Try any of 919 endpoints — live.
Pick an endpoint, load a working example, tweak the params, and send — no signup to try. Results render the way the data deserves; raw JSON, headers & code are one tab away.
Fetch one URL → AI-ready content in the requested formats (markdown/text/html/rawHtml/metadata/links/images/). + trafilatura; honest extraction confidence + render flag. Firecrawl /scrape + Jina Reader parity.
The page URL to extract. Full URL or bare domain (https:// assumed). Only http/https; private/internal/metadata targets are SSRF-blocked.
Which outputs to return (array or comma-string). Any of: markdown, text, html (cleaned main-content), rawHtml, metadata, links, images, jsonld. Default: markdown+metadata. Unknown values are ignored. For screenshots use the web-capture engine.
Browser-render policy. The bare-DC fetch is the fast default; on a 200-but-thin (client-rendered/SPA) page render=auto transparently re-fetches the JS-rendered DOM via the web-capture browser engine and re-extracts (extraction.rendered=true). Never fires on already-rich SSR pages (waste) nor on 401/403/captcha (access-gated). render=force always renders; render=never disables.
Strip nav/header/footer/boilerplate, keep the main article/content (default true). Firecrawl onlyMainContent parity.
CSS selectors to KEEP (only these become the content). Firecrawl parity.
CSS selectors to REMOVE before extraction. Firecrawl parity.
Return only the content inside this CSS selector (Jina Reader x-target-selector parity).
Per-request timeout in seconds (3-60, default 25). (3–60)
curl -X POST https://api.reefapi.com/web-extract/v1/scrape \
-H "x-api-key: $REEF_KEY" \
-H "content-type: application/json" \
-d '{"url":"https://en.wikipedia.org/wiki/Web_scraping","formats":"[\"markdown\",\"metadata\"]"}'Hit Send to run this endpoint live.