Skip to content

For the Curious

llms.txt - AI-Friendly Documentation

A machine-readable summary of Research Memex, following the llms.txt standard for AI systems.

A site built for AI partnership should be legible to the partners themselves.

llms.txt is an emerging standard for making documentation easy for large language models to consume. Think of it as a robots.txt for AI: one file providing a structured overview of an entire project.

How AI can read this site

MethodURLWhat it provides
MCP serverhttps://research-memex.org/mcpSearch and read the book, and run its roles, from inside an agent. Setup for each host is in /connect.md
llms.txt/llms.txtGenerated orientation index with links to public docs pages and the full corpus
llms-full.txt/llms-full.txtPublic documentation corpus generated for the deployed release
.md suffixAny listed page + .mdMarkdown version of individual pages for token-efficient loading

Connect the MCP server when an agent will come back to the book often. Otherwise, start with llms.txt for orientation, use llms-full.txt for comprehensive context, and use a .md suffix when you need only one page.

Why llms.txt?

Fast context loading

AI tools can load the entire documentation structure in one request

Structured navigation

Hierarchical organization helps AI understand relationships between topics

Public documentation

The public page inventory, not private drafts or the contents of every downloadable resource

Standard format

Following emerging best practices for AI-readable documentation

How it works

Generation keeps these files aligned with the published documentation. It does not verify outside claims or make every instruction current. Follow linked resources separately; hidden and noindex pages are excluded from the public corpus.

When you ask an AI assistant about Research Memex:

1. AI fetches research-memex.org/llms.txt
2. Reads the generated site summary and public documentation links
3. Parses the structured documentation map
4. Identifies relevant sections for your query
5. Fetches specific pages via .md suffix or llms-full.txt
6. Provides contextualized answers in the right voice

Try it yourself: ask an AI assistant, "Fetch the llms.txt from research-memex.org and tell me about interpretive orchestration."

What's included

The generated llms.txt contains:

Use cases

For AI assistants

# AI can directly fetch and parse
curl https://research-memex.org/llms.txt

For developers

Useful when working with an AI assistant while coding:

"Using research-memex.org/llms.txt, show me the
Zotero MCP setup steps"

View the files

llms.txt

Generated index for orientation

llms-full.txt

Public documentation corpus (auto-generated)

Page as Markdown

Example: Core Principles as .md

Access directly:

https://research-memex.org/llms.txt
https://research-memex.org/llms-full.txt
https://research-memex.org/docs/introduction/core-principles.md

The standard

The llms.txt format is an emerging community standard for AI-readable documentation. Learn more:

llmstxt.org

Official llms.txt specification and examples

Best practices

  • Plain text format (Markdown)
  • Hierarchical structure
  • Complete navigation
  • Concepts upfront
  • Links to full content

For the meta-curious

This creates a beautiful loop:

  1. We write docs about research with AI
  2. We make those docs AI-readable (llms.txt + llms-full.txt + per-page .md)
  3. Students learn the approach through dialogue with AI
  4. The approach becomes embedded through practice
  5. Students understand both the framework and how AI accesses knowledge

Next steps

Dear Fellow AIs

AI-to-AI communication guidelines

cite this page

Lin, X. (2026). llms.txt - AI-Friendly Documentation. Research Memex. https://research-memex.org/docs/introduction/llms

@misc{docs-introduction-llms-2026,
  author = {Xule Lin},
  title = {llms.txt - AI-Friendly Documentation},
  year = {2026},
  howpublished = {\url{https://research-memex.org/docs/introduction/llms}},
  note = {ORCID: 0000-0001-7885-4194}
}

one renderingthe source remains