This is the nuclear option for website cloning. It doesn't just grab HTML, it spins up parallel builder agents in worktrees while simultaneously extracting computed styles, animations, scroll behaviors, and layered assets from live pages via browser automation. The prompts are obsessive about completeness: it documents not just what elements look like but how they behave, distinguishing between click-driven tabs and scroll-triggered ones, capturing transition timing, even tracking down background layers most clones miss. Use it when "view source" won't cut it and you need a pixel-perfect rebuild that actually feels like the original. Requires browser MCP tools and a Next.js scaffold ready to go.
npx skills add https://github.com/jcodesmore/ai-website-cloner-template --skill clone-website