MCP server

Browser Use MCP Server

An agent that drives a real Chrome to finish tasks: fill forms, click through flows, extract data. Official: built and maintained by Browser Use. It runs as a hosted remote server or locally. Setup below for 13 clients (checked 2026-10-25). Repo and registry data checked 2026-10-24.
Maintainer
Official · Browser Use
Runs
Remote + local
GitHub
★ 119,059, last push 2026-10-24 (checked 2026-10-24)
License
MIT
URL
https://api.browser-use.com/v3/mcp
Registry
com.browser-use/browser-use
Connects to
Chrome

Install the Browser Use MCP Server

Pick your client. Remote first: nothing to install, and you sign in with OAuth where the client supports it. Config shapes checked against each client's docs on 2026-10-25.

The URL also takes X-Browser-Use-API-Key as a header (see its docs), for clients without OAuth sign-in.

Run in a terminal, then /mcp inside Claude Code to sign in. Add --scope project to share it via .mcp.json. docs

claude mcp add --transport http browser-use https://api.browser-use.com/v3/mcp

Security notes

  • The remote server runs at api.browser-use.com, hosted by Browser Use. You sign in with OAuth in clients that support it, or with X-Browser-Use-API-Key header where OAuth isn't available, and it acts with your account's permissions.
  • The local server runs with your user's permissions (uvx downloads it from PyPI on first run).
  • It reads web pages, and a page can carry prompt-injection instructions. Keep approval prompts on and use a browser profile without your personal sign-ins.
  • Chrome DevToolsofficial · ★ 54k

    Control a live Chrome: performance traces, network requests, console and screenshots for debugging.

  • Playwrightofficial · ★ 39k

    Browser automation through the accessibility tree: click, type, navigate and snapshot pages, no screenshots needed.

  • Skyvernofficial · ★ 23k

    AI browser automation in Skyvern's cloud: navigate, fill forms and pull data from any site.

  • Apifyofficial · ★ 13k

    Run any of thousands of Apify scrapers (Actors) for social media, maps, e-commerce and more.

  • Firecrawlofficial · ★ 7.7k

    Scrape, crawl, map and search websites into clean Markdown or JSON.

  • Bright Dataofficial · ★ 2.7k

    Web search and scraping that gets past blocks and CAPTCHAs, with a free monthly quota.

  • Hyperbrowserofficial · ★ 791

    Scrape, crawl and extract with cloud browsers, plus browser-use and Claude computer-use agents.

  • ScrapeGraphAIofficial · ★ 117

    Extract structured data from pages with a prompt, plus smart crawling.

Comparing servers for one product? Chrome MCP servers

FAQ

What does the Browser Use MCP Server do?+

An agent that drives a real Chrome to finish tasks: fill forms, click through flows, extract data.

Is the Browser Use MCP server official?+

Official: built and maintained by Browser Use.

How do I add Browser Use to Claude Code?+

Run: claude mcp add --transport http browser-use https://api.browser-use.com/v3/mcp, then /mcp inside Claude Code to sign in. Add --scope project to share it with your team through .mcp.json.

How do I use the Browser Use MCP server in Cursor or VS Code?+

Open the Cursor or VS Code tab above: it has the mcp.json entry and a one-click install link. Cursor uses the "mcpServers" key; VS Code uses "servers" and needs a type ("http").

Does Browser Use work in ChatGPT and claude.ai?+

Yes, it has a remote URL (https://api.browser-use.com/v3/mcp). In claude.ai add it under Customize > Connectors > Add custom connector; in ChatGPT under chatgpt.com/plugins > Add custom MCP server.

What are the alternatives to the Browser Use MCP server?+

Other browser & web scraping MCP servers: Chrome DevTools, Playwright, Skyvern, Apify and Firecrawl.

All MCP servers · What is an MCP server? · Agent skills