@levelcode/cli

The official CLI for LevelCode - an open-source AI coding agent that outperforms Claude Code.

Features

• Multi-Agent Architecture: Specialized agents work together for precise codebase editing
• 200+ Model Support: Use any model via OpenRouter (Claude, GPT, DeepSeek, Qwen, etc.)
• Terminal-First: Beautiful, responsive UI with real-time streaming
• Custom Workflows: Create specialized agents with TypeScript

Installation

# Using npm
npm install -g @levelcode/cli

# Using bun (recommended)
bun install -g @levelcode/cli

# Using yarn
yarn global add @levelcode/cli

Quick Start

1. Set your API key:

export OPENROUTER_API_KEY="sk-or-v1-..."

Get your key from OpenRouter.

2. Run LevelCode:

cd your-project
levelcode

3. Start coding with AI:

Just tell LevelCode what you want:

• "Fix the SQL injection vulnerability"
• "Add rate limiting to all API endpoints"
• "Refactor the database connection code"
• "Add unit tests for the auth module"

CLI Options

# Use a specific model
levelcode --model anthropic/claude-3.5-sonnet

# Set working directory
levelcode --cwd /path/to/project

# Non-interactive mode
levelcode "Add error handling to all API calls"

# Show help
levelcode --help

Configuration

Create a levelcode.config.ts in your project:

export default {
  model: 'anthropic/claude-3.5-sonnet',
  ignore: ['node_modules/**', 'dist/**'],
};

Development

Running in Development Mode

bun run dev

Testing

bun test

Interactive E2E Testing

For testing interactive CLI features, install tmux:

# macOS
brew install tmux

# Ubuntu/Debian
sudo apt-get install tmux

# Windows (via WSL)
wsl --install && sudo apt-get install tmux

Run the proof-of-concept:

bun run test:tmux-poc

Documentation

• Full Documentation
• SDK Reference
• Custom Agents Guide
• Configuration Options

License

Apache-2.0 - Created by Yethikrishna R