Website RSS & Atom Feed Finder

Find every RSS, Atom and JSON feed a website publishes — and get back only the ones that actually work.

Guessing /feed is easy. This Actor fetches every candidate and parses it before returning it, so the output is a list of confirmed, working feed URLs with their format, title and item count. No dead links to filter out afterwards.

What you get

{
  "siteUrl": "https://overreacted.io",
  "feedsFound": 2,
  "primaryFeedUrl": "https://overreacted.io/atom.xml",
  "primaryFeedFormat": "atom",
  "candidatesChecked": 14,
  "feeds": [
    {
      "url": "https://overreacted.io/atom.xml",
      "format": "atom",
      "title": "overreacted — A blog by Dan Abramov",
      "itemCount": 58,
      "latestItemDate": "2026-07-31T00:00:00.000Z",
      "discoveredVia": "link-tag"
    }
  ]
}

How it finds them

Three routes, tried in order of reliability, then every candidate is validated:

1. <link rel="alternate"> tags — the feed the site declares. Most reliable.
2. On-page anchors that look like feeds — catches sites that forgot the link tag.
3. Well-known paths — /feed, /rss.xml, /atom.xml, /index.xml, /feed.json and others.

Results are ranked so a declared feed always beats a guessed path.

Main use cases

• Build a monitoring list. Give it 500 competitor or industry sites and get back the feeds worth watching.
• Feed-reader onboarding. Turn a URL a user pastes into a working subscription.
• Content aggregation and newsroom monitoring.
• Migration checks — confirm feeds survived a replatform.
• Feeding an RSS pipeline. Pair with RSS Feed Scraper to go from a list of websites to a stream of articles.

Pricing

Event	Price	When
Site checked	$0.003	One website searched, whether or not feeds were found

A site with no feeds is a real answer and costs the same to determine, so it is charged. Sites that could not be reached at all are not charged.

1,000 websites = $3.00.

Input

Field	Type	Default	Notes
websites	array	required	Bare domains work. Up to 2,000.
checkWellKnownPaths	boolean	true	Probe common paths when nothing is declared. Turn off for speed.
maxFeedsPerSite	integer	20	Cap per site.
concurrency	integer	8	Parallel sites.

Limitations

• No JavaScript rendering. A feed link injected client-side will only be found if it also sits at a well-known path.
• Well-known probing costs requests. With it on, a site with no declared feed triggers up to 14 extra checks. That is why it is a toggle.
• Podcast feeds are found but not parsed as podcasts — you get the feed URL, format and item count, not episode metadata.
• Feeds behind authentication or bot protection cannot be validated and are omitted rather than guessed at.

Related Actors

• RSS Feed Scraper ★5.0 — the natural next step: read the feeds this Actor finds.
• Substack Newsletter Monitor — for Substack publications specifically.

Reliability

• Retries on 429 and 5xx with exponential backoff; permanent 4xx errors are not retried.
• Bad input is reported and skipped, never charged.
• One failing target never aborts the rest of the run.
• Concurrency is bounded and configurable — this work is network-bound, so parallel requests, not memory, are what control speed.

Data source and responsible use

Reads only publicly available data. No authentication bypass, no CAPTCHA circumvention, no browser automation.

Not affiliated with any of the services checked.