🏋️ Fitness Tracker — MCP Server

A fully offline Model Context Protocol server that lets any MCP‑compatible client (Claude Code, Claude Desktop, etc.) log workouts, track dietary macros, and retrieve daily summaries — all stored in a local SQLite database.

📂 Project Structure

MCP_Project/
├── server.py           # MCP server — tools, schemas, DB logic
├── test_server.py      # Pytest suite (22 tests)
├── requirements.txt    # pip dependencies
├── README.md           # You are here
└── fitness_tracker.db  # Created automatically on first run

Related MCP server: Nutrition MCP

🚀 Quick Start

1. Install Dependencies

pip install -r requirements.txt

2. Run the Test Suite

pytest test_server.py -v

You should see 22 tests pass, covering:

• ✅ Pydantic schema validation (valid inputs, bad dates, negative numbers, missing fields, SQL injection strings)

• ✅ Database insert and retrieve logic

• ✅ Daily summary aggregation and date isolation

3. Run the Server Standalone (optional)

python server.py

The server starts on stdio — it reads JSON‑RPC from stdin and writes to stdout. You will not see a prompt; this is intended for MCP client consumption.

🔌 Connect to Claude Code

Run this once in your terminal to register the server:

claude mcp add fitness-tracker --transport stdio -- python server.py

Tip: Use the full absolute path to server.py if Claude Code is not launched from this directory:

claude mcp add fitness-tracker --transport stdio -- python "C:\Users\Mayan\OneDrive\Documents\MCP_Project\server.py"

After registering, Claude Code can call the three tools below whenever you ask fitness-related questions.

🛠️ Exposed Tools

log_workout

Log a single workout session.

log_macros

Log dietary macronutrients for a meal or full day.

get_daily_summary

Retrieve a combined workout + nutrition summary for a given date.

Returns aggregated totals: workout count, total duration, total calories burned, macro totals (protein / carbs / fat in grams), and estimated calories consumed (using 4 / 4 / 9 kcal per gram).

🔒 Safety

• Parameterised queries — all SQL uses ? placeholders; user input is never interpolated into query strings.

• Pydantic validation — every tool input is parsed through a strict schema before touching the database, so malformed LLM outputs fail immediately with a clear error.

• Offline by design — stdio transport means zero network traffic.

📝 License

MIT — use freely.