LH

leifmork/hevy-garmin-mcp

Developer tools

このページは AI により自動翻訳されています。技術的な詳細は原文をご確認ください。 原文を見る

1 stars 9 forks 品質 47 トレンド 5

MCP server that connects Garmin Connect and Hevy to ChatGPT — giving your AI full access to your running, recovery, and strength training data.

概要

A secure, production-ready Model Context Protocol (MCP) server built with Next.js that connects and directly to ChatGPT or any other MCP-compatible LLM. By connecting this server to ChatGPT, you instantly give it the ability to act as your personalized, data-driven AI Personal Trainer. It can read your run splits, analyze your sleep and recovery, dive into your 1-rep max progression, and compare your planned workouts against your actual performance. - : Fetch your latest runs, rides, walks, or hikes with distance, duration, pace, and heart rate. - : Deep dive into a specific activity, including per-lap splits and elevation data. - : Read your Garmin Coach planned intervals and target paces for today (or any date). - : Compare what your Garmin Coach told you to do vs. what you actually did. - : Summarize a full Monday-Sunday week of training load. - : Read your sleep stages, HRV, body battery, and resting heart rate to determine training readiness.

README

以下は原文 README の抜粋です。原文を見る。

Fitness AI Coach (MCP Server)

A secure, production-ready Model Context Protocol (MCP) server built with Next.js that connects Garmin Connect and Hevy directly to ChatGPT or any other MCP-compatible LLM.

By connecting this server to ChatGPT, you instantly give it the ability to act as your personalized, data-driven AI Personal Trainer. It can read your run splits, analyze your sleep and recovery, dive into your 1-rep max progression, and compare your planned workouts against your actual performance.


🚀 Features

Garmin Connect Integration

  • get_recent_activities: Fetch your latest runs, rides, walks, or hikes with distance, duration, pace, and heart rate.
  • get_activity_detail: Deep dive into a specific activity, including per-lap splits and elevation data.
  • get_planned_workout: Read your Garmin Coach planned intervals and target paces for today (or any date).
  • get_planned_vs_actual: Compare what your Garmin Coach told you to do vs. what you actually did.
  • get_week_summary: Summarize a full Monday-Sunday week of training load.
  • get_daily_recovery: Read your sleep stages, HRV, body battery, and resting heart rate to determine training readiness.
  • get_weight_trend: Check recent body weight weigh-ins securely.
  • get_heart_rate_zones: Returns your heart rate zone boundaries (Zone 1-5 in bpm) per sport, max HR, lactate threshold, VO2 max, and training method. Essential for properly classifying effort levels in activity analysis.

Hevy (Strength Training) Integration

  • get_lifting_sessions: Pull recent or historical (via date range) strength training logs. Includes sets, reps, weight, RPE, and set types (warmup, dropset, failure).
  • get_exercise_progress: Chart your progress on a specific exercise (like Squats or Bench Press) over a date range.
  • get_exercise_templates: Search the Hevy database by name or muscle group to find the specific template ID needed for tracking exercise progress.
  • get_lifting_routines: Read your saved Hevy workout routines and programs.
  • get_lifting_volume: Calculate total workout volume per exercise to track progressive overload.

Security

  • Capability URL Auth: Integrates natively with ChatGPT’s UI without needing a complex OAuth setup. The MCP server uses a highly secure ?token=YOUR_SECRET_KEY query parameter that acts as a password, fully encrypted over HTTPS.

🤖 Example ChatGPT Prompts

Once connected, you can ask ChatGPT things like:

“Check my sleep and HRV from last night. Based on my recovery, what kind of run should I do today? Check if Garmin Coach has anything planned for me.”

“I’m going to the gym for leg day. Can you pull up my recent Hevy lifting sessions and tell me what weights I should target for Squats today to ensure progressive overload?”

GitHub で README 全文を見る

インストール

This server does not publish a one-line install command.

Open the repository installation guide MCP 初心者?mcpskill.org のガイドから始めましょう →