Browserbase Mcp

Browserbase

Allow LLMs to control a browser with Browserbase and Stagehand for AI-powered web automation, intelligent data extraction, and screenshot capture.

When to use

• Use Browserbase only when the task directly involves the relevant service, SaaS product, platform, or technology.

Connection

Docker stdio

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

MCP instructions

• Use the MCP tools only for the user-requested service workflow and prefer read-only operations by default.
• Confirm the target account, workspace, project, repository, or dataset before actions that can read private data or mutate remote state.
• This server requires authorization or environment-provided credentials; ask the user before connecting or requesting access.

Docker launch notes

• Launch through Docker stdio with docker run --rm -i -e BROWSERBASE_API_KEY -e BROWSERBASE_PROJECT_ID -e GEMINI_API_KEY mcp/browserbase.
• Confirm required environment variables before launch: BROWSERBASEAPIKEY, BROWSERBASEPROJECTID, GEMINIAPIKEY.

References

• https://github.com/browserbase/mcp-server-browserbase
• https://github.com/docker/mcp-registry/tree/main/servers/browserbase

Security policy

• Authentication: Environment variables

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Friz-zy
• Source: Friz-zy/ai-capability-registry
• License: MIT

Install and usage instructions live in the source repository linked above.