langgraph-agent-patterns

To coordinate multiple agents in a LangGraph workflow, you use established multi-agent patterns like Supervisor-Subagent, Router, Orchestrator-Worker, and Handoffs. These patterns provide boilerplate templates to structure centralized coordination, parallel task execution, or sequential routing reliably.

The supervisor-subagent pattern uses centralized coordination where a supervisor delegates tasks, whereas the orchestrator-worker pattern focuses on parallel task execution. Both LangGraph multi-agent patterns provide distinct topologies for routing and executing tasks deterministically.

You validate a LangGraph multi-agent graph topology using graph validation utilities like `validate_agent_graph.py` to ensure correct structure and debug flows. Visualization tools like `visualize_graph.py` also help verify topology and state schemas.

Yes, you can build a router pattern for deterministic task routing in LangGraph. The Skill provides prebuilt boilerplate templates and asset examples to illustrate how to implement and validate deterministic routing within your multi-agent graphs.

Yes, the LangGraph multi-agent boilerplates support both Python and TypeScript. The provided templates for Supervisor-Subagent, Router, Orchestrator-Worker, and Handoff patterns include code examples in both languages to facilitate faster setup.

Use a handoff pattern for sequential task execution where one agent passes control directly to another, whereas an orchestrator-worker graph suits parallel task execution. The provided boilerplates help validate which topology fits your workflow.