Browser CTL MCP Server

SSE Anthropic based MCP server built on Python Playwright, providing a suite of tools that enable AI agents to control and interact with web browsers , this simple version use chrome browser.

Exposed Tools:

1. execute_javascript

Execute JavaScript code in real-time browser using Playwright's evaluate method and return the script's result.

2. go_to_url

Visit a target url in real-time browser.

3. sleep_for

Browser sleep for a random seconds between begin and end .

4. close_browser

Close the opened browser.

5. open_browser

Open browser in real-time.

6. take_browser_screenshot

Capture a screenshot of the current browser page and return a base64-encoded image.

Default log

This MCP server start with info log level by default:

#Configure logging
logging.basicConfig(
    level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s"
)

Default End Point

http://127.0.0.1:8001/sse

INSTALL

First, let’s install uv and set up our Python project and environment:

Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Clone the repo:

git clone https://github.com/biteval/browser_ctl.git

Go to the project dir:

cd browser_ctl

Create virtual environment and activate it

uv init

uv venv

Linux:

source .venv/bin/activate

Windows:

.venv\Scripts\activate

Install dependencies

Linux

uv add "mcp[cli]" httpx playwright

Windows

uv add mcp[cli] httpx playwright

Download playwright browser binaries

python -m playwright install

Start the Browser CTL MCP Server

python browser_ctl.py

Contributions Are Welcome!

Your expertise and ideas can make a significant impact on this project. Thanks!

Learn More

Visit our YouTube channel to learn how to create AI agents: