# RapidChart MCP Server

Model Context Protocol (MCP) server for [RapidChart](https://rapidchart.com) - Generate and manage diagrams directly from Cursor IDE.

## Features

- 🎨 Generate diagrams from code (Class, ER, Sequence, Architecture, etc.)
- 🔄 Update existing diagrams with new code
- 📁 Organize diagrams in workspaces and folders
- 🤖 Support for multiple AI models (GPT-5, Claude, Gemini)
- 🔐 Secure Bearer token authentication
- 💳 Respects your RapidChart credits and API keys

## Installation

```bash
pip install rapidchart-mcp
```

Or install from source:

```bash
git clone https://github.com/Samicostox/rapidchart-mcp.git
cd rapidchart-mcp
pip install -e .
```

## Configuration

### 1. Generate API Token

1. Go to [RapidChart Settings](https://rapidchart.com/settings)
2. Navigate to "API Tokens"
3. Click "Generate New Token"
4. Copy the token (starts with `rc_`)

### 2. Configure Cursor

Add to your Cursor MCP settings (`~/.cursor/mcp.json` or workspace `.cursor/mcp.json`):

```json
{
  "mcpServers": {
    "rapidchart": {
      "command": "python",
      "args": ["-m", "rapidchart_mcp"],
      "env": {
        "RAPIDCHART_API_TOKEN": "rc_your_token_here",
        "RAPIDCHART_API_URL": "https://fastuml-0bb6938ba599.herokuapp.com",
        "RAPIDCHART_DEFAULT_TYPE": "general",
        "RAPIDCHART_FEW_PROMPTS": "false",
        "RAPIDCHART_GUIDELINES": "true"
      }
    }
  }
}
```

## Usage

### In Cursor

Once configured, you can use natural language:

```
User: "Generate a class diagram from src/models.py"
→ Cursor reads the file and calls RapidChart MCP
→ Returns diagram URL

User: "List my available AI models"
→ Shows models with your credit status

User: "Update diagram abc123 with new code from src/models.py"
→ Updates the diagram with context awareness
```

### Available Tools

- `list_models` - Show available AI models and your credits
- `list_workspaces` - List your workspaces
- `list_folders` - List folders in a workspace
- `list_diagrams` - Browse your diagrams
- `create_diagram` - Generate diagram from code
- `get_diagram` - Get specific diagram details
- `update_diagram` - Update diagram with new code
- `delete_diagram` - Delete a diagram
- `move_diagram` - Move diagram to different folder/workspace

## Environment Variables

| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| `RAPIDCHART_API_TOKEN` | ✅ Yes | - | Your RapidChart API token |
| `RAPIDCHART_API_URL` | No | Production URL | API base URL |
| `RAPIDCHART_DEFAULT_TYPE` | No | `general` | Default diagram type |
| `RAPIDCHART_DEFAULT_MODEL` | No | Auto | Default AI model ID |
| `RAPIDCHART_FEW_PROMPTS` | No | `false` | Enable multi-step thinking |
| `RAPIDCHART_GUIDELINES` | No | `true` | Include diagram guidelines |
| `RAPIDCHART_TIMEOUT` | No | `300` | Request timeout (seconds) |

## Development

```bash
# Clone the repo
git clone https://github.com/rapidchart/rapidchart-mcp.git
cd rapidchart-mcp

# Install in development mode
pip install -e .

# Run tests
pytest

# Run the server directly
python -m rapidchart_mcp
```

## Known Issues

### 503 Timeout on Diagram Creation

**Symptom:** You get a `503 Service Unavailable` error, but the diagram is created successfully.

**Cause:** Heroku has a 30-second router timeout. Complex diagrams can take 30-60+ seconds to generate.

**Solution:**
1. The diagram IS being created despite the error
2. Run `list_diagrams` to see your newly created diagram
3. This is a Heroku limitation, not a bug in RapidChart

**Workaround:** Use simpler code or smaller diagrams for faster generation.

---

## License

MIT

## Support

- 📧 Email: hello@rapidchart.com
- 🐛 Issues: [GitHub Issues](https://github.com/Samicostox/rapidchart-mcp/issues)
- 📖 Docs: [docs.rapidchart.com](https://docs.rapidchart.com)