Skip to main content
Glama
Browserbase MCP Server

Browserbase MCP Server

by MesuterPikin

Browserbase MCP Server

cover

The Model Context Protocol (MCP) is an open protocol that enables seamless integration between LLM applications and external data sources and tools. Whether you're building an AI-powered IDE, enhancing a chat interface, or creating custom AI workflows, MCP provides a standardized way to connect LLMs with the context they need.

This server provides cloud browser automation capabilities using Browserbase and Stagehand. It enables LLMs to interact with web pages, take screenshots, extract information, and perform automated actions with atomic precision.

What's New in Stagehand v3

Powered by Stagehand v3.0, this MCP server now includes:

  • 20-40% Faster Performance: Speed improvements across all core operations (act, extract, observe) through automatic caching

  • Enhanced Extraction: Targeted extraction and observation across iframes and shadow roots

  • Improved Schemas: Streamlined extract schemas for more intuitive data extraction

  • Advanced Selector Support: CSS selector support with improved element targeting

  • Multi-Browser Support: Compatible with Playwright, Puppeteer, and Patchright

  • New Primitives: Built-in page, locator, frameLocator, and deepLocator for simplified automation

  • Experimental Features: Enable cutting-edge capabilities with the --experimental flag

For more details, visit the Stagehand v3 documentation.

Related MCP server: Browserbase MCP Server

Features

Feature

Description

Browser Automation

Control and orchestrate cloud browsers via Browserbase

Data Extraction

Extract structured data from any webpage

Web Interaction

Navigate, click, and fill forms with ease

Screenshots

Capture full-page and element screenshots

Model Flexibility

Supports multiple models (OpenAI, Claude, Gemini, and more)

Vision Support

Use annotated screenshots for complex DOMs

Session Management

Create, manage, and close browser sessions

High Performance

20-40% faster operations with automatic caching (v3)

Advanced Selectors

Enhanced CSS selector support for precise element targeting

How to Setup

Quickstarts:

Add to Cursor

Copy and Paste this link in your Browser:

cursor://anysphere.cursor-deeplink/mcp/install?name=browserbase&config=eyJjb21tYW5kIjoibnB4IEBicm93c2VyYmFzZWhxL21jcCIsImVudiI6eyJCUk9XU0VSQkFTRV9BUElfS0VZIjoiIiwiQlJPV1NFUkJBU0VfUFJPSkVDVF9JRCI6IiIsIkdFTUlOSV9BUElfS0VZIjoiIn19

We currently support 2 transports for our MCP server, STDIO and SHTTP. We recommend you use SHTTP with our remote hosted url to take advantage of the server at full capacity.

SHTTP:

To use the Browserbase MCP Server through our remote hosted URL, add the following to your configuration.

Go to smithery.ai and enter your API keys and configuration to get a remote hosted URL. When using our remote hosted server, we provide the LLM costs for Gemini, the best performing model in Stagehand.

Smithery Image

If your client supports SHTTP:

{
  "mcpServers": {
    "browserbase": {
      "type": "http",
      "url": "your-smithery-url.com"
    }
  }
}

If your client doesn't support SHTTP:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": ["mcp-remote", "your-smithery-url.com"]
    }
  }
}

STDIO:

You can either use our Server hosted on NPM or run it completely locally by cloning this repo.

❗️ Important: If you want to use a different model you have to add --modelName to the args and provide that respective key as an arg. More info below.

Go into your MCP Config JSON and add the Browserbase Server:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": ["@browserbasehq/mcp-server-browserbase"],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

That's it! Reload your MCP client and Claude will be able to use Browserbase.

To run 100% local:

Option 1: Direct installation

# Clone the Repo
git clone https://github.com/browserbase/mcp-server-browserbase.git
cd mcp-server-browserbase

# Install the dependencies and build the project
npm install && npm run build

Option 2: Docker

# Clone the Repo
git clone https://github.com/browserbase/mcp-server-browserbase.git
cd mcp-server-browserbase

# Build the Docker image
docker build -t mcp-browserbase .

Then in your MCP Config JSON run the server. To run locally we can use STDIO or self-host SHTTP.

