rea-video
JuneYaooo / rea-video
Basic information
About rea-video
Reverse-engineers a video shot by shot: was each shot filmed, AI-generated, 3D, animated or rendered from code, plus captions, audio and edit rhythm, and how to remake it. Writes a human report and a layered VIDEO.md for agents.
Install rea-video
npx skills add https://github.com/JuneYaooo/rea-video --skill rea-videoHow to use rea-video
- 1
Easiest route from the README: ask your agent to install it and run it in one go, for example "Install https://github.com/JuneYaooo/rea-video, then break down this video: /path/to/video.mp4". The agent installs the CLI and learns the workflow.
- 2
Or install the CLI yourself (needs ffmpeg and Python 3.10+). If pip reports externally-managed-environment, use pipx with the same spec, or run scripts/install_skill.sh from the repo, which also links the skill into your agents.
pip install "rea-video[ocr] @ git+https://github.com/JuneYaooo/rea-video" - 3
Break a video down. Use --lang zh or en for the report language; add --env-file and --vlm only if you configured a vision model.
rea-video decompose ref.mp4 -o out/ref --lang en - 4
Read the result: out/ref/report.html is for you, out/ref/VIDEO.md is the layered version your agent reads. rea-video summary out/ref/breakdown.json flags low-confidence shots.
- 5
Optional: an agent that can see images runs rea-video review out/ref, checks the flagged shots and merges its verdicts with rea-video apply-review out/ref.
What it does
Classifies every shot (live action, AI video, avatar, 3D, 2D, stop-motion, code/template render, screen recording, still) with evidence and confidence
Breaks down captions and styled text (SRT/ASS timing), stickers and picture-in-picture, cuts and beat sync, voice, music BPM and sound effects
Gives a remake plan per shot: shooting setup and lighting for live footage, reverse-engineered prompts for AI shots, a Remotion approach for code shots
Outputs report.html for people and VIDEO.md, a layered breakdown agents read with fewer tokens
Requirements and notes
- What it decides per shot: live action, film clip, AI video, avatar, 3D, 2D animation, stop-motion, code/template render, screen recording, pan-and-zoom still or static frame, plus AI edits on live footage (face swap, lip-sync, restyling).
- Beyond visuals: caption timing and styling (exports to SRT/ASS), stickers and picture-in-picture, cut rhythm and beat sync, human vs AI voice, music BPM and sound-effect timing, and which material you would have to supply yourself.
- Remake plan: a shooting and lighting plan for live shots, a reverse-engineered prompt per AI shot, and a Remotion approach for code shots. With capsule-cinema installed, AI shots can be generated and placed back on the timeline.
- Cost and accuracy: tools-only mode is offline and free; the README reports 98–99.6% accuracy on original files and 88.6% on recompressed WeChat videos, rising to 99.6–100% with a vision model or agent visual review. A vision model (Gemini, Doubao or any OpenAI-compatible endpoint) is optional and billed by that provider.
- Limits: one video at a time (multi-episode dramas: rea-toon); it does not render a new video by itself. Use it to learn structure and technique, then remake with material you have rights to; don't repost others' footage, music or likeness.
- Optional speech recognition: pip install "rea-video[ocr,asr] @ git+https://github.com/JuneYaooo/rea-video" adds local transcription (Chinese, English, Japanese, Korean, Cantonese) when a video has no burned-in captions.
Sources
- GitHub: JuneYaooo/rea-video (README, Apache-2.0)
- rea-video SKILL.md and setup reference
- VIDEO.md format
Checked against the source repository on 2026-11-15.
More skills in Creative & media
vercel-react-native-skills
vercel-labs / agent-skills
React Native and Expo best practices for building performant mobile apps. Use when building React Native components, optimizing list performance, implementing animations, or working with native modules. Triggers on tasks involving React Native, Expo, mobile performance, or native platform APIs.
remotion-to-hyperframes
heygen-com / hyperframes
Port an existing Remotion (React) composition's source to HyperFrames HTML. Use ONLY on an explicit ask to port/convert/migrate/translate a Remotion source — one-way, Remotion-only. A passing Remotion mention, reference-only code, or "make something like my Remotion video" is a fresh build (/general-video). Unclear → /hyperframes.
gsap
heygen-com / hyperframes
GSAP animation reference for HyperFrames. Covers gsap.to(), from(), fromTo(), easing, stagger, defaults, timelines (gsap.timeline(), position parameter, labels, nesting, playback), and performance (transforms, will-change, quickTo). Use when writing GSAP animations in HyperFrames compositions.
website-to-video
heygen-com / hyperframes
Capture a general website/URL and turn it into a video OF the site — tour, showcase, or social clip built from captured screenshots and the site's own brand assets. Use for portfolio / blog / docs / landing-page showcases. Not a product launch or promo, even from a URL (/product-launch-video). Unclear → /hyperframes.
pr-to-video
heygen-com / hyperframes
Turn a GitHub pull request (a PR URL, owner/repo#N, or 'this PR' in a checked-out repo) into a code-change explainer video — changelog, feature reveal, fix, or refactor walkthrough built from the diff, commits, and files: the input is a code change, not a website. Not a product promo (/product-launch-video) or a no-PR topic explainer (/faceless-explainer). Unclear → /hyperframes.
music-to-video
heygen-com / hyperframes
Turn a music track (an audio file, a video to pull audio from, or a track generated from a mood brief) into a beat-synced video — lyric video, slideshow, or kinetic promo. The music drives all pacing; any user-supplied images/videos are cut onto the same beat grid, and a complete video needs zero assets. Narrated pieces → the input-matched workflow (see /hyperframes). Unclear → /hyperframes.
Comments