MCP.so
登录

Chromium Bridge

@dashi96

关于 Chromium Bridge

MCP bridge between Chromium browsers and Claude Code — browser automation tools + Claude chat panel for browsers where the official Claude in Chrome extension doesn't work (Arc, Vivaldi, …)

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "chromium-bridge": {
      "command": "npx",
      "args": [
        "chromium-bridge"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Chromium Bridge?

An MCP bridge between Chromium browsers and Claude Code. It gives Claude Code a set of browser_* automation tools plus a Claude chat panel inside the browser — aimed at Chromium browsers where the official "Claude in Chrome" extension isn't available (Arc, Vivaldi, Brave, …).

Two parts:

  • extension/ — an MV3 extension whose service worker keeps a WebSocket to the server and drives the tab through the Chrome DevTools Protocol (CDP).
  • server/ — a stdio MCP server that exposes the browser_* tools to Claude Code.

How to use Chromium Bridge?

Add it to your MCP client config:

{
  "mcpServers": {
    "chromium-bridge": {
      "command": "npx",
      "args": ["chromium-bridge"]
    }
  }
}

Then install the extension in the browser profile you want to automate. The tools only work while the server is running (typically an active Claude Code session).

Tools

ToolWhat it does
browser_statusCheck the connection to the extension
browser_tabs_listList tabs (id, title, URL)
browser_tab_create / browser_tab_closeOpen / close a tab
browser_navigateNavigate to a URL; back/forward for history
browser_page_textPage title, URL, and visible text
browser_computerMouse/keyboard/screenshots via CDP: click, drag, hover, type, key combos, scroll, zoomed screenshots
browser_read_pageAccessibility tree with ref ids (filter=interactive)
browser_findFind elements by text/role, returns refs
browser_form_inputSet input/textarea/select/checkbox/contenteditable values
browser_clickDOM click by CSS selector
browser_upload_filePut files into an <input type="file">
browser_javascriptRun JS on the page (await supported)
browser_console_messagesTab console (with a regex filter)
browser_network_requestsTab network requests (with a regex filter)
browser_resize_windowWindow size
browser_gif_start / browser_gif_stopRecord a GIF of the tab to a file

Everything except basic tab operations runs through chrome.debugger (CDP): screenshots don't require activating the tab, and clicks are real mouse events.

Limitations

  • The extension lives in one browser profile — install it in the one you want to automate.
  • It only works while the MCP server is running.

Where is the source code?

The source is on GitHub, linked from the repository URL on this page.

评论

浏览器自动化 分类下的更多 MCP 服务器