Taxuspt/garmin_mcp

What it solves

This project provides a Model Context Protocol (MCP) server that allows AI assistants (like Claude) to interact with Garmin Connect. It bridges the gap between your fitness and health data and LLMs, enabling them to analyze your workouts, track health metrics, and even create and schedule training plans directly within the chat interface.

How it works

The server acts as an intermediary between an MCP-compatible client and the Garmin Connect API (via the python-garminconnect library). It exposes over 110 tools that the AI can call to fetch data or perform actions. It supports multiple transport methods, including standard input/output (stdio) for local clients and HTTP/SSE for remote deployments.

Who it’s for

Fitness enthusiasts, athletes, and users of Garmin devices who want to use AI to analyze their health data, manage their training schedules, and automate the workout creation process.

Highlights

• Extensive Toolset: Implements 110+ tools covering activity management, health and wellness, training performance, and nutrition.
• Workout Builders: High-level tools for creating walk/run intervals, Z2 walks, and strength workouts without needing raw JSON.
• ** uma l-click Install**: Supports .dxt extension files for easy installation into Claude Desktop.
• Advanced Analytics: Access to power zones, FIT file analysis, and training load trends (CTL/ATL/TSB).
• Flexible Configuration: Ability to allowlist or denylist specific tools to reduce LLM context window usage.