---
title: "LLMs.txt (llms.txt)"
description: "The site map for language models."
canonical: "https://geordy.ai/formats/llms-txt"
---
# LLMs.txt

**The site map for language models.**

A plain-text directive at your domain root that tells AI engines what your site is about, which pages matter, and where to look first. The signal-to-noise ratio they wish your homepage had.

LLMs.txt is a plain-text format specifically designed for AI language models to quickly understand a website's purpose, structure, and content without parsing HTML or complex markup. It provides a standardized, machine-readable format that tells AI systems what a site is about, how it's organized, and how to use it effectively.

## At a glance

- File: `llms.txt`
- First released: 2023
- Created by: AI Optimization Community
- Specification: https://llmstxt.org
- Read by: ChatGPT (GPTBot), Claude (ClaudeBot), Perplexity (PerplexityBot), Google Gemini, Bing AI

## Why it matters for AI

llms.txt is a single Markdown file at /llms.txt that gives LLMs a curated, token-efficient map of a site's most important documentation, bypassing the noise of full HTML pages. AI engines and the agents feeding them ingest it deterministically, raising the chance your canonical content is the version they cite.

## Example

```
# Geordy
> https://geordy.ai

## Purpose
Optimize websites for AI-powered search engines.

## Paths
/features: Platform features
/pricing: Pricing plans
/formats: All 16 formats
```

## Benefits

- Zero ambiguity - tells AI exactly what your site does
- No HTML noise or visual formatting to confuse models
- Simple format any AI system can parse instantly
- Improves likelihood of being cited in ChatGPT, Claude, and Perplexity answers

## Best practices

- Place the file at the root: https://yourdomain.com/llms.txt (and optionally an expanded /llms-full.txt with inlined content).
- Open with an H1 (site name), then a blockquote one-line summary, then short prose context, then H2 sections of links - strict spec order.
- Each link should be `[Title](url): short description` and point to clean Markdown versions of pages where possible (e.g. /docs/foo.md).
- Curate, don't dump. Use an Optional H2 section for material that can be skipped under tight context.

## Pitfalls

- Treating it like a sitemap.xml dump - listing every URL defeats the curation purpose and wastes the model's context.
- Linking to JS-heavy HTML pages instead of plain Markdown/text equivalents the LLM can actually parse cheaply.
- Skipping the required structure (H1 + blockquote + H2 link sections); ad-hoc Markdown breaks tools that parse the spec strictly.

## Use cases

- **Site purpose declaration** - Communicate site purpose in clear, unambiguous language to AI systems.
- **Content structure mapping** - Help AI understand organization and navigation for multi-section sites and knowledge bases.
- **Topic / keyword signals** - Provide explicit context for niche sites and specialized content categorization.

## Adopters

- Anthropic (https://docs.anthropic.com/llms.txt)
- Cloudflare (https://developers.cloudflare.com/llms.txt)
- Perplexity (https://docs.perplexity.ai/llms.txt)
- Hugging Face (https://huggingface.co/docs/llms.txt)
- FastHTML (Answer.AI) (https://fastht.ml/docs/llms.txt)
---

Source: https://geordy.ai/formats/llms-txt
This is a machine-readable markdown version of that page, generated by Geordy.