[bug] Langgraph error AssertionError: expected list, found <class 'NoneType'>

[ionmincu]

Describe the bug
Error logs in agent.

Traceback (most recent call last):
  File "/tmp/company-research-agent/0.0.3/.venv/lib/python3.11/site-packages/openinference/instrumentation/langchain/_tracer.py", line 327, in wrapper
    yield from wrapped(*args, **kwargs)
  File "/tmp/company-research-agent/0.0.3/.venv/lib/python3.11/site-packages/openinference/instrumentation/langchain/_tracer.py", line 577, in _extract_message_role
    assert isinstance(id_, List), f"expected list, found {type(id_)}"
           ^^^^^^^^^^^^^^^^^^^^^
AssertionError: expected list, found <class 'NoneType'>

To Reproduce

Just use the create_agent method from langgraph and invoke the agent with any prompts.

from langchain.agents import create_agent

agent = create_agent(llm, tools=[search_tool], prompt=SYSTEM_PROMPT)

With the following dependencies.

langchain==1.1.0
langchain-classic==1.0.0
langchain-community==0.4.1
langchain-core==1.1.0
langchain-openai==1.1.0
langchain-text-splitters==1.0.0
langgraph==1.0.3
langgraph-checkpoint==3.0.1
langgraph-checkpoint-sqlite==3.0.0
langgraph-prebuilt==1.0.5
langgraph-sdk==0.2.10
openinference-instrumentation==0.1.42
openinference-instrumentation-langchain==0.1.55
openinference-semantic-conventions==0.1.25

[dosubot]

Hi @ionmincu! I'm Dosu and I’m helping the openinference team.

This is a known issue with OpenInference's LangChain instrumentation when used with LangGraph agents like create_react_agent and LangChain 1.x. The tracer expects every message to have an id field that is a list, but LangChain 1.x sometimes omits this field or sets it to None, causing the assertion to fail and resulting in the error you see (details, code reference).

Workaround:
Patch the _extract_message_role (or _parse_message_data) function to check if id_ is a list before using it, and handle cases where it's None or missing. For example, replace the assertion with:

id_ = message_data.get("id", [])
if not isinstance(id_, list) or not id_:
    # handle gracefully, e.g., skip or set a default role
    return

(example patch)

Alternatively, you can pin to a pre-1.0 version of LangChain, but patching is usually easier and avoids downgrading.

Current status:
The maintainers have acknowledged the issue, but no official fix has been released yet. The error does not prevent traces from being sent, but it does clutter logs and may affect trace completeness (issue thread). This can also happen with other LangGraph agents or custom workflows that feed dict-like messages without the expected serialization metadata.

Monitor for updates:
Keep an eye on future releases of openinference-instrumentation-langchain for a permanent fix.

Let me know if you need a more detailed patch or have questions about where to apply it!

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[caroger]

will be resolved by #2412