shorts-cutter
Skill4 starsChanged 33 days agoMITView on GitHub
Find the best standalone moments in a long-form talking-head video and cut them into vertical YouTube Shorts / TikToks / Reels with a clean caption style. Use when someone asks for shorts or clips from a long form.
In this repository4 other files it gives its agents.
- interview-shot-sheetSkill
- long-form-editorSkill
- video-quality-checkSkill
- youtube-metadataSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Read the full transcript of the long form (with word timestamps).
- Distinctive: line 13
End on a payoff,: not a trailing "so yeah".
- Distinctive: line 18
Use the same word-snapped cutting as the long form: cut at the quietest point near word boundaries, shrink pauses over 0.6 s to 0.3 s.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Shorts cutter
1. Pick the clips
Read the full transcript of the long form (with word timestamps). Pick 5–8 moments, or as few as 3 from a short video. Only keep the ones that are actually good. Each must:
- Stand alone. A stranger understands it with zero context.
- Hook in the first 2 seconds: a number, a question or a bold claim. If the best line is mid-clip, start there.
- Run 20–60 s. Cut setup ruthlessly.
- End on a payoff, not a trailing "so yeah".
For each clip, write: start/end times, the hook line, why it works, and a tag for testing (hook type: number / question / claim; length: short / mid / long; topic: money / how-to / story).
2. Cut
- Use the same word-snapped cutting as the long form: cut at the quietest point near word boundaries, shrink pauses over 0.6 s to 0.3 s.
- Reframe 16:9 → 9:16 around the face (crop centred on the face; keep headroom).
- Process the voice the same way as the long form (about -14 LUFS, true peak ≤ -1 dBTP).
3. Style (keep it clean)
- One caption line at a time, white, bold, centred in the lower third, synced to words.
- One hook line at the top for the first ~2 s.
- Few or no extra graphics. Heavy editing styles fight the footage. One focal point at a time.
4. Deliver
One file per short, named NN - Title.mp4, plus a text file with each short's caption, hashtags and test tags. Post the same file and caption on all platforms so results compare fairly.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.