rea-video: break down how any video was made
JuneYaooo/rea-video
Give your AI assistant a video and it works out, shot by shot, how it was made: filmed, AI-generated, 3D, animated, screen recording or built in code. It also lists the captions, stickers, voice, music and sound effects, and writes a plan to remake the video with your own material.
Pack description by its author: “Give your AI assistant a video and it works out, shot by shot, how it was made: filmed, AI-generated, 3D, animated, screen recording or built in code. It also lists the captions, stickers, voice, music and sound effects, and writes a plan to remake the video with your own material.”
Video, images and mediaGives your assistant new abilities with scriptsCommands you run when needed
1. Changes your files
None of its skills edits your files.
github.com/JuneYaooo/rea-video/blob/4089199/2. Runs code on your machine
Install: pip installs rea-video and its dependencies (numpy, OpenCV, Pillow, PyYAML, videomd, onnxruntime, OCR packages). install_skill.sh may create .venv in the clone, link ~/.local/bin/rea-video and symlink the skill into ~/.claude, ~/.codex, ~/.agents skills folders. First decompose run: Downloads models into ~/.cache/rea-video/models (or $REA_VIDEO_MODELS): UVR MDX-Net (~67 MB) and BandIt v2 (~149 MB) with SHA-256 checks, a YuNet face model from opencv_zoo; with the asr extra, SenseVoice (~160 MB) and speaker models from sherpa-onnx releases; Japanese OCR model from ModelScope when --ocr-lang ja.
github.com/JuneYaooo/rea-video/blob/4089199/3. Sends data
The skill tells the assistant never to put keys into the chat, the repo or the report, and not to open the .env file. One model (BandIt v2) is downloaded from a third-party GitHub account, checked against a SHA-256 hash in the code.
github.com/JuneYaooo/rea-video/blob/4089199/4. Commercial use
Apache-2.0: keep the license and NOTICE
github.com/JuneYaooo/rea-video/blob/4089199/Checked against the pack's files at commit 4089199, on Nov 14, 2026. Tap a verdict to see its source.How we checkedReport an error
The pack has changed since AgentGid checked it on Nov 14, 2026. Some details may be out of date.
Good for you if
- You saw a short video or ad and want to know if it is real footage or AI, and how it was made
- You want to remake a video in the same style with your own product, people or footage
- You use Claude Code, Codex or another assistant that can run commands on your computer
Probably not for you if
- You want a finished video in one click: it writes a breakdown and a plan, it does not render a video
- You want to repost someone else's video: the author says it is for learning and remaking with your own material
What its answer looks like
The author's sample report for a live-action unboxing video
In one line: a top-down unboxing vlog with no voice-over, one pair of hands opening 7 home items on a carpet. How it was made: live action 99% + still image with pan/zoom 1%; 69 shots, about 1.0 s each. Then: what makes it work (the hook in the first 3 s), a picture / packaging / sound table, and the remake steps.
The skill's own example, from docs/cases/unboxing/report_en.md. More examples
What you'll notice
- A report you can open in the browser (report.html, PDF on request) with a shot list, sound and remake plan
- A layered VIDEO.md folder your assistant reads instead of the raw data
- Every uncertain call is marked as to be confirmed, with what is missing
- On the first run, the tool downloads several sound and face models from GitHub
What you need
- Claude Code, Codex
- Mixed, Apache-2.0 license
- Ask your assistant to install it from GitHub; it needs ffmpeg and Python 3.10 or newer
- Works offline and free with its own tools. A vision model key (Gemini, Doubao or similar) is optional and paid by that service; ChatCut and capsule-cinema remakes cost money
Words on this page
- Skill
- A set of written instructions your AI assistant follows. Installing one is like giving it a playbook.
- CLI
- A program you run by typing commands in a terminal.
- API
- A way for programs to talk to a service, usually with a key.
- Repository
- A project folder on GitHub with its files and history.
| Repository | JuneYaooo/rea-video · website |
|---|---|
| GitHub stars | 8,773 |
| Skills | 1 |
| Category | Video, images and media |
| License | Apache-2.0 |
| Last updated | Nov 15, 2026 · card checked Nov 14, 2026 at commit 4089199 |
| Install counts | JuneYaooo/rea-video on skills.sh, Vercel's skills directory, which shows installs and security audits per skill |
JuneYaooo/rea-video popularity over time
What rea-video does
- Splits the video into shots and the sound into voice, music and effects
- Judges each shot as one of 11 types: live action, film clip, AI video, avatar, 3D, 2D animation, stop-motion, code template, screen recording, still image with pan or zoom, still frame
- For live footage: framing, camera position, lighting setup, colour grade and editing pace
- Lists the layers on top: timed captions, styled text, titles, stickers, picture-in-picture
- Marks real material you must supply yourself (screenshots, photos, recordings)
- Writes a remake plan: AI generation units with prompts, filming sessions with a gear list, or a Remotion code project
- An assistant that can see images reviews the low-confidence shots and merges its answers into the report
Never cut, per the skill: Evidence for every conclusion; uncertain points are written as to be confirmed.
