Terms

RSS Amplifier is an open directory of public feeds. Reading it needs no account, no key and no agreement. These terms describe what the service does, what it commits to, and what it deliberately does not.

Using the directory

Every page, feed, JSON endpoint and MCP tool here is public and anonymous. You may read it, crawl it, cache it, index it, and build on it, by hand or with an agent, commercially or otherwise. We do not rate-limit reads and we do not ask you to identify yourself. The one write — submitting a feed — carries a per-IP limit so that one submitter cannot fill the queue.

The content itself is not ours. Titles, summaries and article text belong to the people who published them, and the directory indexes them the way a search engine does. If you are a publisher and you would rather not be listed, say so and the feed comes out.

What we commit to

One thing, and it is the thing an automated reader actually needs: every feed carries an honest freshness signal, and it is never hidden. Each feed page and every API and MCP response says when we last read the publisher, when the publisher last posted, and when we will look again. A feed whose publisher has gone quiet is labelled dormant even though our copy of it is perfectly current.

This matters more than a speed or uptime figure because it is the failure that does real damage. A directory that is briefly down is obvious and recoverable. A directory that silently serves a blog which stopped publishing in 2023, with no indication that anything is wrong, produces confidently wrong answers — and about a sixth of the feeds in here are in exactly that state.

How often we re-read a feed follows the feed’s own publishing rhythm rather than a fixed schedule: something that posts hourly is checked hourly, and something that has posted nothing for two years is checked roughly quarterly. A feed is never left unchecked for more than ninety days, whatever it looks like, because publishers do come back.

Availability

We do not offer a service level agreement, and we do not publish an uptime target. This is a free, open service run on ordinary infrastructure by a small team, and a number like “99.999%” would be a promise we cannot presently keep — five nines is five minutes of downtime a year. Publishing one would be exactly the kind of confident, unverifiable claim this directory exists to protect its readers from.

What we publish instead is what we measure. These figures are live, and they are the same numbers we use to run the service:

86.1%Crawler active31 of 36 hours observed

93.6%Crawl success50,523 of 54,001 in 24h

368,887Feeds indexed63,193 read at least once

18,855Dormantno post in over a year

“Crawler active” is the share of hours in which the crawler recorded work, measured over the 36 hours we have records for rather than over a calendar month — we have not been keeping this record long enough to claim more, and an absent hour is an hour with no evidence rather than a proven outage. It says nothing about whether this website answered, which is a different question we do not yet measure and will not guess at.

“Crawl success” is partly a measure of the open web rather than of us: a feed that has moved, expired or started refusing robots counts as a failure here and is not something we can fix. The crawler status page shows the same figures hour by hour, live.

What we do not promise

That a feed is complete. We store what a feed document offered when we read it. A publisher who serves only their ten most recent posts has an archive here that begins when we first found them.
That a feed is accurate. Titles, dates, categories and authorship are the publisher’s claims, not ours. Where we derive something ourselves — a category, a topic — we say so.
That the service will keep existing. Should it ever shut down, the whole directory is published as OPML and JSON so it can be taken elsewhere, and the code is open source.
Anything, warranty-wise. The service is provided as is. We are not liable for decisions made on the strength of what it returns, which is why the freshness signal above is a commitment and this section is not.

Changes

These terms change when the service does, and the history is public in the repository like everything else here. Questions, corrections and removal requests: get in touch.