long-to-shorts
Skill0 starsChanged 18 days agoMITView on GitHub
Turn a long recording (screen capture with face cams, a podcast, a call, a stream of your own) into ready-to-post vertical 1080x1920 Shorts/Reels/TikToks - pick the best moments from the transcript, cut the dead air, lay out hook, content pane, face cams, captions and CTA, add synthesised music and sound effects, render with ffmpeg. Use when the user wants Shorts or Reels cut from a long video, "clip the best parts", or a vertical version with captions and a hook.
- Installs packages
1 other project uses a version of this.
Adopted by
Most recently taken up by
- vidmoat/skillstheir own version, first seen Oct 1, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- vidmoat/skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository13 other files it gives its agents.
- 2bft-video-editorSkill
- asset-librarySkill
- comedy-editing-timingSkill
- editor-osSkill
- edit-videoSkill
- ffmpeg-edit-toolkitSkill
- local-av-toolsSkill
- motion-graphics-for-videoSkill
- quest-lab-editorSkill
- talking-head-polishSkill
- transcribe-localSkill
- video-editing-craftSkill
- video-sound-and-musicSkill
The file
Files
Long recording to Shorts
Claude reads the transcript, picks the moments, writes a plan, and these scripts do the cutting and rendering. Everything is driven by JSON; times are SOURCE seconds. Only use footage of people who agreed to appear.
Needs
ffmpeg on PATH, Python 3.9+, pip install numpy pillow. Fonts: it uses Segoe UI / Impact / Arial / DejaVu if found; set CLIPKIT_FONTS=/path/to/fonts or "fonts_dir" in the spec to use your own. Music and effects are synthesised by synth_audio.py (no samples, no copyright).
Workflow
- Transcribe the long file with
transcribe-local. Read it all. - Pick 30-60 s moments that stand alone: a hook in the first line, a payoff, no needed context. Prefer a surprising claim, a number, a funny reaction, a clear how-to step.
- Find the layout in the source: pull a frame (
ffmpeg -ss T -i in.mp4 -frames:v 1 f.png), look at it, and note pixel rects for the content area and each face cam. - Write
plan.json(format in the docstring ofscripts/make_shorts_spec.py): per clip a name, hook (a question or claim, under 8 words),ranges([start, end, rect_name, tight]), captions (short lines timed to the audio), a fewsfxon the punchline, acta. - Generate specs:
python -I scripts/make_shorts_spec.py plan.jsonsplits each talk range at its pauses (dead air removed) and writes one spec per Short into_specs/. - Render:
python -I scripts/build_react_short.py _specs/short_<name>.json --out out/(face-cam layout,"cams"in the plan) orscripts/clipkit.pyfor the two-pane layout (top/bottomkeys, see its docstring). You get<name>.mp4(1080x1920, H.264, -14 LUFS) and<name>.srt. - Check: pull frames across the Short and look at them, transcribe the output, proofread every caption, say which sounds are synthesised placeholders. A person must watch and listen before posting: the agent cannot hear the audio or judge whether a joke lands.
Spec features
Hook box sliding in, progress bar, caption pill on the seam, zoom punch-ins (zooms), time-lapse ranges (speed > 1, muted), blurred panes for people who must not be recognisable ("blur": true), whooshes on jumps, side-chain ducked music, CTA, watermark.
Tools in scripts/
| File | Role |
make_shorts_spec.py | plan.json -> one spec per Short, auto-tightening the talk |
build_react_short.py | face-cam + content pane layout renderer |
clipkit.py | two-pane layout renderer and the shared toolkit (captions, hook, timeline) |
synth_audio.py | numpy-synthesised music beds (groove, bounce) and effects (whoosh hit pop ding boing riser airhorn) |
A runnable example on synthetic footage is in examples/04-shorts-from-a-long-recording of this repo.
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.