Give /v1/extract a URL and a prompt or a JSON schema, and get back the data — no CSS selectors to write or maintain, no scraper to rewrite the next time the page's markup changes.
Point it at a URL, describe what you want in plain language (or a schema, if you need consistent field names), and the response comes back as { "data": { ... } }.
curl "https://snapurl.app/v1/extract?url=https://example.com&prompt=Get+the+page+title" \ -H "Authorization: Bearer $SNAPURL_KEY"
A prompt is fastest for one-off lookups. A schema is the better choice once another part of your system expects a fixed shape back — the same field names, request after request, regardless of how the source page is laid out.
Set include_screenshot=true and the response includes a base64-encoded screenshot alongside the extracted data — one request, one bill, and a visual record of exactly what the extraction saw. Most extraction APIs make you run a second request (and pay for it) to get that.
Selector-based scrapers break the moment a site redesigns a class name or restructures a div. Extraction describes the data you want instead of the markup that currently holds it, so it keeps working across the redesigns that would otherwise send you back to rewrite a scraper.
AI extraction is included on the Starter and Growth plans. Free-plan requests to /v1/extract return 403 plan_not_eligible.
Yes — GET /v1/extract takes a url and a prompt or a JSON schema, and returns structured JSON extracted from the page with AI. Available on the Starter and Growth plans. AI data extraction →
More in the full FAQ.