Documentation
Amplify is a free, open-source AI coding assistant you can self-host. It gives you full control over your data, runtime, and AI providers. Build with a browser-based sandbox, connect any LLM, and extend withskillsand MCP.
Start using your favourite platform
Explore the documentation
From quick starts to advanced deployment — everything you need to understand, deploy, customize, and contribute to Amplify.
Architecture
Understand how Amplify works from frontend to sandbox.
Quick Start
Install, configure, and make your first AI request in minutes.
Providers
Connect to OpenAI, Anthropic, Google, and more AI providers.
Sandbox
Run code in a browser-based sandbox with WebContainers.
Templates
Start from Expo, Next.js, React, and other project templates.
Built-in Skills
Extend the AI with purpose-built skills for code generation.
MCP
Integrate with Model Context Protocol for tool communication.
Self Hosting
Deploy Amplify on your own infrastructure with Docker.
Configuration
Configure environment variables, providers, and runtimes.
Starter templates
Spin up a new project from one of the 14 built-in starter templates — or let the AI pick the best one for your prompt. Each template is a GitHub repo cloned into the sandbox on first run.
Expo App
Cross-platform mobile apps with Expo.
Basic Astro
Fast static websites with Astro.
NextJS Shadcn
Fullstack Next.js with shadcn/ui.
Vite Shadcn
Fullstack Vite with shadcn/ui.
Qwik Typescript
Resumable apps with Qwik + TypeScript.
Remix Typescript
Fullstack web apps with Remix + TypeScript.
Slidev
Markdown-based developer presentations.
SvelteKit
Fast, efficient web apps with SvelteKit.
Vanilla Vite
Minimal vanilla JavaScript + Vite.
Vite React
React + Vite + TypeScript for fast dev.
Vite Typescript
Type-safe minimal Vite starter.
Vue
Vue.js with modern tooling and best practices.
Angular
Angular with TypeScript and best-practices config.
SolidJS
Lightweight SolidJS + Tailwind for fast sites.
Advanced guides
Go deeper with deployment, migration, API internals, and contribution guides.
Common questions
Quick answers when you are new to Amplify or deciding how to set it up.
What is Amplify?
Amplify is a free, open-source AI coding assistant you can self-host. It runs in your browser with a WebContainer-based sandbox, connects to any LLM provider (OpenAI, Anthropic, Google, and more), and lets you extend the AI with built-in skills and MCP servers. You own your data, your runtime, and your configuration.
Is Amplify really free?
Yes. Amplify is completely free and open source. You self-host the platform on your own infrastructure. You only pay for the AI provider API keys you choose to use (OpenAI, Anthropic, etc.). The project does not own or license WebContainers — the sandbox runtime is provided by StackBlitz's open-source WebContainer API.
How do I get started?
Follow the Quick Start guide to install Amplify, configure your environment variables and API keys, and make your first AI request. It takes just a few minutes to set up.
What AI providers are supported?
Amplify supports OpenAI (GPT-4o, GPT-4o-mini, o1, o3), Anthropic (Claude 3.5 Sonnet, Claude 3 Opus), Google (Gemini Pro), and more. Each provider has different capabilities for streaming, vision, reasoning, and tool calling. See Providers for the full list.
What is the sandbox?
The sandbox is a browser-based execution environment powered by WebContainers. It lets Amplify run Node.js, install packages, execute shell commands, and edit files — all inside the browser without a remote server. See Sandbox for details on how it works, its limitations, and security model.
Can I self-host Amplify?
Yes. Amplify is designed for self-hosting. Use the provided Docker Compose configuration to deploy on your own server with a reverse proxy and HTTPS. You can also replace the default WebContainer sandbox with Docker, Firecracker, or Kubernetes if you need a different execution model.
How do I contribute?
See the Contributing guide for repository structure, coding standards, and how to add new providers, models, templates, MCP servers, and built-in skills. Pull requests are welcome on GitHub.