Selenium MCP Server
@Raghvendra-Raghuvanshi
Model Context Protocol server for Selenium WebDriver - enables LLMs to control web browsers
Overview
What is Selenium MCP Server?
Selenium MCP Server is a Model Context Protocol (MCP) server for browser automation using Selenium WebDriver. It enables LLMs and teams to automate browsers (Chrome, Firefox, Edge, Safari) via a standard protocol, similar to Playwright MCP.
How to use Selenium MCP Server?
Install globally or use npx, then configure in an MCP client (e.g., Cursor). Run with selenium-mcp-server --browser chrome or via npx selenium-mcp-server@latest. Add a JSON configuration to .cursor/mcp.json or equivalent, specifying the command and arguments.
Key features of Selenium MCP Server
- Multi-browser support (Chrome, Firefox, Edge, Safari)
- Team-friendly: visible browser by default
- MCP-compliant API (navigate, click, type, screenshot, etc.)
- Easy onboarding for teams and CI
- Zero-installation option via npx
- Support for headless and custom browser paths
Use cases of Selenium MCP Server
- Automate web form filling and data extraction
- Take screenshots of web pages for testing or archival
- Run end-to-end browser tests in CI pipelines
- Interact with dynamic content (click, hover, drag, dropdown)
- Generate PDFs of web pages
FAQ from Selenium MCP Server
What are the system requirements?
Node.js 18+ and Java 11+ are required. Browser drivers are automatically managed by WebDriverManager.
How do I run the server without installing?
Use npx selenium-mcp-server@latest --browser chrome. npx automatically downloads and runs the server.
Can I use browsers other than Chrome?
Yes, use the --browser option with firefox, edge, or safari.
How do I share the server with my team?
Publish the npm package to a public or private registry, then team members can install globally or use npx.
What happens if the browser is not found?
Install the browser manually or specify a custom path using --executable-path. Ensure Java is installed and accessible.