---
title: "What Is llms.txt? A Practical Spec for Site Owners"
description: "llms.txt is a Markdown contract for AI systems, not a robots.txt clone. This operator spec covers fields, placement, crawl conflicts, and file ownership."
url: "https://articles.flowcrews.com/what-is-llms-txt-practical-spec-for-site-owners"
category: "AEO"
date_published: 2026-09-16
reading_time_minutes: 10
word_count: 2374
---

# What Is llms.txt? A Practical Spec for Site Owners

*Ship a crawl contract for answer engines—not a robots.txt clone or a citation generator.*

![What Is llms.txt? A Practical Spec for Site Owners](https://pub-07fb5e4955ba485b822d6b388be96d9a.r2.dev/2e474d51-a072-472e-ab2f-1a43a8990566/what-is-llms-txt-practical-spec-for-site-owners/hero-e1b6fda4-9063-4468-bd8a-23a24bb582c9.jpg)

**TL;DR:**

- llms.txt is optional Markdown at a well-known path that briefs AI systems on your site, not a replacement for robots.txt.
- It describes purpose, key pages, and usage notes; it does not grant crawl rights or auto-create citations.
- Place it where fetchers expect it, keep it truthful, and resolve conflicts with robots and sitemaps in robots’ favor.
- Treat it as content ops plus crawl policy, not as a magic AI-visibility plugin.
- Generators can draft structure; a human still owns accuracy and legal claims.

## llms.txt Is a Curated Map, Not a Crawl Policy

## What a Valid llms.txt File Actually Looks Like

## Where llms.txt Lives—and Why robots.txt Can Quietly Kill It

## What belongs in llms.txt (and what generators dump in by mistake)

## Who owns llms.txt after you ship it

## How to tell whether anything actually read your llms.txt

## Conclusion

- Curated map, not crawl policy — llms.txt is a voluntary Markdown file that nominates durable URLs and context for answer engines; it does not replace robots.txt, sitemaps, meta robots, or legal terms, and engines may ignore it.
- A small Markdown contract — A valid file is UTF-8 plain Markdown at a well-known root URL: H1 name, short summary, H2 link sections with one-line descriptions, optional secondary resources, and no HTML, scripts, or tracking params.
- Root URL and robots can hide it — Serve a stable 200 with a text/plain or Markdown type, no auth; redirects, host splits, CDN caches, and Disallowing AI user-agents or listing noindex/login URLs can quietly kill the map.
- Nominate high-intent canonicals only — Include guides, docs hubs, and original research with factual descriptions; omit login walls, thin tags, archives, faceted search, pagination, and generator crawl dumps.
- Named owner and a review gate — A human owns the file, refreshes it on real IA or publish changes, keeps staging/production/locales separate, and adds URLs only by nomination—not automatic scrape.
- Verify with logs, not screenshots — Confirm 200, MIME, HTTPS, and crawlable canonicals; logs show fetches, while blocked bots, wrong hosts, stale caches, noindexed URLs, and overlong files are common failures.

Put a short, owned llms.txt at your site root, pair it with crawlable canonical pages, and check server logs to see whether answer engines actually fetch it.

## Frequently Asked Questions

### Where should llms.txt live?

The common convention is the site origin root, such as https://example.com/llms.txt, sometimes with a longer companion like llms-full.txt. Confirm your host can serve plain Markdown with a stable URL; do not hide it behind login or a random CMS path if you want fetchers to find it.

### Does llms.txt override robots.txt?

No. If robots.txt disallows a path, treat that as binding for crawlers that honor robots. llms.txt can *point* to allowed, high-value pages; it should not list or encourage fetching disallowed URLs.

### Will adding llms.txt improve AI visibility or rankings?

Not by itself. Some answer engines may use a clear brief when they already crawl you; many will ignore it. Visibility still depends on crawl access, page quality, and how each product retrieves sources—not on a filename.

### Should I use an llms.txt generator?

A generator is fine for scaffolding headings and URL lists. You still need a person to verify claims, drop stale URLs, and avoid stuffing keywords. Garbage Markdown is worse than no file.

### Who should own the file inside a company?

Treat it like a public policy doc: content or docs owns the narrative, SEO or platform owns crawl consistency with robots and sitemaps, and legal reviews anything that looks like license or warranty language.

### Is llms.txt a standard I must implement?

It is a community proposal, not an IETF or W3C requirement. Ship it if you want a single, maintainable brief for machines; skip it if you cannot keep it accurate.