# DeepVLab Analytics MCP Server

A Model Context Protocol (MCP) server for querying DeepVLab account statistics and model usage analytics.

## Features

- 🔐 **Browser Authorization** - Automated login and token management.
- 👤 **User Profile** - View basic account information and balance.
- 📊 **Usage Analytics** - Detailed statistics on model usage, daily trends, and quotas.
- 💰 **Cost Calculator** - Estimate and compare costs across different models.
- 📈 **Real-time Data** - Fetched directly from DeepVLab API.

## Quick Start

### 1. Installation

```bash
git clone https://github.com/yourusername/deepvlab-mcp.git
cd deepvlab-mcp
pip install -r requirements.txt
```

### 2. Configuration

Add the server to your `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "deepvlab-analytics": {
      "command": "python",
      "args": ["/path/to/deepvlab-mcp/server.py"],
      "env": {
        "PYTHONPATH": "/path/to/deepvlab-mcp"
      }
    }
  }
}
```

*Note: Replace `/path/to/deepvlab-mcp` with your actual absolute path.*

### 3. Usage

Restart Claude Desktop and simply ask:
- "Login to DeepVLab"
- "Show my account balance"
- "Compare costs for Claude 3.5 Sonnet vs GPT-4o"

## Available Tools

| Tool | Description |
|------|-------------|
| `login` | Launch browser to authorize and get token |
| `get_user_summary` | Get user info and balance overview |
| `get_model_rates` | List all models and their rates |
| `get_credits_usage` | Detailed credit usage stats |
| `get_model_usage` | Usage stats per model |
| `get_daily_usage` | Daily usage trends |
| `get_quotas` | Quota information |
| `calculate_cost` | Calculate cost for specific token counts |
| `estimate_batch_cost` | Estimate costs for batch requests |
| `compare_model_costs` | Compare costs across multiple models |

## License

MIT