Fire in da houseTop Tip:Paying $100+ per month for Perplexity, MidJourney, Runway, ChatGPT and other tools is crazy - get all your AI tools in one site starting at $15 per month with Galaxy AI Fire in da houseCheck it out free

cline-browser-use-mcp

MCP.Pizza Chef: ztobs

cline-browser-use-mcp is a Model Context Protocol server enabling automated browser operations via Python scripts. It supports capturing screenshots, retrieving HTML content, executing JavaScript, and fetching console logs, with customizable interaction steps like clicking and scrolling. Designed for headless browser automation, it integrates seamlessly with Cline to facilitate real-time web environment interaction and data extraction for AI workflows.

Use This MCP server To

Capture full-page or viewport screenshots of web pages Retrieve HTML content for web scraping or analysis Execute custom JavaScript on web pages Collect browser console logs for debugging Automate complex browser interactions with clicks and scrolls Enable headless browser automation in AI workflows Integrate browser data into LLM-powered applications

README

Browser Use Server

smithery badge

A Model Context Protocol server for browser automation using Python scripts. For use with Cline

Browser Use Server MCP server

Features

Browser Operations

  • screenshot: Capture a screenshot of a webpage (full page or viewport)
  • get_html: Retrieve the HTML content of a webpage
  • execute_js: Execute JavaScript on a webpage
  • get_console_logs: Get console logs from a webpage

All operations support custom interaction steps (e.g., clicking elements, scrolling) after page load.

Prerequisites

  1. (Optional but recommended) Install Xvfb for headless browser automation:
# Ubuntu/Debian
sudo apt-get install xvfb

# CentOS/RHEL
sudo yum install xorg-x11-server-Xvfb

# Arch Linux
sudo pacman -S xorg-server-xvfb

Xvfb (X Virtual Frame Buffer) creates a virtual display, allowing browser automation without detection as a bot. Learn more about Xvfb here.

  1. Install Miniconda or Anaconda
  2. Create a Conda environment:
conda create -n browser-use python=3.11
conda activate browser-use
pip install -r requirements.txt
  1. Set up LLM configuration:

The server supports multiple LLM providers. You can use any of the following API keys:

# Required: Set at least one of these API keys
export GLHF_API_KEY=your_api_key
export GROQ_API_KEY=your_api_key
export OPENAI_API_KEY=your_api_key
export OPENROUTER_API_KEY=your_api_key
export GITHUB_API_KEY=your_api_key
export DEEPSEEK_API_KEY=your_api_key
export GEMINI_API_KEY=your_api_key
export OLLAMA_API_KEY=your_api_key

# Optional: Override default configuration
export MODEL=your_preferred_model  # Override the default model
export BASE_URL=your_custom_url    # Override the default API endpoint
export USE_VISION=false  # Enable/disable vision capabilities (default: false)

The server will automatically use the first available API key it finds. You can optionally customize the model and base URL for any provider using the environment variables.

Installation

Installing via Smithery

To install Browser Use Server for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @ztobs/cline-browser-use-mcp --client claude
  1. Clone this repository to /home/YOUR_HOME/Documents/Cline/ directory
  2. Install dependencies:
npm install
  1. Build the server:
npm run build

MCP Configuration

Add the following configuration to your Cline MCP settings:

"browser-use": {
  "command": "node",
  "args": [
    "/home/YOUR_HOME/Documents/Cline/MCP/browser-use-server/build/index.js"
  ],
  "env": {
    // Required: Set at least one API key
    "GLHF_API_KEY": "your_api_key",
    "GROQ_API_KEY": "your_api_key",
    "OPENAI_API_KEY": "your_api_key",
    "OPENROUTER_API_KEY": "your_api_key",
    "GITHUB_API_KEY": "your_api_key",
    "DEEPSEEK_API_KEY": "your_api_key",
    "GEMINI_API_KEY": "your_api_key",
    "OLLAMA_API_KEY": "your_api_key",
    // Optional: Configuration overrides
    "MODEL": "your_preferred_model",
    "BASE_URL": "your_custom_url",
    "USE_VISION": "false"
  },
  "disabled": false,
  "autoApprove": []
}

Replace:

