Is Google ADK's LlmAgent not compatible with deploying to AgentEngine?

[paulrinaldi]

OS: Colab
ADK Version: 1.7.0

I successfully ran https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/gemini/agent-engine/intro_agent_engine.ipynb as is.

Then, I edited the example:
https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/gemini/agent-engine/intro_agent_engine.ipynb

To use an LlmAgent from google-adk

agent = Agent(  # changed from LangChainAgent to Agent (LlmAgent)
    model=model,
    name='exchange_rate_agent', # added
    description='An exchange rate workflow', # added
    tools=[get_exchange_rate],
)

and I get a permissions error on the agent_engines.create( line:

INFO:vertexai.agent_engines:Deploying google.adk.agents.Agent as an application.
INFO:vertexai.agent_engines:Identified the following requirements: {'pydantic': '2.11.7', 'cloudpickle': '3.0.0', 'google-cloud-aiplatform': '1.104.0'}
INFO:vertexai.agent_engines:The final list of requirements: ['google-cloud-aiplatform[agent_engines,google-adk]', 'cloudpickle==3.0.0', 'pydantic>=2.10', 'requests']
INFO:vertexai.agent_engines:Using bucket agentic-blue-grass-flower-1
INFO:vertexai.agent_engines:Wrote to gs://agentic-blue-grass-flower-1/agent_engine/agent_engine.pkl
INFO:vertexai.agent_engines:Writing to gs://agentic-blue-grass-flower-1/agent_engine/requirements.txt
INFO:vertexai.agent_engines:Creating in-memory tarfile of extra_packages
INFO:vertexai.agent_engines:Writing to gs://agentic-blue-grass-flower-1/agent_engine/dependencies.tar.gz
---------------------------------------------------------------------------
_InactiveRpcError                         Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/google/api_core/grpc_helpers.py in error_remapped_callable(*args, **kwargs)
     75         try:
---> 76             return callable_(*args, **kwargs)
     77         except grpc.RpcError as exc:

11 frames
_InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
	status = StatusCode.PERMISSION_DENIED
	details = "Permission denied on resource project PROJECT_ID."
	debug_error_string = "UNKNOWN:Error received from peer ipv4:REDACTED:443 {grpc_message:"Permission denied on resource project PROJECT_ID.", grpc_status:7}"
>

The above exception was the direct cause of the following exception:

PermissionDenied                          Traceback (most recent call last)
/usr/local/lib/python3.11/dist-packages/google/api_core/grpc_helpers.py in error_remapped_callable(*args, **kwargs)
     76             return callable_(*args, **kwargs)
     77         except grpc.RpcError as exc:
---> 78             raise exceptions.from_grpc_error(exc) from exc
     79 
     80     return error_remapped_callable

PermissionDenied: 403 Permission denied on resource project PROJECT_ID. [reason: "CONSUMER_INVALID"
domain: "googleapis.com"
metadata {
  key: "service"
  value: "aiplatform.googleapis.com"
}
metadata {
  key: "containerInfo"
  value: "PROJECT_ID"
}
metadata {
  key: "consumer"
  value: "projects/PROJECT_ID"
}
, locale: "en-US"
message: "Permission denied on resource project PROJECT_ID."
, links {
  description: "Google developers console"
  url: "https://console.developers.google.com"
}
]

Account's Roles: AI Platform Admin, Editor, Owner, Service Usage Admin, Service Usage Consumer, Vertex AI Administrator, Vertex AI User
Enabled APIs: Vertex AI Search for commerce API, Vertex AI API, Notebooks API, Compute Engine API, Dataform API, Cloud Run Admin API, BigQuery Migration API, Cloud Functions API, Gemini for Google Cloud API, AI Platform Training & Prediction API, Analytics Hub API, API Keys API, Artifact Registry API, BigQuery API, BigQuery Connection API, BigQuery Data Policy API, BigQuery Reservation API, BigQuery Storage API, Cloud Asset API, Cloud Build API, Cloud Dataplex API, Cloud Deployment Manager V2 API, Cloud Logging API, Cloud Monitoring API, Cloud OS Login API, Cloud Pub/Sub API, Cloud Resource Manager API, Cloud Spanner API, Cloud SQL Admin API, Cloud Storage, Cloud Storage API, Cloud Trace API, Container Registry API, Data Lineage API, Dataflow API, Dialogflow API, Discovery Engine API, Gemini Cloud Assist API, Generative Language API, Google Cloud Storage JSON API, IAM Service Account Credentials API, Identity and Access Management (IAM) API, Legacy Cloud Source Repositories API, Maps JavaScript API, Recommendations AI (Beta, Recommender API, Vision AI API, Web Security Scanner API

[ankursharmas]

Can you please ensure that your local Colab changes are in line with what is recommended here:

https://google.github.io/adk-docs/deploy/agent-engine/

At the very least i can see that the public colab doesn't have this:

requirements=[
        "google-cloud-aiplatform[adk,agent_engines]",
    ]

[paulrinaldi]

This worked! Thank you for pointing that out. I wonder if google could provide a better error in this situation.

I am trying to duplicate this work locally, and deploy via terminal but having the same error. (further context, I am running a python shim via rye)