁 Generated Output

AutoEngineer creates:

• Source code files in your repository
• Configuration files (requirements.txt, etc.)
• Documentation (README, API docs)
• Final Report with code review and recommendations

🔒 Security

• Path Traversal Prevention: File tools validate paths to prevent escaping repository
• Network Isolation: Podman containers run with --network=none
• Memory Limits: Containers limited to 512MB by default
• Execution Timeouts: 60-second limit on sandboxed code execution

🤝 Contributing

Contributions are welcome! Please read our Contributing Guide first.

📄 License

MIT License - see LICENSE for details.

🙏 Acknowledgments

• CrewAI - Multi-agent orchestration framework
• OpenRouter - LLM API gateway
• LiteLLM - Universal LLM interface
• Podman - Rootless containerization