  • YOUR_HOME with your actual home directory name
  • your_api_key with your actual API keys

Usage

Run the server:

node build/index.js

The server will be available on stdio and supports the following operations:

Screenshot

Parameters:

  • url: The webpage URL (required)
  • full_page: Whether to capture the full page or just the viewport (optional, default: false)
  • steps: Comma-separated actions or sentences describing steps to take after page load (optional)

Get HTML

Parameters:

  • url: The webpage URL (required)
  • steps: Comma-separated actions or sentences describing steps to take after page load (optional)

Execute JavaScript

Parameters:

  • url: The webpage URL (required)
  • script: JavaScript code to execute (required)
  • steps: Comma-separated actions or sentences describing steps to take after page load (optional)

Get Console Logs

Parameters:

  • url: The webpage URL (required)
  • steps: Comma-separated actions or sentences describing steps to take after page load (optional)

Example Cline Usage

Here are some example tasks you can accomplish using the browser-use server with Cline:

Modifying Web Page Elements during Development

To change the color of a heading on a page that requires authentication:

Change the colour of the headline with the text "Alle Foren im Überblick." to deep blue on https://localhost:3000/foren/ page

To check/see the page, use browser-use MCP server to:
Open https://localhost:3000/auth,
Login with ztobs:Password123,
Navigate to https://localhost:3000/foren/,
Accept cookies if required

hint: execute all browser actions in one command with multiple comma-separated steps

This task demonstrates:

  • Multi-step browser automation using comma-separated steps
  • Authentication handling
  • Cookie acceptance
  • DOM manipulation
  • CSS styling changes

The server will execute these steps sequentially, handling any required interactions along the way.

Configuration

LLM Configuration

The server supports multiple LLM providers with their default configurations:

  • GLHF: Uses deepseek-ai/DeepSeek-V3 model
  • Ollama: Uses qwen2.5:32b-instruct-q4_K_M model with 32k context window
  • Groq: Uses deepseek-r1-distill-llama-70b model
  • OpenAI: Uses gpt-4o-mini model
  • Openrouter: Uses deepseek/deepseek-chat model
  • Github: Uses gpt-4o-mini model
  • DeepSeek: Uses deepseek-chat model
  • Gemini: Uses gemini-2.0-flash-exp model

You can override these defaults using environment variables:

  • MODEL: Set a custom model name for any provider
  • BASE_URL: Set a custom API endpoint URL (if the provider supports it)

Vision Support

The server supports vision capabilities through the USE_VISION environment variable:

  • Set USE_VISION=true to enable vision capabilities for browser operations
  • Default is false to optimize performance when vision is not needed
  • Useful for tasks that require visual understanding of webpage content

Xvfb Support

The server automatically detects if Xvfb is installed and:

  • Uses xvfb-run when available, enabling better browser automation without bot detection
  • Falls back to direct execution when Xvfb is not installed
  • Sets RUNNING_UNDER_XVFB environment variable accordingly

Timeout

Default timeout is 5 minutes (300000 ms). Modify the TIMEOUT constant in build/index.js to change this.

Error Handling

The server provides detailed error messages for:

  • Python script execution failures
  • Browser operation timeouts
  • Invalid parameters

Debugging

Use the MCP Inspector for debugging:

npm run inspector

Uses

Browser Use

License

MIT

cline-browser-use-mcp FAQ

How do I install cline-browser-use-mcp?
Install via pip and optionally set up Xvfb for headless operation on Linux systems.
Can I capture screenshots of specific parts of a webpage?
Yes, you can capture full page or just the viewport screenshots.
How do I execute custom JavaScript on a webpage?
Use the execute_js operation to run arbitrary JavaScript code within the page context.
Is it possible to retrieve browser console logs?
Yes, the get_console_logs operation fetches console output for debugging.
Does cline-browser-use-mcp support interaction steps like clicking or scrolling?
Yes, all operations support custom interaction steps after page load.
Can this server run in a headless environment?
Yes, it is recommended to install Xvfb for headless browser automation on Linux.
What browsers are supported?
Typically Chromium-based browsers are supported via underlying automation libraries, but check documentation for specifics.
How does this integrate with Cline?
It acts as an MCP server providing browser context and control to Cline clients for AI-driven workflows.