ToolNode doesn't support state update

[hepengfe]

def call_tool(state):
    messages = state.messages
    # Based on the continue condition
    # we know the last message involves a function call
    last_message = messages[-1]
    # We construct an ToolInvocation from the function_call
    tool_call = last_message.tool_calls[0]
    action = ToolInvocation(
        tool=tool_call["name"],
        tool_input=tool_call["args"],
    )
    # We call the tool_executor and get back a response
    response = tool_executor.invoke(action)
    # We use the response to create a ToolMessage
    tool_message = ToolMessage(
        content=str(response), name=action.tool, tool_call_id=tool_call["id"]
    )
    # We return a list, because this will get added to the existing list
    return {"messages": [tool_message]}

In the old ToolMessage, it supports return customized state but with pre-built ToolNode, there is no way to update other state status by using tool returned values. It could be related to https://github.com/langchain-ai/langgraph/issues?q=is%3Aissue+is%3Aopen+toolnode if custom ToolNode is supported then the issue could be solved.

[hinthornw]

I'd probably use a custom node to execute tools if I wanted to make arbitrary state changes

[hepengfe]

I checked the source code of ToolNode, and it indicates it should be able to make changes on state. Correct me if i am wrong.

Also, if I customize node, does it mean I need to inherit the ToolNode or use old tool_executor and ToolInvocation?

[hwchase17]

_inject_state is used to inject state as input to the tool call, but not to update state based on output

[hepengfe]

Can you provide some examples or examples' pointers on how to customize node to update state?

[vbarda]

We have now added support for updating the state from the tools using the new Command type. Please see this how-to guide for reference https://langchain-ai.github.io/langgraph/how-tos/update-state-from-tools/. Let me know if you have any questions / feedback