Deployed Agent Engine crashes with TypeError: Object of type AuthConfig is not JSON serializable when using any OAuth 2.0 enabled tool.

[ThibaultScalbert]

Describe the bug
When an agent using any tool that requires OAuth 2.0 (e.g., OpenAPIToolset, BigQueryToolset) is deployed to Vertex AI Agent Engine, it crashes during the initial tool call.

To Reproduce

• Error Log:
  1 TypeError: Object of type AuthConfig is not JSON serializable
• Traceback Analysis: The error occurs within the VertexAiSessionService when it attempts to append the OAuth request event to the session history. The service fails because it cannot serialize the AuthConfig object into JSON.
• Steps to Reproduce:
  1. Create an agent that uses any tool requiring OAuth 2.0 (e.g., BigQueryToolset with a client_id and client_secret).
  2. Deploy the agent to Vertex AI Agent Engine using vertexai.agent_engines.create().
  3. Query the deployed agent with a prompt that triggers the OAuth-enabled tool.
  4. Observe the FAILED_PRECONDITION error on the client and the TypeError in the server-side Cloud Logs.

[fufufukakaka]

same situation.

[fufufukakaka]

recreate agent engine resource → work

p.s. → update agent engine resource → the issue happens again

[bashimr]

I face the same issue. The agent is deployed on Cloud Run and is trying to use Vertex AI Session service. The crash happens with the error:

TypeError: Object of type AuthConfig is not JSON serializable

[ryan-szns]

Same here, using BigQueryToolset on an Agent Engine instance

[mertbozkir]

Hey, any update on this? because I'm deploying a simple agent to agent_engine, and it throws the same issue. After created, if I try to create a session.

using this code.

# Initialize client
client = vertexai.Client(
    project="**",
    location="europe-west1"
)

adk_app = client.agent_engines.get(name="projects/***/locations/europe-west1/reasoningEngines/**")

print(adk_app)
# shows up

adk_app.operation_schemas()
# shows up 

session = await adk_app.async_create_session(user_id="USER_ID")
######  this throws the error!  ###### 

ClientError: 400 FAILED_PRECONDITION. {'error': {'code': 400, 'message': 'Reasoning Engine Execution failed.\nPlease refer to our documentation (https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/troubleshooting/use) for checking logs and other troubleshooting tips.\nError Details: {"detail":"Agent Engine Error: (\\"FastAPI could not JSON-encode the response from invocation method %s. Error: %s. Invocation method\'s original response: %r\\", \'async_create_session\', ValueError([TypeError(\\"\'coroutine\' object is not iterable\\"), TypeError(\'vars() argument must have __dict__ attribute\')]), <coroutine object AdkApp.async_create_session at 0x7f90ef6ddc60>)"}', 'status': 'FAILED_PRECONDITION'}}