How to Connect MCP Servers to Google ADK

What is Google ADK?

Google Agent Development Kit (ADK) is Google's framework for building AI agents that can use tools, maintain state, and orchestrate multi-step workflows. It supports MCP natively, meaning ADK agents can connect to any MCP server endpoint for tool access.

Why connect MCP servers to ADK?

ADK agents are powerful at planning and execution, but they need tools to interact with external systems. Connecting MCP servers gives your ADK agent access to CRMs, dashboards, ticketing systems, and any web app without writing custom tool code.

Prerequisites

• Python 3.10+ with google-adk installed
• A DataFaucet account (free tier works)
• An MCP server endpoint (we'll create one below)

Step 1: Create an MCP server with DataFaucet

1. Go to DataFaucet
2. Enter the URL of the app you want your agent to access (e.g., your project management tool)
3. Browse the app while DataFaucet captures the API surface
4. Click Deploy — you'll get a hosted MCP endpoint URL

Your endpoint looks like: https://datafaucet.dev/mcp/your-server-id

Step 2: Configure ADK to use the MCP server

In your ADK agent configuration, add the MCP server as a tool source:

from google.adk import Agent
from google.adk.tools import McpToolset

# Connect to your DataFaucet MCP server
mcp_tools = McpToolset(
    endpoint="https://datafaucet.dev/mcp/your-server-id",
    transport="sse"
)

agent = Agent(
    model="gemini-2.0-flash",
    tools=[mcp_tools],
    instructions="You have access to project management tools. Use them to check ticket status and update tasks."
)

Step 3: Test the connection

Run your agent and ask it to use the connected tools:

response = agent.run("What tickets are assigned to me this sprint?")
print(response)

The agent discovers available tools from the MCP server automatically and calls them as needed.

Step 4: Add multiple MCP servers

ADK agents can connect to multiple MCP servers simultaneously. Each server provides a different set of tools:

crm_tools = McpToolset(endpoint="https://datafaucet.dev/mcp/crm-server")
analytics_tools = McpToolset(endpoint="https://datafaucet.dev/mcp/analytics-server")
slack_tools = McpToolset(endpoint="https://datafaucet.dev/mcp/slack-server")

agent = Agent(
    model="gemini-2.0-flash",
    tools=[crm_tools, analytics_tools, slack_tools],
    instructions="You can access CRM data, analytics dashboards, and Slack channels."
)

Tips

• Start with one server and verify the agent can discover and call tools before adding more
• Use descriptive instructions so the agent knows when to use which tools
• SSE transport is recommended for hosted MCP servers (vs stdio for local)
• DataFaucet handles auth so your agent config stays simple

What tools can you connect?

Any web app DataFaucet can scan becomes available to your ADK agent:

• Project management (Linear, Jira, Asana)
• CRMs (HubSpot, Salesforce)
• Analytics (Amplitude, Mixpanel, Google Analytics)
• Communication (Slack, Discord)
• Databases and dashboards (Metabase, Grafana)

Next steps

Create your first three MCP servers free at DataFaucet, then connect them to your ADK agent using the config above.

For more on Google ADK vs DataFaucet, see our comparison page.