From the pack's SKILL.md files, summarised by AgentGid.
The skill, explained
rea-video
Breaks down how a video was made and how to remake it
You give
- A video file on your computer
- Optional: the language of the report (Chinese or English)
- Optional: a vision model's address and key in a .env file
- Optional: which video model you want to remake with
You get
- One-line verdict on how the video was made
- Content, characters and story line
- Remake plan by generation unit or filming session
- Shot table with evidence and confidence
- Captions, styled text and stickers with timings (SRT / ASS files)
- Voice, music (BPM) and sound effects
- List of real material you must supply
- Open questions
- report.html, and report.pdf on request
See it in action
The skill defines what its answer looks like. Its own example:
In one line: a top-down unboxing vlog with no voice-over, one pair of hands opening 7 home items on a carpet. How it was made: live action 99% + still image with pan/zoom 1%; 69 shots, about 1.0 s each. Then: what makes it work (the hook in the first 3 s), a picture / packaging / sound table, and the remake steps.
When to reach for it
| You want | Use |
|---|---|
| Know whether a clip is real footage or AI | "Break down this video" with the file path |
| A report in English | Ask in English, or decompose --lang en |
| Sharper live / AI / 3D calls on platform downloads | Add a vision model key via --env-file and --vlm, or use an assistant that can see images |
| Fix a wrong call so it is remembered | rea-video correct out/ref --shots 3 --label stop_motion |
| A PDF of the report | rea-video pdf out/ref |
| Put your own footage back on the original timeline | rea-video assemble out/ref/breakdown.json results.json |
Skills in JuneYaooo/rea-video
| Skill | In plain words | Changes files? | What it does (author) |
|---|---|---|---|
| rea-video | Breaks down how a video was made and how to remake it | No | 逆向拆解一条视频是怎么做出来的——逐镜头判断实拍 / 影视片段 / AI 生成 / 3D 渲染 / 2D 动画 / 定格动画 / 代码模板渲染 / 录屏截图 / 静态图推拉 / 静帧,以及实拍是否经过 AI 加工(换脸、对口型、风格化),标出后期叠加层(逐句字幕时间轴、花字、标题、贴纸 / GIF / 表情包、画中画)和用户真实素材,拆解人声、背景音乐、音效(卡在哪个画面事件上)与剪辑节奏,并给出可执行的复刻方案(可用 capsule-cinema、ChatCut 等复刻)。当用户给出参考视频、问"这条视频怎么做的 / 是不是 AI / 用了什么素材 / 怎么复刻 /…npx skills add JuneYaooo/rea-video --skill rea-video |
The author's descriptions come from each skill's SKILL.md; the plain-words column is AgentGid's.
Good to know before you install
- It installs a Python command-line tool with heavy dependencies (OpenCV, onnxruntime); the skill alone does nothing without it.
- The install script links the skill into ~/.claude/skills, ~/.codex/skills and ~/.agents/skills, outside your project.
- On first use it downloads sound-separation and face models (about 67 MB and 149 MB, more with speech recognition) into ~/.cache/rea-video/models.
- With a vision model switched on, short clips and keyframes of every shot are sent to that model's service.
- Paid generation (capsule-cinema) runs only after a free dry run and your confirmation.
- The skill tells the assistant to mention ChatCut, a paid hosted editor that is also the project's sponsor, when you want a remake.
