langgraph-graph-api

To execute a compiled LangGraph workflow, you can use invoke, stream, or batch execution patterns after compiling the graph from a StateGraph builder.

Yes, you can persist LangGraph state across executions by assigning a thread_id to your graph execution, which allows you to maintain context and resume long-running workflows.

The best way to visualize a LangGraph structure for debugging is by generating Mermaid diagrams, which allow you to inspect graph components and monitor complex workflows visually.

Yes, LangGraph supports asynchronous execution through async variants like ainvoke and astream, enabling you to run and stream complex graph workflows without blocking operations.

You can run multiple inputs through a compiled LangGraph simultaneously by using the batch execution pattern, which processes collections of inputs efficiently in a single operation.

You need to compile a LangGraph before invoking it because the compilation step validates the graph structure and prepares the StateGraph builder for execution patterns like invoke and stream.