agentleFS
Sign inStart free
arturmuller25arturmuller25/claude-skills/skills/scroll-cinema/SKILL.md

scroll-cinema

Skill0 starsChanged 46 days agoMITView on GitHub

Universal scroll-driven cinematic hero. One entry point that routes to the best technique for the job and budget. Free by default (Lenis + GSAP engine with SVG/CSS diorama, Three.js/Spline 3D, or Rive layers); for AI-generated photoreal "fly-through-the-world" video it DELEGATES to the scroll-world skill (Higgsfield/Monid pipeline) instead of reimplementing it. Use when the user wants a scroll cinematic, a 3D-world or diorama landing hero, a "browse-through-the-industry" section, or a scroll-driven storytelling page. Optimized for Next.js/React, mobile-safe.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. tdimino/claude-code-minoan★ 41same name
    2. Dhyey2907/claude-skill-scroll-cinema★ 1same name
    3. forever8896/scroll-cinema-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repositoryOne other file it gives its agents.

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask, in plain prose (not fabricated multiple-choice for the open ones):

    line 18
  2. Distinctive:

    Default to lane 1. Escalate only on explicit need.

    line 34
  3. Distinctive:

    Lane 4 (paid, photoreal video): do NOT reimplement the render pipeline.

    line 56

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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

Scroll Cinema (universal)

One skill, four visual lanes. The scroll mechanic is always the same (scroll drives a timeline or camera, no cuts), but the renderer is chosen by fidelity and budget. This skill picks the lane, wires the engine, and for the paid photoreal lane hands off to the specialist skill rather than duplicating it.

Design principle: take the best of each technique, reimplement none. The free lanes share one code engine; the photoreal-video lane is the scroll-world skill's job.

Step 1: Interview

Ask, in plain prose (not fabricated multiple-choice for the open ones):

  • Subject/brand + one-line pitch, brand kit (4-6 hex, name, tone).
  • Beats: 3 to 6 ordered scenes, each with headline + role in the story.
  • Fidelity + budget: this picks the lane (Step 2). State the cost of the paid lane before choosing it.
  • Target framework (Next.js/React assumed; the engine is framework-agnostic).

Step 2: Route to a lane

LaneRendererCostUse whenEngine
1. SVG/CSS diorama (default)DOM layers, parallaxfreemost heroes, editorial, isometric dioramathis skill's assets/scroll-engine.js
2. Three.js / Splinereal-time 3Dfreetrue camera flythrough, depth, lightingthis skill's engine + a 3D layer
3. Rivevector state machinefreeinteractive mascot/character, 2.5Dthis skill's engine + Rive input
4. AI photoreal videopre-rendered seamless flythroughPAID (per clip)client wants photoreal, no manual assets, accepts costdelegate to scroll-world

Default to lane 1. Escalate only on explicit need. Only reach lane 4 when the user wants AI-generated photoreal video AND accepts per-clip spend.

Step 3: Build the chosen lane

Lanes 1-3 (free, code): wire assets/scroll-engine.js (Lenis + GSAP ScrollTrigger). It pins the stage, builds one scrubbed timeline across the beats, and emits progress (0..1) per scene that the visual layer reads. Layer specifics + code: references/visual-layers.md. Next.js/React integration (layout provider, hook, SSR, cleanup): references/react-nextjs.md.

Engine rules, kept in line with my landing-page master prompt so a page built from it can use this engine as is:

  • One Lenis per page. Standalone page: the engine creates its own (lenis: 'own', the default). Next.js: Lenis lives once in the root layout as a provider and every section passes lenis: 'external'. Never two smooth-scroll systems at once.
  • Scrub 0.5 by default (scrub option). Measured in Chromium: pure true jolts the scene on keyboard and anchor jumps, and above 1 the scene trails Lenis visibly. The numbers are in the engine header.
  • React uses useGSAP from @gsap/react with a scope, not a bare useEffect.
  • Load lenis/dist/lenis.css, the stylesheet Lenis recommends.

Lane 4 (paid, photoreal video): do NOT reimplement the render pipeline. Invoke the scroll-world skill: it interviews for art direction/camera, generates the scene stills, the dive-in clips, and the frame-locked connector clips (Higgsfield stills + Monid/Seedance video, or Higgsfield-credits fallback), and ships its own vanilla scrub engine that plays the chain as one flight. This skill's role in lane 4 is only to route there and, if the page also needs code-driven sections, to run them on the same page alongside scroll-world's video hero. scroll-world owns the seams, the cost gate, and the video engine; trust it, don't fork it.

Step 4: Performance & accessibility (all lanes, non-negotiable)

  • Honor prefers-reduced-motion: no smooth-scroll, no scrub; show scenes static.
  • Headlines are real DOM text (SEO + screen readers), never baked into image/video only.
  • Lazy-load heavy layers (Three/Spline/video) below the fold; never block first paint.
  • 3D: stay inside the device budget in references/visual-layers.md (pixel ratio, triangles, draw calls); pause the loop when the stage is offscreen or the tab hidden.
  • Video (lane 4): scroll-world already hardens phones (seek-coalescing, iOS priming); keep it.

Step 5: Render and eyeball

Open the page; scroll slow and fast; check a narrow viewport. Watch for jank, layout shift, wrong scene overlap, and (lane 4) any seam "pop". For 3D, read the renderer stats (draw calls, triangles) on a phone against the budget.

Judge it in a real browser (Chrome on the localhost URL, then a real phone), never in an app's built-in preview pane: those panes are known to stutter on scroll-scrubbed video (reported in the tiagopro.com.br/guia10k guide, 2026-10), so a scrub that looks broken there may be fine, and one that looks fine there proves nothing.

Non-negotiables

  • Free by default. Lane 4's per-clip cost is opt-in, stated before spend.
  • One engine, swappable visuals for the code lanes; the renderer only consumes progress.
  • One Lenis per page. In Next.js it belongs to the layout, not to a section.
  • Delegate, don't duplicate. Photoreal video = the scroll-world skill. This skill never re-creates the Higgsfield/Monid pipeline.
  • Reduced-motion is a real branch. Text is DOM text.

Files

FileWhat it holds
assets/scroll-engine.jsLenis + GSAP ScrollTrigger engine for lanes 1-3 (options lenis and scrub)
references/visual-layers.mdThe code lanes (SVG/Three/Rive), the 3D device budget, and how the video lane delegates
references/react-nextjs.mdLenis provider in the root layout, useScrollCinema hook with useGSAP, SSR and cleanup

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.

Check your setup, free