Extract
Extract API
Read any URL and get the page back as raw HTML, Markdown, or plain text - the read-a-page primitive an agent needs when the answer is behind a link rather than in a search result. Tier-priced by mode: normal and advanced cost 1 credit, ultra costs 2. Only a successful extraction is billed, so a dead link, a bot wall or a timeout costs nothing.
Authorizations
AuthorizationstringheaderrequiredBearer authentication header of the form Bearer <token>, where <token> is your Scavio API key (e.g. Bearer sk_live_your_key).
Body
application/jsonurlstringrequiredThe page to read, 1-2048 characters. http and https only; a bare host like example.com is upgraded to https. Loopback, private, link-local and cloud-metadata hosts are rejected with a 400 before anything is fetched, so they cost nothing.
Example: https://en.wikipedia.org/wiki/Web_scraping
formatenum<string>default:markdownHow to return the page. Whatever you pick, the shape is the same: content is a single string and content_length is its length in characters.
html— The raw page exactly as served, chrome and scripts included.markdown— The page converted to Markdown, with the metadata preamble removed. Navigation and footer text can remain on some sites. The default.text— That same Markdown flattened to plain text. The flattener is deliberately conservative, so snake_case identifiers and inline code survive intact.
modeenum<string>default:normalFetch tier, and the price-bearing parameter - this is what makes the call cost 1 credit or 2. Start at normal and step up only when a site refuses it.
normal— The standard fetch. 1 credit. Right for static pages, docs, articles and most blogs.advanced— Renders JavaScript, for pages whose content is built in the browser. Also 1 credit.ultra— For the hardest pages. 2 credits.
Notes
Need a whole site, not one page? Looping Extract over a site works, but three operations do it in one call, at the same per-page price:
- Map lists a site's page URLs from its sitemaps and start page. 1 credit.
- Crawl discovers and extracts a site's pages in one job, robots.txt obeyed, billed per successful page.
- Batch Extract extracts up to 1,000 URLs you already have in one job, billed per successful page.
Crawl and batch run as jobs: a 202 with an id, results from a status URL or a signed webhook, and no per-request concurrency limit on the pages they fetch.
Request
from scavio import ScavioClient
client = ScavioClient(api_key="sk_live_your_key")
# client.extract(url) reads one page.
page = client.extract("https://en.wikipedia.org/wiki/Web_scraping")
print(page["data"]["content"])Response