🛒 ClaudeCart

ClaudeCart is a multimodal, agent-based retail assistant that uses Claude + MCP to provide intelligent shopping assistance. The application routes user queries between SQL-based inventory lookup and RAG-based product discovery. It runs as a Streamlit Cloud application with product information management capabilities.

Features

• Intelligent Chat Interface: Conversational assistant powered by Claude AI models
• Price Matching: Scrape external product information and analyze competitor pricing
• Product Database: Sample product data with detailed specifications across categories
• Policy Information: Store policies for returns, warranty, shipping, and price matching

Tech Stack

• Frontend: Streamlit for interactive UI
• AI: Anthropic's Claude models via their Python SDK
• Web Scraping: Firecrawl for product information retrieval
• Search: Tavily for competitive price searching
• Observability: Arize and OpenInference for tracing

Setup and Run

# Development environment setup
./dev.sh

# Or manually:
uv venv
uv pip install -e .
source .venv/bin/activate
streamlit run app.py

Project Structure

• /app.py - Main Streamlit application entry point
• /src/claudecart/ - Core application code
  • /backend/ - Claude controller logic
  • /database/ - Database managers (SQLite and vector store)
  • /mcp_tools/ - Model control protocol tools for Claude
  • /utils/ - Utility functions and external API clients
• /data/ - Sample product and policy data
• /mcp_schemas/ - JSON schemas for MCP tools

To-Do

• Database Implementation:
  • Complete SQLite manager for product database
  • Implement vector store for semantic search
• MCP Tool Enhancements:
  • Implement inventory tools
  • Complete search tool functionality
  • Fix the Tavily client import in search_tools.py
• Claude Controller Updates:
  • Add full MCP tool registration and routing
  • Implement structured tool responses
• Additional Features:
  • User account functionality
  • Shopping cart persistence
  • Order history tracking
  • Product recommendations
  • Complete price matching implementation