Cadence¶
AI-powered coaching platform that connects to your fitness wearables and provides personalized training insights, recovery monitoring, and AI-driven coaching conversations.
What It Does¶
Cadence pulls data from your fitness devices and uses AI to help you train smarter:
- Daily briefings — sleep quality, recovery status, training readiness
- Training analytics — load tracking, zone analysis, performance trends, injury risk detection
- AI coaching chat — ask questions about your training, get personalized advice backed by your data
- Workout logging — track strength training sessions with progressive overload tracking
- Multi-device support — connect Garmin, WHOOP, and Strava simultaneously
How It Works¶
Cadence is built on the Model Context Protocol (MCP). It exposes 75 tools that AI assistants can call to fetch your fitness data, analyze trends, and log workouts. It also ships with a built-in web dashboard and chat interface.
┌─────────────┐ ┌─────────────────┐ ┌──────────────┐
│ Claude │ │ │ │ Garmin │
│ Desktop │────▶│ Cadence │────▶│ WHOOP │
│ or Web UI │◀────│ (MCP Server) │◀────│ Strava │
└─────────────┘ └─────────────────┘ └──────────────┘
Quick Start¶
git clone https://github.com/ladkam/coaching-mcp.git
cd coaching-mcp
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env # then fill in your credentials
python server.py --http
See the Getting Started guide for full setup instructions.
Key Sections¶
| Section | What You'll Find |
|---|---|
| Getting Started | Installation, first run, connecting devices |
| Configuration | Environment variables, feature flags |
| Architecture | System design, database schema, project structure |
| Integrations | Device-specific setup and data reference |
| MCP Tools | Complete tool reference (75 tools) |
| AI Chat | LLM provider setup, coaching behavior |
| REST API | HTTP endpoints for building clients |
| Deployment | Railway, Heroku, production config |