calorie-tracker-mcp

0x28F4/calorie-tracker-mcp

3.1

If you are the rightful owner of calorie-tracker-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

The Calorie Tracker MCP is a comprehensive system designed to facilitate calorie tracking and weight management through seamless integration with chat agents using the Model Context Protocol (MCP).

Tools
3
Resources
0
Prompts
0

Calorie Tracker MCP Server

🎯 Project Overview

A calorie tracking system designed as a first-class citizen for chat agents (like Claude), featuring MCP (Model Context Protocol) integration for seamless AI assistant interaction, persistent storage, and comprehensive analytics.

📋 Core Requirements

1. MCP Server Features

  • Remote MCP Server: Accessible from anywhere (not just localhost)
  • Chat-First Design: Optimized for interaction via Claude and other AI assistants
  • Simple API: Add meals/ingredients to storage via MCP tools
  • Persistent Storage: Database for long-term data retention

2. Tracking Capabilities

  • Batch Calorie Tracking: Log single or multiple meals in one operation with atomic transactions
  • Weight Management: Record daily weight measurements with batch operations and date validation
  • Historical Data Import: Support CSV imports for existing data

3. Analytics Features

  • Daily Metrics:

    • Total calories consumed
    • Caloric deficit/surplus
    • Weight measurements

  • Trend Analysis:

    • Moving average weight (configurable N days)
    • Daily weight changes
    • Average/median weight changes over N days
    • Total weight loss over time periods

  • Advanced Analytics:

    • Cumulative deficit calculation over date ranges
    • Metabolic rate estimation based on weight loss vs deficit
    • Ability to update metabolic rate from calculations

  • Export Options:

    • Google Sheets integration for visualization
    • CSV export capabilities

🛠 Technology Stack

Backend (MCP Server)

  • Language: TypeScript/Node.js

    • Excellent MCP SDK support
    • Strong ecosystem for web services

  • MCP Transport:

    • Local: Stdio transport for Claude Desktop
    • Remote: Streamable HTTP (MCP protocol 2025-03-26) via Express.js

  • Database: SQLite

    • Simple file-based database
    • No separate database server needed
    • Supports multi-user with session isolation
    • Easy backup (just copy the file)

  • MCP Integration: @modelcontextprotocol/sdk

    • Official MCP SDK for TypeScript
    • Latest protocol support (2025-03-26)
    • Dual transport capability (stdio + HTTP)

Local Development

  • Requirements: Node.js, npm/yarn
  • Database: SQLite file in project directory
  • MCP Testing: Claude Desktop app for local testing
  • Configuration: Environment-based configuration
    • USER_ID for stdio transport (required)
    • Optional debug middleware for HTTP testing
    • Database path and server settings
  • Build: npm run build compiles TypeScript to JavaScript

Analytics & Export

  • CSV Export: SQLite native .import command
  • Report Generation: Enhanced summary reports with moving averages
  • Moving Averages: Configurable N-day weight trend smoothing
  • Weight Trend Analysis: First-to-last day difference calculations

🏗 Architecture Design

graph TB
    Claude[Claude/Chat Agents] --> MCP[MCP Server<br/>TypeScript + Node.js]
    MCP --> DB[(SQLite Database)]
    
    subgraph "Core Data"
        DB --> Meals[Meals Table]
        DB --> Weights[Weights Table]
        DB --> Settings[User Settings]
    end
    
    subgraph "MCP Tools"
        MCP --> AddMeals[add_meals]
        MCP --> CheckWeight[check_weight]
        MCP --> Summary[get_summary]
        MCP --> Settings[update_user_settings]
    end
    
    subgraph "Future Export"
        MCP -.-> CSV[CSV Export]
        CSV -.-> Sheets[Google Sheets]
    end

📊 Data Model

  • Meals: Calories, optional macros (protein/carbs/fat), timestamps
  • Weights: Daily weight entries with date constraints
  • User Settings: Metabolic rate, timezone preferences

All data scoped by user_id with full multi-user support and session isolation.

🔧 MCP Tools

  • add_meals: Batch meal addition - Add one or more meals in a single transaction with atomic operations (arrays only)
  • add_weights: Batch weight tracking - Add one or more weight entries with date validation and duplicate handling
  • update_user_settings: Update timezone and metabolic rate
  • get_summary: Multi-day summary with moving averages, daily stats, and weight difference analysis in JSON format
  • calculate_metabolic_rate: Calculate metabolic rate from 7-day historical data with 3-day moving averages (pure calculation tool)
  • list_recent_meals: List recent meal entries
  • update_meal: Update existing meal entries
  • delete_meal: Delete meal entries
  • export_csv: Data export functionality (SQLite native support available)

🧪 Testing

✅ Current Testing

  • Comprehensive test suite with Vitest framework including batch meal operation tests
  • Database layer tests with in-memory SQLite and transaction rollback validation
  • MCP server integration tests with InMemoryTransport for client-server communication testing
  • Metabolic rate calculation tests with type-safe result validation helpers
  • CI/CD integration with automated testing pipeline
  • Concurrent access testing validated with multiple clients
  • Manual testing via Claude Desktop and MCP Inspector

📚 Documentation

  • - Complete guide for running and debugging both transport modes
  • - Development guidelines and coding standards
  • - Container deployment configuration

✅ Current Status

  • ✅ TypeScript project with MCP SDK
  • ✅ SQLite database with schema and proper async/await patterns
  • ✅ Enhanced MCP tools with moving average calculations
  • ✅ Dual transport support (stdio + HTTP) with graceful shutdown
  • ✅ User-contextual architecture with concurrent access support
  • ✅ Comprehensive test coverage with CI/CD pipeline
  • ✅ Production-ready containerization with Docker

💡 Why TypeScript?

We chose TypeScript over Go because the official MCP SDK is TypeScript-first, providing better stability, documentation, and long-term support for building MCP servers.

📋 Remaining Todo Items

Core Features

  • Implement graceful shutdown with connection removals
  • Test concurrent access through two different clients
  • Batch meal operations - Replace add_meal with add_meals for atomic bulk insertions
  • Refactor index.ts - Separate MCP server implementation into dedicated module (src/mcp/index.ts)
  • Meal presets system - Solved via CLAUDE.md context instructions to keep recent meals in context (no new tools needed)

Analytics & Reporting

  • Metabolic rate calculation from historical data
  • Enhanced daily summary tool (any date)
  • Weight trend analysis with moving averages
  • Date range statistics tool
  • Weekly/monthly summary reports
  • Weight prediction models

Testing & Quality

  • Set up Vitest testing framework with in-memory SQLite
  • Write comprehensive database layer tests
  • MCP server integration tests with InMemoryTransport and type-safe result validation
  • Metabolic rate calculation testing with live data validation
  • Add tests to CI/CD pipeline

Production & Deployment

  • Set up CI/CD pipeline with automated testing
  • Choose hosting platform and deploy
  • Implement backup strategy
  • Monitoring and alerting
  • Security hardening
  • Add proper authentication when internet-facing

Optional Enhancements

  • Web dashboard
  • Google Sheets direct integration
  • Mobile app considerations
  • Third-party fitness app sync