Agentic Browser Detection
◐
Attributed by network
From an address range Anthropic publishes for ClaudeBot / Claude-User / Claude-SearchBot — attribution, not proof
Operator console
Sign in to inspect traffic and manage the challenge observations.
Traffic Overview Agents Challenge settings
The old signal — User-Agent
claims Anthropic ClaudeBot
User-Agent contains a token for Anthropic ClaudeBot. Anyone can send any string here, so this is a hint, never a fact.
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)
The middle signal — published IP ranges
Anthropic · ClaudeBot / Claude-User / Claude-SearchBot
216.73.216.0/22 from claude.com, list dated 2026-10-20. A range names the operator, never the customer behind it, and only cloud-side operators publish one. The address is announced by AS16509 (Anthropic, PBC), as reported by the edge. A network names the landlord, not the tenant — OpenAI runs in Microsoft’s, Anthropic’s crawler in Amazon’s — so this is context, never proof. Cloudflare has verified this bot itself and files it under AI Crawler.
216.73.217.134
The new signal — Web Bot Auth
no signature
no HTTP Message Signature on this request
Everything this request carried
Every header as received, credentials removed. Kept for 28 days, then deleted. The client address is what the edge reported in CF-Connecting-IP; the User-Agent is verbatim.
Show all 11 request headers 
accept: */*
accept-encoding: gzip, br
cf-connecting-ip: 216.73.217.134
cf-ipcountry: US
cf-ray: a4e273b61c77dbb4
cf-visitor: {"scheme":"https"}
connection: Keep-Alive
host: botauth.info
user-agent: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)
x-forwarded-proto: https
x-real-ip: 216.73.217.134
Response
Verdict attributed → action allow. Purpose Training (documented by the operator for this product) — established, so a purpose rule in the policy would apply. The mapping is set by the site operator and takes effect on the next request.
Point something at this page
Every request to this URL is classified live and answered with its own verdict, so any agent, crawler or script can be pointed here and read what the site made of it.
node agent.js demo https://botauth.info/     # run every scenario
node agent.js verified https://botauth.info/
node agent.js spoof    https://botauth.info/
Standards implemented here
Web Bot Auth is IETF work in progress; an IETF working group was chartered in early 2026 and adopted its first document in September. This site verifies against the versions below.