This is a CLI wrapper around notte.cc's browser automation platform that lets Claude control a real browser session for scraping, screenshots, and web interactions. You authenticate once, spin up a session, and then can navigate pages, observe content, and grab screenshots through simple commands. It's basically giving Claude headless browser capabilities without you having to manage Playwright or Puppeteer yourself. The audit results are mixed (Snyk failed), and with only 16 GitHub stars it's pretty early stage, but if you need Claude to interact with live web pages rather than just reading static content, this handles the plumbing. The sessions model means you're not spinning up a new browser for every task.
npx skills add https://github.com/nottelabs/notte-cli --skill notte-browser