Seedance Prompt — Short

What it does

Seedance Prompt — Short provides a concise, production-friendly workflow for authoring short prompts tailored to Seedance 2.0. It prescribes a Five-Layer Stack (Core Intent, Visuals, Audio, Technical) and a Compression Engine that prioritises verbs, film language, and model-friendly shorthand. The skill helps creators convert high-level ideas into tight prompts aimed at 30–100 words (or a 2000-character hard limit), making prompts predictable and repeatable across T2V, I2V, V2V, and R2V pipelines.

When to use it

Use this skill when you need short, robust prompts for quick iterations, production-ready shots, or constrained-caption scenarios (ads, thumbnails, short-form scenes). It's useful during prompt authoring, debugging prompt failures, and prepping inputs for downstream models that require compact, deterministic prompts.

What's included

• Scripts: none bundled in the skill directory (has_scripts=false).
• References: none bundled (has_references=false).
• Instructions: outlines the Five-Layer Stack, practical compression heuristics (verbs over adjectives, show-not-tell, film language), and guidance for delegating constraints via @Tags. The frontmatter also links to companion workflows (e.g., seedance-interview) for guided prompt building.

Compatible agents

Best used with agents or CLIs that assemble and validate prompts (Gemini CLI, Copilot-style assistants, or custom prompt toolchains). The guidance is model-agnostic but optimised for quad-modal Seedance pipelines.