Appwit

An AI-powered code generation platform built as a Turborepo monorepo. Users describe what they want to build, and the system generates production-ready code files in an isolated sandbox environment.

Table of Contents

• Overview
• Architecture
• Tech Stack
• Project Structure
• Getting Started
• Development
• API Reference
• License

---

Overview

Appwit enables users to generate complete web applications from natural language prompts. The platform leverages:

• LangGraph State Machine for orchestrating multi-step AI workflows
• E2B Sandboxes for secure, isolated code execution and file management
• LLM Integration supporting Anthropic Claude and OpenAI models
• Real-time Streaming for progressive file generation feedback

---

Architecture

System Architecture

graph TB
    subgraph Client["Frontend (Next.js)"]
        UI[Web Interface]
        Editor[Monaco Editor]
        Preview[Sandbox Preview]
    end

    subgraph Server["Backend (Express/Bun)"]
        API[REST API]
        Auth[Auth Middleware]
        Orchestrator[LangGraph Orchestrator]
    end

    subgraph Agent["AI Agent Pipeline"]
        Planner[Planner Node]
        Coder[Coder Node]
        Theme[Theme Applicator]
        Cmd[Command Handler]
        Writer[Writer Node]
        Reviewer[Reviewer Node]
    end

    subgraph External["External Services"]
        LLM[LLM Provider<br/>Claude / OpenAI]
        E2B[E2B Sandbox]
        DB[(PostgreSQL)]
        S3[S3 Storage]
    end

    UI --> API
    API --> Auth
    Auth --> Orchestrator
    Orchestrator --> Agent
    Agent --> LLM
    Agent --> E2B
    API --> DB
    API --> S3
    Preview --> E2B

Loading

UML Sequence Diagram: Code Generation Flow

sequenceDiagram
    actor User
    participant Frontend
    participant API as Backend API
    participant Orchestrator
    participant LLM as LLM Provider
    participant E2B as E2B Sandbox

    User->>Frontend: Enter prompt
    Frontend->>API: POST /prompt/stream
    API->>E2B: Create sandbox
    E2B-->>API: Sandbox instance
    API->>Orchestrator: Start streaming pipeline

    Note over Orchestrator: Planning Phase
    Orchestrator->>LLM: Analyze prompt
    LLM-->>Orchestrator: Task plan
    Orchestrator-->>Frontend: Stream: plan_complete

    Note over Orchestrator: Code Generation Phase
    loop For each batch
        Orchestrator->>LLM: Generate code
        LLM-->>Orchestrator: File contents
        Orchestrator->>E2B: Write files progressively
        Orchestrator-->>Frontend: Stream: file_created
    end

    Note over Orchestrator: Command Execution
    Orchestrator->>E2B: Run npm commands
    E2B-->>Orchestrator: Command output
    Orchestrator-->>Frontend: Stream: executing

    Note over Orchestrator: Review Phase
    Orchestrator->>E2B: List files
    E2B-->>Orchestrator: File list
    Orchestrator-->>Frontend: Stream: review_complete

    alt Issues Found AND retryCount < 1
        Orchestrator->>LLM: Regenerate missing files
    end

    Orchestrator-->>Frontend: Stream: done
    Frontend->>E2B: Connect to preview URL
    E2B-->>Frontend: Live preview
    Frontend-->>User: Display result

Loading

---

Project Structure

appwit/
├── apps/
│   ├── backend/                 # Express API server
│   │   ├── agent/               # LangGraph orchestrator
│   │   │   ├── nodes/           # Pipeline nodes
│   │   │   │   ├── planner.ts   # Task planning
│   │   │   │   ├── coder.ts     # Code generation
│   │   │   │   ├── writer.ts    # File writing
│   │   │   │   └── reviewer.ts  # Validation
│   │   │   ├── orchestrator.ts  # State machine
│   │   │   └── systemPrompts/   # LLM prompts
│   │   ├── routes/              # API endpoints
│   │   ├── middleware/          # Auth middleware
│   │   ├── prisma/              # Database schema
│   │   └── lib/                 # Utilities
│   └── frontend/                # Next.js application
│       ├── app/                 # App router pages
│       ├── components/          # React components
│       ├── hooks/               # Custom hooks
│       └── lib/                 # Utilities
├── docker/                      # Docker configurations
├── ops/                         # Deployment configs
├── turbo.json                   # Turborepo config
└── package.json                 # Root workspace

---

Getting Started

Prerequisites

• Bun v1.1.38 or higher
• PostgreSQL database
• E2B API key
• LLM provider API key (Anthropic or OpenAI)

Installation

1. Clone the repository:

git clone https://github.com/your-org/appwit.git
cd appwit

2. Install dependencies:

bun install

3. Configure environment variables:

cp .env.example .env
cp apps/backend/.env.example apps/backend/.env
cp apps/frontend/.env.example apps/frontend/.env

4. Set up the database:

cd apps/backend
bunx prisma generate
bunx prisma db push

5. Start development servers:

bun run dev

---

Development

Commands

Command	Description
bun run dev	Start all apps in development mode
bun run build	Build all apps for production
bun run lint	Run linting across all workspaces
bun run clean	Remove build artifacts
bun run format	Format code with Prettier

Run Specific Workspace

# Backend only
bun run dev --filter=@appwit/backend

# Frontend only
bun run dev --filter=@appwit/frontend

Turborepo Features

• Build Caching: Caches build outputs to speed up subsequent builds
• Task Orchestration: Runs tasks in the correct order based on dependencies
• Parallel Execution: Runs independent tasks in parallel

---

API Reference

Authentication

Endpoint	Method	Description
/auth/signup	POST	Register new user
/auth/signin	POST	Login with credentials
/auth/google	POST	Google OAuth login

Projects

Endpoint	Method	Description
/project	GET	List user projects
/project	POST	Create project
/project/:id	GET	Get project details
/project/:id	DELETE	Delete project

Code Generation

Endpoint	Method	Description
/prompt/stream	POST	Stream code generation
/prompt/files/:sandboxId	GET	List sandbox files
/prompt/file/:sandboxId	GET	Read file content

Health Check

Endpoint	Method	Description
/health	GET	Service health status

---

Environment Variables

Backend

Variable	Description
PORT	Server port
DATABASE_URL	PostgreSQL connection string
JWT_SECRET	JWT signing secret
E2B_API_KEY	E2B sandbox API key
ANTHROPIC_API_KEY	Anthropic API key
OPENAI_API_KEY	OpenAI API key
AWS_ACCESS_KEY_ID	AWS credentials
AWS_SECRET_ACCESS_KEY	AWS credentials
AWS_S3_BUCKET	S3 bucket name

Frontend

Variable	Description
NEXT_PUBLIC_API_URL	Backend API URL

---

License

This project is private and proprietary.