Feeds Fun

Feeds Fun aggregates RSS‑style news feeds and presents entries in a single interface where each item is automatically tagged. Tags are generated by configurable processors, ranging from simple domain extraction to LLM‑based analysis using OpenAI or Gemini APIs. Users can define scoring rules that assign numeric values to tags, then filter or sort the feed by those scores, dates, or other criteria, allowing them to surface the most relevant articles while hiding the rest.

The system supports both single‑user and multi‑user deployments, tracks which items have been read, and includes specialized handling for popular sources such as Reddit, GitHub, YouTube, arXiv, and Hacker News. All configuration is exposed via environment variables or .env files, and the backend and frontend are available as separate packages on PyPI and npm, respectively.

Being self‑hostable, free‑tier, and open‑source under a BSD‑3‑Clause license, Feeds Fun is intended for users who want full control over their news aggregation pipeline without subscription constraints, while optionally leveraging LLMs for richer tag generation.