Hey HN community,
I've created an agentic AI system using a Model Context Protocol (MCP) server, based on my ongoing tutorial series:
https://github.com/zkzkGamal/Agentic-AI-Tutorial.
The system uses LangChain for orchestration and LangGraph for stateful workflows, with the MCP server handling tool integrations, data access, and real-time decision-making. It's designed for autonomy—agents can reason, plan, and execute tasks like API calls or database interactions without constant human input.
Chapter 5 dives into production deployment (still in progress, but this project is a practical case study). Open to feedback on improving security, scalability, or adding features like multi-agent collaboration.
What agentic setups are you working on?