STDIO:

Using Direct Installation

To your MCP Config JSON file add the following:

{
  "mcpServers": {
    "browserbase": {
      "command": "node",
      "args": ["/path/to/mcp-server-browserbase/cli.js"],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Using Docker

To your MCP Config JSON file add the following:

{
  "mcpServers": {
    "browserbase": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e",
        "BROWSERBASE_API_KEY",
        "-e",
        "BROWSERBASE_PROJECT_ID",
        "-e",
        "GEMINI_API_KEY",
        "mcp-browserbase"
      ],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Then reload your MCP client and you should be good to go!

Configuration

The Browserbase MCP server accepts the following command-line flags:

Flag

Description

--proxies

Enable Browserbase proxies for the session

--advancedStealth

Enable Browserbase Advanced Stealth (Only for Scale Plan Users)

--keepAlive

Enable Browserbase Keep Alive Session

--contextId <contextId>

Specify a Browserbase Context ID to use

--persist

Whether to persist the Browserbase context (default: true)

--port <port>

Port to listen on for HTTP/SHTTP transport

--host <host>

Host to bind server to (default: localhost, use 0.0.0.0 for all interfaces)

--browserWidth <width>

Browser viewport width (default: 1024)

--browserHeight <height>

Browser viewport height (default: 768)

--modelName <model>

The model to use for Stagehand (default: gemini-2.0-flash)

--modelApiKey <key>

API key for the custom model provider (required when using custom models)

--experimental

Enable experimental features (default: false)

These flags can be passed directly to the CLI or configured in your MCP configuration file.

NOTE:

Currently, these flags can only be used with the local server (npx @browserbasehq/mcp-server-browserbase or Docker).

Using Configuration Flags with Docker

When using Docker, you can pass configuration flags as additional arguments after the image name. Here's an example with the --proxies flag:

{
  "mcpServers": {
    "browserbase": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e",
        "BROWSERBASE_API_KEY",
        "-e",
        "BROWSERBASE_PROJECT_ID",
        "-e",
        "GEMINI_API_KEY",
        "mcp-browserbase",
        "--proxies"
      ],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

You can also run the Docker container directly from the command line:

docker run --rm -i \
  -e BROWSERBASE_API_KEY=your_api_key \
  -e BROWSERBASE_PROJECT_ID=your_project_id \
  -e GEMINI_API_KEY=your_gemini_key \
  mcp-browserbase --proxies

Configuration Examples

Proxies

Here are our docs on Proxies.

To use proxies, set the --proxies flag in your MCP Config:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": ["@browserbasehq/mcp-server-browserbase", "--proxies"],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Advanced Stealth

Here are our docs on Advanced Stealth.

To use advanced stealth, set the --advancedStealth flag in your MCP Config:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": ["@browserbasehq/mcp-server-browserbase", "--advancedStealth"],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Contexts

Here are our docs on Contexts

To use contexts, set the --contextId flag in your MCP Config:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": [
        "@browserbasehq/mcp-server-browserbase",
        "--contextId",
        "<YOUR_CONTEXT_ID>"
      ],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Browser Viewport Sizing

The default viewport sizing for a browser session is 1024 x 768. You can adjust the Browser viewport sizing with browserWidth and browserHeight flags.

Here's how to use it for custom browser sizing. We recommend to stick with 16:9 aspect ratios (ie: 1920 x 1080, 1280 x 720, 1024 x 768)

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": [
        "@browserbasehq/mcp-server-browserbase",
        "--browserHeight 1080",
        "--browserWidth 1920"
      ],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Experimental Features

Stagehand v3 includes experimental features that can be enabled with the --experimental flag. These features provide cutting-edge capabilities that are actively being developed and refined.

