RSS Feed Aggregator: Merge, Dedupe & Filter Multiple Feeds

Turn any number of RSS, Atom, RDF and JSON feeds into one ranked digest where each story appears once, with every source listed. Filter by keywords, get only new stories on scheduled runs, and send them to a webhook, with no browser and no article scraping.

Use cases

• A daily industry or competitor digest from 20 to 200 sources without reading the same news five times.
• A Slack or email newsletter pipeline: schedule the actor hourly and send only new stories through the webhook.
• Brand and keyword monitoring across news sites, blogs, Google News and Bing News searches, release feeds and podcasts.
• Feeding an LLM or a database with a deduplicated, ranked stream instead of raw feed items.

How to use it

1. Paste your feed URLs into Feed URLs (RSS, Atom, RDF or JSON Feed, up to 500).
2. Optionally add include and exclude keywords, a webhook URL and a stateKey for this digest.
3. Click Start, or create an Apify schedule (for example hourly) so each run returns only the stories you have not seen yet.

Input

{
  "feedUrls": [
    "https://hnrss.org/frontpage",
    "https://lobste.rs/rss",
    "https://feeds.arstechnica.com/arstechnica/index",
    "https://www.theverge.com/rss/index.xml"
  ],
  "maxAgeHours": 24,
  "keywords": ["rust", "open source"],
  "excludeKeywords": ["sponsored"],
  "onlyNew": true,
  "stateKey": "dev-news",
  "webhookUrl": "https://hooks.zapier.com/hooks/catch/123/abc/"
}

Field	Default	Meaning
feedUrls	required	Feed URLs (max 500 per run).
maxAgeHours	48	Skip items older than this. 0 = no age limit. Items without a date are kept.
publishedAfter	none	Skip items published before this date (2026-10-01 or 2026-10-01T08:00:00Z).
maxItemsPerFeed	50	Newest items taken from each feed.
keywords / excludeKeywords	none	Include / exclude words or phrases (title, summary, categories).
onlyNew	true	Skip stories already returned by an earlier run with the same stateKey (remembered 30 days).
stateKey	default	Separate memory per digest, for example per client or topic.
webhookUrl	none	POST new digest items here as JSON.
includeDuplicates	false	Add one free duplicate row per merged copy, with dupOf.
includeContent	false	Add the full HTML content the feed itself contains.
timeoutSecs	20	Per-feed timeout.
concurrency	5	Feeds fetched in parallel.

Output

One dataset row per story, in rank order (real row from a run on Hacker News + Lobsters):

{
  "recordType": "item",
  "rank": 1,
  "title": "Valen's Memory Safety: A New Kind of Borrow Checking",
  "url": "https://verdagon.dev/blog/valen-group-borrowing",
  "source": "Hacker News: Front Page",
  "feedTitle": "Hacker News: Front Page",
  "feedUrl": "https://hnrss.org/frontpage",
  "publishedAt": "2026-10-11T12:06:22.000Z",
  "sourceCount": 2,
  "matchedKeywords": [],
  "dupOf": null,
  "sources": [
    { "source": "Hacker News: Front Page", "feedTitle": "Hacker News: Front Page", "feedUrl": "https://hnrss.org/frontpage", "title": "Valen's Memory Safety: A New Kind of Borrow Checking", "url": "https://verdagon.dev/blog/valen-group-borrowing", "publishedAt": "2026-10-11T12:06:22.000Z" },
    { "source": "Lobsters", "feedTitle": "Lobsters", "feedUrl": "https://lobste.rs/rss", "title": "Valen's Memory Safety: A New Kind of Borrow Checking", "url": "https://verdagon.dev/blog/valen-group-borrowing", "publishedAt": "2026-10-11T12:23:01.000Z" }
  ],
  "summary": "Article URL: https://verdagon.dev/blog/valen-group-borrowing Comments URL: https://news.ycombinator.com/item?id=50042362 Points: 5 # Comments: 3",
  "author": "verdagon",
  "categories": [],
  "imageUrl": null,
  "enclosures": [],
  "guid": "https://news.ycombinator.com/item?id=50042362",
  "fetchedAt": "2026-10-11T13:08:40.249Z"
}

