LLM Support
Provide AI-friendly endpoints to assist large language models
Introduction
Large language models waste tokens crawling HTML, JS, and site chrome. By shipping a few simple endpoints, you can provide a more efficient way for LLMs to access your documentation.
/llms.txt
: a concise sitemap with titles, URLs, and summaries/llms-full.txt
: a full Markdown dump of every document/llms.mdx/*
: the raw MDX/Markdown content of a requested page
/llms.txt
- Lists each section and page in your documentation
- Includes optional one-line descriptions from frontmatter
- No boilerplate or styling—just the outline
This file is generated at the root of your deployed site, enabling agents to discover it at https://<your-domain>/llms.txt
.
/llms-full.txt
- Concatenates the raw MDX/Markdown of every page
- Preserves headings, paragraphs, code samples, and frontmatter
- Lets an LLM ingest your entire documentation corpus in a single fetch
This file is generated at build time by globbing content/docs/**/*.mdx
and joining files in sidebar order.
/llms.mdx/*
- Provides the raw MDX/Markdown content of a requested page
- Preserves headings, paragraphs, code samples, and frontmatter
Example: /llms.mdx/ui