> ## Documentation Index
> Fetch the complete documentation index at: https://docs.platform.statista.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect to Google Gemini

> Connect the Statista MCP Server to Gemini via the google-genai SDK

<Info>
  Gemini supports MCP servers natively through the `google-genai` Python SDK — no FastMCP wrapper required. Google currently labels their MCP integration as **experimental**, so the exact API surface may evolve over time.
</Info>

The recommended way to connect Gemini to the Statista MCP server is to use the official [`mcp`](https://pypi.org/project/mcp/) Python package's streamable HTTP transport, then pass the resulting MCP session directly to Gemini as a tool.

<Steps>
  <Step title="Install dependencies" icon="python">
    ```sh theme={null}
    pip install google-genai mcp
    ```
  </Step>

  <Step title="Connect via the google-genai SDK" icon="key">
    Pass your Statista API key as the `x-api-key` header, then hand the MCP session directly to Gemini.

    ```python theme={null}
    from google import genai
    from mcp import ClientSession
    from mcp.client.streamable_http import streamablehttp_client

    mcp_server_url = "<PROVIDED_MCP_SERVER_URL>"
    YOUR_STATISTA_API_KEY = "<YOUR_STATISTA_API_KEY>"
    YOUR_GEMINI_API_KEY = "<YOUR_GEMINI_API_KEY>"

    async with streamablehttp_client(
        mcp_server_url,
        headers={"x-api-key": YOUR_STATISTA_API_KEY},
    ) as (read_stream, write_stream, _):
        async with ClientSession(read_stream, write_stream) as session:
            await session.initialize()

            client = genai.Client(api_key=YOUR_GEMINI_API_KEY)
            response = await client.aio.models.generate_content(
                model="gemini-3.5-flash",
                contents="What is the ice cream market in Japan?",
                config=genai.types.GenerateContentConfig(
                    temperature=0,
                    tools=[session],
                ),
            )
            print(response.text)
    ```

    The `ClientSession` object is passed directly into the `tools` list — Gemini introspects the MCP session's available tools and invokes them automatically as needed.
  </Step>
</Steps>

## Alternative: using FastMCP

If you're already using [`fastmcp`](https://pypi.org/project/fastmcp/) elsewhere in your stack, you can pass its session to Gemini the same way. This is functionally equivalent to the native pattern above, just with FastMCP's higher-level wrapper.

```python theme={null}
from google import genai
from fastmcp import Client
from fastmcp.client.transports import StreamableHttpTransport

mcp_client = Client(
    transport=StreamableHttpTransport(
        mcp_server_url,
        headers={"x-api-key": YOUR_STATISTA_API_KEY},
    ),
)

client = genai.Client(api_key=YOUR_GEMINI_API_KEY)

async with mcp_client:
    response = await client.aio.models.generate_content(
        model="gemini-3.5-flash",
        contents="What is the ice cream market in Japan?",
        config=genai.types.GenerateContentConfig(
            temperature=0,
            tools=[mcp_client.session],
        ),
    )
    print(response.text)
```