Question 1

Does it work on pages that need JavaScript?

Accepted Answer

Yes. Every page is loaded in a real Chromium instance from a warm pool, so a site that assembles its article with script is read the same as one served as HTML. Fetching the source with an HTTP client sees an empty div on most of the modern web.

Question 2

Do tables survive?

Accepted Answer

As markdown tables, with their header row. Flattened into sentences, a specification table is unreadable and a model asked about any value in it has nothing to work with. Links keep their href and are made absolute, and code blocks keep their language.

Question 3

How are cookie banners removed?

Accepted Answer

The consent platform's script is blocked at the network layer, so the banner never renders at all. Deleting the element afterwards is too late: it has already run and already fetched its trackers. A first-party banner served from the site's own domain cannot be blocked by host, so those few are removed from the page instead.

Question 4

Is this a crawler?

Accepted Answer

No. It fetches the one URL you give it and follows no links. There is no queue, no sitemap discovery and no recursion, which is also why it is priced per page rather than per crawl.

Question 5

What does it cost?

Accepted Answer

One credit per page, flat, whatever the article's length — one page load is one page load. Credits are bought in packs and never expire, and there is no subscription.

URL to markdown

Related