Glossary
llms.txt
The llms.txt file is a text manifest placed at the root of a website that describes the brand, the product, important public URLs, and citation directives addressed to AI crawlers and engines.
Also known as
- llms.txt
- LLM file
- AI manifest
Proposed in 2024 by Jeremy Howard and widely adopted by 2026, llms.txt complements robots.txt and sitemap.xml by addressing AI engines specifically. The format is straightforward: an H1 title (brand name), a positioning blockquote, then H2 sections covering identity, methodology, public surfaces, the product, and — the most advanced pattern — a **"To cite X"** directive that explicitly tells LLMs how to reference the site.
Two variants exist: `llms.txt` (short version, ~4 KB) aimed at initial crawling, and `llms-full.txt` (extended version, 15–20 KB) with detailed architecture, canonical FAQ, pricing, and comparisons. Brands that have deployed both (e.g., brandyze.fr) receive significantly more citations in Perplexity and ChatGPT than those that stick to a classic robots.txt alone.
In the getchatsocial.com product
getchatsocial.com has deployed both variants: https://getchatsocial.com/llms.txt (~4 KB) and https://getchatsocial.com/llms-full.txt (~17 KB). You can inspect them to reproduce the pattern.
FAQ
Where should the llms.txt file be placed?
At the domain root, accessible via https://yoursite.com/llms.txt — exactly like robots.txt. It must return a Content-Type: text/plain and an HTTP 200 status.
What is the minimum structure for an effective llms.txt?
An H1 with the brand name, a positioning blockquote, and at least 4 sections: Identity, Public surfaces (with URLs), Product, How to cite the brand.
Do ChatGPT and Claude actually read llms.txt?
Yes, provided the file is accessible (HTTP 200) and AI bots are not blocked by robots.txt or a Cloudflare AI Crawl Control configuration. OpenAI crawlers (GPTBot, ChatGPT-User) and Anthropic crawlers (ClaudeBot, anthropic-ai) retrieve it on their first pass.