Spawn.

Spawn your own websites with just a prompt.

About the Project

Spawn is an AI-native cloud development environment that turns natural language prompts into full-stack React applications in seconds. Unlike standard chatbots, Spawn runs a dedicated Firecracker microVM (via E2B) for every project, allowing the AI to write code, install dependencies, and run a dev server in real-time.

It features a robust ReAct Agent architecture capable of researching the web for design inspiration, planning component structures, and iteratively building production-ready UIs using Tailwind v4 and Shadcn. Users can preview their apps instantly via a live iframe, edit code manually in the browser, and download the source code as a ZIP file.

spawn-video-1766085104862.mp4

System Architecture

The following flowchart demonstrates how data flows between the User, the AI Agent, the Database, and the Sandboxed Environment.

flowchart TD
    User["User / Frontend"] -->|1. Prompt| API["Backend API Express"]
    
    subgraph AI_Orchestration ["AI Orchestration Layer"]
        API -->|2. Stream Request| Agent["LangGraph Agent"]
        Agent <-->|3. Loop| Model["LLM OpenRouter"]
        Agent -->|4. Tool Call| Tools["Tool Manager"]
    end
    
    subgraph Data_Context ["Data & Context"]
        Tools -->|Read/Write Files| DB[("(PostgreSQL)")]
        Tools -->|Web Search| Exa["Exa.ai API"]
    end
    
    subgraph Infrastructure ["Infrastructure Layer"]
        Tools -->|5. Sync Files & Run| E2B["E2B Sandbox MicroVM"]
        E2B -->|6. Host Port 5173| Preview["Live Preview URL"]
    end
    
    API -.->|7. Stream Tokens SSE| User
    Preview -.->|8. Iframe Render| User

Loading

Technologies Used

• Frontend: Next.js 14 , Tailwind CSS v4, Framer Motion, Shadcn UI, Monaco Editor
• Backend: Node.js, Express.js
• AI & Orchestration: LangGraph, OpenRouter
• Infrastructure: E2B (Sandboxed Firecracker VMs), Docker
• Database: PostgreSQL, Drizzle ORM
• Tools: Exa.ai, BetterAuth, Turborepo

Getting Started

To start Spawn locally, use these following steps:

Clone the repository

git clone https://github.com/Spandan-Mishra/spawn.git

cd spawn

Installation

Since this is a Turborepo, install dependencies at the root:

pnpm install

Configuration

Create .env files in apps/web, apps/api and packages/db (or a global one if configured). Use the example below:

# Database (Local Docker or Neon)
DATABASE_URL="postgresql://postgres:password@localhost:5433/spawn"

# AI Providers
OPENROUTER_API_KEY=sk-or-v1-...
EXA_API_KEY=...
E2B_API_KEY=...

# Authentication (BetterAuth)
BETTER_AUTH_SECRET=...
BETTER_AUTH_URL=http://localhost:3000
GITHUB_ID=...
GITHUB_SECRET=...
GOOGLE_CLIENT_ID=...
GOOGLE_CLIENT_SECRET=...

# API Connectivity
NEXT_PUBLIC_BACKEND_URL=http://localhost:3001
PORT=3001

Infrastructure Setup

1. Start the Database: Spawn uses a local Docker container for PostgreSQL.

   docker-compose up -d

2. Push Database Schema: Sync the Drizzle schema with your local database.

   cd packages/db
   npx drizzle-kit push

Starting the project

Go to the root directory and run the development command. This will start the Next.js frontend (Port 3000) and the Express backend (Port 3001) concurrently.

pnpm run dev

Visit http://localhost:3000 to start your local instance of Spawn.