Seventeen sources that have no RSS
Telegram, Bluesky, GitHub, app stores, Twitch, Bandcamp and a dozen more now read natively, plus the feeds of sites that never link them. How it works.
The promise of a feed reader is that everything comes to you. The catch is that a growing share of what people follow publishes no feed: a Telegram channel, a Bluesky account, the releases of a plugin, the free game of the week. Over the past days the engine learned to read seventeen such places directly. The Built-in sources pages list every one with examples; here is what is behind them.
A source is a parser
Each source is a small reader of its own inside the engine. It knows which links belong to the platform, makes one or two requests to the platform's own page or API, and hands back entries shaped like a feed's. You paste the link you would open in a browser, MicroRSS recognises it without fetching anything, and the result sits in the list like any feed, with the platform's own icon.
The sources so far: Telegram channels (from the public preview page, with photos, documents and polls), Bluesky profiles (with or without replies and reposts, link cards and quoted posts rendered), GitHub trending, issues, pull requests and workflow runs, Bandcamp artists and labels, Docker Hub repository tags, the Epic Games Store's free games, WordPress.org plugins and themes, the App Store, Firefox Add-ons, Twitch videos, the YouTube Posts tab, Mixcloud, Modrinth, Comics Kingdom, GOG new releases, MangaDex and Apple Music artists.
A source's options live in the feed's address: a Docker Hub tag filter, the country of the Epic store, a Bluesky feed with replies. That keeps them in an OPML export, and two readers who choose the same options share one feed.
Sites that hide their feed
A second group needed no parser at all. Reddit, Mastodon, Medium, Steam, Dailymotion, Pinterest, ArtStation, dev.to, GitLab, Codeberg, Wikipedia, Substack and a row of news sites all serve a feed; the page just never links it. The engine now knows the address pattern for each, so pasting a subreddit or a Mastodon profile finds the feed straight away, without reading the page.
The news sites were the interesting measurement: of the sixty-odd sites with a known feed we probed from the server, seven wall their home page behind a bot check and eighteen more declare no feed in their markup, while the feeds themselves answer fine. The Economist, The New York Times, NPR, Deutsche Welle, heise, tagesschau and the others on the news sites page are mapped by section.
Discovery that says why
When MicroRSS cannot find a feed, it now says what happened instead of «No RSS feed found»: the site put a bot check in front of the page, the page is gone, the address answered with an error. The usual feed paths are tried in any case, so a site that walls its HTML and leaves /rss.xml open still works. For the rest there is the page feed.
What stays out
A few platforms refuse any server address whatever the request looks like: Patreon and Rumble answer 403 to a datacenter, uNoGS's API answers 500 to everything, NewOnNetflix sits behind a challenge. Those are noted on their pages rather than pretended.