Content accessibility
Explore checks for server-rendered content, structured data and Markdown alternatives that help AI agents read and understand your site.
Once an agent has found a page, these checks look at how cheaply and reliably
it can read it: whether the content is in the HTML the server sends, whether a
Markdown version is on offer, whether /llms.txt points somewhere useful, and
whether structured data says what the page and its publisher are.
Content accessibility carries a weight of 3 in the overall score, three times that of Discoverability or Agent experience. It is one of the two categories where a site only scores by doing something deliberately for agents.
Checks in this category
- Content without JavaScript
Agents read the HTML your server sends and do not run JavaScript. Whatever is not in that HTML does not exist to them. - Markdown content negotiation
Serving a Markdown version when an agent sends Accept: text/markdown makes your content cheaper and cleaner for it to read. - Markdown negotiation caching (Vary)
Without Vary: Accept, a cache may return the wrong response, sending browsers Markdown instead of HTML. - llms.txt links resolve
An llms.txt should not contain links that return a 404. - JSON-LD entity linking (sameAs)
sameAs ties your entity to ones an agent already knows, such as Wikipedia, Crunchbase, and your own profiles. - Organization schema completeness
The Organization node is what an agent reads to tell who publishes a site. - Schema type breadth
Generic types tell an agent less than specific ones such as articles, products, and FAQs. - Structured data (JSON-LD)
Schema.org JSON-LD gives agents typed data about your page that is easy to process. - Emerging checkMarkdown URL fallback (.md)
A .md twin of each page lets an agent fetch Markdown by convention when the server does not negotiate. - Emerging checkMarkdown alternate link
A rel="alternate" link pointing to text/markdown tells an agent a Markdown copy exists, so it does not have to guess. - Emerging checkllms.txt present
An /llms.txt file gives agents a hand-picked Markdown map of your most important content. - Emerging checkModular llms.txt
Per-section llms.txt files let an agent pull more granular context. - Emerging checkSpeakable content markup
speakable marks the sections worth reading aloud, which is what a voice agent reaches for first.