cd /news/ai-tools/show-hn-i-turned-apple-style-scroll-… · home topics ai-tools article
[ARTICLE · art-107229] src=github.com ↗ pub= topic=ai-tools verified=true sentiment=· neutral

Show HN: I turned Apple-style scroll-video websites into a reusable AI skill

Musoyangrigor released an open-source AI skill that converts videos into Apple-style scroll-video websites, available via npx. The skill, named scroll-video-website, generates optimized WebP frame sequences and builds responsive, full-viewport canvas pages that scrub smoothly with scrolling. It supports optional visual direction and interactive optimization, and is MIT licensed.

read1 min views1 publishedAug 22, 2026
Show HN: I turned Apple-style scroll-video websites into a reusable AI skill
Image: Michielbdejong (auto-discovered)
npx skills add musoyangrigor/scroll-video-website-skill --skill scroll-video-website

Live Demo · Install · Usage · Features

The demo shows a simple product website made from a single video. The video fills the screen, and scrolling forward or backward smoothly moves through its frames.

You can create the same kind of website with:

$scroll-video-website ./media/product-demo.mp4

The skill converts the video into optimized frames and builds the responsive scroll-controlled page for you.

Give your agent the skill name followed by a video path:

$scroll-video-website ./media/product-film.mp4

Optionally add visual direction after the path:

$scroll-video-website ./media/product-film.mp4 use a dark editorial style

Optimize an existing generated frame sequence interactively:

$scroll-video-website optimize

The command takes no arguments. The agent discovers generated sequences in the current project, measures the real assets, trial-encodes representative frames, and shows separate estimates for format, frame count, quality, and other supported settings. It then shows their combined estimated impact and asks for confirmation before rewriting anything.

  • Converts video into an optimized WebP frame sequence.
  • Scrubs smoothly in both directions with progressive frame .
  • Renders a responsive, full-viewport canvas with adjacent-frame blending.
  • Preserves the existing project stack and supports reduced-motion preferences.
  • Builds a minimal canvas-only experience by default, or follows added design direction.
  • Interactively optimizes existing sequences with file-specific size estimates before processing.

MIT licensed. See LICENSE.

── more in #ai-tools 4 stories · sorted by recency
── more on @musoyangrigor 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/show-hn-i-turned-app…] indexed:0 read:1min 2026-08-22 ·