Skip to main content

Google ADK

Build autonomous data agents with Google ADK that are grounded in your enterprise data context from DataHub — ownership, lineage, documentation, quality signals, and more.

The integration works two ways:

  • Python tools — embed DataHub tools directly in your ADK agent
  • MCP server — connect via ADK's built-in McpToolset

Prerequisites​

  • Python 3.10+
  • Google ADK (pip install google-adk)
  • A DataHub instance and access token
  • A Google API key (Gemini Developer API) or Google Cloud credentials (Vertex AI)

Installation​

pip install datahub-agent-context[google-adk]

Quick Start​

from datahub.sdk.main_client import DataHubClient
from datahub_agent_context.google_adk_tools import build_google_adk_tools

client = DataHubClient.from_env()

# Read-only tools by default; set include_mutations=True for write operations
tools = build_google_adk_tools(client, include_mutations=False)

Wire the tools into an ADK Agent:

from google.adk.agents import Agent

agent = Agent(
    model="gemini-2.5-flash",
    name="datahub_agent",
    description="A data discovery assistant with access to DataHub.",
    instruction="Use the available tools to search for datasets, get entity details, and trace lineage. Always include URNs in your answers.",
    tools=tools,
)

Full working examples: basic_agent.py · simple_search.py

Connecting via MCP Server​

Instead of embedding tools, you can connect ADK to the DataHub MCP server using McpToolset:

from google.adk.tools.mcp_tool import McpToolset
from google.adk.tools.mcp_tool.mcp_session_manager import StreamableHTTPConnectionParams

toolset = McpToolset(
    connection_params=StreamableHTTPConnectionParams(
        url="https://<tenant>.acryl.io/integrations/ai/mcp"
    ),
    headers={"Authorization": f"Bearer {YOUR_TOKEN}"},
)
# Initialize eagerly so the AsyncExitStack is owned in this task
await toolset.get_tools()

agent = Agent(
    model="gemini-2.5-flash",
    name="datahub_agent",
    instruction="You help users find datasets in DataHub.",
    tools=[toolset],
)

Call await toolset.close() when done.

Using Vertex AI Instead of Gemini Developer API​

Don't set GOOGLE_API_KEY — ADK falls back to Application Default Credentials automatically. Make sure you've run gcloud auth application-default login.

Troubleshooting​

  • Tool execution errors? Verify your DataHub connection (client.config) and token permissions.
  • Agent not using tools? Strengthen the instruction prompt, or try a model with better tool-calling (Gemini 2.0+).
  • AsyncExitStack / task errors with McpToolset? Call await toolset.get_tools() in the same async task that owns the toolset, and await toolset.close() in a finally block.
  • Import errors? Run pip install datahub-agent-context[google-adk] google-adk.

Links: Google ADK Docs · Agent Context Kit · MCP Server Guide