Skip to content
E-GEO

What is llms.txt?

llms.txt is a proposed convention: a Markdown file served at /llms.txt that gives AI systems a clean, prioritized map of your site’s most useful content — instead of forcing them to parse navigation, ads, and boilerplate from raw HTML.

The format

Section titled “The format”

A llms.txt file is Markdown with three parts:

  1. H1 title — the project or site name.
  2. Blockquote summary — one short paragraph stating what the site/product is. This is the entity definition an LLM reads first.
  3. Link sections — ## Docs, ## Concepts, ## Optional — with - [Title](url): one-line description entries.

A companion file, /llms-full.txt, contains the full text of the key pages so an agent can ingest everything in one fetch.

Honest status: proposal, not standard

Section titled “Honest status: proposal, not standard”
  • llms.txt was proposed at llmstxt.org in 2024. No major AI engine has committed to honoring it.
  • It does not replace robots.txt (access control) or sitemaps (crawl discovery). It is a content curation layer.
  • Agentic tools and IDE assistants already consume it; adoption by developer-documentation sites is broad and growing.
  • Cost of shipping one: minutes. Risk: none. Expected value: positive but unproven at engine level — treat it as cheap insurance, not a ranking factor.

A real example

Section titled “A real example”

This site’s own llms.txt declares the entity in the blockquote and links every docs, concept, and comparison page with one-line descriptions. The full-text companion lives at /llms-full.txt.

How to write a good one

Section titled “How to write a good one”
  • One canonical entity sentence, byte-identical to your README and listings.
  • Curate, don’t dump — link your 10–20 most useful pages, not your whole sitemap.
  • One-line descriptions that carry the claim an engine would need to cite you accurately.
  • Keep it fresh — regenerate or review it whenever you ship pages.

How E-GEO uses it

Section titled “How E-GEO uses it”

E-GEO’s schema/affordance step generates llms.txt and JSON-LD alongside rewritten content, and this site ships both files at the root. The toolkit’s position is the honest one: llms.txt is a low-cost affordance for agents, complementary to sitemaps and structured data — not a substitute for being cited by real sources.

In short: llms.txt is a Markdown index at your site root that tells AI systems what content matters and what your entity is. It’s an unofficial proposal with zero cost and growing agent-side adoption — ship it, keep it curated and fresh, but don’t expect it to substitute for real citations. E-GEO generates it as part of its optimization pipeline: github.com/mverab/eGEOagents.

Related: How to rank in ChatGPT search — crawler access and Bing indexation: the steps llms.txt does not replace.