rss-feed-generator
A self-hosted feed generator for sites that don't publish RSS, with pluggable extractors for HTML parsing, full-text enrichment, and JSON API sources.
A Seattle Times Huskies feed in Reeder, generated from a topic page that doesn't ship its own RSS.
rss-feed-generator is a self-hosted service that produces RSS feeds for sites that don't publish their own. The original itch was Seattle Times topic pages: I wanted Mariners and Huskies football coverage in my reader, but the Times only ships a handful of section feeds, none of them topic-scoped. The service started with the Times pages, then expanded to whatever else I needed feeds for.
Different sources need different strategies. Some pages need only the index parsed: the title and link from each entry is all a feed needs. Others need a second pass: pull the article list from an index page, then visit each article to grab the description, reading time, and metadata. Single-page applications that load their content client-side get an API-backed fetcher instead. Take the Associated Press: their newswire photo search runs on a JSON API, so the fetcher posts a query and outputs Media RSS. Each feed picks the strategy it needs through a pluggable extractor interface.
The service feeds a few of my other projects. claudenotes pulls from the Anthropic engineering blog and official Claude blog since Anthropic doesn't ship feeds. e-ink-scoreboard reads the Mariners feed for game data, and the AP newswire photo feed drives the scoreboard's screensaver, rotating through fresh game-day photos when nothing's actively playing.