News Article Extractor: RSS, URLs & Clean Text

News Article Extractor turns news article links, news sites, RSS/Atom feeds and news sitemaps into clean article text with title, author, publish date, language, tags and main image.

Use it for media monitoring, research datasets and AI pipelines. The smallest run is one article link: you get one row with the article text in a few seconds.

Table of contents

• What it does
• Quick start
• Input
• Output
• Pricing and limits
• API, schedules and AI agents
• Troubleshooting
• FAQ
• Support

What it does

The Actor works in two ways, and you can use both in one run:

• Article links (articleUrls): it opens each link and returns the article body without menus, ads, bylines, photo captions and newsletter boxes.
• News sites (siteUrls): give a homepage like https://www.theverge.com. The Actor finds the site's RSS/Atom feed or news sitemap, takes the newest articles and extracts each one. You can also give the feed or sitemap URL directly.

Every article becomes one Dataset row. The body comes as plain text, Markdown or both.

Good fit

• Daily or hourly news monitoring: schedule the Actor with Only new articles and get only articles you have not received before.
• Building a text dataset from a list of article links for analysis, summaries or RAG.
• Getting publish dates, authors and tags for a list of article URLs.

Not a fit / limits

• Paywalls and logins are not bypassed. For subscriber-only pages you get only the public part (see isAccessibleForFree).
• Sites that block automated requests (for example some large publishers with bot protection) return a BLOCKED error. The Actor does not use a browser or proxies.
• Sites without an RSS/Atom feed or a sitemap cannot be used in siteUrls. Use their article links in articleUrls instead.
• Comments, videos and image galleries are not extracted.

Quick start

1. Open the Input tab.
2. Keep the prefilled news site https://www.theverge.com in News sites, RSS feeds or sitemaps, or add your own article links to Article URLs.
3. Keep Max articles from sites at the prefilled 10.
4. Click Start.
5. Open the Output tab. The Articles view shows title, author, date and site; Full text shows the article body.

Expected result: 12 rows in a few seconds: the 2 prefilled BBC article links plus the 10 newest articles from the site, each with title, publishedAt and text.

Input

Input fields

Field	Plain name	Type	Default	Allowed values	What it does
articleUrls	Article URLs	array of URLs	empty	up to 500 public http(s) links	Each link is extracted. maxArticles and dateFrom do not apply here.
siteUrls	News sites, RSS feeds or sitemaps	array of URLs	empty	up to 50	Homepage, RSS/Atom feed or sitemap. The Actor finds the feed itself for a homepage.
maxArticles	Max articles from sites	integer	20 (prefill 10)	1–500	Total articles taken from all siteUrls, newest first, sites take turns.
dateFrom	Only articles published after	string	none	2026-10-01 or 7 days	Skips site articles older than this date. Articles without a date are skipped when it is set.
onlyNew	Only new articles	boolean	false	true / false	Remembers delivered article URLs and skips them in the next runs.
stateKey	Memory name for 'Only new'	string	default	letters, digits, -, _	Keeps separate memory for different schedules.
outputFormat	Article body format	string	text	text, markdown, both	Which body fields are filled.
minWords	Minimum words	integer	50	0–5000	Pages with fewer words are skipped (site pages) or reported as TOO_SHORT (article links).

At least one of articleUrls or siteUrls is required. The same article found twice (also through AMP or canonical links) is returned once.

Input examples

Newest articles from several news sites

{
  "siteUrls": [
    { "url": "https://www.theverge.com" },
    { "url": "https://feeds.bbci.co.uk/news/rss.xml" }
  ],
  "maxArticles": 20,
  "dateFrom": "2 days"
}

Expected behavior: the Actor reads both feeds and takes up to 20 newest articles from the last 2 days, alternating between the sites.

Scheduled monitoring without repeats

{
  "siteUrls": [{ "url": "https://techcrunch.com" }],
  "maxArticles": 50,
  "onlyNew": true,
  "stateKey": "techcrunch-hourly"
}

Expected behavior: the first run returns the newest articles. Next runs return only articles that appeared since the last run. A run with nothing new ends with status not_found and no article charges.

