ScrapingBee → APITube migration kit

For teams using ScrapingBee to collect news — either scraping publisher sites with the HTML API, or scraping Google News with search_type=news — and wondering whether a news API would be simpler.

This kit is not a parameter table pretending the two products are the same. ScrapingBee is a general-purpose scraper; APITube is a news index. Most of ScrapingBee's parameters exist to make scraping work, and after this migration there is no scraping to make work.

Every APITube claim here was executed against the live API before being written down.

Which of the two are you doing?

A. Scraping publisher sites with the HTML API (url=, extract_rules=, render_js=). You maintain a URL discovery step, a fetcher, extraction rules per publisher, and the breakage that comes with all three. → reference/from-html-api.md

B. Scraping Google News with the Google Search API (search_type=news). You get titles, links and snippets, then fetch each link to get the text. → reference/from-google-search.md

Most teams doing news at any scale end up doing both.

The arithmetic, using ScrapingBee's own numbers

Credit costs are from ScrapingBee's knowledge base:

Request	Credits
Classic proxies without JS rendering	1
Classic proxies with JS rendering	5
Premium proxies without JS rendering	10
Premium proxies with JS rendering	25
Stealth proxies	75
Google Search API, light request	10
Google Search API, normal request	15

News sites are the case that pushes you up that table: paywalls, consent walls, lazy-loaded body text and aggressive bot detection are the norm, so render_js and often premium_proxy become non-optional.

One article, scraped: find the URL, then one HTML API request at 5–25 credits, then extraction rules that you keep working per publisher.

One article, from APITube: part of a response that returns 250 of them, body included.

Verified 27 July 2026: title=tesla&per_page=250 returned 250 articles, all 250 with a non-empty body, averaging 4 818 characters each. One request.

What is in here

Path	What it gives you
reference/from-html-api.md	Scenario A — extract_rules mapped onto fields APITube already returns
reference/from-google-search.md	Scenario B — search_type=news parameters and news_results fields
reference/what-you-keep-scraping.md	Honest list: what APITube does not cover, so keep ScrapingBee for it
reference/limitations.md	APITube quirks that will bite during the migration
shim/node/	ScrapingBeeNewsShim — accepts search_type=news parameters, returns the ScrapingBee response shape
shim/python/	Same shim for Python
examples/	Before/after, including the code that disappears
tools/ai-migration-prompt.md	System prompt for Claude/ChatGPT that converts your queries

Quick start

# ScrapingBee, Google News
curl "https://app.scrapingbee.com/api/v1/store/google?search=tesla&search_type=news&country_code=us" \
  -H "Authorization: Bearer YOUR_KEY"

# APITube
curl "https://api.apitube.io/v1/news/everything?title=tesla&source.country.code=us&per_page=250" \
  -H "X-API-Key: YOUR_API_KEY"

The second one includes the article text. The first one includes a link to it.

The honest part: keep ScrapingBee for these

A migration kit that claims you can delete your scraper is lying. APITube replaces the news-collection half of the job. It does not do:

• Sites outside the index. APITube indexes news publishers. A competitor's pricing page, a government portal, a forum thread — still a scrape.
• Pages behind a login. No session handling, no cookies, no js_scenario.
• Rendering a specific page as a user sees it. No screenshots, no viewport control.
• Arbitrary DOM extraction. extract_rules against a page structure APITube does not model.
• Non-news content on news domains. Job listings, event calendars, e-commerce sections.

The realistic outcome is that your ScrapingBee bill shrinks to the non-news work rather than going to zero. reference/what-you-keep-scraping.md is the detailed version.

What you stop maintaining

Stage	With ScrapingBee	With APITube
URL discovery	RSS parsing, sitemap crawling, or SERP scraping	The query is the discovery
Fetching	One request per article, credits per request	Included
Extraction	Per-publisher extract_rules, revisited when a site redesigns	Structured fields
Body text	return_page_text and hope the selectors held	body and body_html
Enrichment	Your own NLP on top	Entities, sentiment, IPTC categories, topics, industries, readability
Deduplication	Yours	is_duplicate and story.id
Publisher metadata	Yours	Domain, country, political bias, Open Page Rank

License

MIT. Use it, fork it, ship it.