Write video prompts with AI
Develop a video brief into timed visual and sound instructions, preserving approved story details and carrying a consistent state between generated clips.
Write video prompts with AI that describe how a scene unfolds, not just how a frame looks. The video-prompt-writer skill combines action, staging, camera, lighting and sound. Existing story facts and approved dialogue remain fixed unless you authorise changes.
It delivers prompt text rather than generating a film. A longer story may need a connected pack of clips, but the plan must respect the actual target tool. Unknown duration or reference capabilities remain explicit conditions, not invented support.
What it does
- Turn a complete story into a prompt pack
- Give a short clip meaningful action timing
- Separate camera movement from subject movement
- Keep handover states consistent between clips
- Define which sound comes from which source
- Revise a bounded section without changing the rest
Install and use
- Provide the complete folder and references to a file-capable agent. Writing a prompt does not require a video-service account or submit a generation job. Reference inspection depends on the host’s actual media tools.
- Try: “Write a 12-second prompt for a notebook opening on a desk. Use a gentle camera push and audible page movement, with no added lettering.” Specify the output language and any known target.
- Identify fixed details and real reference assets, then choose one clip or a connected sequence. Assign action, camera and sound deliberately; do not claim an asset was uploaded or a platform capability verified when it was not.
- Review story coverage and handover states. Each independently submitted clip uses local time from zero; check the actual generated material separately before accepting continuity.
FAQ
Source
Made by JinUnique. Original skill: video-prompt-writer, published on GitHub under the MIT licence.
View on GitHub