Morten Olsen.pro

Ideas. Systems. Humans.

← Project archive

Open system / project file

Editions

A self-hosted news reader that assembles RSS feeds into finite, curated magazines using on-device ML for article classification.

alpha 08 components

System vocabulary

TypeScriptReactViteTailwind CSSFastifySQLiteTransformers.jsDocker

Editions is a self-hosted news reader that takes a different approach to RSS: instead of an infinite scroll, it assembles feeds into finite, curated “magazines” with reading-time budgets and per-source limits.

How it works

  • Sources — subscribe to any RSS feed
  • Focuses — define topics using natural language descriptions
  • Editions — create magazine configs with per-section article budgets and reading-time limits
  • Source budgeting — prevent any single feed from dominating a section

Classification is powered by zero-shot NLI via Hugging Face Transformers.js, running entirely on-device. Nothing phones home.

Additional features

  • Ranked feed — a traditional feed view with voting that propagates through semantic similarity
  • Bookmarks — save articles for later
  • Single container — ships as one Docker image with Node.js, Fastify, SQLite, and the React frontend