Skip to content

Using the site with AI

This site is built to be read by both people and AI. Every page is available as clean Markdown, and the site is openly discoverable by search engines and AI assistants — so you can bring MCP Description (mcpdesc) content into whatever tool you already use.

  • Open to search and AI assistants. The site welcomes search engines and AI crawlers for indexing and for answering questions grounded in its content. (Using the content to train models is not permitted while the specification is a Draft — it is a reference, not training data.)
  • A site map for machines. A machine-readable index of key pages lives at /llms.txt, so an assistant can quickly orient itself to the most important content.
  • A Markdown version of every page. Each page has a Markdown twin: the same content, without the site chrome, ready for an assistant to read with fewer wasted tokens.

Use the Copy button in the site header (next to the GitHub icon) to copy the current page as Markdown, then paste it into your assistant (ChatGPT, Claude, Copilot, or any other) and ask questions grounded in it. Prefer to see the raw text first? Use the .md link next to the Copy button.

Every page answers to a .md address. Add .md to a page’s path to fetch its Markdown directly — handy for scripts, prompts, or “read this URL” features:

https://mcpdesc.org/docs/getting-started (the page)
https://mcpdesc.org/docs/getting-started.md (its Markdown)

Give an assistant https://mcpdesc.org/llms.txt as a starting point. It lists the format overview, documentation, specification, tools catalog, and more, so the assistant can follow links (and their .md twins) to whatever it needs.

Once you have a page’s Markdown in your assistant, you can ask things like “summarize what mcpdesc describes,” “explain this section in plain terms,” or “draft an mcpdesc document for a server with these tools” — with the assistant working from the actual content rather than guessing.