video-to-shorts
Skill0 starsChanged 2 days agoMITView on GitHub
Turn a long video (podcast, interview, webinar, tutorial, livestream VOD) into captioned 9:16 YouTube Shorts, TikToks and Reels, an open-source Opus Clip alternative for Claude Code, Codex and Cursor. Transcribes once, lists every sentence with timestamps so the agent picks the best moments (no extra LLM key), snaps cuts to whole sentences, reframes to 1080x1920 (centre crop or blurred fill), burns 1-3 word captions, normalises to -14 LUFS, and can open each clip with a spoken hook in the creator's own cloned voice. Use when the user asks to clip a long video, make shorts from a podcast or stream, repurpose a video, find viral moments, or wants an AI clip maker. Speech recognition and voice by Miragea; stdlib Python + ffmpeg.
- Reads credentials
2 other projects use a version of this.
Adopted by
Most recently taken up by
- WhiteTowerAI/cut-as-codetheir own version, first seen Oct 13, 2026
- augchan42/inkstonetheir own version, first seen Jun 7, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- WhiteTowerAI/cut-as-code★ 16same name
- augchan42/inkstone★ 5same 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 repository10 other files it gives its agents.
- llms.txtllms.txt
- audiogram-clipsSkill
- auto-brollSkill
- hook-variantsSkill
- remotion-narrationSkill
- screen-recording-voiceoverSkill
- shorts-voiceoverSkill
- ugc-video-adsSkill
- video-captionsSkill
- video-dubbingSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 24
Check MIRAGEA_API_KEY is set (never print it).
- Very specific: line 30
Read talk.moments.md and write the final talk.clips.json yourself.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Video to Shorts (AI clip maker)
Long video in, a folder of ready-to-post vertical clips out. You (the agent) are the "virality model": read the timestamped sentences and choose moments a viewer would understand with no context. The script does the precise work: sentence-safe cuts, 9:16 framing, captions, loudness, and an optional voiced hook in the creator's own voice.
When to use
- "Make 5 Shorts from this podcast episode."
- "Clip the best moments of my webinar / stream VOD for TikTok and Reels."
- "Find the viral parts of this interview and caption them."
Steps
- Check
MIRAGEA_API_KEYis set (never print it). Checkffmpeg -version. - Transcribe and list moments (billed once, then cached in
<video>.transcript.json, shared with the video-captions skill):python scripts/clips.py moments talk.mp4 --suggest 5This writestalk.moments.md(one sentence per line,[start-end]seconds) and a keyword-scored drafttalk.clips.json. - Read
talk.moments.mdand write the finaltalk.clips.jsonyourself. Good clips:- 20-60 s, start on a claim, question or number, never on "and / so / but";
- one complete idea with its payoff; no inside references ("like I said earlier");
- a
title(under 60 chars, for the upload) and, if the opening is slow, ahook: 5-9 words, under ~3 s spoken, that states the payoff ("This one rule doubled our retention.").[{"start": 12.3, "end": 48.0, "title": "...", "hook": "..."}]
- Preview for free:
python scripts/clips.py render talk.mp4 talk.clips.json --dry-runFix any warnings (too short, too long, long hook). - Render:
python scripts/clips.py render talk.mp4 talk.clips.json -o shorts_out--frame crop(default) for one person on camera;--frame fitfor slides, screen recordings or two people side by side (whole frame over a blurred fill).- Hooks are voiced only if a voice is set; only clone a voice the user owns or may use.
--no-hookskips them.
- Report each clip: file, length, title, and the hook. Offer to re-cut any clip by editing its times in clips.json and re-running (the transcript is cached, so only hooks are billed).
Output
shorts_out/clip_01.mp4 (1080x1920, H.264/AAC, 30 fps, -14 LUFS), clip_01.srt, and
manifest.json: {"id","title","start","end","hook","seconds","hook_seconds","warnings","file","srt"}.
Honest limits
- Framing is a centre crop or blurred fill, not face tracking. For a speaker off-centre, use
fit. - Speech recognition is cloud-based (Miragea); the video itself is never uploaded, only a 16 kHz mono audio track.
- English is the focus.
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.

