langgraph-human-in-the-loop

Three things are required for interrupts to work: interrupt(value) pauses the graph. The value surfaces in the result under __interrupt__ . Command(resume=value) resumes — the resume value becomes the return value of interrupt() . Critical : when the graph resumes, the node restarts from the beginning — all code before interrupt() re-runs. class State(TypedDict): approved: bool def approval_node(state: State): # Pause and ask for approval approved = interrupt("Do you approve this action?") # When resumed, Command(resume=...) returns that value here return {"approved": approved} checkpointer = InMemorySaver() graph = ( StateGraph(State) .add_node("approval", approval_node) .add_edge(START, "approval") .add_edge("approval", END) .compile(checkpointer=checkpointer) ) config = {"configurable": {"thread_id": "thread-1"}} result = graph.invoke({"approved": False}, config) print(result[" interrupt "]) result = graph.invoke(Command(resume=True), config) print(result["approved"]) # True A common pattern: interrupt to show a draft, then route based on the human's decision. class EmailAgentState(TypedDict): email_content: str draft_response: str classification: dict def human_review(state: EmailAgentState) -> Command[Literal["send_reply", " end "]]: """Pause for human review using interrupt and route based on decision.""" classification = state.get("classification", {}) Use interrupt()...