Article Extractor – News Articles to Text, Markdown & Metadata

📰 Article Extractor

Turn any news site, blog, RSS feed or sitemap into clean, structured articles. Paste a home page such as theguardian.com and get the latest articles with title, author, publish date, full text, LLM-ready markdown, lead image, language and tags in seconds. No login, no coding, and you pay per article.

✨ Why use Article Extractor?

• 🔎 Finds the articles for you: give it a home page, a section page, an RSS/Atom feed or a sitemap. It reads the page's article links, the site's feeds, its news sitemaps and a few section pages, and skips tag, author, video and login pages.
• 🧹 Clean article text: boilerplate, menus, ads and related-story boxes are removed. You get the body as plain text and as markdown with headings, lists and tables, ready for LLMs, RAG and search indexes.
• 🗞️ Real news metadata: author(s), publish and update time with time zone, section, tags, lead image, publisher, language, canonical URL, paywall flag and embedded videos, read from schema.org NewsArticle data, Open Graph and meta tags.
• 📅 Date and keyword filters: keep only articles from the last 24 hours, a date range or that mention your keywords. Feed and sitemap dates are checked before a page is downloaded, and filtered articles are never charged.
• 🔁 Only new articles: schedule it every hour and get each article once. Articles already delivered in earlier runs are skipped.
• ⚡ Fast and cheap: plain HTTP with browser fingerprints, many pages in parallel, and a real browser only for pages that need JavaScript. In our tests 20 articles from two news sites took about 10 seconds, and 1,000 articles about 3 minutes.

🎯 Use cases

• 📡 News monitoring: track what outlets publish about your brand, competitors or industry, every hour, with keyword filters.
• 🤖 LLM and RAG pipelines: feed fresh, clean markdown articles into embeddings, summarizers and AI agents.
• 📚 Research corpora: build datasets of articles per outlet, topic or date range for NLP, media studies and journalism research.
• 📰 Newsletters and aggregators: collect the latest stories from many sources with title, summary, image and author.
• 📈 Content analysis: compare publishing volume, authors, sections and article length across sites.

🚀 How to use

1. Add news sites, section pages, RSS feeds or sitemaps under Start URLs, and/or direct article links under Article URLs.
2. Set Max articles (default 20) and, if you like, a date window such as Published from = 24 hours.
3. Click Start. Articles appear in the dataset while the run is going.
4. Download the results as JSON, CSV, Excel or HTML, or read them through the API.

📥 Input

{
  "startUrls": ["https://www.theguardian.com", "https://techcrunch.com"],
  "articleUrls": ["https://techcrunch.com/2024/05/13/openais-newest-model-is-gpt-4o/"],
  "maxArticles": 20,
  "dateFrom": "3 days",
  "keywords": ["AI", "climate"]
}

📤 Output

One row per article.

{
  "url": "https://techcrunch.com/2026/10/15/ramp-reaches-60b-valuation-as-it-pushes-deeper-into-ai/",
  "canonicalUrl": "https://techcrunch.com/2026/10/15/ramp-reaches-60b-valuation-as-it-pushes-deeper-into-ai/",
  "title": "Ramp reaches $60B valuation as it pushes deeper into AI",
  "author": "Ram Iyer",
  "authors": ["Ram Iyer"],
  "publishedAt": "2026-10-15T14:47:09Z",
  "modifiedAt": "2026-10-15T15:29:11Z",
  "description": "The funding comes as Ramp's annualized revenue crosses $2B, and its customer count reaches 85,000.",
  "siteName": "TechCrunch",
  "publisher": "TechCrunch",
  "domain": "techcrunch.com",
  "language": "en",
  "section": "Fintech",
  "categories": ["Fintech", "Fundraising", "Startups"],
  "tags": ["ai tokens", "expense management", "Ramp"],
  "image": "https://techcrunch.com/wp-content/uploads/2026/10/eric-glyman.jpg",
  "wordCount": 541,
  "readingTimeMinutes": 3,
  "text": "Expense management company Ramp said on Thursday that it had raised $1.85 billion in a new funding round ...",
  "markdown": "Expense management company Ramp said on Thursday that it had raised $1.85 billion in a new funding round ...",
  "isPaywalled": null,
  "pageType": "article",
  "status": "ok",
  "discoveredFrom": "page",
  "startUrl": "https://techcrunch.com/",
  "extractedWith": "http",
  "scrapedAt": "2026-10-15T20:53:20Z"
}

💰 Pricing

You pay $0.90 per 1,000 articles ($0.0009 per article), plus a small amount of platform usage (compute and proxy traffic). Only extracted articles are charged. Pages that turn out not to be articles, articles outside your date range or keywords, and failed URLs are free.

Example: 1,000 articles ≈ $0.90 in article fees plus about $0.07 of proxy traffic and compute (measured on a 1,000-article run), so about $0.97 in total. A typical 20-article run costs about $0.02. Pages that need the browser fallback add browser usage for those pages only.

❓ FAQ

Is it legal to extract articles? Collecting publicly available pages is generally allowed, but articles are protected by copyright. Check each site's terms before you republish content. Use the data for analysis, monitoring and research.

Do I need a login, API key or proxy? No. The actor uses datacenter proxies by default and retries blocked pages over a residential proxy automatically.

How many articles can I get? As many as the site exposes. Home pages and feeds give the newest 20 to 200 articles; sitemaps let you go back thousands of articles. Set Max articles to 0 for no limit.

What if a field is empty? Fields such as author, tags or section are filled only when the site publishes them. The publish date is taken from structured data, meta tags, the page or the URL; if a site shows none it stays empty (or enable "Must have a publish date").

Does it work on paywalled sites? It extracts what the site sends to a normal visitor. On hard paywalls that is often the teaser only, and isPaywalled is set when the site declares it. Some sites serve more to the Googlebot user agent option.

Can I schedule it or use the API? Yes. Schedule runs with "Only new articles" and a date window for hourly monitoring, and read the results through the API, webhooks or integrations.

⚠️ Limits

• Sites behind strong anti-bot walls (for example Reuters) may stay blocked even after the residential retry and browser fallback. Those pages are reported and not charged.
• Content behind a login or hard paywall is not unlocked.
• Comments, live-blog updates loaded later by JavaScript and image galleries are not extracted as separate items.
• Section and sitemap discovery is automatic. For full control, give the RSS feed, sitemap or article URLs you want.