OpenAI agent readiness report

Score history

OpenAI's products are a large part of why agent readiness is now a question every company has to answer. Its own homepage is one of the first places a curious agent, or a person asking an agent about OpenAI, would try to read. How that page performs is a useful signal for the rest of the industry watching what agent-friendly infrastructure is supposed to look like.

The scan found a homepage that passes the basics: HTTPS with a forced redirect, a reachable sitemap, full meta tags, and an open robots.txt. Past that, most of the agent-specific checks fail. There is no agents.md, no llms.txt, no llms-full.txt, and the homepage ignores requests for a markdown version of the page. No JSON-LD structured data was found at all. The text-to-markup ratio came in far below the readability threshold, and the scan flagged that key content only appears after JavaScript runs, meaning agents that do not execute scripts see a largely blank page.

For a company whose product is built on language models reading and reasoning over content, the irony is direct: an agent trying to read openai.com today gets less structured signal than it would from many far smaller sites.

Same category