A feed that fails adds a row like this (free):

{ "recordType": "feed-error", "feedUrl": "https://example.com/", "finalUrl": "https://example.com/", "status": 200, "error": "Not an RSS, Atom, RDF or JSON feed: the URL returned an HTML page", "fetchedAt": "2026-10-11T13:08:40.249Z" }

The run summary (key-value store record OUTPUT) has per-feed status (feeds[]: format, item count, error) and the counts itemsFetched, itemsInWindow, clusters, duplicatesMerged, filteredOut, alreadySeen, itemsEmitted, plus the webhook result.

Webhook payload

Sent only when the run has at least one new digest item: { "generatedAt", "stateKey", "itemCount", "feedsOk", "feedsFailed", "items": [first 100 item rows], "feedErrors": [...] }.

Pricing (pay per event)

Event	Price	When
Run started (apify-actor-start)	$0.005	Once per run, charged by Apify when the run starts.
feed-polled	$0.001	Per feed fetched and parsed successfully. Broken feeds are free.
digest-item	$0.001	Per story in the digest, after merging, filters and only-new. Duplicate rows and error rows are free.

Example 1, one-off digest: 20 feeds with about 340 stories after merging = $0.005 + 20 x $0.001 + 340 x $0.001 = $0.365.

Example 2, hourly monitoring: the same 20 feeds every hour with only-new, about 5 new stories per run = $0.005 + $0.020 + $0.005 = $0.03 per run, about $21.60 per month (720 runs).

If you set a maximum charge per run, the actor stops adding stories at the limit and the unpaid ones stay new for the next run.

FAQ and limits

Which feeds does it read? RSS 2.0, Atom, RDF (RSS 1.0) and JSON Feed. Gzipped feeds, redirects, non-UTF-8 feeds (for example windows-1251 or ISO-8859-1) and relative links are handled. Give it feed URLs: a homepage URL is reported as "not a feed" instead of being crawled for feed links.

How does it detect duplicates? Two items are the same story when either:

1. their links point to the same page after cleanup (tracking parameters such as utm_*, fbclid or ref, AMP, www./m., trailing slash and #fragment removed), or
2. their headlines contain the same words after cleanup (case, accents, punctuation, word order, small words such as "the" or "with", "Updated:" / "Breaking:" prefixes, and the feed's own name at the end are ignored). Long headlines (8+ words) may differ by one extra word. Numbers must match exactly, so "Rust 1.90" and "Rust 1.91" stay separate, and headlines more than 7 days apart are never merged by title.

The rules are deliberately strict. A false merge would hide a story from you, while a missed merge only shows a duplicate. The earliest published copy becomes the main row; set Also output duplicate rows to get the others too, each with dupOf pointing to the main story.

How are stories ranked? Stories covered by more sources come first, then the newest.

Does it work with Google News and Bing News? Yes. The same headline from different publishers ("... - Reuters", "... - Yahoo") becomes one row, source is the publisher (from the feed's <source>), feedTitle is the feed it came from, and sourceCount counts publishers.

How do keyword filters match? Whole words only (rust does not match "trust"), case and accent insensitive, any language. Keywords are searched in the title, summary and categories.

What does "only new" do? With the same stateKey, each run returns only stories you have not received yet, including late copies of an old story in another feed. Stories are remembered for 30 days.

What happens when a feed is broken? The run continues. Each broken feed adds a free feed-error row with the reason (HTTP 404, timeout, not a feed, rate limited).

Limits.

• Feed content only: full article text is not fetched (the actor never visits article pages).
• Up to 500 feeds and 1000 items per feed per run; a feed larger than 20 MB is reported as an error.
• Private and local network addresses are blocked, for feed URLs and for the webhook URL.
• Runs use at most 1 GB of memory. 500 feeds x 50 items (25,000 items) needs under 400 MB; for hundreds of feeds with 1000 items each, split the feeds over several runs or tasks.

More

Real output, prices and changelog: RSS Feed Aggregator, merge and dedupe feeds.