browser-automation / mcp
Chrome MCP Server
Capability: Chrome MCP Server
Use it when
- An agent needs to automate pages behind logins the user already has, and launching a fresh Playwright browser forces re-authentication every run
- You want AI browser control without installing Playwright dependencies and separate browser binaries, using the Chrome instance already open on the desktop
What it solves
Not the fit when
- Headless servers or CI environments without desktop Chrome or Chromium
- Firefox; broader browser support remains a roadmap item
- Exposing the MCP bridge beyond its documented loopback URL; endpoint authentication is not implemented
- Unsupervised consequential actions in logged-in sessions, because the tool surface can navigate, inject scripts, send requests, fill forms, click, and access history
- Users requiring a mature stability guarantee; the project describes itself as early-stage and under intensive development
- Large-scale crawling fleets, CAPTCHA evasion, or clean-room reproducible browser environments
- large-scale headless crawling fleets
- bot detection or captcha evasion
- mobile app automation
- clean-room reproducible browser environments
Install
Download the extension from https://github.com/hangwin/mcp-chrome/releases, run npm install -g mcp-chrome-bridge, then load the unpacked extension at chrome://extensions/ with Developer mode enabled (Node.js >= 20 required)
Invoke
Click the extension icon and connect, then add to your MCP client config: {"chrome-mcp-server": {"type": "streamableHttp", "url": "http://127.0.0.1:12306/mcp"}}; stdio via the installed mcp-server-stdio.js is the fallbackAlternatives
No reviewed alternatives recorded yet.