LLM Access
Feed the documentation to AI assistants and coding agents as machine-readable text.
These docs are also served as machine-readable text so an LLM-backed tool (Claude, ChatGPT, Cursor, your own RAG pipeline) can consume them without HTML parsing.
The site is public. No auth, no token, no headers required.
Endpoints
Three endpoints expose the same docs in three shapes:
| Endpoint | Returns | Use it for |
|---|---|---|
/llms.txt | Page index, every doc title and URL | Discovery / link maps |
/llms-full.txt | All pages concatenated as one markdown blob | One-shot context dump |
/<path>.md | Single page as markdown (/foundation/what-is-radio.md, …) | Pulling a specific page |
The /<path>.md form mirrors the live site. Any URL you can open in
the browser has a .md twin you can hand to an agent.
Examples
curl https://docs.radionemiers.com/llms.txtcurl https://docs.radionemiers.com/llms-full.txtcurl https://docs.radionemiers.com/foundation/what-is-radio.mdFeeding Claude / ChatGPT manually
Run the llms-full.txt curl, pipe the output into a file, drag it
into the chat as context. The whole doc set is one markdown blob
optimized for model intake.
Cursor, Continue, Cline, etc.
Most coding assistants accept a URL as a documentation source. Point
them at https://docs.radionemiers.com/llms-full.txt. No auth
configuration needed.
Programmatic RAG ingestion
const res = await fetch("https://docs.radionemiers.com/llms-full.txt");
const markdown = await res.text();
// chunk, embed, store...Caching
The endpoints are statically rendered (revalidate = false), so each
deploy produces a fresh snapshot. If you cache the response on your
side, invalidate on every docs release.
What's included
Only pages under content/docs/ are exposed. Drafts, internal notes,
and source code never reach these endpoints. What you see in the
sidebar is exactly what an LLM gets.