AI crawlers stop being invisible: Hatamoto logs them server-side and spots 16 agents in 30 days
Synopsis
Hatamoto released a content management system with built-in AI crawler analytics, running as one Go binary beside Postgres with no PHP, Node or plugins. Because AI crawlers do not run JavaScript, script-based analytics never see them; Hatamoto logs each fetch on the server and checks it against operators' published IP ranges and reverse DNS. On its own site over the last 30 days, 16 of 19 known agents were recorded fetching pages, and 3,301 spoofed fetches posing as ClaudeBot, GPTBot and others were flagged.
Interpretation
A content management system now treats AI crawler visits as something to measure, not only human visitors. Conventional analytics scripts depend on JavaScript, and AI crawlers do not execute JavaScript, so those scripts never record them. Two analytics screens from a live Hatamoto site over the last 30 days show 46,191 AI fetches, with 16 of 19 known agents recorded.
Counting happens on the server, and each fetch has its origin checked and is grouped by why the bot came. The check uses each operator's published IP ranges and reverse DNS, covering nine operators. Bots using a fake user agent are flagged and left out of the totals; the same 30-day window flags 3,301 spoofed fetches pretending to be ClaudeBot, GPTBot and others.
Fetch records are tied to individual pages, and the editor shows which agents fetched the post being edited and when they first did. The same data reports 11,456 error responses served to AI crawlers with the failing paths listed, and 863 published pages no AI has fetched yet. These figures come from the same live site's 30-day analytics screens.
Perspective
The demonstrated scope is server-side crawler analytics from one live Hatamoto site over the last 30 days, aimed at content-site operators and developers who need to know how AI crawlers reach their pages.
Spoof detection depends on operators' published IP ranges and reverse DNS, and the material says this covers nine operators; all figures come from a single site over 30 days, with no results from other sites or longer periods.