miragea-skills
llms.txt0 starsChanged 2 days agoMITView on GitHub
Open-source Agent Skills (SKILL.md folders) for Claude Code, Codex, Cursor and other agents that add voiceover, captions and video-narration abilities, powered by the Miragea voice and speech-recognition API (https://miragea.ai/v1). Python standard library + ffmpeg; MIT licensed.
- Reads credentials
In this repository10 other files it gives its agents.
- audiogram-clipsSkill
- auto-brollSkill
- hook-variantsSkill
- remotion-narrationSkill
- screen-recording-voiceoverSkill
- shorts-voiceoverSkill
- ugc-video-adsSkill
- video-captionsSkill
- video-dubbingSkill
- video-to-shortsSkill
The file
Files
# miragea-skills > Open-source Agent Skills (SKILL.md folders) for Claude Code, Codex, Cursor and other agents that add voiceover, captions and video-narration abilities, powered by the Miragea voice and speech-recognition API (https://miragea.ai/v1). Python standard library + ffmpeg; MIT licensed. ## Skills - [video-captions](https://github.com/yyert123/miragea-skills/tree/main/video-captions): caption any video or audio into readable SRT/VTT (whole words, 42-char lines) and burn captions into MP4, including 1-3 word Shorts/Reels/TikTok captions. `python3 scripts/captions.py input.mp4 --burn` - [remotion-narration](https://github.com/yyert123/miragea-skills/tree/main/remotion-narration): script to per-scene voiceover MP3s, measured durations as Remotion durationInFrames, word-level Remotion Caption[] JSON and SRT. `python3 scripts/narrate.py script.md --voice VOICE_ID` - [screen-recording-voiceover](https://github.com/yyert123/miragea-skills/tree/main/screen-recording-voiceover): add a timed AI voiceover to a silent screen recording; finds steps, voices one line per step, auto speed-up or freeze-frame when a line runs long, ducked music, SRT/VTT, burned captions. `python3 scripts/voiceover.py demo.mp4 demo.script.txt --voice VOICE_ID` - [shorts-voiceover](https://github.com/yyert123/miragea-skills/tree/main/shorts-voiceover): text script to a finished 1080x1920 YouTube Short / TikTok / Reel with voiceover, captions timed from the real speech, Ken Burns images or B-roll or gradients, ducked music, -14 LUFS. `python3 scripts/short.py script.txt --voice VOICE_ID` - [video-to-shorts](https://github.com/yyert123/miragea-skills/tree/main/video-to-shorts): AI clip maker and open-source Opus Clip alternative; long video (podcast, webinar, stream VOD) to captioned 1080x1920 YouTube Shorts/TikToks/Reels, agent-picked moments, sentence-safe cuts, crop or blurred fill, -14 LUFS, optional spoken hook in a cloned voice. `python3 scripts/clips.py render talk.mp4 talk.clips.json` - [hook-variants](https://github.com/yyert123/miragea-skills/tree/main/hook-variants): A/B test the first 3 seconds of a Short, TikTok, Reel or ad; voices 4-6 hooks in the same AI voice, trimmed, -14 LUFS, measured length with a budget flag. `python3 scripts/hooks.py hooks.txt --budget 3` - [video-dubbing](https://github.com/yyert123/miragea-skills/tree/main/video-dubbing): AI video dubbing and video translation in your own cloned voice (own-voice dubbing); agent translates each line within a time budget, lines start on the original, max 1.2x speed-up, music kept in pauses, translated SRT, -14 LUFS, full-length audio track for YouTube multi-language audio. Open-source alternative to ElevenLabs Dubbing / HeyGen video translation. `python3 scripts/dub.py render talk.mp4 talk.es.dub.json` - [auto-broll](https://github.com/yyert123/miragea-skills/tree/main/auto-broll): automatic B-roll for talking-head videos, podcasts, tutorials and faceless videos; the agent picks cutaways from a word-timed transcript, every cut lands on the spoken word, speaker audio untouched, cover-crop 16:9/9:16, Ken Burns stills, editorial checks (hook, flash cuts, coverage), MP4 + CMX 3600 EDL for Premiere/Resolve; own clips or free Pexels stock; faceless mode voices a script in a cloned voice. `python3 scripts/broll.py render talk.mp4 talk.broll.json` - [ugc-video-ads](https://github.com/yyert123/miragea-skills/tree/main/ugc-video-ads): UGC-style video ads for TikTok, Reels, YouTube Shorts and Meta without a video model; agent writes 3-5 hooks and a direct-response body, every hook x body voiced in your own cloned voice, captions and overlays in platform safe zones, end card + CTA, 3-second hook budget, ad-claim lint, 9:16/4:5/1:1, -14 LUFS, manifest for Ads Manager. `python3 scripts/ads.py render ad.json --ratios 9:16,4:5` - [audiogram-clips](https://github.com/yyert123/miragea-skills/tree/main/audiogram-clips): podcast to video / audiogram maker; captioned animated waveform MP4 from any audio clip for TikTok, Shorts, Reels, LinkedIn, X; Miragea ASR captions, cover art, 4 aspect ratios. `python3 scripts/audiogram.py clip.mp3 -o clip.mp4` ## Facts - Requires MIRAGEA_API_KEY (environment variable). Keys are never printed or stored. - Captions cost 3 Miragea points per audio second; speech is billed per character; cached re-runs are free. - Voice: a saved Miragea voice id, or a public reference clip URL (with consent). - Install: `npx skills add yyert123/miragea-skills` or copy a skill folder into your agent's skills directory.
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.

