# letta-code

**Repository Path**: veuns/letta-code

## Basic Information

- **Project Name**: letta-code
- **Description**: Letta Code
面向开发者与团队的 AI 代码生成工作台

Letta Code 是一个类 bolt.new / v0.dev 的 AI 代码生成平台，支持通过自然语言对话快速生成、编辑、运行和管理完整代码项目。它集成了 AI 对话、代码编辑器、文件管理、实时预览、终端运行、模型配置和管理后台，让用户可以在一个工作台中完成从想法到可运行项目的完整流程。


- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 4
- **Forks**: 2
- **Created**: 2026-05-18
- **Last Updated**: 2026-05-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Letta Code

A bolt.new / v0.dev style AI code generation platform powered by OpenAI-compatible APIs.

## Features

### User Application (Port 3000)

- Chat-based code generation using AI
- Real-time streaming responses
- Monaco code editor with syntax highlighting
- File tree navigation
- Live preview with WebContainer (basic HTML preview available)
- Multi-model support (OpenAI, Anthropic, DeepSeek, local models, etc.)

### Admin Dashboard (Port 3001)

- API Key management for multiple providers
- Custom base URL support for any OpenAI-compatible API
- Model configuration (enable/disable models)
- Usage analytics and statistics
- System settings

## Tech Stack

- **Framework**: Next.js 15 (App Router)
- **Styling**: Tailwind CSS + shadcn/ui
- **AI Backend**: OpenAI-compatible APIs
- **Code Editor**: Monaco Editor
- **Preview**: WebContainer API
- **State Management**: Zustand
- **Database**: SQLite with Drizzle ORM for users/projects/messages
- **File Store**: Shared JSON-backed store for admin-managed API keys and model configuration
- **Monorepo**: pnpm + Turborepo

## Supported Providers

The application supports any OpenAI-compatible API:

| Provider     | Base URL                                 | Notes                         |
| ------------ | ---------------------------------------- | ----------------------------- |
| OpenAI       | `https://api.openai.com/v1`              | Default                       |
| Azure OpenAI | `https://your-resource.openai.azure.com` | Set custom base URL           |
| Anthropic    | Via proxy                                | Use OpenAI-to-Anthropic proxy |
| DeepSeek     | `https://api.deepseek.com/v1`            | Set custom base URL           |
| Moonshot     | `https://api.moonshot.cn/v1`             | Kimi AI                       |
| Zhipu AI     | `https://open.bigmodel.cn/api/paas/v4`   | GLM models                    |
| Ollama       | `http://localhost:11434/v1`              | Local models                  |
| LM Studio    | `http://localhost:1234/v1`               | Local models                  |
| OpenRouter   | `https://openrouter.ai/api/v1`           | Multi-provider gateway        |

## Project Structure

```
letta-code/
├── apps/
│   ├── web/          # User-facing application
│   └── admin/        # Admin dashboard
├── packages/
│   ├── shared/       # Shared code (OpenAI client, DB, file store, types)
│   └── ui/           # Shared UI components
└── package.json      # Monorepo root
```

## Getting Started

### Prerequisites

- Node.js 18+
- pnpm 9+
- OpenAI API key (or any compatible API key)

### Installation

```bash
# Install dependencies
pnpm install

# Copy environment file
cp .env.example .env

# Add your API key to .env or configure it in the admin panel
```

### Configuration

Edit `.env` file:

```env
# Optional: Your API key. The admin panel can also store provider keys.
OPENAI_API_KEY=sk-xxx

# Optional: Custom base URL for other providers
OPENAI_BASE_URL=https://api.openai.com/v1

# Optional: Proxy settings
PROXY_ENABLED=false
PROXY_URL=
```

### Development

```bash
# Run all apps
pnpm dev

# Run only web app
pnpm dev:web

# Run only admin app
pnpm dev:admin
```

### Database Setup

```bash
# Generate database schema
pnpm db:generate

# Push schema to database
pnpm db:push

# Open Drizzle Studio (database viewer)
pnpm db:studio
```

## Usage

1. Open http://localhost:3000 for the main application
2. Open http://localhost:3001 for the admin dashboard
3. Add your API key in the admin panel (or via environment variable)
4. Select a model and start chatting!

## API Endpoints

### Web App

- `POST /api/chat` - Send message to AI (streaming)
- `GET /api/models` - Get available models

### Admin App

- `GET/POST/DELETE /api/api-keys` - Manage API keys
- `GET/PATCH /api/models` - Model configuration
- `GET /api/dashboard` - Dashboard summary
- `GET /api/analytics` - Usage analytics
- `GET/PATCH /api/settings` - System settings

## Quality Gates

```bash
# Run unit tests
pnpm test

# Run ESLint across apps
pnpm lint

# Run production build
pnpm build
```

## Custom Models

You can add custom models directly in the UI:

1. Click on the model selector in the header
2. Click "Add custom model..."
3. Enter the model identifier (e.g., `llama3.1:70b` for Ollama)

## Migration from Letta

If you were using the previous Letta-based version:

1. Update your `.env` file:
   - Replace `LETTA_API_KEY` with `OPENAI_API_KEY`
   - Add `OPENAI_BASE_URL` if using a custom provider

2. Run database migration:

   ```bash
   pnpm db:generate
   pnpm db:push
   ```

3. Re-add your API keys in the admin panel

## License

MIT