rea-video by JuneYaooo — tool skill for research & trends
For AI agents: find the right skill in one call —
curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>"returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add.mdto its path. Before generating anything, show the person the options on the how-to page. From a terminal:curl -fsSL https://varg.ai/install | shinstalls the vargai CLI (https://varg.ai/how-to/cli.md):vargai searchandvargai install <slug>with no key, every varg model aftervargai login.
Breaks a video down shot by shot: filmed, AI-generated or coded, with prompts, camera, light, audio and a remake plan.
- Type: Tool · Research & trends
- Slug:
rea-video - Status: live
- Publisher: JuneYaooo (community)
- License: Apache-2.0
- Source: https://github.com/JuneYaooo/rea-video
- Works in: Claude Code, Codex
- Requires: FFmpeg, Python 3.10–3.14, Optional: a vision model endpoint (REA_VLM_BASE_URL, REA_VLM_API_KEY, REA_VLM_MODEL)
- Cost: Free, Apache-2.0: offline tools; an optional vision model bills your own key
- Adoption: 62 installs, 6,388 GitHub stars (GitHub + skills.sh, 2026-11-10)
What it does
A CLI plus agent skill that reverse-engineers a reference video. It splits shots and audio stems, gathers evidence (frame cadence, sensor noise, AIGC metadata, OCR, BPM) and decides for every shot whether it was filmed, AI-generated, an avatar, 3D, 2D animation, a code template or a screen recording, then writes a remake plan split into what to film, generate and code, with a reverse-engineered prompt per shot, camera and lighting notes, captions as SRT/ASS and a report.html. A vision-capable agent reviews the uncertain shots itself; an optional VLM endpoint (Gemini, Doubao) sharpens the calls, and the README reports 98–100% accuracy on its own eval sets. Riley Brown's post about it (99.9k views, 3,289 bookmarks, 2026-11-09) is the most-saved skill post of the Fable 5.2 launch. Study a reference, then make your own.
Use when
- how was this video made: filmed, AI or code?
- check whether a video is AI-generated, shot by shot
- break a reference video down before remaking it
- prompts, camera and lighting notes for each shot of a reference
- a remake plan split into what to film, generate and code
Not for
- Rendering the new video: it plans, it does not render (pair it with capsule-cinema, varg or HyperFrames)
- Reposting someone else's footage, music, characters or likeness
Install
- Claude Code:
git clone https://github.com/JuneYaooo/rea-video && cd rea-video && bash scripts/install_skill.sh - Codex:
git clone https://github.com/JuneYaooo/rea-video && cd rea-video && bash scripts/install_skill.sh - Any agent, with the vargai CLI:
vargai install rea-video
More research & trends skills
- Video Skills Finder: Finds the right skill for any AI video, image, voice or music job here and shows you the visual options.
- Apify MCP: Thousands of ready scrapers behind one MCP: TikTok, Instagram, YouTube, hashtags, sounds.
- TwelveLabs Video Understanding: Semantic search and analysis across your video library from Claude Code.
- Civitai MCP: Top AI images and videos with prompts and base models, plus model search.
- Hugging Face MCP: Trending models, Spaces and papers from Hugging Face inside your agent.
- Exa Search MCP: Neural web search and crawling for research agents.