Does the website knowledge base scrape the whole site or just the url?