chrisdoc/hevy-mcp

What it solves

Hevy MCP Server allows AI assistants (like Claude, Cursor, and Codex) to interact with your Hevy fitness and workout tracking data. It bridges the gap between your workout history and LLMs, enabling you to analyze training progress, ask questions about your exercise history in plain language, and manage your routines and body measurements directly through an AI interface.

How it works

The project implements the Model Context Protocol (MCP), providing a set of 22 tools that the AI assistant can call to interact with the Hevy API. It can be deployed in two ways:

• Hosted: A Cloudflare Worker acts as a stateless proxy that forwards requests from the AI assistant to the Hevy API using a provided API key.
• Local: A Node.js server runs on your machine via npx, bunx, or Docker, communicating with the AI assistant via stdio.

Who it’s for

Fitness enthusiasts and Hevy PRO users who want to use AI to analyze their training data, automate the creation of workouts and routines, and track body measurement trends using natural language.

Highlights

• Comprehensive Toolset: 22 tools for reading and updating workouts, routines, exercise templates, and body measurements.
• Flexible Deployment: Supports hosted Cloudflare Worker, local Node.js/Bun, or Docker containers.
• ** uma AI-friendly search:** Compact results for routines and exercise templates to avoid overwhelming the LLM context window.
• Guided Prompts: Pre-defined workflows for analyzing workout progress (1-12 weeks) and creating workouts from saved routines.
• Broad Client Support: Compatible with Claude Desktop, Cursor, Codex, and Google Antigravity.