If you're working with Seedance 2.0 and tired of manually wrangling multiple images, videos, and audio files into coherent prompts, this handles that translation layer for you. It takes your scattered assets and vague creative direction and outputs a structured JSON prompt with optimized parameters. The internal recognition and mapping logic is hidden, so you just feed it your Mona Lisa image, a Coke can, and "make it cinematic," and it figures out the rest. It's narrow in scope but solves a real problem if you're doing multimodal video generation and don't want to become a prompt engineering expert yourself.
npx skills add https://github.com/pexoai/pexo-skills --skill seedance-prompter