xuanxuan321/xuanxuan-prompts — explained in plain English
Analysis updated 2026-05-18
Copy a prompt file and hand it to Claude Code or Codex to generate a polished landing page.
Preview a page's finished look before generating it, using the included animated clip.
Recreate a real existing website offline using a website-archiving skill and a matching prompt.
Use the mp4-to-gif script to turn your own recordings into preview clips for a new prompt.
| xuanxuan321/xuanxuan-prompts | dockur/proxmox | k-dense-ai/science-superpowers | |
|---|---|---|---|
| Stars | 148 | 149 | 140 |
| Language | Shell | Shell | Shell |
| Setup difficulty | easy | — | easy |
| Complexity | 2/5 | — | 2/5 |
| Audience | vibe coder | ops devops | researcher |
Figures from each repo's GitHub metadata at analysis time.
Requires access to an AI coding assistant like Claude Code or Codex to run the prompts against.
xuanxuan-prompts is a collection of detailed prompts you can hand to an AI coding assistant to recreate polished, visually elaborate web pages. Each subfolder in the repository represents one example website and contains a prompt file spelling out every detail needed to rebuild it: fonts, colors, animations, asset links, and how the page's pieces fit together. Alongside each prompt sits a short preview video, shown first as a lightweight cover image so the page does not load every animation at once, that you can click to see what the finished site actually looks like. Using it is simple. You pick a folder whose style you like, copy the entire contents of its prompt file, and paste it into a coding assistant such as Claude Code or Codex, asking it to build the project from that prompt. The assistant then generates the full project, sometimes a React and Vite application and sometimes a single self-contained HTML file, and running it produces a site matching the preview. The collection currently includes examples like a dark, glass-styled landing page for an AI design studio, a scroll-triggered hero section for a geology brand, a minimalist AI product page with a scrolling video background, and a three dimensional basketball storefront built with Three.js. A few entries take a different approach: rather than describing a page from scratch, they instruct the assistant to install a website-archiving skill first and use it to recreate an existing real website offline, such as a Formula One driver's site or a WebGL-heavy creative studio homepage. All videos, images, and animated previews referenced in the prompts are hosted on public CDN links that generated code can pull directly, and a small script in the repository can convert any local video into the animated preview format used throughout. The author says the collection will keep growing and welcomes contributions of new prompts and preview clips. This is useful for developers and designers who want ready-made, detailed prompts to quickly produce impressive-looking landing pages with an AI assistant rather than writing one from a blank page.
A collection of detailed prompts for AI coding assistants that recreate polished, animated landing pages, each paired with a preview clip.
Mainly Shell. The stack also includes Shell, React, Vite.
No license is stated in the README, so reuse and redistribution terms are unclear.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
double-check against the repo, no cap.