Invisible Playwright MCP: Stealth Browser Automation for AI Agents
The open-source repository `feder-cr/invisible_playwright_mcp` gained over 1,600 GitHub stars this month. It offers a Python-based Model Context Protocol (MCP) server that enables AI agents to automate browsers and scrape web data undetected by anti-bot protections and CAPTCHAs using stealth Firefox. As AI agents are increasingly tasked with web browsing and autonomous computer use, traditional headless browsers are easily blocked by modern anti-bot security. By combining MCP with stealth browser automation, this tool provides a standardized interface for LLMs to reliably interact with protected websites. The project utilizes an anti-detect Firefox setup in Python to evade browser fingerprinting techniques while scraping or executing web tasks. Built on Anthropic's Model Context Protocol, it seamlessly connects stealth browser actions directly to compatible LLM clients like Claude.
## BACKGROUND
The Model Context Protocol (MCP) is an open standard introduced by Anthropic in late 2024 to standardize how large language models integrate with external tools and environments. Playwright is a widely used browser automation library, but standard automated browsers trigger anti-bot mechanisms, necessitating specialized stealth evasions for autonomous agent web access.