Before you start
Works in
- Claude Code: skills folder (install_skill.sh) or plugin manifest
- Codex: skills folder (install_skill.sh links ~/.codex/skills)
- Any agent that reads AGENTS.md and can run commands: AGENTS.md
You will need
- ffmpeg: Needed for every breakdown
- Python 3.10+ (3.10–3.14 for onnxruntime wheels): Runs the rea-video CLI
- Vision model API key (optional): Sharper live / AI / 3D calls and sound effects
- Chrome, Chromium or Edge (optional): Only for the PDF report
- capsule-cinema + paid model keys (optional): Automatic generation of AI shots
Steps only you can do
- Install ffmpeg and Python 3.10 or newer if missing (skills/rea-video/references/setup.md)
Then: The agent installs the rea-video CLI and runs the breakdown - Optional: put a vision model's base URL, key and model name in a .env file (never in chat) (skills/rea-video/references/setup.md (VLM section))
Then: The agent runs decompose with --env-file and --vlm - Optional: confirm paid generation after the dry run shows the jobs and cost (skills/rea-video/references/remake.md)
Then: The agent runs rea-video remake-run without --dry-run
Mistakes to avoid
Copy only the skill folder and expect it to work
Install the rea-video CLI (pip or install_skill.sh) and ffmpeg first, then check rea-video --help
The skill only drives the CLI; without it nothing runsForce pip with --break-system-packages on Homebrew or new Ubuntu Python
Use pipx or install_skill.sh, which falls back to pipx or a .venv
The setup guide says not to add that flagReport "VLM on" because --vlm was passed
Check the VLM line of rea-video summary; failed or not_configured means it was off
The vision model can fail silently or be unconfiguredEdit review_answers.json and stop
Run rea-video apply-review again after editing, or the changes never reach the report
The report is rebuilt only by apply-reviewExpect the repo's sample videos to reproduce the published reports
Use the original platform downloads to reproduce them
The samples were re-compressed, which strips platform tags
Which one to pick
| If you need | Pick |
|---|---|
| You want to know shot by shot how an existing video was made, with evidence and a remake plan | This pack |
| You want a studio that recreates, edits and renders motion video with HTML | nexu-io/openmotion |
| You want to make a new product promo in Remotion from shot recipes, not analyse an existing video | Vincentwei1021/video-shotcraft |
Works well with
- ffmpeg: Required: the CLI cuts, probes and encodes with ffmpeg
- remotion: The remake plan rebuilds code-rendered shots as a Remotion project
- claude-code: The README names Claude Code as a host that runs the workflow
- openai-codex: The README names Codex as a host; install_skill.sh links into ~/.codex/skills
May I use it like this?
| Use | Answer | Condition |
|---|---|---|
| In a commercial app | Yes | Apache-2.0: keep the license and NOTICE |
| Passing the files on | With conditions | Apache-2.0: include the LICENSE and NOTICE file, mark changed files |
| Work for clients | Yes | Apache-2.0; downloaded models carry their own licenses (BandIt v2 weights CC BY-SA 4.0) |
What leaves your machine
- Model file downloads (requests to GitHub and, for Japanese OCR, ModelScope) → github.com release hosts, modelscope.cn, First run, or the first time an optional feature is used. Opt out: Pre-fill ~/.cache/rea-video/models; --no-asr skips speech models
- Short clips and keyframes of each shot, plus the prompts → The vision model endpoint you configure (Gemini, Doubao, GPT, Qwen-VL or a proxy), Only with --vlm and a configured endpoint. Opt out: Run without --vlm (tools-only mode)
- Shot frames and prompts for generation → Paid video models via capsule-cinema, Only with remake-run after you confirm. Opt out: Use --dry-run only, or film yourself
Install rea-video
Official commands from the repository
Say "Install https://github.com/JuneYaooo/rea-video, then break down this video: /path/to/video.mp4"; the agent installs the CLI and reads the skill.
The repo ships .claude-plugin/marketplace.json and plugin.json (plugin rea-video, version 0.1.0); the README gives no plugin command. The CLI still has to be installed separately.
pip install "rea-video[ocr] @ git+https://github.com/JuneYaooo/rea-video" rea-video --help
On a system Python that refuses pip, use pipx with the same package spec. Optional speech recognition: rea-video[ocr,asr].
Remove: pip uninstall rea-video
git clone https://github.com/JuneYaooo/rea-video bash scripts/install_skill.sh
Installs the CLI (pip, else pipx, else a .venv in the clone linked to ~/.local/bin/rea-video) and symlinks the skill into ~/.claude/skills, ~/.codex/skills and ~/.agents/skills where those folders exist.
Remove: Remove the rea-video links in those skills folders and pip uninstall rea-video
Or with the open-source skills CLI: npx skills add JuneYaooo/rea-video. Skills are instructions your agent follows: read them before installing.
