Create structured Garmin Connect workouts (pace, HR, cadence targets, repeat blocks) from natural language — MCP server for Claude
- ✓Open-source license (MIT)
- ✓Actively maintained (<30d)
- ✓Clear description
- ✓Topics declared
- ✓Documented (README)
claude mcp add garmin-connect-workouts-mcp -- npx -y garmin-connect-workouts-mcp{
"mcpServers": {
"garmin-connect-workouts-mcp": {
"command": "npx",
"args": ["-y", "garmin-connect-workouts-mcp"]
}
}
}MCP Servers overview
# Garmin Connect Workouts MCP
[](https://glama.ai/mcp/servers/slaat/garmin-connect-workouts-mcp)
[](https://glama.ai/mcp/servers/slaat/garmin-connect-workouts-mcp)
An MCP (Model Context Protocol) server that turns natural-language workout
descriptions into structured workouts in Garmin Connect. Describe a session
in plain English and the calling model builds a typed step-by-step workout
that this server encodes into Garmin's workout API format and creates
directly in your account.
## Features
- **Pace, heart-rate, power and cadence targets** - pace ranges, HR zones 1-5,
explicit bpm ranges, cycling power, and cadence, all encoded against
Garmin's real `workoutTargetType` table.
- **Real repeat blocks** - `5x1km` becomes a single repeat group, not five
duplicated steps, so the watch shows an actual rep counter.
- **Swim workouts** - per-step strokes (freestyle, backstroke, breaststroke,
fly, drill, individual medley, or any stroke) and equipment (fins,
kickboard, paddles, pull buoy, snorkel), plus a workout-level pool length
in meters or yards.
- **Full workout lifecycle** - create, list, get, update, delete, and
schedule workouts onto the Garmin Connect calendar.
- **Structured input, not text parsing** - the server never parses free text
itself; every duration and target is an explicit, unit-tagged field, so
there's nothing for a string like "10 min" to be silently misread as.
## Quick Start
### Claude Code
```bash
claude mcp add garmin-workouts npx garmin-connect-workouts-mcp
```
### Claude Desktop
Add to your configuration file:
- **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
```json
{
"mcpServers": {
"garmin-workouts": {
"command": "npx",
"args": ["-y", "garmin-connect-workouts-mcp"]
}
}
}
```
Then restart Claude Desktop.
## Authentication
Garmin Connect's web client authenticates with session cookies plus a CSRF
token, not a password grant this server could hold on to. So the first time
you create a workout (or whenever the session has expired):
1. The server checks the stored session with a cheap authenticated request.
2. If it's missing or rejected, run the `authenticate_garmin` tool. It opens
a real browser window at `connect.garmin.com` for you to log in normally.
3. Once login completes, the session is captured and stored at
`~/.config/garmin-connect-workouts-mcp/auth.json` with `600` permissions
(owner read/write only).
No password is ever stored - only the resulting session cookies and CSRF
token. Garmin sessions expire, so you may be prompted to re-authenticate
between sessions; the server detects this automatically and tells you when
to run `authenticate_garmin` again. The login browser uses a persistent,
isolated profile under `~/.config/garmin-connect-workouts-mcp/browser-profile/`
so Garmin can remember the device across logins - it never touches your
personal browser profile.
