dLazy Happy Horse 1.0 Video Model

Problem it solves

When building script or agent pipelines for video generation, teams often hit three issues: different tasks—text-to-video, first-frame-to-video, reference-to-video, and video editing—require different parameter shapes; local image or video assets must be uploaded to media storage; and asynchronous jobs need polling, status handling, and error recovery. dlazy-happyhorse-1.0 condenses these concerns into a single dLazy CLI entry point.

How it works

The skill is a thin client over the dLazy hosted API. When happyhorse-1.0 is triggered, it calls dlazy happyhorse-1.0, sending prompts, parameters, and local media references to api.dlazy.com. Local files are uploaded to files.dlazy.com, and generated output URLs are served from the same media host. Core capabilities include:
- Text-to-video (t2v): generate video from a textual description.
- First-frame-to-video (i2v): continue generation from a starting frame.
- Reference-to-video (r2v): generate with reference assets.
- Video editing (edit): perform editing tasks on existing video.
Requests can wait synchronously. With --no-wait, the response returns outputs: [] and a task.generateId; the job can then be polled with dlazy status <generateId> --wait.

Boundaries and notes

The skill requires a dLazy API key. unauthorized responses indicate missing or invalid credentials, while insufficient_balance means the account needs credits. It does not run local inference and does not hide the SaaS upload boundary, so it fits workflows that already accept dLazy’s cloud file path.