Browser Automation

Puppeteer

Browser automation for agents — navigate pages, click, fill forms, and take screenshots through a headless Chrome instance.

Publisher

Community

View repo ↗

Install

npx -y @modelcontextprotocol/server-puppeteer

The Puppeteer server gives an agent control of a real (or headless) Chrome browser — navigating to URLs, clicking elements, filling in forms, and capturing screenshots as it goes.

Why it matters

Some tasks can only be verified by actually looking at rendered output — a UI change, a broken layout, a JavaScript-heavy page that a plain HTTP fetch can’t render. Browser automation closes that verification gap.

Good for

End-to-end testing, scraping content from JS-rendered sites, and visually confirming that a frontend change actually looks right.