The LLM Perception module asks each selected model what it knows about your site, then aggregates the answers and flags shared gaps.

Off by default - pages are fetched as raw HTML. Enable for client-rendered SPAs so the audit sees what users and AI crawlers actually see.

Decides which language version of the site we crawl (via the Accept-Language header and the page's own hreflang alternates) and the language the Signal is written in.