home / mcp / stealth browser mcp server
The only browser automation that bypasses anti-bot systems. AI writes network hooks, clones UIs pixel-perfect via simple chat.
Configuration
View docs{
"mcpServers": {
"vibheksoni-stealth-browser-mcp": {
"command": "C:\\path\\to\\stealth-browser-mcp\\venv\\Scripts\\python.exe",
"args": [
"C:\\path\\to\\stealth-browser-mcp\\src\\server.py"
]
}
}
}You install and run the Stealth Browser MCP server to enable undetectable browser automation for MCP-compatible AI agents. It lets you automate browser tasks, inspect and modify network traffic, clone UI elements, and interact with dynamic web pages through a ready-made MCP interface. You can run it locally as a stdio server or deploy via your MCP client to control a browser from conversations with your agent.
You use this MCP server by connecting it to your MCP client and sending commands via a standard MCP protocol. Start the local server, then issue commands from your agent to navigate web pages, interact with elements, clone UI components, inspect network traffic, and apply dynamic hooks. You can operate with the full 90-tool arsenal or opt for a minimal core set of 22 tools to keep interactions lean. The server automatically detects your browser environment and provides a rich set of tools for browser management, element interaction, CDP function execution, and network debugging.
To perform common tasks, begin by starting the local server, then instruct your agent to navigate to a target page, extract elements, or capture a screenshot. Use the modular loading options to tailor the available tools to your workflow, enabling faster responses and reducing overhead during your sessions.
Prerequisites: ensure you have Python installed and a Python virtual environment tool ready. You will also need an MCP client to connect to the local server.
# 1. Clone the project
git clone https://github.com/vibheksoni/stealth-browser-mcp.git
cd stealth-browser-mcp
# 2. Create and activate a virtual environment
python -m venv venv
# Windows
venv\Scripts\activate
# Mac/Linux
source venv/bin/activate
# 3. Install dependencies
pip install -r requirements.txtWhen integrating with an MCP client, you typically add a reference to this server as a local stdio server. The provided examples show the exact runtime commands required to launch the server from the client’s perspective. Use the command that matches your operating system to run the server in a dedicated Python virtual environment.
Windows users often run the server with a path to a Python executable inside the virtual environment and the server script path, for example: the command is the Python launcher in the virtual environment, with the server script as an argument.
If the server cannot detect a compatible browser, install Chrome, Chromium, or Microsoft Edge. The server will auto-detect the first available browser.
If tools hang or return malformed JSON, ensure you are using the latest stable version. Restart the MCP client after updating.
For Linux/CI environments, run with sandboxing considerations in mind. Use --sandbox=false if necessary or ensure your environment supports sandboxing.
If you want a smaller tool surface, launch with the minimal mode or disable specific sections, then restart the server to apply changes.
Ask your agent to navigate to a page, extract a pixel-accurate clone of a UI element, and inspect the resulting DOM and assets. Or request a network interception session to inspect headers and payloads for a particular request.