Clips Machine

Transform long videos into viral short-form clips. Auto-detect best moments, add trendy captions, export for TikTok/Reels/Shorts. Self-contained, no external modules. 100% free tools.

Security Scan

The skill's code, runtime instructions, and required binaries are consistent with its stated purpose of creating short clips from long videos and do not request unrelated credentials or hidden network endpoints.

Purpose & Capability

Name/description (video -> short clips) match the actual behavior: it invokes yt-dlp, ffmpeg, and whisper-cpp to download, transcribe, cut, crop, caption and export clips. The declared required binaries are appropriate and necessary for the stated functionality.

Instruction Scope

SKILL.md and the script restrict actions to downloading permitted domains, local transcription, clip detection, cutting, captioning, and exporting into a local output folder. No instructions ask the agent to read unrelated system files, secrets, or to send data to unknown endpoints.

Install Mechanism

There is no install spec (instruction-only install), and the script itself uses only standard subprocess calls to system binaries. The user must install ffmpeg, yt-dlp and whisper-cpp manually as documented; no risky third-party download/execute steps are embedded.

Credentials

The skill declares no required environment variables or credentials. The code optionally reads OUTPUT_DIR (to set an output path) and resolves local whisper model paths; no secrets or unrelated credentials are requested.

Persistence & Privilege

always is false and the skill does not request persistent system-wide changes or modify other skills' configurations. It runs as a standalone script and writes outputs only to a user-writable directory.

Assessment

This skill appears coherent with its stated purpose, but before installing/run it: (1) ensure ffmpeg, yt-dlp and whisper-cpp are obtained from trusted sources and installed in a secure environment; (2) confirm you have rights to download/process any remote video you give it; (3) review or run the script in a sandbox/VM if you're concerned about executing third-party code; (4) note it writes outputs to ~/Videos/OpenClaw (or OUTPUT_DIR if set) and will download remote media—monitor disk/network usage. If you want higher assurance, review the full clips_machine.py file (the visible portions are standard) or run it on a non-sensitive machine first.

Transform long videos into viral short-form clips. Auto-detect the best moments, add trendy captions, export for TikTok/Reels/Shorts.

100% FREE tools - Runs entirely on your machine.

Love this skill? Support the creator and help keep it free: Buy Me a Coffee

What This Skill Does

1. Input any long video (YouTube URL, podcast, stream, local file)
2. Transcribe with timestamps using Whisper (free, local)
3. Detect viral-worthy moments using AI analysis
4. Cut the best 30-60 second segments
5. Caption with animated, trendy text styles
6. Export in vertical 9:16 format ready for upload

Quick Start

Turn this podcast into viral clips: https://youtube.com/watch?v=xyz

Extract the 5 best moments from my-interview.mp4 and add captions

Commands

From YouTube URL

/clips-machine https://youtube.com/watch?v=VIDEO_ID

From Local File

/clips-machine /path/to/video.mp4

Custom Number of Clips

/clips-machine VIDEO --clips 10

Caption Styles

/clips-machine VIDEO --style [style]

Available styles:

• hormozi - Alex Hormozi style (bold, word-by-word highlight) - Most Viral
• minimal - Clean white text
• karaoke - Word-by-word color change
• news - Lower third style
• meme - Impact font, top/bottom

How Viral Detection Works

The AI analyzes the transcript looking for:

1. Hook Potential - Strong opening statements
2. Emotional Peaks - Passion, excitement, surprise
3. Quotable Lines - Memorable one-liners
4. Controversial Takes - Debate-worthy opinions
5. Surprising Facts - "Did you know" moments
6. Actionable Advice - Clear takeaways

Each moment gets a "virality score" from 1-100.

Output Structure

~/Videos/OpenClaw/clips-[video-name]/
├── transcript.json      # Full transcript with timestamps
├── viral_moments.json   # Detected moments with scores
├── clip_001.mp4         # First viral clip (vertical, captioned)
├── clip_002.mp4         # Second viral clip
├── clip_003.mp4         # ...
└── summary.md           # Overview of all clips

Supported Sources

Source	Example
YouTube	https://youtube.com/watch?v=...
TikTok	https://tiktok.com/@user/video/...
Twitter/X	https://twitter.com/user/status/...
Twitch VOD	https://twitch.tv/videos/...
Local MP4	/path/to/file.mp4

Requirements

• FFmpeg (brew install ffmpeg)
• yt-dlp (brew install yt-dlp)
• Whisper.cpp (brew install whisper-cpp)

Setup

# Install dependencies
brew install ffmpeg yt-dlp whisper-cpp

# Or on Linux
sudo apt install ffmpeg
pip install yt-dlp
# Build whisper.cpp from source

Monetization

Method	Potential
Clip service for creators	$50-150/video
Monthly retainer	$500-2,000/client
Podcast clipping agency	$2,000-5,000/mo
Sell this skill	$100-300 on ClawHub

Examples

Podcast to Clips

Take this 2-hour podcast and find the 10 best moments:
https://youtube.com/watch?v=PODCAST_ID
Make them Hormozi-style with bold captions.

Interview Highlights

/clips-machine interview.mp4 --clips 5 --style minimal

Gaming Stream

Extract funny moments from my Twitch VOD:
https://twitch.tv/videos/12345
Add meme-style captions

---

Support This Project

If this skill saved you time or made you money, consider buying me a coffee!

Buy Me a Coffee

Every coffee helps me build more free tools for the community.

---

Built for OpenClaw | Powered by Whisper + FFmpeg | Support the Creator