🍋 Why "sqzd"?. Just like squeezing a piece of fruit to get the pure juice while leaving the rind behind, sqzd takes bloated videos and extracts only the concentrated, high-value information
sqzd.mp4
sqzd is a hands-free YouTube wrapper that extracts high-signal moments from long videos. Instead of reading long summaries or manually scrubbing through timelines, sqzd does the work for you—automatically skipping the filler so you can watch just the gold.
- ⏭️ Auto-Skip: No manual clicking required. The app automatically jumps to the core takeaways.
- 🎙️ Voice-Guided: A natural voice announces the topic, letting you stay informed without needing to look at your screen.
- 🎧 Background Audio: Keep listening with your screen off, fully integrated with native media controls.
- 📲 Quick-Share: Send links directly from the official YouTube app to start processing instantly.
- Direct APK: Download the latest compiled
app-release.apkdirectly from our Releases page.
- AltStore: Sideload easily without a computer by adding our custom source to your AltStore app. Tap
+in the Sources tab and paste this URL:
https://raw.githubusercontent.com/icnahom/sqzd/main/app.json
- Direct IPA: Download the unsigned
app.ipafrom our Releases page for manual sideloading.
If you want to contribute or modify the codebase, follow these simplified steps to compile and run sqzd on your machine:
You need the Flutter SDK installed on your computer to build this app.
- Install Flutter (Official Guide) (Select your Operating System: Windows, macOS, or Linux).
Open your computer's terminal (or command prompt) and run these commands to clone the code and enter the project folder:
git clone https://github.com/yourusername/sqzd.git
cd sqzd
Run this command to download the software libraries the app needs to run:
flutter pub get
Connect your phone via USB (with Developer Mode enabled) or open a virtual device emulator, then run:
flutter run --profile
We do not use unofficial web-scraping tricks to fetch video transcripts. Instead, we use the official Gemini Video Understanding API.
Note
Because the AI model analyzes the video content directly, initial generation can take some time. We hope to transition to an official, fast YouTube Transcript API whenever one becomes publicly available.
You might notice the entire app lives in one file — lib/main.dart. This is intentional.
- 🧑💻 Optimized for AI agents. This codebase is designed to be read, understood, and edited by AI coding agents. A single file means the agent gets the full picture in one shot.
- 🪙 Input tokens are cheap. Throwing a ~20k token file at an LLM costs pennies. There's no need to split things up just to save on token cost.
- 📦 Context windows are big. Modern models comfortably handle 100k+ tokens. In a world of 1M context windows, the codebase can be the memory.
Note
Since the agent has full context, we can split into multiple files anytime. The single file stays until there's a concrete reason to split.
This project is licensed under the PolyForm Noncommercial License 1.0.0.
You are free to view, fork, modify, and use this software for personal, educational, and non-commercial purposes. Commercial use, including publishing this application to any app store for profit, or using it within a commercial business, is strictly prohibited.
For commercial licensing inquiries, please contact me directly.