Agent readiness reads what your site tells AI crawlers and assistants. It is a checklist of facts, not a score. It does not predict citations or rankings, and its results are never added to any AI traffic figure. AI Visibility shows how assistants reach your pages; this page only shows what your site says to them.
What is checked
Each run makes read-only requests to your public site. A request that fails or times out is shown as "not measured", never as a failure.
- robots.txt per AI crawler. OAI-SearchBot, GPTBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended, CCBot and Bytespider. Each is read with the RFC 9309 rules: the group that names the crawler wins over
*, the longest matching rule wins, and Allow wins a tie. The answer is allowed, blocked, partly blocked or not mentioned, with the exact rule line. Each crawler is tagged training, search or user fetch. Google-Extended does not affect Google Search. - llms.txt. Present or not found, for information only. It has no weight in anything on this page. Google says llms.txt neither helps nor hurts Search.
- Main text in the raw HTML. The number of words in the home page HTML before any script runs. Below 100 words the text may appear only after JavaScript. This is a SerpUp heuristic: the page is not rendered.
- Markdown delivery. A Markdown version at the home page's
.mdaddress (/index.md) or atext/markdownalternate link. Serving Markdown through anAccept: text/markdownheader is not tested. - /.well-known documents. security.txt, ai-plugin.json and mcp.json, present or not found.
Reading the result
"Not mentioned" means no robots.txt group names the crawler; the row says what applies instead. Allowing or blocking a crawler is your decision, so the page reports the rule and nothing more. The MCP tool get_agent_readiness returns the same rows from the last finished run.