Pet Video Script Domain

Problem

Pet short-video production often stalls in three places: the pet’s appearance drifts across shots; story, shot scale, camera movement, and duration are designed in isolation; and AI video prompts lack a consistent structure, making Seedance, Kling, and similar tools harder to use predictably. This skill is not an execution framework. It acts as a domain payload repository that structures what to do and what the output should look like, then delegates execution to Universal Task OS.

How It Works

The core capabilities include task lists, dependency topology, content requirements, and an exemplar index:
- P1 character profiling: turns breed, appearance, personality, and behavior into reusable character data.
- P2 story design: constrains premise, emotional style, conflict, and ending, avoiding flat “pure recording” content.
- P3 storyboarding: splits shots by target duration and validates shot type, camera movement, and timeline.
- P4 prompt generation: produces a shared subject description, Seedance-style English prompts, and audio notes.

The key workflow is to read pet-video-catalog.md for domain classification and dependencies, then read pet-video-requirements.md for component requirements, and finally pass the list, samples, and topology to UTOS. Character consistency is the cross-cutting guardrail: every shot reuses the same subject description and explicitly states maintain exact appearance across shots.

Boundaries

This skill depends heavily on universal-task-os. Without UTOS, it can only be used to browse task lists, requirements, and exemplars; it cannot produce storyboards, orchestrate pipelines, or generate prompts. The exemplar library is mostly marked [to be filled], so usable samples must be added by the user. The Seedance references are tuned to 2.0 prompt habits, and migration to other AI video tools still requires calibration to the target tool’s syntax and constraints.