In this article
video-to-gif
| Field | Value |
|---|---|
| Kind | skill |
| Source | .github/skills/experimental/video-to-gif |
| Invocation | Invoked directly as /video-to-gif, or loaded on demand by referencing agents |
| Interactive | No |
What it does
Video-to-GIF conversion with FFmpeg two-pass optimization
When to use it
Use this skill for a short, silent animation of an existing video, such as a UI
interaction for documentation. Its default two-pass process generates and applies
a palette; single-pass trades color quality for speed. Frame rate, width, time
range, dithering, and loop behavior are configurable. Use demo-video instead
when narration and an MP4 are required.
FFmpeg must be available, with ffprobe used for HDR detection. Explicit input and output paths avoid selecting the wrong similarly named recording.
Example usage
Ask: /video-to-gif Convert seconds 5 through 15 of the attached synthetic UI recording to output/search-demo.gif, 640 pixels wide at 10 fps. Keep two-pass palette generation. On Windows, these inputs map to the skill's convert.ps1
parameters -InputPath, -OutputPath, -Start 5, -Duration 10, -Width 640,
and -Fps 10.
Expect a linked GIF with proportional height and the selected interaction only.
Success includes visual inspection for readable text, acceptable colors, timing,
and file size. HDR footage is tone-mapped; -Tonemap can select a different
algorithm when needed. Keep recordings free of secrets and personal information,
confirm before replacing an existing output, and do not promise a fixed compression
ratio before inspecting the result.