A list of article links as Markdown

{
  "articleUrls": [
    { "url": "https://www.bbc.co.uk/news/articles/cm2d6e62p49do" },
    { "url": "https://www.bbc.co.uk/news/articles/ck054nvzr676o" }
  ],
  "outputFormat": "markdown"
}

Expected behavior: two rows with the article body in markdown. A link that fails is listed in the run summary with an error code; the other links still return rows.

Output

Output fields

Field	Meaning	Can be empty?
url	Article URL after redirects	no
canonicalUrl	Canonical URL declared by the page	falls back to url
title	Headline	rarely null
author	Author name(s), comma-separated	null when the page names no author
publishedAt / modifiedAt	Publish and update time, ISO 8601 UTC	null when unknown
lang	Language code, e.g. en, en-GB	null when not declared
siteName	Publisher name	can be null
description	Short summary	can be null
tags	Publisher keywords and sections	empty array when none
image	Main image URL	can be null
text	Body as plain text, paragraphs separated by a blank line	null when outputFormat is markdown
markdown	Body as Markdown with headings, links and lists	null when outputFormat is text
wordCount	Words in the body	no
isAccessibleForFree	Publisher paywall flag: false = subscriber-only page, only the public part returned	null when not declared
discoveredVia	input, rss, atom or sitemap	no
feedUrl	Feed or sitemap where the article was found	null for articleUrls
fetchedAt	When the page was fetched	no

Output example

{
  "url": "https://www.theverge.com/tech/1010990/fujifilm-instax-square-2-analog-instant-film-camera-filters",
  "canonicalUrl": "https://www.theverge.com/tech/1010990/fujifilm-instax-square-2-analog-instant-film-camera-filters",
  "title": "Fujifilm’s new square format instant camera adds analog color filters",
  "author": "Andrew Liszewski",
  "publishedAt": "2026-10-14T13:43:36.000Z",
  "lang": "en-US",
  "siteName": "The Verge",
  "tags": ["Cameras", "Gadgets", "News", "Tech"],
  "text": "Continuing the analog approach of the Instax Square SQ6 that launched in 2018…",
  "markdown": null,
  "wordCount": 250,
  "isAccessibleForFree": false,
  "discoveredVia": "atom",
  "feedUrl": "https://www.theverge.com/rss/index.xml"
}

This example is trimmed (fields modifiedAt, description, image, fetchedAt and most of text are left out).

Run summary and statuses

Every run also saves RUN_SUMMARY in the Key-value store (the Run summary link in Output). It has status, reason, counts of extracted and skipped pages, feeds found per site and failed URLs with error codes.

Status	What it means	What to do
complete	All requested articles were extracted	Use the Dataset
partial	Some rows were extracted, but some links failed, a site had no feed or the spending limit was reached	Check failedUrls and reason
not_found	The sources worked, but no new or matching articles were found (for example with onlyNew or dateFrom)	Nothing to fix; no articles were charged
error	No article could be extracted because sites failed or had no feed. The run is marked as failed	Check the first error code in the status message

Common error codes: NOT_FOUND (404), BLOCKED (403, bot protection or login), TIMEOUT, HTTP_5xx, NOT_AN_ARTICLE (section page, video page, feed or file), TOO_SHORT, NO_FEED_FOUND, PRIVATE_URL, INVALID_URL.

The run also stops early, keeping every article already saved, with one of these reasons: DEADLINE (10-minute limit), CHARGE_LIMIT (your spending limit), TOO_MANY_FAILURES (8 network failures in a row), TOO_MANY_UNREACHABLE_SITES (6 sites in a row did not respond; the remaining sites get the warning NOT_SCANNED and are not charged). With saved articles the status is partial, without them error.

Pricing and limits

Pricing unit

You pay per event. Platform usage is included.

