clipify-video-clip-generator

Convert long videos into vertical clips with face tracking and captions.

4|Updated May 16, 2026
One-click install
npx skills add https://github.com/Aradotso/devtools-skills --skill clipify-video-clip-generator
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: clipify-video-clip-generator
Source: https://github.com/Aradotso/devtools-skills/tree/main/skills/clipify-video-clip-generator
Command: npx skills add https://github.com/Aradotso/devtools-skills --skill clipify-video-clip-generator

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ffmpeg, openai-whisper, numpy, libx264, ass (ASS subtitles via ffmpeg).

What problem does it solve?

Turning long-form videos into social-ready clips is time-consuming because it requires transcription, finding highlight moments, reframing for vertical platforms, and generating engaging captions.

Core Features & Use Cases

  • Auto-detect clip-worthy moments: Uses Whisper transcription to propose 3–5 candidate clips based on punchlines, reversals, awkward pauses, and audio peaks.
  • Reframe 16:9 to 9:16 with face-tracking panning: Creates vertical crops using motion-energy speaker region analysis (no ML face models) and generates ffmpeg crop/pan expressions.
  • Burn opus-style word-level captions: Produces karaoke/opus-like ASS subtitles with word-level highlighting and burns them into the final clip.
  • Hardware-accelerated rendering (macOS VideoToolbox): Uses platform-friendly decoding/encoding flags for faster turnaround.

Quick Start

Clone the clipify repo into your Claude Code skills directory and install ffmpeg plus the required Python packages, then restart Claude Code so you can run the /clipify command.

Frequently Asked Questions about clipify-video-clip-generator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert long videos into vertical clips with word-level captions?

To convert long videos into vertical clips, this tool transcribes audio with Whisper, detects highlight segments, reframes 16:9 to 9:16 using motion-energy face tracking, and burns word-level ASS captions into the output.

How does 16:9 to 9:16 video reframing work without machine learning face models?

16:9 to 9:16 reframing uses motion-energy speaker region analysis to track subjects without ML face models, generating ffmpeg pan crop expressions to keep the speaker centered in the vertical frame.

Do I need ffmpeg and openai-whisper installed to generate social media clips?

Yes, you need ffmpeg and the openai-whisper plus numpy toolchain installed to transcribe audio, generate ASS subtitle files, and render the final vertical clips with burned-in captions.

Can I use hardware acceleration to speed up video clipping and rendering on macOS?

Yes, hardware-accelerated rendering is supported via macOS VideoToolbox, using platform-friendly ffmpeg decoding and encoding flags to achieve faster turnaround for clip generation.

What types of videos work best for automated highlight detection and clipping?

Automated highlight detection works best for interviews, podcasts, and multi-speaker conversations, where Whisper transcription identifies punchlines, reversals, awkward pauses, and audio peaks to propose candidate clips.

Are there limitations when using motion-energy face tracking for vertical reframing?

Motion-energy tracking relies on speaker region movement rather than ML face models, so limitations arise when subjects are stationary or multiple speakers overlap, potentially causing less accurate vertical crop framing.