Is the ensure_config() function work?

[xiaochaohit]

Python version: 3.12
langchain-core:0.2.33
langgraph:0.2.4
llm: AzureChatOpenAI

Hi,I encountered a problem
When I was testing with the langgraph chatbot tutorial(link：https://langchain-ai.github.io/langgraph/tutorials/customer-support/customer-support/ ).I can't get the runnableconfig using ensure_config function.
The tutorial code is

@tool
def fetch_user_flight_information() -> list[dict]:
    """Fetch all tickets for the user along with corresponding flight information and seat assignments.

    Returns:
        A list of dictionaries where each dictionary contains the ticket details,
        associated flight details, and the seat assignments for each ticket belonging to the user.
    """
    config = ensure_config()  # Fetch from the context
    configuration = config.get("configurable", {})
    passenger_id = configuration.get("passenger_id", None)
    if not passenger_id:
        raise ValueError("No passenger ID configured.")

    conn = sqlite3.connect(db)
    cursor = conn.cursor()

    query = """
    ...

when I run the case, I get the result like this:

================================ Human Message =================================

Hi there, what time is my flight?
================================== Ai Message ==================================
Tool Calls:
  fetch_user_flight_information (call_dMELZC2UM7gWMatoPwTZQwpx)
 Call ID: call_dMELZC2UM7gWMatoPwTZQwpx
  Args:
================================= Tool Message =================================
Name: fetch_user_flight_information

Error: ValueError('No passenger ID configured.')
 Please fix your mistakes.
================================== Ai Message ==================================

It looks like there's been an issue retrieving your flight information, as there seems to be a missing passenger ID for your query. Could you provide me with your booking reference or flight details to assist you further?

I print the return of ensure_config function like this, the configurable is empty

{'tags': [], 'metadata': {}, 'callbacks': <langchain_core.callbacks.manager.CallbackManager object at 0x7f94fd682630>, 'recursion_limit': 25, 'configurable': {}}

I just used a different llm model than the tutorial, I use AzureChatOpenAI not ChatAnthropic.
How do I get the RunnableConfig in the tool function？

[levi-katarok]

I'm facing the same issue, it was working a couple weeks ago and now my configurable object is always empty.

config = ensure_config()
configuration = config.get("configurable", {})

[vbarda]

this is not the right way to pass config into the tool -- you should just have config as one of the arguments in the tool function, see here https://langchain-ai.github.io/langgraph/how-tos/pass-config-to-tools/#define-tools-and-model

[levi-katarok]

I've tested it using that method as well, unfortunately it still provides an empty config.

[vbarda]

can you provide a minimal reproducible example where passing config directly is failing?

[levi-katarok]

Hi Vadym,

As I was working through this in a existing project I built this (mostly) minimal example to demonstrate the config not passing.
https://github.com/levi-katarok/langgraph-tools-config/blob/main/minimal-tool-config-case.ipynb

It was consistently failing to pass config to the tool. However when I pulled it out and put it in a new repo to post here and freshly installed all the packages it worked. Returning back to my main project and upgrading the packages it worked.

pip install -U langgraph langchain-core typing-extensions langchain-openai

Hope this helps someone else!

[vbarda]

Glad it worked!