Agents

About

This repository serves as a personal laboratory for understanding the intricate workings of LLMs and AI agents from the ground up. It provides a modular framework to experiment with various agent architectures, featuring a universal web UI for seamless chat, streaming responses, and thread management. The project includes a LangGraph-based planning and execution agent enhanced with web search capabilities and Phoenix tracing for deep observability. Additionally, it offers a specialized MCP server designed for Claude Desktop to manage pantry inventory, complete with receipt image parsing, expiration date tracking, and full CRUD operations for grocery items.

Key Features

• Integrated web search and task management capabilities
• LangGraph-based planning and execution agent
• 0 GitHub stars
• Observability for agent execution with Arize Phoenix tracing
• Universal web UI for chat and agent interaction
• MCP server for pantry inventory management with receipt parsing

Use Cases

• Automating pantry inventory management through receipt processing and expiration tracking
• Developing and experimenting with custom AI agents and LLM applications
• Building intelligent planning and execution workflows with LangGraph