Selenium MCP Server

Automate browser interactions including navigation and clicking via Selenium WebDriver.

Data last scanned yesterday · Reviewed last month

235 tools★ 4TypeScriptMIT

Overview

This server provides a bridge for AI assistants to control web browsers using the Selenium WebDriver protocol. By exposing browser-based actions as Model Context Protocol tools, it enables LLMs to perform automated tasks such as navigating to specific URLs, interacting with page elements, and retrieving page source content.

Our verdict

With a security score of 23/100, this server is currently high-risk and not recommended for critical production environments. The project is community-maintained rather than vendor-backed, has not undergone an editorial review, and has seen no commits in over 480 days, suggesting it may lack recent security patches or maintenance. It is best suited for experimental local testing rather than sensitive or automated workflows.

Uses only four direct dependencies
Provides documented tools for browser automation
Uses standard MIT license

No activity for 482 days
Lacks official vendor maintenance
Authentication methods remain unreviewed

Setup

Claude Desktop

{
  "mcpServers": {
    "selenium-mcp-server": {
      "args": [
        "-y",
        "selenium-mcp-server"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
selenium_navigate	Navigates the browser to a specific URL.	medium
selenium_findElement	Finds an element on the page using a CSS selector.	low
selenium_click	Clicks an element.	medium
selenium_sendKeys	Sends keystrokes to an element.	medium
selenium_getPageSource	Retrieves the current page source HTML.	low

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›What software prerequisites are required to run this server?

You must have Node.js (v18 or later), npm (v7 or later), a compatible web browser like Chrome or Firefox, and the corresponding WebDriver executable installed and available in your system's PATH.

›How can I customize the WebDriver configuration?

You can modify the src/seleniumService.ts file to specify custom browser options or point to specific WebDriver installation paths.

›What transport mechanism does this server use for MCP integration?

The server is designed to communicate with MCP hosts like Cursor or Claude Desktop using the stdio transport.

›How do I ensure the browser session closes properly?

The server includes a signal listener for SIGINT that triggers driver.quit() to ensure the WebDriver session is shut down gracefully when the server exits.