A little about this skill

Create subtitle screenshot collages from a video file or accessible URL. The Agent automatically chooses a passage, extracts frames, and stacks the original burned-in subtitles into one quote image. Requires Python, Pillow, and FFmpeg.

What you’ll get

• An English Agent Skill that starts from only a video or URL, automatically chooses a passage and preserves native subtitle pixels; post-rendered dialogue remains an explicit optional request.
• A deterministic Python renderer for candidate contact sheets, subtitle-band guides, and JPEG quote collages.
• Natural-width and fixed-ratio composition instructions, timestamp JSON schemas, and output review guidance.
• Dependency diagnostics, an authorized URL acquisition reference, and the upstream MIT license.

Good fit

• Creators making dialogue collages from their own interviews, lessons, or recordings.
• Editors preserving the original look of burned-in video subtitles.
• Teams placing reviewed dialogue or translations on clean video frames with explicit disclosure.

Not for

• Inventing quotations, silently replacing native subtitles, or unreviewed automatic translation.
• Text-only Agent hosts, editable layered design exports, or automatic social publishing.
• Guaranteed access to protected online videos or guaranteed complex-script font rendering.

What you need

An Agent with local command execution, file access, and visual inspection; Python 3.10+, Pillow, and FFmpeg or imageio-ffmpeg. Scripted multilingual text requires an appropriate font. Optional URLs use yt-dlp and may require a supported JavaScript runtime and authorized access. Fonts, videos, and transcription tools are not bundled. No image-provider account is required for rendering.

Make it yours

Tell your agent about your project, audience, and the result you want. The skill provides a direction and a practical workflow; your idea makes it unique. Review its instructions before running it and make sure the required tools are available.