## Tools
| Tool | Description |
|---|---|
| `create_garmin_workout` | Create a structured workout with steps, targets, and repeat blocks. |
| `list_garmin_workouts` | List workouts in the Garmin Connect account. |
| `get_garmin_workout` | Fetch the full structure of one workout. |
| `update_garmin_workout` | Replace an existing workout's contents, keeping its id. |
| `delete_garmin_workout` | Delete a workout. |
| `schedule_garmin_workout` | Put a workout on the Garmin Connect calendar for a date. |
| `check_garmin_auth` | Check whether the stored Garmin session still works. |
| `authenticate_garmin` | Authenticate with Garmin Connect (opens a browser). |
## Usage Examples
- "Create a 30 minute easy run in zone 2"
- "10 min warmup zone 3, then 5x1km threshold intervals at 4:00/km with 2 min
rest between, then 10 min cooldown zone 2"
- "8x400m at 5k pace with 60 second recovery, then check what workouts I
already have scheduled this week"
- "Update workout 123456 to add a 5 minute cooldown, then schedule it for
next Monday"
- "Create a 25m pool swim: 200m warmup any stroke, 4x100m freestyle with 30s
rest using a kickboard, 100m breaststroke cooldown"
## How targets work
A stated pace is treated as the **fast edge** of a range: Garmin stores pace
as a range, not a single point, so `4:00/km` on its own becomes a
`4:00-4:10/km` range, widened by 10 seconds. Give both `fast` and `slow`
explicitly for a narrower or different range.
## Development
```bash
pnpm install
pnpm test # vitest: encoders + golden-file tests against real captured payloads
pnpm run build # compile TypeScript to dist/
pnpm run dev # run the server directly with tsx
```
Tests validate the workout encoders (targets, durations, repeat nesting)
against fixtures captured from Garmin's own web client, so payload shape
regressions get caught before they reach a real account.
## Disclaimer
This is an unofficial, third-party project. It is not affiliated with,
endorsed by, or sponsored by Garmin Ltd. or any of its affiliates, and it
talks to Garmin Connect's web API by reverse-engineering the same requests
the official web client makes. It may stop working if Garmin changes that
API. Use at your own risk, and at your own responsibility for the workouts
and data you send to your account.
## Acknowledgements
Thanks to [garmin-workouts-mcp](https://github.com/charlesfrisbee/garmin-workouts-mcp)
by Charles Frisbee for the original idea of driving Garmin Connect workouts
through MCP.
## License
MIT
What people ask about garmin-connect-workouts-mcp
What is slaat/garmin-connect-workouts-mcp?
+
slaat/garmin-connect-workouts-mcp is mcp servers for the Claude AI ecosystem. Create structured Garmin Connect workouts (pace, HR, cadence targets, repeat blocks) from natural language — MCP server for Claude It has 0 GitHub stars and its last recorded update is dated 2026-09-02.
How do I install garmin-connect-workouts-mcp?
+
You can install garmin-connect-workouts-mcp by cloning the repository (https://github.com/slaat/garmin-connect-workouts-mcp) or following the README instructions on GitHub. ClaudeWave also provides quick install blocks on this page.
Is slaat/garmin-connect-workouts-mcp safe to use?
+
Our security agent has analyzed slaat/garmin-connect-workouts-mcp and assigned a Trust Score of 95/100 (tier: Verified). See the full breakdown of passed checks and flags on this page.
Who maintains slaat/garmin-connect-workouts-mcp?
+
slaat/garmin-connect-workouts-mcp is maintained by slaat. The last recorded GitHub activity is dated 2026-09-02, with 0 open issues.
Are there alternatives to garmin-connect-workouts-mcp?
+
Yes. On ClaudeWave you can browse similar mcp servers at /categories/mcp, sorted by popularity or recent activity.
Deploy garmin-connect-workouts-mcp to your cloud
Ship this repo to production in minutes. Each platform spins up its own environment with editable env vars.
Maintain this repo? Add a badge to your README
Drop the badge into your GitHub README to show it's tracked on ClaudeWave. Each badge links back to this page and reflects the live Trust Score.
[](https://claudewave.com/repo/slaat-garmin-connect-workouts-mcp)<a href="https://claudewave.com/repo/slaat-garmin-connect-workouts-mcp"><img src="https://claudewave.com/api/badge/slaat-garmin-connect-workouts-mcp" alt="Featured on ClaudeWave: slaat/garmin-connect-workouts-mcp" width="320" height="64" /></a>More MCP Servers
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)
An open-source AI agent that brings the power of Gemini directly into your terminal.
Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface
The fastest path to AI-powered full stack observability, even for lean teams.
🕷️ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl!