Garmin Health MCP Server
The Garmin Health MCP Server enables you to query and analyze your Garmin health and fitness data through natural language conversations in Claude Desktop.
Key capabilities:
Query in plain English - Ask questions like "What's my VO2 max?" or "Am I overtraining?" and get answers from your Garmin data
Get comprehensive health overviews - View all metrics including VO2 max, activities, sleep, and race predictions in one summary
Track cardiovascular fitness - Monitor VO2 max history and trends, filtered by sport type and date ranges
Analyze workout patterns - Break down activities by type with statistics on counts, distances, and durations over custom time periods
Monitor sleep quality - Review sleep statistics including average duration and total nights tracked
View race predictions - Get predicted finish times for 5K, 10K, half marathon, and marathon based on current fitness
Optimize training intensity - Access personalized heart rate training zones calculated from max HR and lactate threshold
Assess training load - Monitor acute/chronic workload ratios to identify overtraining or undertraining risk
Compare activities - Run comparisons like "Compare my running vs cycling this year" to analyze performance across sports
The server integrates with Claude Desktop as an MCP server and queries data stored in Supabase, supporting historical analysis across multiple years of fitness tracking.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Garmin Health MCP Server show me my sleep statistics from last week"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Garmin Health MCP Server
Query your Garmin health data in plain English through Claude Desktop.
Quick Start
Prerequisites
Node.js 18+
Claude Desktop
Garmin data in Supabase
Install
git clone https://github.com/willc121/garmin-health-mcp-server.git
cd garmin-health-mcp-server
npm install
npm run buildConfigure Claude Desktop
Add to your Claude Desktop config:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"garmin": {
"command": "node",
"args": ["/absolute/path/to/dist/index.js"],
"env": {
"SUPABASE_URL": "your-supabase-url",
"SUPABASE_ANON_KEY": "your-supabase-anon-key"
}
}
}
}Restart Claude Desktop. The Garmin connector should appear in your connectors menu.
Related MCP server: garmin-mcp
Available Tools
Tool | Description |
| Overview of all health data |
| VO2 max history and trends |
| Activity breakdown by type |
| Sleep statistics |
| Predicted race times |
| Personalized HR training zones |
| Acute/chronic workload ratio |
Example Queries
What's my VO2 max?Am I overtraining?Compare my running vs cycling this yearGetting Your Garmin Data
1. Request your data export
Navigate to Account Settings → Data Management
Click "Export Your Data" and request all data
2. Wait for the email
Garmin sends a download link within 48 hours. The export can be several GB depending on how long you've been tracking.
3. Download and extract
You'll get a zip with JSON files for activities, sleep, VO2 max, heart rate, and more.
4. Load into a database
The raw export is too large to query directly (mine was 9 years of data), so I loaded it into Supabase (free tier works fine).
You'll need these tables:
vo2_max— VO2 max readings by date and sportactivities— Activity records with type, duration, distance, HRsleep_summary— Aggregated sleep statsrace_predictions— Garmin's predicted race timesheart_rate_zones— HR zone boundariestraining_load— Daily training load metrics
I wrote Python scripts to parse the Garmin JSON and insert into Supabase. Happy to share if there's interest — open an issue.
Troubleshooting
Problem | Fix |
Connector doesn't appear | Check JSON syntax, use absolute path, fully restart Claude |
Connection errors | Verify Supabase credentials |
License
MIT
🏃 Powered by questionable cardio and an unhealthy amount of VO2 max anxiety. No treadmills were harmed in the making of this server.
Available Tools
7 toolsget_activitiesC
Get activity breakdown by type, including counts, distances, and durations
Inputs: start_date, end_date, activity_type.
get_health_summaryB
Get an overview of all health data including VO2 max, activities, sleep, and race predictions
Inputs: no named parameters.
get_heart_rate_zonesB
Get personalized heart rate training zones based on max HR and lactate threshold
Inputs: no named parameters.
get_race_predictionsB
Get predicted race times for 5K, 10K, half marathon, and marathon based on current fitness
Inputs: no named parameters.
get_sleepC
Get sleep statistics including average duration and total nights tracked. Optionally filter by date range.
Inputs: start_date, end_date.
get_training_loadB
Get training load data including acute/chronic workload ratio to assess overtraining risk
Inputs: days.
get_vo2maxC
Get VO2 max history and trends. VO2 max measures cardiovascular fitness in ml/kg/min.
Inputs: start_date, end_date, sport.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
Each tool has a clearly distinct purpose targeting specific health and fitness data domains, such as activities, health summary, heart rate zones, race predictions, sleep, training load, and VO2 max. There is no overlap in functionality, and an agent can easily differentiate between them based on their descriptive names and scopes.
All tool names follow a consistent 'get_' prefix pattern with descriptive nouns, such as get_activities, get_health_summary, and get_heart_rate_zones. This uniformity makes the tool set predictable and easy to navigate, adhering to a clear verb_noun convention throughout.
With 7 tools, the server is well-scoped for a health and fitness data domain, covering key aspects like activities, sleep, training metrics, and predictions. Each tool serves a distinct purpose without redundancy, making the count appropriate and manageable for the server's intended use.
The tool set provides comprehensive read-only coverage for retrieving health and fitness data, including summaries, detailed metrics, and predictions. A minor gap exists in the lack of write or update operations, but for a data retrieval-focused server, this is reasonable and agents can work effectively with the available tools.
Maintenance
Related MCP Connectors
Explore Garmin activity and health data with your AI assistant.
Ask your AI about your fitness: activities and data from Garmin, COROS, Strava, GPX and more
Your Garmin, Oura, WHOOP, Hevy and 35 more fitness apps, queryable from your own AI.
Garmin watch data for AI assistants: activities, sleep, HRV, recovery. Never writes to Garmin.
Related MCP Servers
- FlicenseNot gradedqualityCmaintenanceEnables querying and analyzing personal Strava activity data through natural language, including activities, segments, gear, and training trends.-
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to query live Garmin Connect health and fitness data, including daily metrics, activities, sleep analysis, and trends via natural language.MIT
- FlicenseNot gradedqualityDmaintenanceEnables querying your Garmin Connect data including recent activities, detailed metrics, and daily stats through natural language.-
- FlicenseNot gradedqualityBmaintenanceEnables AI assistants to retrieve and analyze Garmin Connect health data, including daily summaries, sleep, HRV, workouts, and raw FIT files, while also supporting SQL queries on the underlying database.10-