You focus on building products and getting customers. We abstract everything related to scraping: captchas, proxies, sessions, browsers, and the rest.
Result preview
https://example.com/private-report
{
"title": "Private market report",
"company": "Northwind Labs",
"signals": [
"pricing changed",
"new enterprise plan",
"fresh hiring activity"
],
"confidence": 0.94
}Job details
Scrapunk is for teams that need web data without babysitting browsers, proxy failures, captcha retries, and brittle extraction scripts.
Feed your model clean context from JavaScript-heavy, login-protected, and interaction-gated pages. Get Markdown for prompts and typed JSON when you need reliable tool output.
Start with a single direct request, graduate to full browser automation, then schedule flows later. The same account, credits, API keys and webhook model carry across the workflow.
A single scraping API for the work that usually requires five moving parts: browser rendering, session replay, captcha handling, extraction logic, and result delivery.
Every successful scrape can return Markdown for prompts, raw HTML for parsers, page metadata, screenshots, and structured JSON from your schema.
Save a browser profile once and reuse cookies, fingerprint, locale, timezone, and session state across pages that require authentication.
Click, type, scroll, wait, run JavaScript, branch on conditions, switch iframes, and solve captchas before extraction starts.
Handle reCAPTCHA, Turnstile, hCaptcha, FunCaptcha, GeeTest, image, and text challenges inline and itemize the cost per solved challenge.
Queue long-running browser scrapes, poll job state, or receive a Stripe-style HMAC-signed webhook when the result is ready.
Buy credits, spend them as requests complete, keep your balance forever, and automatically recover the reserve when a scrape fails.
You don't need to code and orchestrate. Model the whole workflow: trigger with an API call or schedule, pass variables and receive a signed webhook call with the resulting data.
Example flow
Monitor a pricing page and notify only when it changes
Use the SDK, call the REST API directly, or let an MCP client invoke the same tools. Your integration stays small because Scrapunk owns the browser work.
curl -X POST https://api.scrapunk.com/scrape \-H "X-API-Key: $API_KEY" \-H "Content-Type: application/json" \-d '{"url": "https://example.com/article","mode": "browser","autoparse": true,"structure": {"title": {"type": "string","description": "Article title","required": true},"summary": {"type": "string","description": "Short factual summary"}}}'
Prefer to build it visually? Try the builder — see the response live, then export the JSON.
Pricing should not force you into a plan before you know your data pipeline works. Buy credits, run scrapes, keep unused balance, and only spend credits when the request succeeds.
Single HTTP request through the proxy for APIs, feeds, static HTML, and pages that do not need a browser.
0.25credit
Full browser execution with JavaScript, profiles, optional screenshots, and actions for modern app-like pages.
1.00credit
Optional add-on for captcha flows. Charged only when a challenge is actually solved.
0.50 - 4.00credits
Scrapunk is a pay-per-success web scraping API for modern websites. It runs direct HTTP requests or full browser scrapes, handles JavaScript rendering, logged-in sessions, browser actions, and captchas, then returns Markdown, HTML, metadata, screenshots, and optional structured JSON.
Direct requests cost €0.25 per 1,000 calls. Full browser scrapes cost €1 per 1,000 successful scrapes. You buy credits, spend them as jobs complete, and keep unused balance without a subscription or minimum commit.
You don't pay for the failed scrape. Scrapunk reserves the estimated credits when the job is queued and releases the reserve automatically if the job fails.
Yes. Save a browser profile once from the dashboard, including cookies, fingerprint, locale, timezone, screen size, and user agent. Scrapunk can reuse that profile when scraping authenticated pages.
Yes. Use browser actions to click, type, scroll, wait for elements, run JavaScript, branch on conditions, switch iframes, and solve captchas. Build the flow visually or send the same actions as JSON.
reCAPTCHA, Cloudflare Turnstile, hCaptcha, FunCaptcha, GeeTest, and image/text challenges resolve inline as part of the scrape. You decide which captchas to solve and where; the fee is itemized per solve and only charged when one actually triggers.
Yes. Pass webhookUrl on any scrape and Scrapunk POSTs the terminal job to your endpoint with an HMAC-SHA256 signature over a timestamped payload. The SDK includes verifyWebhookSignature for verification.
Use the hosted MCP server for agent clients, the TypeScript SDK for application code, or the JSON API from any workflow engine. Built-in AI extraction can return typed JSON against a schema you define.
Clean Markdown for LLM prompts, optional raw HTML for custom parsing, structured metadata, an optional AI-extracted JSON object against your schema, and an optional full-page screenshot URL.
Start with the visual builder, copy the JSON payload, then move the same scrape into your app, agent, or automation workflow.