AgentReady scanner

What we scan
AgentReady checks whether a website is readable and usable by AI agents. On request of a user, we fetch a fixed
list of public URLs on the submitted site: the home page, /robots.txt, /sitemap.xml, /llms.txt, /llms-full.txt,
a markdown variant of the home page, about/contact/privacy/terms pages linked from the home page, two other pages
linked from the home page, /openapi.json, /.well-known/ai-plugin.json and /.well-known/mcp.json. We do not crawl,
submit forms or log in. One scan is about 25 requests, with a 10 second timeout per request and a 45 second budget.

How we identify ourselves
Every request carries the header X-AgentReady-Scan: https://agentcan.dev/bot
Most requests use a desktop browser User-Agent:
  Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36
Exactly one request, made to compare responses, uses the GPTBot User-Agent string. It is sent by us, not by OpenAI:
  Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.1; +https://openai.com/gptbot
Five more requests to the home page in a row measure rate limiting.

What we store
Nothing. Scan results are produced per request and are not saved by this service.

Opt-out and contact
To stop scans of your domain, write to abuse@agentcan.dev with the domain name. Abuse reports go to the same address.
Do not scan sites you have no right to test.