An agent skill for Claude Code and Codex that turns AI-generated motion into web animations driven by scroll, mouse, drag, touch or device orientation.
oso95/scroll-world
An agent skill that generates a scroll-scrubbed 3D landing page where the camera flies from outside each scene into its interior, with no cuts.
- Estrellas
- 10,261
- Forks
- 1,128
- Issues abiertos
- 10
- Último commit
- hace 3 meses
¿Qué es?
scroll-world is an agent skill for Claude Code, Codex, and other SKILL.md-compatible agents. It builds a landing page for any brand or industry in which scrolling moves a camera from outside each scene into its interior, then continues to the next scene as one continuous flight without cuts.
The skill generates the art with AI: isometric diorama stills via GPT Image 2 (through Higgsfield or the Codex CLI), and image-to-video camera clips via Seedance or Kling. Connector clips are generated from the rendered frames of neighbouring scenes so that each seam matches frame for frame. Scroll position drives playback time, while the camera itself moves in the generated footage.
The scroll engine is a framework-agnostic vanilla JavaScript module that can be used in plain HTML, Next.js, Vue, or a Python-served page. The skill also covers an optional mobile 9:16 chain rendered separately for phones, and it states estimated generation costs before any rendering begins.
Funciones clave
- 01Generates isometric diorama scene stills and camera flight clips with AI image and video models.
- 02Builds connector clips from neighbouring rendered frames so scene transitions are seamless.
- 03Provides a framework-agnostic vanilla JavaScript scroll engine with lazy loading and seam crossfade.
- 04Supports an optional separately composed 9:16 mobile version.
- 05Shows estimated credit or dollar costs and waits for approval before generating.
Casos de uso
- Creating a scroll-driven product or brand landing page for a logistics or industrial company.
- Prototyping an immersive 3D-style marketing page without a 3D modeling pipeline.
- Adding a cinematic scroll section to an existing Next.js, Vue, or plain HTML site.
Temas
Preguntas frecuentes
- ¿Qué es scroll-world?
- An agent skill that generates a scroll-scrubbed 3D landing page where the camera flies from outside each scene into its interior, with no cuts. scroll-world is an agent skill for Claude Code, Codex, and other SKILL.md-compatible agents. It builds a landing page for any brand or industry in which scrolling moves a camera from outside each scene into its interior, then continues to the next scene as one continuous flight without cuts.
- ¿scroll-world es gratuito? ¿Qué licencia tiene?
- scroll-world es open source bajo la licencia MIT. Revisa los términos de la licencia antes de usarlo en productos comerciales.
- ¿scroll-world sigue mantenido?
- Actualizado en los últimos 6 meses. El último commit de scroll-world fue el 29 jul 2026.
- ¿Para qué puedo usar scroll-world?
- Creating a scroll-driven product or brand landing page for a logistics or industrial company. Prototyping an immersive 3D-style marketing page without a 3D modeling pipeline. Adding a cinematic scroll section to an existing Next.js, Vue, or plain HTML site.
- ¿Dónde consigo scroll-world?
- El código fuente y las instrucciones de instalación están en GitHub: https://github.com/oso95/scroll-world
Resumen generado por IA a partir del README del proyecto. Las estadísticas de GitHub se actualizan a diario.
Proyectos relacionados
An Agent Skill that converts a video into an optimized WebP frame sequence and builds a scroll-controlled canvas website around it.
Agent Skills para que los agentes de programación diseñen interfaces distintivas, con dirección de arte, fotos reales y una revisión en el navegador, en lugar de páginas genéricas de IA.
An AI agent skill that turns an article into a 3:4 vertical cover image prompt for WeChat Official Accounts and Xiaohongshu, with optional face consistency.
A Codex and Claude Code skill that turns a topic, article or story into a narrated hand-drawn explainer video, with stroke-by-stroke drawing and Chinese handwriting.
Interfaz gráfica de nodos, API y backend para ejecutar modelos de difusión localmente y generar imágenes, video, audio y contenido 3D.