| name | short-video-assembly |
| description | Use when the user wants to assemble a short-form MP4 from narration audio, stock/local clips, subtitles, music, or images; render vertical/horizontal/square shorts; concatenate clips to match audio; mux audio; burn subtitles; or build reusable video composition code. |
Short Video Assembly
Create a finished short video from script assets.
Inputs
- Audio file, usually narration.
- One or more validated video/image materials.
- Optional
.srt subtitles.
- Target aspect:
9:16, 16:9, or 1:1.
- Optional background music, title card, watermark, or platform constraints.
Workflow
- Probe inputs with
ffprobe; reject unreadable media.
- Determine target duration from narration unless the user specifies otherwise.
- Select and trim clips to cover the full duration.
- Normalize dimensions:
9:16: 1080x1920
16:9: 1920x1080
1:1: 1080x1080
- Crop/scale with center or subject-aware framing. Avoid stretching.
- Concatenate clips with ffmpeg concat or MoviePy, depending on the project stack.
- Mux narration audio. Duck background music if present.
- Burn subtitles only when requested or when platform output needs them.
- Export H.264/AAC MP4 unless the target platform requires another codec.
- Run
video-render-qa before delivery.
Defaults
- Short social videos should default to
9:16.
- Keep transitions simple unless the user asks for a style.
- Avoid auto-posting. Rendering and publishing are separate actions.