Introducing MicroRSS
A web RSS reader with an engine of its own, built to read the places that publish no feed and to hide the noise before it reaches you. What it is and why.
MicroRSS started as a small, fast interface to Miniflux, a reader we had used for years. Over the summer the interface grew the things we kept wanting from a reader: a page feed wizard for sites with no feed, rules that hide what we do not read, page archives of images, a research workspace. The one thing it could not grow was the engine underneath, so we built one.
As of this week MicroRSS runs on its own engine at microrss.com. Here is what that means.
One fetch for everyone
A feed is one row in the engine however many readers follow it, and it is fetched once for all of them, on an interval that adapts to how often it posts: a feed that publishes hourly is read hourly, one that posts twice a year is read a few times a day and no more. Your subscription holds what is yours — the title you gave it, the category, the rules — and the entries are shared.
That is also what makes plans possible. A free account follows up to 30 feeds refreshed hourly; Pro raises the count, shortens the interval, keeps history longer and adds the full-text crawler and API keys. The numbers are on the pricing table.
Sources that publish no feed
Besides feeds, the engine reads Telegram channels from their public preview pages, with the posts' photos, documents and polls. More sources follow the same shape, each a parser of its own: paste the page's link, MicroRSS recognises it, and the result looks like any other feed. The Built-in sources pages list them.
For everything else there is the page feed: point at the repeated items on a page, and MicroRSS serves them as a feed. The configuration is the signed address of that feed; nothing is stored, and it keeps working from any reader.
Rules on arrival
Keep and block rules used to run in the browser, hiding entries after they had been downloaded. On the engine a rule runs when an entry arrives, so a match is marked read before anyone sees it. Change a rule and it is re-applied to what is already stored; remove it and the hidden entries come back unread, because a rule hid them, not a person. The grammar is Miniflux's, so existing rules carry over as they are.
Accounts
MicroRSS now has accounts of its own: sign up with an email address, confirm it, log in. A Miniflux installation can still be read through the same interface; the login page has a switch for it.
What is next
The sources. A long list of places people follow that have no feed is being worked through, and the ones that make sense as a parser are being added one by one. Then a catalogue of working feeds by topic, so that a new reader has something to read on day one.