langgraph/how-tos/human-in-the-loop/

[giscus[bot]]

langgraph/how-tos/human-in-the-loop/

Build language agents as graphs

https://langchain-ai.github.io/langgraph/how-tos/human-in-the-loop/

[HiraveBapu]

Hello,

using "customizing the state" for Human-in-the-loop for following code

class State(TypedDict):
    messages: Annotated[list, add_messages]
    tool_call_message: Optional[AIMessage]


def call_model(state):
    messages = state["messages"]
    if messages[-1].content == "y":
        return {
            "messages": [state["tool_call_message"]],
            "tool_call_message": None,
        }
    else:
        response = model.invoke(messages)
        if response.tool_calls:
            verification_message = generate_verification_message(response)
            response.id = str(uuid.uuid4())
            return {
                "messages": [verification_message],
                "tool_call_message": response,
            }
        else:
            return {
                "messages": [response],
                "tool_call_message": None,
            }

it always interrupt for every tool call regardless that node/tool was configured.

[HiraveBapu]

also, can you show example on how to handle this over LangServe / api for client side interactions on web etc