agentleFS
wrg32786wrg32786/aigent-design-system/skills/aigent-3d-scroll-system/SKILL.md

aigent-3d-scroll-system

Skill4 starsChanged 3 months agoMITView on GitHub

Build premium AIgent-style 3D scroll web experiences with Spline, GSAP, Three.js, or Blender frame sequences. Use when creating or improving immersive landing pages, scroll-driven 3D backgrounds, asset galleries, cinematic product pages, Spline embeds, GSAP reveals, or mobile-safe premium web experiences that need strong taste and real browser verification.

The file

Show its rules and text

We split this skill into the rules it gives an agent

22 rules. About 0 are common, 20 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈20

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill to build immersive web pages that feel designed, not generated.

    line 8
  2. Distinctive:

    Use Spline embed for fast immersive backgrounds and interactive scenes.

    line 14
  3. Distinctive:

    Use Blender frame sequence for photoreal art direction and mobile-safe cinematic scroll.

    line 15
  4. Very specific:

    Use Three.js only when the page needs live interaction, procedural geometry, or real-time lighting.

    line 16
  5. Distinctive:

    Do not add random floating objects.

    line 18
  6. Distinctive:

    Start with a full-viewport scene.

    line 22
  7. Distinctive:

    Keep copy short and plain.

    line 26
  8. Distinctive:

    Verify mobile before calling it done.

    line 27
  9. Distinctive:

    Use CSS variables to keep the animation simple and inspectable:

    line 31
  10. Distinctive:

    If the background effect is only half done when the page ends, increase the multiplier.

    line 43
  11. Distinctive:

    Keep x/y translation at zero.

    line 49
  12. Distinctive:

    Use scale and rotation for movement.

    line 50
  13. Distinctive:

    Do not let the center drift diagonally as the user scrolls.

    line 52
  14. Distinctive:

    Avoid explaining features twice.

    line 58
  15. Distinctive:

    Use plain labels over technical labels when the audience is not technical.

    line 59
  16. Distinctive:

    Do not place dense copy over a busy part of the scene.

    line 60
  17. Distinctive:

    Use GSAP for hero load and scroll reveals.

    line 64
  18. Distinctive:

    When a hero loop hands off into a pinned video-scrub deck:

    line 72
  19. Distinctive:

    Use one source of truth for opening copy.

    line 78
  20. Very specific:

    Treat videos as scrub-ready on canplay or readyState >= 2, not metadata alone.

    line 82
  21. Distinctive:

    Use 1080p scrub exports with frequent keyframes or all-intra encoding when clarity matters.

    line 83
  22. Distinctive:

    Use browser tooling or Playwright, not visual guessing.

    line 105

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

AIgent 3D Scroll System

Use this skill to build immersive web pages that feel designed, not generated.

First Decision

Choose the lightest 3D technique that can carry the idea:

  • Use Spline embed for fast immersive backgrounds and interactive scenes.
  • Use Blender frame sequence for photoreal art direction and mobile-safe cinematic scroll.
  • Use Three.js only when the page needs live interaction, procedural geometry, or real-time lighting.

Do not add random floating objects. The scene should support the story.

AIgent Pattern

  1. Start with a full-viewport scene.
  2. Anchor a clear focal point.
  3. Map scroll progress to scene progress.
  4. Reveal text and cards only when they are readable.
  5. Keep copy short and plain.
  6. Verify mobile before calling it done.

Spline Scroll Mapping

Use CSS variables to keep the animation simple and inspectable:

const max = document.documentElement.scrollHeight - window.innerHeight;
const p = Math.min(1, Math.max(0, window.scrollY / max));
const scene = Math.min(1, p * 1.8);
root.style.setProperty("--scene-scale", (1 + scene * 1.4).toFixed(3));
root.style.setProperty("--scene-rotate", `${(scene * -72).toFixed(2)}deg`);
root.style.setProperty("--scene-x", "0px");
root.style.setProperty("--scene-y", "0px");

If the background effect is only half done when the page ends, increase the multiplier. If it finishes too early, lower it.

Focal Point Rule

For tunnels, vortexes, and control-room backgrounds:

  • Keep x/y translation at zero.
  • Use scale and rotation for movement.
  • Set transform-origin to the visual center.
  • Do not let the center drift diagonally as the user scrolls.

Text Rules

  • H1 should fit the first viewport.
  • Body copy should be 1-2 sentences.
  • Avoid explaining features twice.
  • Use plain labels over technical labels when the audience is not technical.
  • Do not place dense copy over a busy part of the scene.

Animation Rules

  • Use GSAP for hero load and scroll reveals.
  • Trigger reveals earlier on mobile.
  • Animate cards one at a time for long lists.
  • Use prefers-reduced-motion.
  • Hover states should feel premium: slight lift, border glow, no cartoon scaling.

Cinematic Deck Rules

When a hero loop hands off into a pinned video-scrub deck:

  • Target the deck boundary on click, not a nested sticky scene anchor.
  • For presented sponsorship decks, block free scrolling after entry and drive the internal scroll timeline with clear Next and Previous controls.
  • Land guided moves on approved pause points where the clip is near its end state and copy is fully rendered.
  • Reveal opening copy immediately on the first scrub frame or first approved pause point.
  • Use one source of truth for opening copy. Do not add a duplicate overlay unless the scene copy is removed or hidden.
  • Hide generic Scroll cues on the first deck slide after the hero.
  • Make deck copy enter from the side closest to its layout anchor, not bottom-up by default.
  • Give video-to-video transitions enough scroll distance to read. Push transitions should move opaque video, not behave like crossfades.
  • Treat videos as scrub-ready on canplay or readyState >= 2, not metadata alone.
  • Use 1080p scrub exports with frequent keyframes or all-intra encoding when clarity matters. Do not downgrade to blurry media to mask seek problems.
  • Load the hero poster and hero loop first, warm the first scrub clip on CTA intent, then warm later clips only near their transition windows.
  • Match video seek cadence to the source frame rate instead of seeking on every possible scroll tick.
  • Confirm MP4 range requests return 206 and Accept-Ranges: bytes.
  • Pin sticky facts only after their source slide has left.
  • On mobile, use object-fit: contain for a hero loop that contains a readable brand mark.

For the full checklist, read docs/cinematic-scroll-deck-playbook.md in this repo.

Verification

Before final:

  • Desktop screenshot at 1440px.
  • Mobile screenshot around 390px.
  • Confirm no horizontal scroll.
  • Confirm active sticky nav does not cover content.
  • Confirm scene reaches its intended end state by the page bottom.
  • Confirm links and external-link cues are actually clickable.
  • For guided scrub decks, click through every Next and Previous pause point and inspect active video currentTime.
  • Confirm the first page load does not request all 1080p scrub videos before interaction.

Use browser tooling or Playwright, not visual guessing.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.