Browser Control MCP
An MCP server and Firefox extension: an agent manages tabs, searches history and reads pages in your personal browser
- 346 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The extension gives an agent access to your browser's tab list and history
- On allowed sites the agent reads page content and takes screenshots
Install
- Works in
Manual install
{"mcpServers": {"browser-control": {"command": "node", "args": ["/path/to/repo/mcp-server/dist/server.js"], "env": {"EXTENSION_SECRET": "<secret>", "EXTENSION_PORT": "8089"}}}}For a source build: npm install and npm run build. Replace the path with yours.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The extension gives an agent access to your browser's tab list and history
- On allowed sites the agent reads page content and takes screenshots
This is third-party code. Review the repository files before installing.
What it does
Browser Control MCP consists of an MCP server and a Firefox extension. The server gives an agent tab tools: open and close, list and reorder, group with a name and color. The agent also reads and searches browser history and pulls the text and links of a web page. Inside a tab it finds and highlights text and captures a screenshot of the visible area. The main idea is safety for a personal browser. The server does not modify pages, click or run scripts. Reading content requires consent in the browser for each domain, a limit written into the extension manifest. A screenshot is allowed per tab with a toolbar button, and the permission lapses when the tab navigates or closes. The server and extension talk over a local connection. The extension has a call log and a setting that turns individual tools on and off, and it has no third-party runtime dependencies. The author calls the project experimental.
Who it is for. For people who keep dozens of tabs in Firefox and want Claude to sort them, search history and turn what it read into tables.
Good fit when
- You want to close every tab unrelated to work and group the rest
- You are looking in history for an article you only remember by topic
- An agent should open a few pages, read the text and build a comparison table
- You want tabs reordered into a sensible order from a description
Not a fit when
- You need to fill forms and click through sites
- You work in Chrome: the extension is built for Firefox only
- You need test automation or unattended scraping
Example request
Group all development tabs into a group called Development and close the ones I have not opened for more than a dayMCP
- Transport
- stdio
- Authentication
- API key
Environment variables
EXTENSION_SECRETrequired, secret- Secret from the Firefox extension settings page
EXTENSION_PORT- Port for talking to the extension, 8089 by default
Limitations
Works with Firefox only. The project is experimental, and the author advises watching tool calls. Reading pages and highlighting text require consent for each domain, and a screenshot requires permission for each tab. The secret from the extension settings page must go into the server configuration. The server listens on a local port, 8089 by default, and that port must be free. Claude Desktop has a ready DXT package, while other clients need a build from source.
How to disable. Remove the browser-control entry from mcpServers and disable or remove the extension on the about:addons page. You can turn off individual tools in the extension settings.
FAQ
How do I install Browser Control MCP in Claude Desktop?
Install the Firefox extension from addons.mozilla.org, then download the DXT file from the releases page and open it in Claude Desktop settings. Enable the DXT extension and paste the secret from the extension settings page.
What can the server not do?
Modify pages, click, fill forms or run scripts. It only manages tabs, reads history and, with your consent, page text and screenshots.
How is Browser Control MCP different from web automation servers?
Web automation servers such as Playwright MCP drive a separate browser and click through pages. Browser Control MCP works in your personal Firefox and only tidies tabs and reads.
How do I build the server from source?
Run npm install and npm run build in the repository root, then put the path to mcp-server/dist/server.js into the client configuration.
Can I run the server in Docker?
Yes. Build the image with `docker build -t browser-control-mcp .` and launch it from the client configuration with port 127.0.0.1:8089 and the EXTENSION_SECRET and CONTAINERIZED variables.
Discussion
No questions or reviews yet.
Related
MCP serversA Chrome extension with an MCP server: the agent drives your everyday browser with its sessions, tabs, bookmarks and history
An MCP server and Chrome extension that let the agent drive your already open browser with your logged-in sessions, all locally
An MCP server that drives your already open Chrome and stops to ask you about a 2FA code, a CAPTCHA, or a choice only you can make
An MCP wrapper over the browser-use engine: a single tool drives a browser from a natural-language task.
An MCP server and Firefox extension that let Claude Code open pages, click, type and take screenshots without Chrome
A plugin and MCP server letting an agent drive your open, already logged-in Chrome through an extension with no debugging port.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Browser Control MCP (this card) | no | API key | 346 | |
| Playwright MCP Server | no | not required | 5,681 | browser automation with clicks and typing through Playwright |
| Chrome DevTools MCP | yes | not required | 55,199 | debugging Chrome pages through DevTools |
README in short
The author shows the server through example requests: tidy tabs, find an article in history, compare several pages. A separate section compares the project with web automation servers, and a Docker option closes the document.