Documentation

Getting Started

Lume is a voice-native AI coding environment that generates working code from natural language. Install and start building in under 60 seconds.

Installation

Install Lume globally with npm:

npm install -g lume-cli

Verify the installation:

lume --version

Your First Project

Launch Lume and describe what you want to build:

# Launch Lume
lume

# Start with a prompt
> "Create a React app with a dashboard and dark theme"

# Or use voice mode
lume --voice

This will:

• Generate a complete project structure
• Create all components and pages
• Set up styling and configuration
• Start a live preview server

Voice Commands

Use voice mode to describe changes naturally. Lume listens, interprets, and generates code in real time.

# Start voice mode
lume --voice

# Example voice commands:
# "Add a sidebar navigation"
# "Make the header sticky"
# "Create an API route for user authentication"
# "Fix the layout on mobile"

Desktop App

Lume Desktop gives subscribers a standalone macOS app for running multi-terminal AI workspaces directly against local repos.

• Included with every active or trialing Lume subscription
• Supports local terminals, file editing, and direct project access
• Download the latest `.dmg` from the desktop download page

# Recommended desktop prerequisites
macOS 13+
gh auth login
gh extension install github/gh-copilot # if needed for your local setup

Configuration

Customize Lume with your preferred AI model and output settings.

# Choose AI model
lume --model claude

# Set output directory
lume --output ./my-project

# Available models: gpt-4o, claude, gemini
lume config models

Workspace Management

Organize your projects into workspaces. Each workspace maintains its own context, history, and configuration.

# List workspaces
lume workspaces

# Create new workspace
lume new "my-project"

# Switch workspace
lume switch "my-project"