happyhorse

happyhorse is an integration skill for generating and editing video with Alibaba's HappyHorse 1.0 models through the inference.sh CLI (the belt tool). It exposes four related models — text-to-video (T2V), image-to-video (I2V), reference-to-video (R2V) for character preservation, and a natural-language Video Edit model — solving the task of producing or editing physically realistic video from the command line without managing model infrastructure directly.

The skill documents the belt workflow (belt login, then belt app run against a model app ID with a JSON input payload) and provides worked examples for each capability: text-to-video with duration, resolution, and aspect-ratio controls; animating a still image from a first frame; preserving characters from up to nine reference images; multi-character scenes; editing existing videos with instructions like changing a background; swapping a person for a reference character; and audio control during edits. Per-model parameter tables cover prompt, duration (3 to 15 seconds), resolution (720P/1080P), aspect ratios, seed, watermark, reference images, and audio settings, and a pricing table lists per-second costs by resolution.

Target users are content creators, marketers, and developers who want scriptable, character-consistent video for product demos, social media, and creative work. It depends on the external belt CLI (installed via npx skills add belt-sh/cli) and a paid inference.sh account. The skill only runs the belt CLI against a hosted generation service and performs no destructive local operations or unsafe execution; the main considerations are cost per second and standard media-generation content responsibilities.