Architecture
LangGraph
A graph-based framework for stateful, checkpointed agent workflows.
LangGraph is a framework for building stateful agent workflows with graph nodes, typed state, checkpointing, and interrupt patterns. It is useful for supervised workflows where an agent must pause for review.
Why it matters
Governed agents need more than a single prompt. They need state, recoverability, explicit nodes, and approval pauses.
How HARNEXA uses it
HARNEXA uses LangGraph patterns for PHANTOM and governed workflow orchestration.
FAQ
Questions this definition answers.
Why use a graph for agents?
A graph makes workflow state, control points, and approval boundaries explicit.
Does LangGraph provide governance by itself?
No. It provides workflow primitives; governance comes from the harness, policies, sensors, and audit model.
Need these definitions mapped to a live commerce workflow, agent estate, or protocol-readiness path?
Book diagnostic sprint