Google ADK Setup

Configure Browserbase MCP Server for Google Agent Development Kit

Overview

The Browserbase MCP Server can be integrated with Google ADK agents to provide browser automation capabilities. We support both local STDIO and hosted Streamable HTTP (SHTTP) transport methods.

We recommend using Streamable HTTP with our remote hosted URL to take advantage of the server at full capacity and avoid managing local processes.

Prerequisites

1. Get your Browserbase credentials

   Get your Browserbase API key from the Browserbase Dashboard.

   

   Then copy your API Key directly from the input.

2. (Optional) Get your Gemini API Key

   Get your Gemini API key from Google AI Studio for AI-powered browser automation with Stagehand. Only required if you’re using the local MCP server.

Setup Methods

Remote Hosted

Recommended: When using our remote hosted server, we provide the LLM costs for Gemini, the best performing model in Stagehand.

from google.adk.agents import Agent
from google.adk.tools.mcp_tool.mcp_session_manager import SseConnectionParams
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset

root_agent = Agent(
    model="gemini-2.5-pro",
    name="browserbase_agent",
    instruction="Help users get information from web pages using Browserbase",
    tools=[
        MCPToolset(
            connection_params=SseConnectionParams(
                url=(
                    "https://mcp.browserbase.com/mcp"
                    "?browserbaseApiKey=YOUR_BROWSERBASE_API_KEY"
                ),
                timeout=300,
            ),
        )
    ],
)

Local STDIO

For local development or when you need more control over the server configuration, use the STDIO transport method.

from google.adk.agents import Agent
from google.adk.tools.mcp_tool.mcp_session_manager import StdioConnectionParams
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset
from mcp import StdioServerParameters

BROWSERBASE_API_KEY = "YOUR_BROWSERBASE_API_KEY"
GEMINI_API_KEY = "YOUR_GEMINI_API_KEY"

root_agent = Agent(
    model="gemini-2.5-pro",
    name="browserbase_agent",
    instruction="Help users get information from web pages using Browserbase",
    tools=[
        MCPToolset(
            connection_params=StdioConnectionParams(
                server_params=StdioServerParameters(
                    command="npx",
                    args=[
                        "-y",
                        "@browserbasehq/mcp-server-browserbase",
                    ],
                    env={
                        "BROWSERBASE_API_KEY": BROWSERBASE_API_KEY,
                        "GEMINI_API_KEY": GEMINI_API_KEY,
                    }
                ),
                timeout=300,
            ),
        )
    ],
)

When using STDIO, you’ll need to provide your own Gemini API key and will incur LLM costs for Stagehand operations.

Optional Runtime Query Params

You can append these query parameters to the MCP URL:

• keepAlive=true|false
• proxies=true|false
• advancedStealth=true|false

Verify Installation

Test your integration by running your agent:

# Run your agent
result = root_agent.run("Navigate to google.com and take a screenshot")
print(result)

Monitor your browser sessions in real-time on the Browserbase Dashboard.

Available Tools

Once configured, your Google ADK agent will have access to these Browserbase tools:

• start: Create or reuse a Browserbase session
• end: Close the current Browserbase session
• navigate: Navigate to a URL
• act: Perform actions using natural language
• observe: Find actionable elements on a page
• extract: Extract data from a page

See MCP Setup for tool details.

Configuration Reference

For full query parameter behavior (required/optional fields, model behavior, and validation), see MCP Setup.

Additional Resources

• Google ADK Documentation
• ADK Quickstart Guide
• Stagehand