RSS & Atom Feed Reader + Auto-Discovery (News Monitor)

Read RSS 2.0, RSS 1.0 (RDF), Atom and JSON Feed from feed URLs — or just paste a website URL and the Actor auto-discovers its feed. Every item comes back as the same clean, stable JSON no matter the feed format: title, link, author, ISO dates, plain-text summary, categories, images and podcast enclosures. Filter by keywords and date, and switch on only-new mode to get just the items you have not seen yet — ideal for scheduled news alerts, RSS-to-Slack/email automations, content monitoring and AI agents (MCP).

Part of IntegriApps MicroApps. Support Sunday to Friday.

What it does

• All feed formats, one output: RSS 2.0 (incl. content:encoded, dc:creator, Media RSS, iTunes enclosures), RSS 1.0/RDF, Atom 1.0 (incl. xhtml content and xml:base) and JSON Feed 1.0/1.1, normalized to one schema.
• Feed auto-discovery: give a normal web page (e.g. https://github.blog/) and it finds the feed from <link rel="alternate">, then tries common paths (/feed, /rss.xml, /atom.xml, /feed.xml, /index.xml, /rss, /feed.json). Comment feeds are tried last.
• Filters: include/exclude keywords (title, summary and categories, case-insensitive), sinceHours or sinceDate, max items per feed and per run. Items are sorted newest first.
• Only-new mode for alerts: remembers item IDs between runs in a named key-value store in your own Apify account, so a scheduled run returns (and charges) only new items. It also sends If-None-Match / If-Modified-Since, so unchanged feeds cost almost nothing.
• Per-feed errors never crash the run: a broken, blocked or missing feed becomes one record with a status (error, blocked_by_robots_txt, no_feed_found, not_a_feed, parse_error) and the other feeds keep going. Errors are not charged.
• Tolerant parser: handles BOMs, comments before the root, CDATA, HTML entities, relative links and many date formats; malformed feeds return whatever items can be read instead of failing.
• Polite by design: identifiable User-Agent (IntegriAppsRssFeedReader), respects robots.txt (incl. Crawl-delay), retries with exponential backoff and Retry-After, timeouts, limited concurrency. Plain HTTP only — no browser, no proxies, no bypassing of blocks, logins or paywalls. Runs in 128–256 MB.
• No article scraping: it returns only what the feed itself publishes. contentHtml is included only if you ask for it and only when the feed contains full content.

Use cases

• News and brand monitoring: watch tech blogs, newsrooms, government and science agencies (e.g. NASA), and get only new items that mention your keywords.
• Alerts: schedule the Actor every 15–60 minutes with only-new mode and send new items to Slack, Teams, email or a webhook via Zapier, Make or n8n.
• AI agents and RAG: give an agent an MCP tool that turns "what's new on this site?" into structured JSON; feed fresh items into a vector database.
• Podcast and release tracking: collect episodes (audio enclosures) or project release feeds that allow automated access.

Input

Field	Description	Default
startUrls	Feed URLs or website URLs (up to 1,000)	— (empty = free 5-item demo with the NASA news releases feed)
maxItemsPerFeed	Newest items per feed after filters (1–1,000)	50
maxTotalItems	Cap of items in the whole run	5,000
includeKeywords / excludeKeywords	Words or phrases matched in title, summary and categories	—
sinceHours / sinceDate	Keep only recent items (items without a date are dropped when used)	—
onlyNew	Return only items not returned before (state kept between runs)	false
stateStoreName	Named key-value store used by only-new mode	rss-feed-reader-state
autoDiscover	Discover feeds from HTML pages	true
maxFeedsPerSite	Feeds to read when a site lists several	1
includeContent	Add contentHtml with the full HTML that the feed publishes	false
maxSummaryChars	Plain-text summary length (0 = no limit)	500
maxConcurrency	Feeds fetched in parallel (1–10)	5

Example — AI news alert from three sources, only new items from the last 3 days:

{
  "startUrls": [
    { "url": "https://feeds.bbci.co.uk/news/technology/rss.xml" },
    { "url": "https://feeds.npr.org/1019/rss.xml" },
    { "url": "https://simonwillison.net/" }
  ],
  "includeKeywords": ["AI", "LLM", "model"],
  "excludeKeywords": ["sponsored"],
  "sinceHours": 72,
  "onlyNew": true,
  "stateStoreName": "ai-news-alert"
}

Output

One dataset item per feed entry. Real item from a run (NPR technology RSS):

{
  "status": "ok",
  "feedUrl": "https://feeds.npr.org/1019/rss.xml",
  "feedTitle": "NPR Topics: Technology",
  "feedFormat": "rss2",
  "siteUrl": "https://www.npr.org/templates/story/story.php?storyId=1019",
  "language": "en",
  "guid": "https://www.npr.org/2026/11/04/nx-s1-6026597/review-legend-of-zelda-ocarina-of-time-remake-switch-2",
  "title": "Review: The Legend of Zelda: Ocarina of Time remake is superb, streamlined and safe",
  "link": "https://www.npr.org/2026/11/04/nx-s1-6026597/review-legend-of-zelda-ocarina-of-time-remake-switch-2",
  "author": "James Perkins Mastromarino",
  "published": "2026-11-04T11:00:00.000Z",
  "updated": null,
  "summary": "Nintendo's new Ocarina of Time remake enshrines its place in the gaming canon for generations to come. But despite its stupendous production values, the game left me feeling oddly bereft.",
  "categories": [],
  "enclosures": [],
  "imageUrl": "https://npr.brightspotcdn.com/dims3/default/strip/false/crop/1920x1080+0+0/resize/1920x1080!/...jpg",
  "commentsUrl": null,
  "inputUrl": "https://feeds.npr.org/1019/rss.xml",
  "scrapedAt": "2026-11-04T17:06:58.890Z"
}

