Yost

Overview

Yost is a local-first dashboard for YouTube community posts. It collects posts from multiple channels into a single feed, with search and filters, so you can follow what creators are saying without the video feed or the recommendation engine.

The challenge

Community posts are not exposed through a simple public API in the way videos are. Yost needed a scraping layer robust enough to collect posts from multiple channels, plus a front end that made the results feel like a feed rather than a data dump.

The approach

A React 19 + Vite front end talks to an Express backend that uses Puppeteer with a stealth plugin to collect posts. The app is designed to run on your machine, keeping the data local. The interface focuses on reading: a unified feed, search, and filters, with no video player and no algorithmic ranking.

What it does

• Collect community posts from multiple channels
• Unified, chronological feed
• Search and filter posts
• Local-first design
• No videos, no recommendations

Technical notes

• React 19 + Vite + Tailwind CSS front end
• Node.js + Express backend
• Puppeteer with stealth plugin for scraping
• Local-first data flow

What I learned

Yost taught me to respect the boundary between a demo and a tool. Scraping is fragile, so the backend has to fail gracefully and the UI has to stay honest about what it can load. It also made me think harder about local-first software: the value is not just privacy, it is ownership.