About

A Claude Code skill that turns long videos into social-ready clips. Point it at any video file and it finds the clip-worthy moments, reframes them 16:9 → 9:16, and burns opus-style word-by-word captions.

Most “auto-clip” tools are expensive SaaS, slow, or produce slop. This is what I actually use to clip my long-form videos for LinkedIn and TikTok. Built for talking-head dialogue — interviews, podcasts, two-person setups.

No cloud APIs, no OpenCV. Runs entirely on your machine — about 20s of work for a 20s clip on Apple Silicon.

What it does

• Finds clip-worthy segments — scans the transcript for punchlines, reversals, awkward pauses and audio peaks
• Reframes 16:9 → 9:16 with hard-cut pans that follow whoever is speaking (or split-screen for both faces)
• Face-pan uses frame-difference motion energy — no face-detection model, a few seconds of ffmpeg per clip
• Burns opus-style captions — big bold white with a yellow active-word highlight
• Match any caption look by pasting a reference image

How to use it

1. Run /clipify and paste a video file path.
2. It proposes 3–5 candidate clips with timestamps and titles — pick one.
3. Choose 9:16 / 16:9 / 1:1, then pan vs split-screen, then a caption style. It renders and opens the result.
4. Final clips land in <source-video-dir>/clipify_out/.