agentleFS
yyert123yyert123/miragea-skills/shorts-voiceover/SKILL.md

shorts-voiceover

Skill0 starsChanged 2 days agoMITView on GitHub

Make a finished vertical short video (YouTube Shorts, TikTok, Instagram Reels; 1080x1920 MP4) from a plain-text script in one command, with AI voiceover in a saved or cloned voice, word-timed pop captions taken from the real speech, Ken Burns motion on images, cropped and looped B-roll clips or animated gradient backgrounds, auto-ducked background music and -14 LUFS loudness. Use when the user asks to turn a script, tips list, story or blog post into a Short/Reel/TikTok, make a faceless short with voiceover and captions, or batch short videos. No GPU, no stock-footage API key, no Docker. English-first; voice and caption timing by Miragea.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 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.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Check MIRAGEA_API_KEY is set (never print it).

    line 24
  2. Very specific:

    Write the script: one short paragraph per scene, blank line between scenes, 120-170 words total for a 45-60 s Short.

    line 27
  3. Very specific:

    Use only music and footage you have the rights to.

    line 85

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Shorts voiceover

Script in, upload-ready Short out. Unlike tools that guess caption timing from text length, this skill transcribes the finished voice track, so every caption pops exactly when the word is spoken.

When to use

  • "Turn this script / these tips / this post into a Short (or Reel, or TikTok)."
  • "Make a faceless short with voiceover and captions." "Use my voice."
  • "Make 5 shorts from these scripts" -> run once per script file.

Steps

  1. Check MIRAGEA_API_KEY is set (never print it). Voice: the user's Miragea voice id, or a public URL of a clip of their voice plus its exact words. Suggest export MIRAGEA_VOICE_ID=.... Only clone a voice the user owns or may use.
  2. Write the script: one short paragraph per scene, blank line between scenes, 120-170 words total for a 45-60 s Short. Open with a hook in scene 1. Optional @file at the start of a scene picks its background:

    Three quick tips to make any voiceover sound professional.
    
    @broll/desk.mp4
    First, record in a small room with soft furniture. Echo is harder to fix than noise.
  3. Render:

    python3 scripts/short.py tips.txt --voice VOICE_ID                       # gradient backgrounds
    python3 scripts/short.py tips.txt --media ./broll --music bed.mp3        # user's images/clips + music
  4. Give the user video and srt from the printed JSON. Offer edits: change a scene's text and re-run; only that scene is re-voiced and the rest comes from the cache.

What is automatic

StepHow
VoiceEach scene voiced with Miragea; 0.18 s breath between scenes
Caption timingFinished voice track transcribed by Miragea; 1-3 word UPPERCASE cues, lower third
ImagesCropped to 9:16, slow zoom in/out alternating per scene
ClipsCropped to 9:16, looped or trimmed to the scene's length, original sound dropped
No mediaAnimated gradient (4 palettes, rotating)
MusicLooped, at 30%, sidechain-ducked under the voice
Loudness-14 LUFS integrated, -1 dBTP (Shorts/TikTok/Reels target)
Re-runsVoice per scene and transcript cached; unchanged scenes are free

Options

FlagDefaultWhat it does
--voice ID / --ref-url URL --ref-text TEXTenvSaved Miragea voice, or clone from a public clip
--media DIRnoneImages/clips for scenes without @file, in name order
--music FILEnoneBackground music, auto-ducked
--lowercaseoffSentence-case captions
--no-captionsoffSkip captions (no transcription cost)
--modelexpressiveor standard
-o FILESCRIPT.short.mp4Output path

Cost and limits

Voice: per character (a 150-word Short is ~850 characters). Caption timing: 3 points per second of the finished voice (~75 points for 25 s); cached afterwards. Live rates: GET https://miragea.ai/v1/capabilities. Scenes must be under 2,000 characters. A warning is printed above 3 minutes (YouTube's Shorts limit).

Troubleshooting

  • media file not found -> paths after @ are relative to the script or --media folder.
  • No such filter: gradients -> ffmpeg is older than 6.1; upgrade, or pass --media with images.
  • 401 key wrong, 402 top up at miragea.ai, 429 wait and re-run (nothing cached is re-billed).

Use only music and footage you have the rights to. See examples/ for a real script and its output.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.