Garmin Health MCP Server
# Garmin Health MCP Server
> **Access your Garmin Connect health data directly in Claude Desktop and Claude Code CLI**
An [MCP (Model Context Protocol)](https://modelcontextprotocol.io) server that provides Claude with access to your Garmin wearable health data including sleep, recovery, HRV, workouts, and more.
## Features
- 🛌 **Sleep Analysis**: Hours, stages (light/deep/REM), quality scores
- 🔋 **Body Battery**: Garmin's proprietary recovery metric (0-100)
- ❤️ **HRV Tracking**: Heart rate variability trends for recovery monitoring
- 🏃 **Activity Data**: Workouts by type, calories, duration, pace
- 📊 **Health Metrics**: Resting heart rate, stress levels, VO2 max
- 📈 **Interactive Charts**: Generate beautiful HTML dashboards
- 🔐 **Secure**: Credentials stored locally, connects only to Garmin's API
## Quick Start
### Prerequisites
- **Claude Desktop** or **Claude Code CLI** installed
- **Node.js** 18+ and npm
- **Python** 3.8+
- Active Garmin Connect account with a connected device
### Installation
```bash
# 1. Clone this repository
git clone https://github.com/eversonl/garmin-health-mcp-server.git
cd garmin-health-mcp-server
# 2. Install Node.js dependencies
npm install
# 3. Install Python dependencies
pip3 install garminconnect fitparse gpxpy
# Or on managed systems:
pip3 install --user garminconnect fitparse gpxpy
# 4. Create .env file with your Garmin credentials
cp .env.example .env
# Edit .env and add your GARMIN_EMAIL and GARMIN_PASSWORD
# 5. Authenticate with Garmin Connect
npm run auth
```
### Configure Claude Desktop
Edit your Claude Desktop configuration file:
- **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
- **Linux**: `~/.config/Claude/claude_desktop_config.json`
Add this MCP server configuration:
```json
{
"mcpServers": {
"garmin-health": {
"command": "node",
"args": ["/absolute/path/to/garmin-health-mcp-server/index.js"],
"env": {
"GARMIN_EMAIL": "your-email@example.com",
"GARMIN_PASSWORD": "your-password"
}
}
}
}
```
**Important**: Replace `/absolute/path/to/garmin-health-mcp-server/` with the actual full path where you cloned this repo.
### Restart Claude Desktop
Completely quit and restart Claude Desktop for the MCP server to load.
## Using with Claude Code CLI
For Claude Code CLI users, this repository includes a `.mcp.json` configuration file that enables automatic MCP server loading.
### Quick Setup for Claude Code
1. **Set environment variables** with your Garmin credentials:
```powershell
# Windows PowerShell
setx GARMIN_EMAIL "your-email@example.com"
setx GARMIN_PASSWORD "your-password"
```
```bash
# Linux/macOS
export GARMIN_EMAIL="your-email@example.com"
export GARMIN_PASSWORD="your-password"
```
2. **Authenticate with Garmin**:
```bash
npm run auth
```
3. **Start Claude Code** from any directory:
```bash
claude
```
4. **Verify the MCP server is loaded**:
```
/mcp
```
You should see `garmin-health` listed as connected.
5. **Start querying your health data**:
> "How did I sleep last night?"
> "What's my body battery level today?"
**📘 For detailed Claude Code setup instructions**, see [CLAUDE_CODE_SETUP.md](CLAUDE_CODE_SETUP.md).
**Note**: The `.mcp.json` file uses environment variable expansion for security. Update the `args` path if you installed the server in a different location.
## Usage
Once configured, ask Claude questions like:
> "How did I sleep last night?"
> "What's my Body Battery looking like this week?"
> "Show me my activities from the past month"
> "Is my HRV improving?"
> "Generate a health dashboard for the last 30 days"
Claude will use the MCP tools to fetch your Garmin data and provide insights.
