VertexAiSearchTool is not able to access the datastore vertexaisearch which i have created in the Gemini Enterprise

[Hetul-Patel-itp]

Hey everyone 👋

I'm trying to use VertexAiSearchTool with an LlmAgent to query a SharePoint Online federated connector datastore, but it’s not working properly.

The agent simply returns an empty response with no errors, which makes debugging quite difficult since it's a silent failure.

Datastore path:

projects/{project}/locations/global/collections/default_collection/dataStores/{id}-sharepoint-federated_{id}_file

Environment setup:

GOOGLE_CLOUD_LOCATION = us-central1
GOOGLE_GENAI_USE_VERTEXAI = True
Model: gemini-2.0-flash

---

Observed behavior:

When running the agent getting this below output.

--- Running Vertex AI Search Agent ---
Query: Receiving the company asset (Laptop)
Agent Response: ```

--- Full Event Details ---
model_version='gemini-2.0-flash' content=Content(
  parts=[
    Part(
      text="""```
"""
    ),
  ],
  role='model'
) grounding_metadata=GroundingMetadata(
  retrieval_metadata=RetrievalMetadata(),
  retrieval_queries=[                                                                                                 pted=None custom_metadata=None usage_metadata=GenerateContentResponseUsageMetadata(
  candidates_token_count=2,
      modality=<MediaModality.TEXT: 'TEXT'>,
      modality=<MediaModality.TEXT: 'TEXT'>,
      token_count=82
    ),
  ],
  total_token_count=84,
  traffic_type=<TrafficType.ON_DEMAND: 'ON_DEMAND'>
) live_session_resumption_update=None input_transcription=None output_transcription=None avg_logprobs=None logprobs_result=None cache_metadata=None citation_metadata=None interaction_id=None invocation_id='e-634215e7-5ea8-4ea3-a77e-228ed0aaa346' author='docs_qa_agent' actions=EventActions(skip_summarization=None, state_delta={}, artifact_delta={}, transfer_to_agent=None, escalate=None, requested_auth_configs={}, requested_tool_confirmations={}, compaction=None, end_of_agent=None, agent_state=None, rewind_before_invocation_id=None) long_running_tool_ids=None branch=None id='6c185527-0bf6-4ead-a1eb-c30696b09bca' timestamp=1774464897.586301

------------------------------
   

**Questions:** 

1. Does **VertexAiSearchTool** support federated connectors like SharePoint and OneDrive? 
2. Could this behavior indicate a configuration issue, or is this a known limitation? 
3. Is there any additional authentication required for a federated SharePoint connector? 

Any guidance would be really helpful. Thanks!

[adk-bot]

Response from ADK Answering Agent (experimental, answer may be inaccurate)

TLDR: While the documentation doesn't explicitly mention federated connectors, VertexAiSearchTool is designed to query any Vertex AI Search datastore. The silent failure you're experiencing is likely due to a configuration or authentication issue, not a fundamental limitation of the tool.

---

Hello! Thanks for reaching out with your question. Let's break down the possible issues.

1. Support for Federated Connectors

VertexAiSearchTool is built to query data that has been indexed into a Vertex AI Search datastore. The tool itself is somewhat agnostic to the original source of the data (be it a website, unstructured documents, or a federated source like SharePoint). As long as your SharePoint data is successfully indexed and searchable within your Vertex AI Search datastore in the Google Cloud console, the VertexAiSearchTool should be able to query it. [1]

2. Configuration and Authentication

The empty response with no errors strongly suggests an authentication or configuration problem. Here are a few things to verify:

• Datastore Health: Before debugging the agent, go to your datastore in the Google Cloud Console for Vertex AI Search and use the "Preview" or "Test" feature to run a query directly. If you get no results there, the issue is with the datastore setup and its connection to SharePoint, not the ADK tool.

• Authentication: The environment running your ADK agent must be properly authenticated to Google Cloud. The documentation for VertexAiSearchTool states that you should authenticate your environment, for example, by running gcloud auth login. [1]

• IAM Permissions: The principal (your user account or the service account) running the agent code needs the correct IAM roles to access Vertex AI Search. Ensure it has at least roles/aiplatform.user and roles/discoveryengine.viewer or a more specific role that allows reading from your datastore.

• Environment Variables: Double-check that your .env file or environment variables (GOOGLE_CLOUD_PROJECT, GOOGLE_CLOUD_LOCATION, GOOGLE_GENAI_USE_VERTEXAI) are correctly set and accessible to the script.

To summarize, the problem is less likely to be a limitation of VertexAiSearchTool with federated connectors and more likely an issue in the chain of authentication between your agent, Vertex AI, and the datastore itself. Verifying that the datastore works in the Cloud Console is the best first step.

---

[1] https://google.github.io/adk-docs/grounding/vertex_ai_search_grounding/