The terminal on the WebReaper homepage used to play a recording. It still opens with one, a known climb against a bot-protected site so you can see the shape of the thing. But now there is a second button under it: Try it on a real site. Click it, type any URL, and the same terminal runs an actual scrape and streams each step back as it happens.

No sign-up to watch. One email to run your own.

What you are watching

WebReaper treats bot protection as a ladder it climbs only as far as a page forces it to. The live terminal shows every rung in real time:

HTTP. A plain request goes out first. On a protected site it comes back a 403 or a challenge page, and the rung lights up red.
Headless browser. WebReaper escalates to a real browser, through a residential exit so the request looks like a person. Most Cloudflare and DataDome challenges clear here, and you get the page back as clean Markdown.
Stealth. The top rung, a hardened stealth browser, for the sites the vanilla browser cannot pass.

The climb stops the moment a rung gets through, so a page that never needed a browser is served from the HTTP rung in well under a second, and the higher rungs read not needed. That is the whole point of escalation: you pay for a browser only when the site makes you.

It tells the truth when it loses

The interesting part of a live demo is what happens when it does not work, and this is where most scraping demos quietly cheat. The hardest managed challenges, the ones that read your actual GPU pixels rather than just your user agent, are not beatable from a commodity cloud host. When WebReaper hits one of those, the terminal does not hand you a challenge page dressed up as success. It climbs to the top, reports the block, and says so.

You will also see the honest shape of real pages. Point it at a marketing homepage and you often get a thin result, because that page is a JavaScript shell with no text to extract. Point it at a product page or an article and you get the real content. The terminal shows you which you got, rather than padding an empty result.

That honesty is deliberate. A scraper that returns a Cloudflare interstitial as "the page" is worse than useless in a pipeline, because you do not find out until the bad data is three steps downstream.

The same engine you run locally

There is nothing special about the playground's scraper. It is the same webreaper you install and run on your own machine, driving the same escalating page loader, emitting the same per-rung progress. The live terminal just forwards that progress over the wire so you can watch it climb.

Locally it is one command:

# A plain scrape auto-climbs to a browser when a page looks blocked
webreaper scrape https://example.com
 
# Bot-protected? Add the stealth backend and let it escalate on a challenge
webreaper scrape https://example.com --browser --auto-stealth

Or in code, the climb is built into the one page loader, so a crawl gets it for free:

using WebReaper.Builders;
 
var engine = await ScraperEngineBuilder
    .Crawl("https://example.com")
    .AsMarkdown()
    .WriteToConsole()
    .BuildAsync();
 
await engine.RunAsync();

Try it

Open webreaper.ai, watch the recorded climb, then run a real one. Pick one of the suggested bot-protected sites or paste your own URL. It is the fastest way to see what WebReaper actually does, on a site you care about, in about the time it takes to read this sentence.