Garmin Coach MCP

Garmin Coach MCP

Connects Claude to Garmin Connect data for personalized running coaching, including morning readiness checks, post-run analysis, weekly reviews, and goal tracking.

Category
Visit Server

README

🏃 Garmin AI Coach — MCP Server

A fully cloud-hosted MCP server that gives Claude direct access to your Garmin Connect data. Runs free on Vercel + Upstash. Works on any device — laptop, phone, tablet.

What it does

  • Morning check-in: Claude reads your HRV, sleep, body battery, and stress, then tells you whether to train hard, go easy, or rest — with actual reasoning.
  • Post-run debrief: Analyzes running dynamics (cadence, ground contact time, vertical oscillation), pacing, and HR data.
  • Weekly review: Training load trends, volume, intensity distribution, recovery quality.
  • Goal tracking: Progress toward sub-38 10km via VO2max trend, race predictions, PRs.
  • Coach profile: Claude interviews you over a few sessions to build deep knowledge of your training history, goals, injury history, and preferences — then uses it every session.

Stack (100% free)

Service What it does Cost
Vercel hobby Hosts the MCP server Free
Upstash Redis Stores Garmin tokens + cached data + coach profile Free (10k req/day)
garminconnect 0.3.4 Talks to Garmin Connect API Free
claude.ai Pro Your AI coach brain Your existing plan

Deploy in 5 steps

Step 1 — Fork and deploy to Vercel

  1. Fork this repo to your GitHub account
  2. Go to vercel.com → New Project → Import your fork
  3. Vercel auto-detects Python. Click Deploy.
  4. Note your deployment URL: https://your-project.vercel.app

Step 2 — Set up Upstash Redis

  1. Go to upstash.com → Create account → Create Redis database
  2. In Vercel: your project → Storage tab → Connect Store → select Upstash Redis
  3. This automatically injects UPSTASH_REDIS_REST_URL and UPSTASH_REDIS_REST_TOKEN into your env vars

Step 3 — Set environment variables in Vercel

In your Vercel project → SettingsEnvironment Variables, add:

ADMIN_SECRET = some-random-secret-only-you-know

Generate a strong random value, e.g. from 1password.com/password-generator. You'll need this once for setup, then you can forget it.

Click Redeploy after adding env vars.

Step 4 — Register yourself and connect Garmin

Install dependencies locally (one-time):

pip install httpx

Run the setup script:

MCP_URL=https://your-project.vercel.app/api/mcp python scripts/setup.py

Follow the prompts:

  1. Enter your ADMIN_SECRET
  2. Enter a label for yourself (e.g. "Alex")
  3. The script generates your API key — save it somewhere safe
  4. Enter your Garmin Connect email and password (used once, then discarded)

Repeat for your friend — they run the same script with the same MCP_URL.

Step 5 — Connect to Claude

  1. Open claude.ai on any device
  2. Go to Settings → Connectors → Add custom connector
  3. Name: Garmin Coach
  4. URL: https://your-project.vercel.app/api/mcp
  5. Click Add

Done. The connector is now available across claude.ai, Claude Desktop, and the mobile app.


First session

Start a conversation with Claude, enable the Garmin Coach connector, then paste your system prompt:

Open COACH_SYSTEM_PROMPT.md and paste its full contents into Claude → Settings → Custom Instructions.

Then say:

My API key is: [your-api-key]
Let's start my onboarding — interview me as my new coach.

Claude will ask you questions one at a time, building up your athlete profile across the session.


Daily use examples

Morning:

Morning check-in. API key: [key]

After a run:

Just got back from a tempo run. API key: [key] — debrief me.

Weekly:

Weekly review please. API key: [key]

Goal check:

How am I tracking for sub-38? API key: [key]

For your friend

They need:

  1. Claude account (free plan works)
  2. Run python scripts/setup.py with the same MCP_URL
  3. Ask you for the ADMIN_SECRET to register their own API key
  4. Add the connector URL to their Claude account
  5. Paste the system prompt into their custom instructions

Each person has completely separate data in Redis. No cross-contamination.


Security notes

  • Your Garmin email/password is sent once during setup, used to get OAuth tokens, then never stored
  • OAuth tokens are stored in Redis with a 1-year TTL and auto-refresh
  • API keys are stored as SHA-256 hashes in Redis — never the raw key
  • All traffic is HTTPS (Vercel enforces TLS)
  • The ADMIN_SECRET is only needed for initial setup — share it with your friend briefly, then change it in Vercel env vars if you want to lock it down

Troubleshooting

"No Garmin tokens found": Re-run scripts/setup.py — your Redis tokens may have expired.

"Invalid API key": Double-check you're using the exact key from setup. Generate a new one if needed.

Timeout errors: The Vercel free tier has a 10s limit. Historical data tools (get_running_history with many weeks) may time out — try smaller date ranges.

Garmin login fails: Garmin sometimes blocks programmatic login. Try again after a few minutes, or check if your Garmin account has 2FA enabled (disable it temporarily for setup).

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured