Writes structured Seedance video-generation prompts with image-reference anchoring, timecoded shot lists, identity-continuity rules, cinematic color grading, and natural-sound direction. Use when the user asks for help with Seedance prompts, AI video prompts, text-to-video scene descriptions, cinematic video prompts, or any request to script a short generated film shot-by-shot. Trigger whenever the user mentions Seedance, video generation, text-to-video, AI video, shot list, scene prompt, or wants to turn a still concept into a timed video brief — even if they don't say "Seedance" by name.
91
88%
Does it follow best practices?
Impact
96%
2.28xAverage score across 3 eval scenarios
Passed
No known issues
Reference anchoring, color grade, sound negation
Image anchor syntax
0%
100%
Anchor at top
0%
100%
Locks stated
0%
100%
No re-citation in beats
100%
100%
Timecoded beats format
0%
100%
Duration within bounds
0%
100%
Named color grade at top
0%
100%
Color grade restated in closer
0%
100%
Diegetic sound only
100%
100%
Music negation
100%
100%
Continuity clause at end
0%
100%
Quality closer present
25%
100%
No marketing adjectives
100%
100%
Beat structure, aftermath hold, sound direction
4–6 beat count
0%
0%
Monotonic timestamps
0%
100%
Aftermath hold beat
70%
100%
Total duration ≤ 15s
50%
100%
Sound as directive action
8%
100%
Crowd/music negation
0%
100%
Consolidated closing paragraph
25%
100%
Continuity clause once
0%
100%
Quality closer with negation
25%
100%
Named color grade
25%
100%
No marketing adjectives
100%
100%
Identity continuity, image real-work, terse prose
Two image anchors declared
75%
100%
Each anchor locks specifics
100%
100%
Images do real work
90%
100%
Continuity clause prohibitions
66%
100%
Continuity clause placement
37%
100%
Terse beat prose
25%
100%
No marketing adjectives
100%
100%
Aftermath beat present
100%
100%
Named color grade
37%
100%
Music/crowd negated
25%
100%
Six layers all present
66%
100%
78df912
Table of Contents
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.