Skip to main content
This tutorial runs a LangGraph plan→act agent under Noesis and captures the cognitive artifacts. Governance is off in this demo; the focus is traceability and verification. Why this matters: after one run you have a sealed evidence bundle you can diff and audit.

​
What you’ll build

  • A LangGraph-backed cognitive run (observe → plan → act → reflect → learn)
  • Workspace verification (verify) and immutable artifacts
  • A clean episode you can inspect with noesis view

​
Prerequisites

  • Python with noesis installed
  • LangGraph installed (uv add langgraph or pip install langgraph)
  • OpenAI API key (for the planner/actor)

​
Run the tutorial

uv run python -m tutorials.langgraph_episode

​
Source

The source file is located at examples/noesis-quickstart/tutorials/langgraph_episode.py.