RSS Feed Monitor: only new items, RSS / Atom / JSON Feed

Read any news, blog or podcast feed as clean JSON. Give a feed URL or just a website URL: the feed is found for you. Turn on Only new items and schedule the Actor: each run returns only what was published since the previous run, ready for Slack, e-mail, Google Sheets, n8n, Make, Zapier or an AI agent.

What it does

• All formats: RSS 2.0, RSS 1.0 (RDF), Atom 1.0 and JSON Feed 1.x. Gzip is supported.
• Feed auto-discovery: from <link rel="alternate"> tags, then common paths (/feed, /rss.xml, /atom.xml, /index.xml, /feed.json, /blog/rss.xml…).
• Only new items: remembers the items it already returned (per feed, in a named key-value store in your account). Use stateKey to keep separate memories per schedule.
• Filters: include/exclude keywords (title, summary, categories) and publishedAfter (a date or "24 hours", "7 days").
• Clean output: plain-text summary (HTML removed, length you choose), absolute links, ISO dates, categories, and podcast/media enclosures.
• Polite: plain HTTP (no browser), respects robots.txt on every request and redirect, refuses private network addresses.
• Privacy: author names are not returned, and e-mail addresses in summaries are removed.

Input example

{
  "startUrls": ["https://github.blog/feed/", "https://crawlee.dev"],
  "onlyNewItems": true,
  "stateKey": "my-daily-digest",
  "includeKeywords": ["security", "release"],
  "maxItemsPerFeed": 20
}

Output example

{
  "title": "Example post title",
  "link": "https://example.com/blog/post",
  "guid": "https://example.com/?p=123",
  "published": "2026-11-02T14:00:00.000Z",
  "updated": null,
  "summary": "First characters of the post as plain text…",
  "categories": ["News"],
  "enclosure": null,
  "itemId": "3f1c…",
  "feedUrl": "https://example.com/feed/",
  "feedTitle": "Example Blog",
  "feedFormat": "rss-2.0",
  "input": "example.com",
  "fetchedAt": "2026-11-03T22:00:00.000Z"
}

(The values above are illustrative.) A per-feed summary (how the feed was found, items in the feed, items returned, errors) is saved in the key-value store record RUN_REPORT.

Pricing

Pay per event: US$ 0.30 per 1,000 items returned (US$ 0.0003 per item) plus a tiny start fee (US$ 0.00005 per run). Items removed by filters, duplicates, already-seen items and errors are free. The Actor stops cleanly when your spending limit is reached.

Tips

• For alerts, schedule it every hour or day with Only new items on, and connect an integration (webhook, Slack, e-mail).
• The first run with a new stateKey returns the latest items (up to maxItemsPerFeed); after that, only new ones.
• If a site lists several feeds (posts, comments…), turn on Read all feeds found on a website.

Limits

• Up to 15 MB per feed. Feeds behind a login or blocked by robots.txt are skipped and reported in RUN_REPORT.
• Item dates come from the feed itself; if the feed has no date, the field is null (never invented).

Questions or a missing feature? Open an issue on the Actor page.