Website URLs work too: the input https://simonwillison.net/ was resolved to https://simonwillison.net/atom/everything/ (Atom) and returned items with categories such as ["machine-learning", "ai", "llms"].

A feed that could not be read is one record like this (not charged):

{ "status": "no_feed_found", "inputUrl": "https://example.com/", "feedUrl": null, "httpStatus": null,
  "error": "No RSS/Atom/JSON feed found on the page (<link rel=\"alternate\">) or at common paths (...)" }

Fields: status, feedUrl, feedTitle, feedFormat (rss2, rdf, atom, jsonfeed), siteUrl, language, guid (feed ID, or a stable hash if the feed has none), title, link, author, published / updated (ISO 8601 UTC or null), summary (plain text), contentHtml (only with includeContent), categories, enclosures (url, type, length), imageUrl, commentsUrl, inputUrl, scrapedAt.

The key-value store record RUN_SUMMARY has per-feed counts (items in feed, returned, filtered, duplicates, not modified, errors). Export the dataset as JSON, JSONL, CSV, Excel, XML or RSS from Apify Console or the API.

Pricing

Pay per event: USD 0.0005 per item returned (USD 0.50 per 1,000 items), plus Apify's standard tiny Actor start fee. No monthly rental and no per-run fee from us. Not charged: feed errors, robots.txt blocks, feeds without new items, items removed by your filters and items skipped as already seen in only-new mode. You can cap spending with the run's maximum charge setting; the Actor stops cleanly at the limit and items it could not return are kept as "not seen" for the next run.

Example: 20 feeds checked every hour in only-new mode that produce ~100 new items per day cost about USD 1.50 per month in item events.

Use it from your code

curl -X POST "https://api.apify.com/v2/acts/integriapps~rss-feed-reader/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"startUrls":[{"url":"https://www.nasa.gov/news-release/feed/"}],"maxItemsPerFeed":5}'

• JavaScript / Python: use the official apify-client (client.actor("integriapps/rss-feed-reader").call(input)).
• Zapier, Make, n8n: use the Apify app/node "Run Actor" (or a saved Task with a schedule), then loop over the dataset items and post them to Slack, email, Sheets or a webhook. Use onlyNew: true so each run only sends new items.
• AI agents (MCP): add integriapps/rss-feed-reader as a tool in the Apify MCP server so Claude, ChatGPT, Cursor or any MCP client can ask "what's new on these sites about X?".
• Schedules and webhooks: create a Task with your feeds, schedule it in Apify, and add a webhook on "run succeeded" to push results anywhere.

FAQ

Does it download the full articles? No. It returns only what the feed publishes. If the feed contains full content you can get it in contentHtml; otherwise you get the feed's summary and the link. Respect each publisher's terms when you reuse content.

Why was a feed blocked? Some sites disallow automated access to their feeds in robots.txt (for example, GitHub's robots.txt disallows *.atom paths), and we respect that. Those inputs come back as blocked_by_robots_txt and are free. If robots.txt cannot be fetched because of a server or network error, the feed is skipped to stay on the safe side.

How does only-new mode remember items? It stores item IDs (the feed's guid/id, else the link) per feed and per keyword set in the named key-value store you choose, in your account. Up to 5,000 IDs per feed are kept. On the first run all current items are new. Items cut by maxItemsPerFeed are also marked as seen, so later runs do not trickle old items. Use a different stateStoreName per alert.

What about items without a date? They are kept (sorted last) unless you use sinceHours or sinceDate, in which case they are dropped.

A feed works in my browser but returns error (e.g. HTTP 403/419/429). Some sites rate-limit or block requests from cloud servers. The Actor retries with backoff and then reports the feed as error (free) instead of trying to get around the block. Running less often or with lower concurrency usually helps.

Does it render JavaScript, use proxies or get around blocks? No. Feeds are plain XML/JSON, so a fast HTTP request is enough. It never uses proxies or headless browsers and never tries to bypass bot protection, logins or paywalls.

Which websites can be auto-discovered? Any site that links its feed in the HTML <head> with the standard <link rel="alternate"> tag (tested with WordPress sites such as github.blog and techcrunch.com, and with simonwillison.net; most blog platforms and news sites add it) or serves one at a common path. Sites without any feed return no_feed_found.

Does it parse podcasts? Yes, audio/video enclosures (RSS enclosure, Media RSS, Atom rel="enclosure", JSON Feed attachments) are returned in enclosures.

Changelog

See ./CHANGELOG.md.

Part of IntegriApps MicroApps — small tools by IntegriApps, a SaaS for SMBs. Support Sunday to Friday.