CoolFace
Apppublic

MCP-1st-Birthday/baby-chef

sourceHugging Faceupdated 10mo agoView on Hugging Face
0likes
App README
Track: MCP in Action – Consumer Tag: mcp-in-action-track-consumer

<div align="center">

👶 Baby Chef

AI-Assisted Baby-Feeding Safety Companion

<a href="https://huggingface.co/MCP-1st-Birthday"> <img src="https://img.shields.io/badge/Track-MCPinAction-FFD21E?style=for-the-badge&logo=huggingface&logoColor=black" alt="Track: MCP in Action"> </a> <a href="https://huggingface.co/spaces/MCP-1st-Birthday/baby-chef"> <img src="https://img.shields.io/badge/Status-Live_Demo-green?style=for-the-badge" alt="Status Live"> </a>

<br>

<a href="https://youtu.be/t8rirmL9-qA" target="_blank"> <img src="https://img.youtube.com/vi/t8rirmL9-qA/maxresdefault.jpg" alt="Watch the Baby Chef Demo" width="600" style="border-radius: 10px; box-shadow: 0 4px 8px 0 rgba(0, 0, 0, 0.2);"> </a> <p><i>👆 Click to Watch the Demo</i></p>

</div>

📖 Overview

Baby Chef helps parents understand how to safely serve foods to their baby using a 4-step autonomous MCP pipeline: Vision → Safety → Recipe → Milestone Logging.

It turns a simple meal photo into actionable, age-appropriate guidance, helping parents make safe, informed choices with clarity and care.

📸 App Screenshot

<div align="center"> <img src="https://huggingface.co/spaces/MCP-1st-Birthday/baby-chef/resolve/main/static/screenshot_ready.jpg" alt="Baby Chef Interface" width="75%"> </div>

🎯 What This App Does

  1. 1.Upload a photo of any food.
  2. 2.Enter your baby’s age (in months).
  3. 3.Select allergens (optional).

The Supervisor Agent then runs the 4-step MCP pipeline to generate:

  • —✅ Detection: Identifies exactly what is in the image.
  • —🛡️ Safety Check: Validates against age & allergy rules.
  • —🍲 Recipe: Provides a baby-safe serving suggestion.
  • —📓 Milestone: Logs the food exposure for your history.

🏗️ Architecture: The 4-Step Pipeline

We use a deterministic, rule-based pipeline to ensure clarity and auditability. No black-box safety predictions.

<div align="center"> <img src="https://huggingface.co/spaces/MCP-1st-Birthday/baby-chef/resolve/main/static/Architecture.png" alt="Architecture Diagram" width="90%"> </div>

🧩 How the Agents Work

  • —Vision MCP: Identifies food items and flags potential non-food hazards.
  • —Safety MCP: Checks strict age-based rules (e.g., "Honey forbidden under 12 months").
  • —Recipe MCP: Generates texture-appropriate serving advice (e.g., "Puree for 6mo").
  • —Milestone MCP: Logs the exposure for the parent's history.

✨ Key Features

  • —Full MCP-Native Design: Demonstrates modular agentic architecture.
  • —Safety First: Logic is grounded in pediatric guidelines, not hallucinated advice.
  • —Privacy Focused: BYOK (Bring-Your-Own-Key) architecture keeps user API keys private.
  • —Mobile-First UI: Clean, card-based interface designed for busy parents.

🚀 Quick Start (Local)

To run this completely locally using Docker:

bash
git clone https://huggingface.co/spaces/MCP-1st-Birthday/baby-chef
cd baby-chef
docker build -t babychef:prod .
docker run -p 7860:7860 --env-file .env babychef:prod

Open your browser at `http://localhost:7860`

Requirements:

  • —Python 3.11 / Docker
  • —.env file with GEMINI_API_KEY (or relevant environment variables)

☁️ Running on Hugging Face Spaces

This repo includes a custom Dockerfile with the correct port configuration (7860). When pushed to the MCP-1st-Birthday org Space, the container builds automatically, and the app goes live instantly.

👨‍💻 Team & Social

Karabolamani (Solo Developer)

⚠️ Disclaimer

Important: This application provides general baby-feeding guidance based on standard rules. It is not medical advice. Always supervise your baby while eating. If in doubt, consult a qualified healthcare professional.

Built for the Hugging Face MCP 1st Birthday Hackathon 2025