Opening

The biggest surprise today is rea-video, a brand‑new project that hit 8.2k stars in 30 days. It classifies every shot as live‑action, AI‑generated, or code‑rendered, then breaks down lighting, camera movement, and color grading for you. Also notable: modern-web-guidance, a joint effort from Chrome and three other major browsers, which cures AI assistants of writing 2019‑era JavaScript. It sits at 12k stars and is ready to grab.

Today’s Open‑Source Picks

1. rea-video: AI Video Reverse Engineering, Shot‑by‑Shot Breakdowns with Executable Re‑creation Recipes

  • What it is: Solves the real pain point of “I want to copy this video’s style but don’t know how to deconstruct it.” Targeted at video creators, technical artists, and AI researchers. It identifies whether each shot is live‑action, AI‑generated, virtual human, 3D animation, code‑rendered, screen recording, or a mashup.
  • Recent buzz: 8.2k GitHub stars (new project, 30 days), Apache‑2.0 license.
  • Highlight features: Classifies 7 video types. Outputs an open VIDEO.md format containing shot classification, lighting setups, camera moves, color grades, audio separation (vocals / music / SFX), subtitle timeline, and edit points. Offline free mode hits 98%+ accuracy; adding Gemini or Doubao vision models pushes it to 99.6%+. Includes the capsule‑cinema auto‑recreation pipeline (AI shot generation + Remotion code rendering + live‑action shoot plan).
  • Quick start: In Claude Code or Codex, run Install https://github.com/JuneYaooo/rea-video, then break down this video: /path/to/video.mp4 (requires ffmpeg and Python 3.10+).
  • Vs. alternatives: Unlike one‑click remake tools such as Hypit that just generate new footage from visuals, rea‑video focuses on “how was the original made” and hands you an executable recipe (shoot plan, prompts, code approaches). Ideal for creators who want to learn technique, not just apply templates.
  • Best for: Creators and technical artists who want to learn filming techniques or deconstruct reference videos.
  • Link: https://github.com/JuneYaooo/rea-video

2. modern-web-guidance: Dragging AI Assistants Out of 2019 into Modern Native Web Development

  • What it is: A skill pack built for frontend engineers to stop AI from generating legacy JavaScript. Maintained jointly by the Chrome, Edge, Firefox, and MDN teams. You drop it straight into an AI agent’s context window so it writes modern native code.
  • Recent buzz: 442 points on Hacker News (146 comments), 12k GitHub stars, Apache‑2.0 license.
  • Highlight features: Guidance for 250 modern web capabilities (CSS Layout, View Transitions, Anchor Positioning, Popover API, <dialog>, field‑sizing, CSS.highlights, etc.). CLI scans projects for legacy patterns (legacy-scan) or fetches guidance to inject into the AI agent context. Token‑efficient design, evaluated to strip out what the model already knows, covering performance (INP, scheduler.yield), accessibility, built‑in AI (translation, summarization, language detection), and more.
  • Quick start: Try instantly with no install via npx modern-web-guidance@latest search "animate a dialog modal backdrop"; for real use, run npx modern-web-guidance@latest install to launch the interactive installer into your project.
  • Vs. alternatives: Unlike MDN Web Docs or caniuse which are reference lookups, this is a skill pack for AI agents, it stuffs best practices directly into the context window. Compared to hand‑crafted prompts or DIY cheatsheets, it carries browser‑vendor endorsement and continuous updates.
  • Best for: Frontend engineers who need their AI assistant to write modern native web code.
  • Link: https://github.com/GoogleChrome/modern-web-guidance

Closing

Both tools today tackle the same class of problem: “AI doesn’t know the latest domain knowledge.” rea-video helps humans understand video grammar; modern-web-guidance helps AI understand modern browser capabilities. If you’re a frontend engineer, install modern-web-guidance tonight and run legacy-scan to see how much legacy code remains in your project. If you make videos or research visual language, rea‑video’s VIDEO.md format is perfect for feeding downstream agents. See you tomorrow, keep digging for gems.

Author

Mark Ku

10 年以上的軟體工程師,做過北美電商與 AI SaaS 訂閱收費系統。Read More

Found this useful?

The author's free tools, daily podcasts and newsletter are all here.

Mark Ku · This article is licensed under CC BY 4.0. Credit the author and link back to the original when reusing it.

Comments

Subscribe to Newsletter

Subscribe to get new posts delivered instantly — never miss a tech share.

By submitting, you agree to receive emails. You can anytime.

Popular Posts

View all
Mark Ku
··633

Oracle Cloud Always Free Tier: Linux Host and Static IP for a $0 Cloud Solution

Oracle Cloud Always Free Tier: Linux Host and Static IP for a $0 Cloud Solution
Mark Ku
··464

Say Goodbye to Postman's Fee Trap! A Hands-on Guide to Bruno, the Open-Source Git-Native API Testing Powerhouse.

Say Goodbye to Postman's Fee Trap! A Hands-on Guide to Bruno, the Open-Source Git-Native API Testing Powerhouse.
Mark Ku
··249

A Free, Open-Source, Notion-like Knowledge Base — A Complete Guide to Deploying and Backing Up Outline Wiki

A Free, Open-Source, Notion-like Knowledge Base — A Complete Guide to Deploying and Backing Up Outline Wiki
Mark Ku
··205

Building an Efficient API Management Platform: Deploying Kong Gateway from Scratch - Part 1

Building an Efficient API Management Platform: Deploying Kong Gateway from Scratch - Part 1
Mark Ku
··202

Setting Up Samba on Ubuntu to Share Folders with Windows 11

Setting Up Samba on Ubuntu to Share Folders with Windows 11
Mark Ku
··198

Training Your Own AI Voice: Hardware Requirements, Open-Source Model Comparison, and LoRA Fine-Tuning

Training Your Own AI Voice: Hardware Requirements, Open-Source Model Comparison, and LoRA Fine-Tuning