Scrolldio v1.0.0

Turn an ordinary website into a cinematic story that plays as people scroll.

You already have a page. It says the right things and nobody remembers it. Scrolldio rebuilds it as one long scene that moves as the visitor scrolls: the product turning, the story arriving a beat at a time, motion that carries the reader to the end instead of asking them to keep reading.

You say what the business is, who it is for, and what you want visitors to do. From there it runs the job in order: agree the look, settle it on one test image before making the rest, generate the matching scenes, turn them into clips that loop without a visible jump, compress them so the page still loads fast, build the page, then open it in a browser and check the real thing on a desktop and a phone.

You approve at each step, and the expensive steps wait for you. You finish holding the page, the images behind it, and a written record of how every one of them was made.

Validated: complete and repeatable

Distribution details are being finalized. Read how we admit skills →

Who it’s for

Who it’s for and what your agent needs

Designers and developers who build websites for a living
Freelancers and small studios delivering landing pages for clients
Founders who build their own site with an AI assistant
Product marketers putting together a launch page
Builders who work mostly in visual tools and are comfortable directing an AI assistant

Minimum (stills-only)

An AI coding agent with Agent Skills support, or the ability to load SKILL.md
Filesystem and shell access
One usable image-generation tool
A browser for visual verification
ImageMagick or another image inspection and conversion tool

Full motion

An image-to-video backend
ffmpeg and ffprobe
Enough storage for source clips and web encodes

What’s inside

What’s inside the bundle

Scrolldio Agent Skill
A router that checks the image and video tools you already pay for before suggesting new spend
A setup guide for running inside the Google Antigravity CLI and the Gemini CLI
An optional adapter for Google's Gemini image and Veo video APIs, if you choose to pay for them
A backup media route through the Higgsfield MCP connector
32 visual directions to choose from, each a different way the page can look and move
Every prompt the skill uses, for the look, the scenes, and the motion
The JavaScript that ties the artwork to the scrollbar, ready to drop into your own site
A plain HTML starter page that needs no framework
Recipes for turning raw clips and images into small, fast-loading web files
Templates for the tool check, the scene plan, the production log, and the record of what you approved
Checklists for desktop, mobile, reduced motion, accessibility, and the final release pass
Installation and troubleshooting guides

Proof

What the evidence shows

Three sample briefs were each run in two configurations, once with the skill loaded and once without it, and all 24 assertions passed in both configurations. What that shows is completeness and standardization: each run with the skill produced the same four named document types, covering the whole planning job. It is not evidence that the skill performs better than a capable assistant working without it.

A freight and logistics brief, planned as one continuous flight through the page, using the image and video tools the session already had
A community science museum brief, planned with still images only, on a session with no video subscription
A modular synth brief, planned as a hop between separate scenes, written so it did not depend on any one agent program, with the Higgsfield MCP connector already attached
Runs with the skill produced the same four documents every time: a tool check, a scene plan, a production log, and a record of what was approved, each matching a template in the bundle
Every check was graded by reading the files the run produced, not by asking the skill how it did

This test shows the skill is complete and repeatable. It does not show that the skill is faster or better than an assistant working without it, and we will not present it that way.

In production

vois.so Built with the skill: the Vois product site was built using Scrolldio, from its illustration system to its production pipeline.

The film at the top of this page was made by running Scrolldio from start to finish. Every image and clip in it has its prompt, its source, and its quality checks written down.

Not included

What it doesn’t include

Subscriptions to the image or video services you choose
Paid API credit for those services
Hosting for the finished site
Generated media: the images and clips are produced during the run, through your own accounts
Legal clearance for the pictures, fonts, and brand assets you supply

The bundle contains standard Markdown, Python, JavaScript, HTML, CSV, and JSON. It avoids tool names and file paths that only work inside one particular assistant. It is built for assistants that support the Agent Skills format, or that can be pointed at SKILL.md and told to follow it, with the capabilities listed above; exact skill installation steps vary by client.

Install, updates, support

Install, updates, and support

Download the zip and extract it into a directory named scrolldio, with SKILL.md at the directory root.
Copy the complete scrolldio directory into the skills directory your AI coding assistant recognizes, at the user level or the project level.
Restart or refresh your assistant, confirm scrolldio appears in the available skills, then ask the agent to use it for a cinematic scroll website task.

Updates. Shipped version: 1.0.0.

Support covers

Installation of the skill bundle, help understanding what a step means, the included scripts, and reproducible defects in bundled assets.

Outside the boundary

Provider outages, provider billing, custom website implementation, creative production services, hosting, legal clearance, and guarantees that a third-party model will accept a prompt.