Prepare web video with AI
Encode a browser-ready clip locally, generate its poster and reserve the right amount of space for it on the page.
Prepare web video with AI when a high-resolution screen recording is destined for a much narrower article column. The web-video skill uses a local encoding script and reports the dimensions and file sizes it produced. An agent can use those results to adapt the embed to your existing video component.
A silent loop, a narrated demo and a transparent animation need different settings. A quick draft helps check the framing before a longer encode. Transparent footage also needs special handling for Safari, so the source export and the available encoders matter.
What it does
- Choose a sensible maximum width for the actual page.
- Remove the audio track from a genuinely silent background loop.
- Make a poster image for the moment before playback.
- Keep the source order when adapting the embed to the site.
- Reserve the display dimensions to reduce movement during loading.
Install and use
- Provide a video you have permission to process, FFmpeg, ffprobe and a separate output directory. Transparent HEVC encoding needs macOS avconvert. Reusing an output location can overwrite earlier results.
- Ask, “Prepare this muted product demo for an article column, keep the original and show me a quick framing draft first.”
- Check the final poster and playback in the target browsers, including the background behind any transparent areas.
FAQ
Source
Made by Evil Martians. Original skill: web-video, published on GitHub under the MIT licence.
View on GitHub