robonuggets/seedance-skill — reverse-engineered prompt

Build me a Claude Code skill that can turn screenshots or an app website into a short cinematic promo video using Seedance 2.0 through Fal AI.

I want to be able to give it something simple like a URL, App Store link, screenshots, or just a text idea, and have it create a glossy liquid glass style motion graphics clip with floating devices, translucent panels, dark backgrounds, reflections, and smooth camera movement. It should know how to grab screenshots when needed, upload the images to Fal, call the Seedance model, and download the finished video.

Please include clear setup instructions, especially where to put the skill file and how to add the Fal API key. Add practical defaults for quick test clips and nicer 10 second promo videos, with rough cost guidance. Also include prompting rules so it avoids messy generated text or logo descriptions, uses reference images properly, and keeps the workflow reliable. Look up current Fal docs online if you need to.