Machine-readable docs
Every page on this site is available as raw markdown — via a per-page content.md URL, llms.txt, and llms-full.txt.
This site is built to be read by agents as much as by people. Every page has a raw-markdown form, and the whole site is indexed for machine readers. No scraping, no HTML parsing.
Site-wide indexes
| Route | What it returns |
|---|---|
/llms.txt | An index of every docs page — titles and absolute URLs. Start here to discover what exists. |
/llms-full.txt | The full content of every page, concatenated into one document. Use it to load the entire docs corpus in one fetch. |
Per-page markdown
Every docs page is also published as a static markdown file: take the page
path and fetch it under /llms.mdx/ with /content.md appended —
/docs/agents/mcp becomes:
curl https://<docs-host>/llms.mdx/docs/agents/mcp/content.mdThe response is the processed page content, prefixed with the page title and
its canonical URL — so a fetched page always identifies itself. (The site is
statically hosted, so there is no Accept-header content negotiation; the
markdown files are plain static assets.)
Copy Markdown button
Every docs page has a Copy Markdown button at the top, next to a view options popover that links the same raw-markdown URL and the page's source on GitHub. Useful when a human is pasting a page into an agent's context by hand — and the easiest way to discover a page's markdown URL.
Suggested reading order for agents
- Fetch
/llms.txtto see the page index. - Fetch the specific pages you need via their
content.mdURLs — or/llms-full.txtif you want everything. - Check Capability status before acting on any capability claim — it is the honest map of what is live in the published alpha.