Question about form of llms.txt file

I checked some examples from here https://llmstxt.site/ and people sites do not have the link pointing to MD format actually. But what is the best approach? Will LLM find the MD format itself with link rel="alternate"?

Drupal's ability to serve content in Markdown format appears to be unique among CMS platforms as highlighted in this Pronovix article about making developer portals AI-ready.

However, I haven't seen evidence that current LLM crawlers automatically fetch .md versions of content or interpret link rel attribute values. Most crawlers already have established methodologies for mass site scanning, and the llms.txt specification is still relatively new in terms of widespread adoption.

My recommendation:Explicitly include the Markdown version of pages in your llms.txt file whenever possible, rather than relying on crawlers to discover them automatically.

You're correct that the module's menu token doesn't currently expose URLs with enforced MD format. This connects to another requested feature for embedding entity links via tokens that point to MD versions ✨ Support dynamic entity linking using tokens Active , which would address this limitation.

Part of the overall llms.txt proposal is having all relevant site content available as MarkDown.

We furthermore propose that pages on websites that have information that might be useful for LLMs to read provide a clean markdown

version of those pages at the same URL as the original page, but with .md appended.

Even though many example files do not contain MD links, my strong conviction is that all links in llms.txt should directly point to the MD version of the page, as the proposal implies. While current crawlers have been optimized to deal with HTML, future AI agents should not have to and ought to be able to fetch and parse MD only.

Thank you all, now it is more clear.

Automatically closed - issue fixed for 2 weeks with no activity.