## Available Tools
The server exposes these tools to Claude:
| Tool | Description |
|------|-------------|
| `get_sleep_data` | Sleep hours, stages, quality scores |
| `get_body_battery` | Recovery metric (0-100) |
| `get_hrv_data` | Heart rate variability trends |
| `get_heart_rate` | Resting, max, min heart rate |
| `get_activities` | Workout/exercise data |
| `get_stress_levels` | All-day stress tracking |
| `get_summary` | Combined health overview |
| `get_user_profile` | Account and device info |
| `generate_chart` | Create interactive HTML visualizations |
## Troubleshooting
### "MCP server not found"
- Verify the path in `claude_desktop_config.json` is absolute (starts with `/` or `C:\`)
- Check that `index.js` exists at that path
- Ensure you ran `npm install` in the repo directory
### "Authentication failed"
- Run `npm run auth` to refresh tokens
- Check your credentials in the `.env` file or environment variables
- Try logging into [Garmin Connect](https://connect.garmin.com) website to verify your account
### "Tools not appearing in Claude"
- Completely quit Claude Desktop (don't just close the window)
- Check Console.app (macOS) or Event Viewer (Windows) for MCP errors
- Verify Node.js is accessible: `node --version`
- Check Python dependencies are installed: `python3 -c "import garminconnect"`
### "Missing data"
- Some metrics require specific Garmin devices (e.g., Body Battery needs HRV-capable watches)
- Historical data may have gaps if the device wasn't worn
- New accounts may have limited history
### "Too many requests" / Rate limiting
- Garmin limits API requests - wait a few minutes before retrying
- Use the `get_summary` tool instead of calling multiple individual tools
## Privacy & Security
- ✅ Credentials stored locally in `.env` and `claude_desktop_config.json`
- ✅ Session tokens cached locally and auto-refresh
- ✅ Connects only to Garmin's official API servers
- ✅ No cloud storage or third-party data sharing
- ✅ Open source - audit the code yourself
## Related Projects
- **[Clawdbot Skill](https://github.com/eversonl/ClawdBot-garmin-health-analysis)** - Automated health monitoring and proactive check-ins for Clawdbot
- **[python-garminconnect](https://github.com/cyberjunky/python-garminconnect)** - The Python library we use to access Garmin's API
## Development
```bash
# Run the server directly (for testing)
npm start
# Test authentication
npm run auth
# Run tests
npm test
```
## License
MIT © EversonL
## Credits
Built with:
- [@modelcontextprotocol/sdk](https://github.com/modelcontextprotocol/sdk) - MCP SDK
- [python-garminconnect](https://github.com/cyberjunky/python-garminconnect) - Garmin API wrapper
- [fitparse](https://github.com/dtcooper/python-fitparse) - FIT file parsing
- [gpxpy](https://github.com/tkrajina/gpxpy) - GPX file handling
---
## 💖 Support This Project
If you find this MCP server useful, consider supporting its development:
<a href="https://buymeacoffee.com/leeev" target="_blank"><img src="https://cdn.buymeacoffee.com/buttons/v2/default-yellow.png" alt="Buy Me A Coffee" style="height: 60px !important;width: 217px !important;" ></a>
---
**Questions or issues?** [Open an issue](https://github.com/eversonl/garmin-health-mcp-server/issues) on GitHub!
TDQS
Scored across 9 tools
Each tool has a clearly distinct purpose targeting specific health metrics or functions, with no overlap in scope. The tools cover different data types like activities, body battery, heart rate, HRV, sleep, stress, summaries, and profiles, making them easily distinguishable.
All tools follow a consistent verb_noun pattern with 'get_' or 'generate_' prefixes, using snake_case uniformly. This predictable naming scheme enhances readability and reduces confusion for agents.
With 9 tools, the server is well-scoped for a health data domain, covering key metrics without being overwhelming. Each tool serves a specific, necessary function, making the count appropriate for comprehensive health monitoring.
The toolset provides broad coverage for retrieving health metrics, including core data like activities, sleep, and stress, with a summary tool for aggregation. A minor gap exists in update or management tools for user data, but retrieval operations are largely complete for the domain.