Skip to main content
When you’re ready to deploy your ChatGPT App to production, Skybridge provides two commands: skybridge build to compile your app, and skybridge start to run it in production mode.

​
Build for Production

Before deploying, you need to build your application. Run the build command from your project root: 
npm run build
This runs the skybridge build command, which compiles your widgets and server code for production.

​
Start Production Server

Once your app is built, start the production server: 
npm start
This runs the skybridge start command, which starts your production server.

​
How it works

The production server runs your compiled application from the dist/ directory with NODE_ENV=production. Unlike the development server, it:
  • Serves pre-built assets - Widgets and styles are served as static files from /assets (compiled during skybridge build)
  • MCP endpoint - Exposes your MCP server at /mcp for ChatGPT to connect to
  • No dev tools - DevTools and Vite dev server are excluded in production for better performance
  • Optimized rendering - Uses production templates that load pre-bundled JavaScript and CSS files
The server listens on port 3000 by default. When deployed, ChatGPT connects to your MCP endpoint (e.g., https://your-domain.com/mcp) to access your tools and widgets.

​
What’s Next?

​
Core Concepts

Learn how Skybridge extends the raw APIs with React hooks

Explore →

​
API Reference

Browse the complete API documentation

View API →