Event	When it is charged	Free plan	Bronze	Silver	Gold+
Actor start	Once per run (one per GB of memory, at least one)	$0.003	$0.003	$0.003	$0.003
Article extracted	One row in the Dataset	$0.0021	$0.002	$0.0019	$0.0018
Site scanned	One entry of siteUrls searched for feeds, once per run, also when the site has no feed or does not respond	$0.002	$0.002	$0.002	$0.002
Article URL failed	One entry of articleUrls that the site answered with an error, a non-article page or a page shorter than minWords	$0.0005	$0.0005	$0.0005	$0.0005

Free: duplicates, skipped site pages (too short, older than dateFrom, not an article), articles already returned with onlyNew, invalid and private URLs, URLs skipped after repeated network failures.

A valid run with no new or matching articles pays only the start and the sites scanned. Set a spending limit in the run options: the Actor stops before it and ends with status partial, reason CHARGE_LIMIT.

Cost examples (Free plan prices)

Scenario	Input	Charged events	Price
Two article links	2 articleUrls	start + 2 articles	$0.0072
Prefilled run	1 site, maxArticles 10	start + 1 site + 10 articles	$0.026
Daily digest	5 sites, maxArticles 100	start + 5 sites + 100 articles	$0.223
Hourly check, nothing new	5 sites, onlyNew	start + 5 sites	$0.013
Large batch	500 articleUrls, 480 extracted, 20 failed	start + 480 articles + 20 failed	$1.02

Limits

• Up to 500 articleUrls, 50 siteUrls and 500 articles from sites per run.
• A run works for at most 10 minutes and stops with status partial (reason DEADLINE) when time runs out. Split very large lists into several runs.
• Memory is set automatically: 512 MB, or 1 GB when you ask for more than 200 articles.
• Public pages only. Private network addresses and links with passwords are rejected.

API, schedules and AI agents

• API: start the Actor with the Apify API and read rows from the default Dataset.
• Export: download the Dataset as JSON, CSV, Excel or XML from the Output tab.
• Schedules: create a schedule in Apify Console with onlyNew: true and a stateKey per schedule to receive only new articles.

Use with AI agents (MCP)

• MCP server with only this Actor: https://mcp.apify.com?tools=cliqtomedia/news-article-extractor (see Apify MCP server).
• Smallest useful input: {"articleUrls": [{"url": "https://www.bbc.co.uk/news/articles/cm2d6e62p49do"}]}
• Key output fields: title, publishedAt, author, text (or markdown), url.
• A run with one site and 10 articles usually finishes in under a minute and costs about $0.026 on the Free plan. Up to 500 article links or 50 sites per run.

Troubleshooting

A site returns BLOCKED

Symptom: failedUrls shows BLOCKED for every link of one site. Likely reason: the site uses bot protection or requires login. Action: this Actor cannot read that site. Blocked links are charged as failed URLs ($0.0005 each), so remove such sites from scheduled runs.

NO_FEED_FOUND for a site

Symptom: the site warning says NO_FEED_FOUND. Likely reason: the homepage does not link a feed and robots.txt lists no sitemap. Action: find the feed URL yourself (often /feed, /rss or /rss.xml) and put it in siteUrls, or use article links in articleUrls.

Fewer articles than maxArticles

Likely reason: the feed has fewer recent items, dateFrom or onlyNew removed some, or the site's pages were videos or short teasers. The run summary shows skipped counts. When a site returns many skipped pages in a row, the Actor stops reading it (warning TOO_MANY_SKIPPED).

The text contains extra lines

Some sites mix related links or author bios into the article body. The Actor removes common bylines, captions and newsletter boxes, but layouts differ. Report the URL in Issues.

FAQ

Do I need an API key or login?

No. The Actor reads public pages only.

Does it bypass paywalls?

No. For subscriber-only pages it returns the public part and isAccessibleForFree: false.

How do I get only new articles every day?

Set onlyNew to true and run the Actor on a schedule. Use a different stateKey for each schedule. You are charged only for articles you have not received yet.

Why is publishedAt null?

The page and the feed did not give a publish date. When dateFrom is set, such site articles are skipped.

Support

For bugs, questions or feature requests, open an issue in the Issues tab of this Actor. Include the run ID, your input (without secrets) and the error code from RUN_SUMMARY. Never send tokens, cookies or passwords.