Web results for a query: Google's results page read through the ladder, Bing's when Google refuses. Organic results only -- title, URL, snippet -- each saying which engine it came from.

Waits for the results like a one-URL /scrape -- up to timeout seconds (60 by default, 120 at most), then 202 with the id to poll at GET /search/{id}. scrape hands the results' URLs to a batch, and each result gains its page as it lands. A refused results page is free; the same search within the hour is answered from the last one at no charge.

webhook (a URL, or {url, events, metadata}) is told once how the search ended -- search.completed or search.failed, an answer from the cache included -- signed with webhookSecret, returned here only.