Search engines and AI crawlers say what they are. Loghound calls them bots, because they are, but always keeps them apart from evasive traffic. “A search engine read four hundred pages” and “someone is scraping you from two hundred home addresses” are opposite facts.
- The rule Declared crawler (100 points) fires, so the verdict is bot with threat none.
- The class says which kind: Declared crawler, AI crawler or Monitor (an uptime checker).
- A crawler name with a failed reverse DNS check is classed Spoofed User-Agent instead: an impersonator, not a crawler.
- A client that announces itself and then asks for something like
/etc/passwdis classed Scripted client. Being honest about the name does not make a probe honest. - Verified crawlers are exempt from the proxy fleet and User-Agent rotation rules, because a real search crawler fetches from many addresses with one fingerprint.
About thirty named agents carry the AI crawler flag, so “how much of my content is taken by AI companies” is one filter away. Some of them are user-triggered fetchers, not training crawlers: they fetch a page because a person asked an assistant about it. They are in one group because operators want them together, but they are not the same event.
A search prefetch proxy, a load balancer health check, an Android app’s networking library and similar tools also name themselves. They are filed with the declared crawlers, not as undeclared clients.
Bot forensics shows declared and evasive traffic side by side, and SEO Tools compares what crawlers fetch with what people visit. The split is made by the scorer, not by a checkbox, so every view and every query sees the same answer.