RSS and Atom Feed Reader: Any Feed to JSON

Give it a list of RSS, Atom or JSON feeds, or just website addresses, and get every item back as clean, consistent JSON: title, link, publish date in ISO format, a short plain-text summary, the full content as Markdown, categories, image and podcast files.

It is built for news monitoring, content pipelines and AI agents: one schema for every feed format, only-new mode for scheduled runs, and $1 per 1,000 items with only Apify's standard $0.00005 per run on top. Feeds that cannot be read are free.

What it does

• Reads every common feed format: RSS 2.0, RSS 0.9x, RSS 1.0 (RDF), Atom and JSON Feed, including feeds with legacy character sets, odd date formats and sloppy markup.
• Finds the feed for you: paste https://www.nasa.gov/ instead of hunting for the feed address. The reader follows the feed link the site declares, or tries the usual addresses (/feed, /rss.xml, /atom.xml, /feed.xml, /index.xml).
• Only new items: turn on onlyNew and schedule the run. Each run returns only items it has not returned before, so you can send them straight to Slack, a spreadsheet or your database without duplicates.
• AI-ready content: full item content as Markdown with absolute links, or the original HTML, or plain text.
• Polite: plain HTTP, no proxy, no browser. It reads each site's robots.txt and skips feeds the site disallows for bots.

Input

Field	Required	Default	Notes
feeds	yes		Feed or website addresses, up to 5,000 per run
maxItemsPerFeed	no	50	Newest items per feed, 1 to 10,000
onlyNew	no	false	Return only items not returned by an earlier run
maxAgeDays	no		Only items published in the last N days
contentFormat	no	markdown	markdown, html, text or none
includeAuthor	no	false	Add author names (email addresses are always removed)

{
    "feeds": [
        "https://www.federalreserve.gov/feeds/press_all.xml",
        "https://www.nasa.gov/"
    ],
    "maxItemsPerFeed": 20,
    "onlyNew": true
}

Output example

One dataset item per feed item. A real item from the input above, with the summary and content shortened:

{
    "input": "https://www.nasa.gov/",
    "feedUrl": "https://www.nasa.gov/feed/",
    "feedTitle": "NASA",
    "feedFormat": "rss",
    "status": "ok",
    "error": null,
    "itemId": "https://www.nasa.gov/?post_type=press-release&p=1060823",
    "title": "NASA Releases MAX POWER Key Events for Media, Participants",
    "link": "https://www.nasa.gov/news-release/nasa-releases-max-power-key-events-for-media-participants/",
    "publishedAt": "2026-10-30T19:31:02.000Z",
    "updatedAt": null,
    "author": null,
    "summary": "... NASA has announced key events beginning Friday, Nov. 6, and running through Sunday, Nov. 8, hosted on and near the agency’s Kennedy Space Center in Florida ...",
    "content": "[![Looking up, three white and black fighter jets fly in formation across a bright, blue sky.](https://www.nasa.gov/wp-content/uploads/2026/10/grc-2026-c-06389large.jpg?w=1920)](https://www.nasa.gov/wp-content/uploads/2026/10/grc-2026-c-06389large.jpg)\n\n...",
    "categories": ["General", "Aeronautics", "Kennedy Space Center"],
    "imageUrl": "https://www.nasa.gov/wp-content/uploads/2026/10/grc-2026-c-06389large.jpg?w=1920",
    "enclosures": [],
    "fetchedAt": "2026-10-31T17:44:47.646Z"
}

• itemId is the feed's own id (guid), else the link, else a hash of title and date. It is stable across runs.
• summary is plain text of up to 1,000 characters. content is never cut.
• enclosures lists attached files such as podcast audio: url, type and length in bytes.
• Items come newest first when every item in the feed has a date; otherwise in feed order.
• Fields the feed does not have are null, never guessed.

A feed that cannot be read still gets one row (shown here without its empty fields), so you always know what happened. It is not charged:

{
    "input": "https://www.reddit.com/r/programming/.rss",
    "status": "blocked-by-robots",
    "error": "https://www.reddit.com/robots.txt disallows https://www.reddit.com/r/programming/.rss for crawlers, so it was skipped."
}

Other statuses: http-error, not-a-feed, no-feed-found, feed-failed, invalid-feed, too-many-redirects, fetch-failed, invalid-url, failed (an unexpected error).

Pricing

Event	Price	When
Feed item	$0.001	Every item in your dataset
Actor start	$0.00005	Once per run

That is $1 per 1,000 items. The start charge is Apify's standard $0.00005, with no extra start fee on top, so reading one 20-item feed costs $0.02005. Error rows are free, and with onlyNew a run that finds nothing new costs only the $0.00005.

The reader checks your run's maximum charge before writing items and stops there, so a run never costs more than the limit you set.

Use it on a schedule

1. Save your input as a task with onlyNew on.
2. Add a schedule (every hour, every 15 minutes, daily).
3. Connect the dataset to Slack, Google Sheets, Zapier, Make or a webhook, or read it through the Apify API.

Each saved task has its own memory, so two tasks can watch the same feed for different purposes without hiding items from each other.

Limits

• Up to 5,000 feeds per run and 10,000 items per feed.
• Feeds larger than 8 MB (a few long-running podcasts) are read up to their first 8 MB, which holds the newest episodes.
• An item with more than 300,000 characters of HTML comes back as plain text instead of Markdown.
• Dates are what the feed states. Dates in non-English formats come back as null.
• The content is whatever the feed carries. Many feeds only include a teaser; to get the full article, send the link values to a web page to text tool.

FAQ

Can I paste a normal website address? Yes. If the address is a web page, the reader looks for the feed the page links to, then tries the usual feed addresses.

Why did a feed come back as blocked-by-robots? The site's robots.txt tells bots not to read that address, and this reader respects it. Some large sites (for example YouTube, Reddit and Google News) disallow their feeds for automated readers. These rows are free.

How does only-new mode remember what I have seen? It keeps the ids of returned items in a key-value store named rss-feed-reader-seen in your own Apify account, one record per feed and task. Items are remembered while they stay in the feed and for 90 days after they leave it. The first run returns the newest items; later runs return only items published since.

Am I charged for feeds that fail? No. Only items in your dataset are charged.

Does it collect personal data? Author names are off by default, and email addresses are always removed, even from author fields. You get what the publisher put in the public feed and nothing else.

Can AI agents use it? Yes. Every feed format comes out in the same flat schema with ISO dates and Markdown content, which is easy to pass to an LLM. The actor can be called from the Apify API or as a tool in Apify's MCP server.

Do I need the right to use the content? Yes. Feeds are published to be read by feed readers, but the articles belong to their publishers. Check the publisher's terms before you republish their content.