Captchas, browser checks and JavaScript-built sites no longer return an empty analysis. And the full redirect chain is visible.
Page loading gained a second method: rendering in a real browser. It helps where a plain request returned nothing — Cloudflare and ddos-guard checks, "checking your browser" pages, reCAPTCHA, and sites whose content is assembled by scripts and isn't in the source. The assistant decides when to switch: if it sees a captcha or an empty result, it retries through the browser.
Ordinary loading now reports the response code and the whole redirect chain — from the requested address to the final one, hop by hop. That removes a whole class of wrong advice, where a recommendation was given for an address the server quietly moves away from.
Separately: robots.txt, sitemaps, .json, .csv, .pdf and compressed files no longer return a service error — they are read as-is.
Anyone analyzing sites they don't control: script-driven shops and protected platforms come up constantly.
Paste the address of any such page and ask for an analysis.