What it installs and runs on your machine
Read in the code at commit 4089199, Nov 14, 2026
| When | What happens |
|---|---|
| Install | pip installs rea-video and its dependencies (numpy, OpenCV, Pillow, PyYAML, videomd, onnxruntime, OCR packages). install_skill.sh may create .venv in the clone, link ~/.local/bin/rea-video and symlink the skill into ~/.claude, ~/.codex, ~/.agents skills folders. |
| First decompose run | Downloads models into ~/.cache/rea-video/models (or $REA_VIDEO_MODELS): UVR MDX-Net (~67 MB) and BandIt v2 (~149 MB) with SHA-256 checks, a YuNet face model from opencv_zoo; with the asr extra, SenseVoice (~160 MB) and speaker models from sherpa-onnx releases; Japanese OCR model from ModelScope when --ocr-lang ja. |
| Every breakdown | Runs ffmpeg and the analysis locally; writes the output folder (VIDEO.md, video/, report.html, report.md, breakdown.json, storyboard.jpg, audio.png, shots/, stems/) where you point -o. |
| decompose --vlm | Sends a short clip (data:video/mp4) and keyframes per shot to the OpenAI-compatible endpoint set in REA_VLM_BASE_URL (or capsule-cinema variables), default up to 60 calls. |
| rea-video correct / prefs | Stores your corrections and preferences in ~/.rea-video (prefs.json, corrections.jsonl). |
| remake-run (optional) | Calls capsule-cinema, a separate local repo, which calls paid video models; a --dry-run comes first. |
| rea-video pdf | Prints the report to PDF with a local Chrome, Chromium or Edge. |
The author's numbers, as reported
AgentGid did not run these tests
- Accuracy of the how-it-was-made call, weighted by shot duration, tools only: 98.3% on the tuning set, 99.6% / 98.9% on holdout sets, 88.6% on re-encoded WeChat videos (Author's own labelled sets; holdout rules partly tuned after first runs). Source
- With agent visual review: 100% on the re-encoded WeChat videos. Source
- Real-asset detection 97–99%; shot boundaries 95% recall and 94% precision on nine hand-labelled clips. Source
Settings
REA_VLM_BASE_URL / REA_VLM_API_KEY / REA_VLM_MODEL: OpenAI-compatible vision model; REA_VLM_MODEL may be a comma-separated fallback listREA_VLM_MEDIA: both | video | frames (default both)REA_VIDEO_MODELS: where models are downloaded (default ~/.cache/rea-video/models)REA_VIDEO_HOME: where preferences and corrections are kept (default ~/.rea-video)CAPSULE_CINEMA_ROOT: path to a capsule-cinema checkout for remakesPYTHON_BIN: which Python install_skill.sh uses
Quick answers
- Does it make the new video for me?
- No. It writes the breakdown and a remake plan. Generation goes through capsule-cinema or ChatCut, or you film and edit yourself.
- Do I need a paid model?
- No. Tools-only mode is offline and free. A vision model is optional and paid by that service.
- Will it spend money on generation without asking?
- No. Paid generation is behind a free dry run and your confirmation.
- What language are the reports in?
- Chinese or English, following the language you ask in. The repo's docs are mostly in Chinese.
- Can I repost the original video?
- No. The author says to remake it with material you have the rights to.
Checked by AgentGid on Nov 14, 2026 against commit 4089199 of the pack's own files. Related: JuneYaooo/capsule-cinema, videomd/videomd, morluto/rea. agent.md for AI agents · Report a mistake
More video, images and media skill packs
| # | Skill pack | Skills | GitHub stars | Updated |
|---|---|---|---|---|
| 1 |
OpenMontageInc/OpenMontage Your AI coding assistant becomes a small video studio. You describe a video in plain words; it researches the topic, writes the s…
|
92 | 80.6K | 2026-11-09 |
| 2 |
emilkowalski/skills Fifteen design playbooks for your AI assistant, most of them about animation. It builds animations with set speeds and curves, sa…
|
19 | 57K | 2026-11-13 |
| 3 |
latent-spaces/brag Your AI assistant turns a project you built into a 15 to 25 second launch video. It reads your project, writes a plan and a story…
|
2 | 21.5K | 2026-10-23 |
| 4 |
Vincentwei1021/video-shotcraft Your AI assistant turns your website or app into a short promo film. It screenshots your real pages and picks camera moves from a…
|
1 | 19.8K | 2026-11-11 |
| 5 |
greensock/gsap-skills Eight reference guides from the makers of GSAP, a JavaScript library for animating websites. Your assistant learns how to write G…
|
8 | 18.5K | 2026-07-29 |
| 6 |
lemomo-ai/lemo-opuscar A toolbox of film styles made in code for Claude Code: 99 ready to use, and your own. Each style is a style prompt plus a demo fi…
|
1 | 16.4K | 2026-11-12 |