browser-kit MCP server

Commit Nov 13, 2026

ndthanhdev/mcp-browser-kit · typescript · MIT

Enables AI assistants to interact with and control your local web browsers through an MCP server and browser extension.

Also known as ndthanhdev/mcp-browser-kit

Browser Automation No API key

GitHub ↗

Stars: 55
Tools: —
Downloads, 30 days: —
Latest version: —
Last commit: Nov 13, 2026
License: MIT

What is browser-kit MCP?

browser-kit enables AI assistants to interact with and control your local web browsers through an MCP server and browser extension.

Summary generated from the project's README

About browser-kit MCP server

Controlling your local web browser through an AI assistant is what mcp-browser-kit enables. This tool allows coding agents and various AI clients to interact directly with your active browser sessions by using a combination of an MCP server and a dedicated browser extension. It serves as a flexible alternative to browser-specific AI integrations, working across different browsers including Chrome, Brave, Edge, Firefox, and Safari.

To set it up, you run the server via npx using stdio transport or configure an http transport on a specific port like 21082. You must also install a compatible extension build, such as the M2 or M3 versions, depending on your browser. While M2 builds offer more functionality, M3 builds are required for Chrome. Users should consider using a separate browser profile to protect sensitive data during these interactions.

What you can do with it

1Automate social interactionsAsk your AI assistant to perform specific actions on a website you currently have open, such as starring a repository.
2Control multiple browser typesUse a single MCP setup to manage different browsers like Firefox or Safari using specific extension builds.
3Connect via network transportConfigure your AI client to communicate with the browser controller over an HTTP connection instead of standard input/output.

Written from the project's README. Every use case is checked against a quote from it.

Automated checks

Ran Nov 14, 2026Full report →

58/100

Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points

Show this score in your README

Source repository linked and reachablendthanhdev/mcp-browser-kit is public and reachablegithub.com/ndthanhdev/mcp-browser-kit ↗Pass
Package names this repository as its sourceNot checked
Build provenance (Sigstore attestation)Not checked
Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
Commit or release in the last yearLast commit or release 2026-11-13github.com/ndthanhdev/mcp-browser-kit ↗Pass
OpenSSF ScorecardNot checked
Package name not close to a popular packageNot checked
No critical advisory in the package or its dependenciesNot checked
No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
Destructive tools are labelledNot checked
Remote endpoint answers the MCP handshakeNot checked

README

View on GitHub

MCP Browser Kit

🚀 On the main branch? You're viewing the bleeding-edge dev docs! For other versions, check out the release/* branches.

An MCP Server that enables AI assistants to interact with your local browsers. It's an alternative to "Claude in Chrome" that works with any AI assistant or coding agent.

https://github.com/user-attachments/assets/1fbf87fd-06d1-42bf-a06f-cc2bbdf375a8

Usage

Add MCP Browser Kit to your MCP Client:
```
{
	"mcpServers": {
		"browser-kit": {
			"command": "npx",
			"args": ["@mcp-browser-kit/server@latest"]
		}
	}
}
```
By default the server use stdio transport, alternatively you can use http transport:
```
npx @mcp-browser-kit/server@latest --transport http
# optional flags (defaults shown):
#   --http-host 127.0.0.1
#   --http-port 21082
```
Then point your MCP client at the URL instead of a command:
```
{
	"mcpServers": {
		"browser-kit": {
			"url": "http://127.0.0.1:21082/mcp"
		}
	}
}
```
Security: there's no auth yet, so --http-host defaults to 127.0.0.1 (loopback only). Only pass --http-host 0.0.0.0 (or a LAN address) if you understand that anyone who can reach that host/port can control your browser.
Choose the right extension build for your browser:
- Check the Compatibility table below to see which build (M2 or M3) is compatible with your browser
- Note that M2 builds offer more functionality than M3 builds, so prefer M2 if your browser supports both
Download the latest compatible build from The Latest Release.
Load the extension into a compatible browser and enable it.
Refresh open tabs to ensure extension scripts are loaded.

Start messaging on your MCP Client:

Use browser-kit, star the last open GitHub repo on my browser

Optional, check User Notes for more information.

Compatibility table

The M2 build supports more functionalities than the M3 build. Extension file name patterns:

M2 build:

Chromium: mcp_browser_kit_m2-<version>.zip (e.g., mcp_browser_kit_m2-5.0.0.zip)

Firefox: mcp_browser_kit_m2-<version>.xpi (e.g., mcp_browser_kit_m2-5.0.0.xpi)

M3 build: mcp_browser_kit_m3-<version>.zip (e.g., mcp_browser_kit_m3-5.0.0.zip)

Browser	M2 build	M3 build
Chrome	❌	✅
Brave	✅	✅
Edge	✅	✅
Firefox	✅	❌
Safari	✅	❌

Load Extension

Chromium-based browsers (Chrome, Brave, Edge)
1. Unzip the downloaded .zip file
2. Open chrome://extensions/
3. Enable Developer mode
4. Click Load unpacked
5. Select the unpacked extension folder
Firefox
1. Open about:addons
2. Click the gear icon and select Install Add-on From File
3. Select the downloaded .xpi file
Safari
1. Unzip the downloaded .zip file
2. Open Develop > Show Extension Builder
3. Click + and select Add Extension
4. Select the unpacked extension folder
5. Click Install to install the extension

User Notes

Note for Alpha stage: Consider using a separate browser profile or dedicated browser instance with this MCP to prevent sensitive data from being unintentionally exposed to AI model providers.

Security: BrowserKit currently lacks authentication for incoming requests. Keep ports 21082 and 2769-2799 on your device network only.

Hosted deployment

A hosted deployment is available on Fronteir AI.

Read full README

About browser-kit MCP server

What you can do with it

1Automate social interactionsAsk your AI assistant to perform specific actions on a website you currently have open, such as starring a repository.
2Control multiple browser typesUse a single MCP setup to manage different browsers like Firefox or Safari using specific extension builds.
3Connect via network transportConfigure your AI client to communicate with the browser controller over an HTTP connection instead of standard input/output.

Written from the project's README. Every use case is checked against a quote from it.

Browser

M2 build

M3 build

Chrome

❌

✅

Brave

✅

Edge

✅

Firefox

✅

❌

Safari

✅

❌