Project Overview
Nutritionix-MCP Chat is a prototype chat application that lets users log meals in natural language. It parses food items via an MCP server backed by the Nutritionix API and returns detailed calorie and macronutrient breakdowns. Core technologies:
- FastAPI backend with WebSocket/chat endpoints
- FastMCP server exposing Nutritionix tools
- LangChain-Ollama orchestrator for tool‐calling and reasoning
- Google Mesop frontend for a lightweight chat UI
Features
- Natural-language meal logging (“I had 1 katori dahi and 2 chapatis”)
- Automatic tool-calling to Nutritionix’s natural nutrients API
- Regional unit conversion (e.g. ounces vs. litres via
alt_measures) - Macro tracking: calories, protein, carbs, fats
- Modular LLM orchestration: swap Ollama models via
llm_provider.py - Lightweight Python UI using Mesop’s
mel.chat
