LangGraph Agents

The problem

Agent frameworks are easy to demo and hard to reason about. I wanted working implementations of each core pattern rather than a single tutorial pipeline.

What I built

Built a ReAct reasoning agent, a retrieval-augmented agent, and a drafting agent with human-in-the-loop revision, plus graph exercises covering state machines and conditional edges.

Impact

The agent patterns here are the same ones running in NeuroDesk's retrieval orchestration.

Key features

• ReAct reasoning agent
• Retrieval-augmented (RAG) agent
• Drafting agent with human-in-the-loop revision
• State machine and conditional edge exercises