metadata

title: FoodWise Remote MCP
emoji: 🥗
colorFrom: green
colorTo: blue
sdk: docker
app_port: 7860
hf_oauth: false

FoodWise

A smart food management system with recipe organization and meal planning, designed for detailed inventory tracking.

Core Concepts

FoodWise is built around a detailed data model to track food items across different storage areas: the pantry, refrigerator, and freezer. The goal is to provide a clear, organized overview of your food inventory to reduce waste and improve meal planning.

Key Features

Pantry Tracking: Keep a detailed inventory of your pantry items, including purchase and best-by dates, location, and whether a package has been opened.
Refrigerator Tracking: Manage refrigerated items with attention to food safety, including temperature zones and sensitivity flags.
Freezer Tracking: A detailed system for your multi-drawer freezer, tracking freeze dates, cooked/raw status, and preparation notes.
Notion Integration: Seamlessly sync your food inventory with a Notion database for ultimate organization.
LangGraph Agent: Conversational AI agent for natural language inventory management using LangGraph's React agent framework.
Discord Bot Ready: Architecture designed for Discord integration (future implementation).

Development Setup

This project uses uv for fast and efficient package and virtual environment management.

Prerequisites

Python 3.12+
uv package manager

Installation

Clone the repository:

git clone https://github.com/your-username/FoodWise.git
cd FoodWise

Create and activate the virtual environment:
```
uv venv
source .venv/bin/activate
```
Install dependencies:
```
uv pip install -e ".[dev]"
```

This command installs the project in editable mode (-e) along with all main and development dependencies.

Notion Integration Setup

Create a Notion integration:
- Visit Notion Integrations
- Create a new integration and copy the "Internal Integration Token"
- Share your food inventory database with the integration

Configure environment variables:

cp env.example .env  # or create .env file

Add your credentials to .env:

NOTION_SECRET=your_notion_integration_token
NOTION_INVENTORY_DB_ID=your_inventory_database_id
NOTION_SHOPPING_DB_ID=your_shopping_database_id  # NEW! For shopping lists
OPENAI_API_KEY=your_openai_key  # for LangGraph agent

# MCP Server Configuration (for FastMCP implementation)
MCP_TRANSPORT=stdio
MCP_SERVER_NAME=FoodWise

Verify setup:

# Discover your Notion databases
python scripts/list_databases.py

# Verify schema compatibility  
python scripts/verify_schema.py

Inventory demo (uses TEST DB if provided, auto-cleanup)

python scripts/inventory_demo.py

Shopping list demo (uses TEST DB if provided)

python scripts/shopping_list_demo.py ```

Quick Start

Test the Notion Integration

# Activate virtual environment
source .venv/bin/activate

# Run database verification
python scripts/verify_schema.py

# Test inventory CRUD (Create, Read, Update, Delete)
python scripts/inventory_demo.py

# Run comprehensive unit tests
python -m pytest tests/ -v

Use the LangGraph Agent

from src.foodwise.agent.agent import run_agent_query

# Query your inventory
response = run_agent_query("What items are expiring soon?")
print(response)

# Add new items
response = run_agent_query("Add 2 lbs of ground beef to the refrigerator")
print(response)

# Update existing items
response = run_agent_query("Update the milk quantity to 1 gallon")
print(response)

# Search for specific items
response = run_agent_query("Search for apples in my inventory")
print(response)

Test the Agent Locally

# Quick agent functionality test
python scripts/agent_smoke.py

Use with Claude Desktop (MCP) ✅ WORKING

Local MCP Server

# The FastMCP server runs automatically via Claude Desktop configuration
# Just update ~/.config/claude/claude_desktop_config.json with:

{
  "mcpServers": {
    "foodwise": {
      "command": "/path/to/FoodWise/.venv/bin/python",
      "args": ["/path/to/FoodWise/mcp_server_entry.py"],
      "cwd": "/path/to/FoodWise"
    }
  }
}

Remote MCP Server ✅ NEW!

# Access the remote FoodWise MCP server deployed on HuggingFace Spaces
# Add this to your ~/.config/claude/claude_desktop_config.json:

{
  "mcpServers": {
    "foodwise-remote": {
      "command": "npx",
      "args": ["supergateway", "--streamableHttp", "https://leowalker-foodwise-remote-mcp.hf.space/mcp/"]
    }
  }
}

🌐 Remote Access: The FoodWise MCP server is now available as a cloud service! Use the remote server for:

✅ Access from multiple devices
✅ No local setup required
✅ Always up-to-date
✅ Shared inventory access (when using shared Notion databases)

💡 Tip: You can have both foodwise (local) and foodwise-remote configured simultaneously for maximum flexibility!

# In Claude Desktop, you'll now have access to:
# 🔧 Tools: 12 total (5 inventory + 7 shopping management)
# 📂 Resources: 4 focused (inventory items, expiring items, shopping list, by-store)
# 📝 Prompts: 5 concise (add_item_helper, meal_planning_assistant, expiration_manager,
#                        shopping_list_optimizer, smart_shopping_assistant)

✅ Production Ready: Complete MCP integration tested and working with Claude Desktop. Both local and remote servers automatically load environment configurations and provide full functionality.

Development Status

🎯 Current Status: ✅ STREAMLINED! Clean, focused MCP server optimized for learning and production use! ✅ Completed Features:

Complete Notion integration with full CRUD (Create, Read, Update, Delete) for inventory & shopping
LangGraph React agent with 5 inventory management tools
Streamlined shopping list management system 🛒 OPTIMIZED!
Store-specific optimization for Costco, Trader Joe's, Safeway, Whole Foods, Target
Pydantic validation for data integrity and partial updates
Comprehensive testing suite (manual + automated)
Schema compatibility verification
FastMCP server with secure environment handling
Claude Desktop integration tested and working
Clean, focused MCP architecture (42% code reduction!)
🌐 Remote MCP Server deployed on HuggingFace Spaces ✅ NEW!
Docker containerization with uv and Python 3.12

🔧 Available Tools (Working in Claude Desktop):

Inventory Management:

get_inventory - Query inventory with filtering options ✅ Tested
add_inventory_item - Add new food items with full field support
search_inventory_items - Search for specific items by name/category
update_inventory_item - Update existing items with validation ✅ Tested
remove_inventory_item - Archive/remove items from inventory ✅ Tested

Shopping List Management: 🛒 STREAMLINED!

get_shopping_list - Retrieve shopping items with store/priority filtering
add_shopping_item - Add items with store preferences and pricing
update_shopping_item - Modify items and track purchase status
remove_shopping_item - Archive completed shopping items
create_shopping_from_recipe - Generate shopping lists from recipes
optimize_shopping_by_store - Organize by store for efficient shopping
move_purchased_to_inventory - Auto-move purchased items to inventory

📋 Streamlined MCP Architecture:

✅ FastMCP server with 12 essential tools (5 inventory + 7 shopping)
✅ 4 focused resources for live data (2 inventory + 2 shopping)
✅ 5 concise prompts for key workflows (mixed domains)
✅ 42% code reduction while maintaining full functionality
✅ Claude Desktop integration working and tested
✅ Secure environment variable handling (no secrets in config files)
✅ Full inventory + shopping management through conversational AI
✅ Store-specific optimization (Costco, Trader Joe's, Safeway, Whole Foods, Target)
✅ Clean, learning-focused design - perfect for understanding MCP patterns

🚀 Ready for:

Enhanced documentation and deployment guides
Additional MCP client integrations (OpenAI, LiteLLM, etc.)
Discord bot implementation (architecture ready)

Last updated: August 2025 - Remote MCP deployment optimized and simplified