To enable experimental features:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": ["@browserbasehq/mcp-server-browserbase", "--experimental"],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Note: Experimental features may change or be removed in future releases. Use them at your own discretion.

Model Configuration

Stagehand defaults to using Google's Gemini 2.0 Flash model, but you can configure it to use other models like GPT-4o, Claude, or other providers.

Important: When using any custom model (non-default), you must provide your own API key for that model provider using the --modelApiKey flag.

Here's how to configure different models:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": [
        "@browserbasehq/mcp-server-browserbase",
        "--modelName",
        "anthropic/claude-sonnet-4.5",
        "--modelApiKey",
        "your-anthropic-api-key"
      ],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": ""
      }
    }
  }
}

Note: The model must be supported in Stagehand. Check out the docs here. When using any custom model, you must provide your own API key for that provider.

Resources

The server provides access to screenshot resources:

  1. Screenshots (screenshot://<screenshot-name>)

    • PNG images of captured screenshots

Key Features

  • AI-Powered Automation: Natural language commands for web interactions

  • Multi-Model Support: Works with OpenAI, Claude, Gemini, and more

  • Screenshot Capture: Full-page and element-specific screenshots

  • Data Extraction: Intelligent content extraction from web pages

  • Proxy Support: Enterprise-grade proxy capabilities

  • Stealth Mode: Advanced anti-detection features

  • Context Persistence: Maintain authentication and state across sessions

For more information about the Model Context Protocol, visit:

For the official MCP Docs:

License

Licensed under the Apache 2.0 License.

Copyright 2025 Browserbase, Inc.

Available Tools

9 tools
browserbase_screenshotC

Capture a full-page screenshot and return it (and save as a resource).

Inputs: name.

browserbase_session_closeA

Close the current Browserbase session and reset the active context.

Inputs: no named parameters.

browserbase_session_createC

Create or reuse a Browserbase browser session and set it as active.

Inputs: sessionId.

browserbase_stagehand_actB

Perform a single action on the page (e.g., click, type).

Inputs: action (required), variables.

browserbase_stagehand_agentB

Execute a task autonomously using Gemini Computer Use agent. The agent will navigate and interact with web pages to complete the given task.

Inputs: prompt (required).

browserbase_stagehand_extractC

Extract structured data or text from the current page using an instruction.

Inputs: instruction (required).

browserbase_stagehand_get_urlA

Return the current page URL (full URL with query/fragment).

Inputs: no named parameters.

browserbase_stagehand_navigateA

Navigate to a URL in the browser. Only use this tool with URLs you're confident will work and be up to date. Otherwise, use https://google.com as the starting point

Inputs: url (required).

browserbase_stagehand_observeB

Find interactive elements on the page from an instruction; optionally return an action.

Inputs: instruction (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.5/5.0

    Scored across 9 tools

    Disambiguation4/5

    Most tools have distinct purposes, such as session management (create/close), navigation (navigate/get_url), and interaction (screenshot/act/extract/observe). However, 'browserbase_stagehand_agent' and 'browserbase_stagehand_observe' could cause some confusion as both involve autonomous or guided interaction, though their descriptions help differentiate them.

    Naming Consistency4/5

    The naming follows a consistent pattern with 'browserbase_' prefix and snake_case throughout. Most tools use descriptive verb_noun combinations, but there is a minor deviation with 'browserbase_stagehand_agent' which uses 'agent' as a noun instead of a verb, slightly breaking the pattern.

    Tool Count5/5

    With 9 tools, the count is well-scoped for a browser automation server, covering essential operations like session handling, navigation, interaction, and data extraction. Each tool serves a clear purpose without redundancy, making the set appropriately sized for the domain.

    Completeness4/5

    The tool surface provides comprehensive coverage for browser automation, including session lifecycle, navigation, interaction, and data extraction. A minor gap exists in lacking explicit tools for handling browser contexts like tabs or windows, but core workflows are well-supported and agents can work around this.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers