llms.txt is a plain-text (usually Markdown) file at the root of your domain. Language-model crawlers look for it the same way search crawlers look for robots.txt: a known path, a short briefing, no JavaScript required.
What it is for
AI answer engines do not only scrape your homepage H1. They try to work out which entity you are. If your brand name is already used by a Shopify app, an agency, or a dead company, a model will happily merge you. llms.txt is how you say, in a file machines expect: we are this product, at this URL, not that other one.
What it is not
- Not a ranking factor for classic Google results.
- Not a substitute for robots.txt (crawling) or noindex (indexing).
- Not a way to hide content you already published — if it is on the web, it can still be trained on unless you use the crawler-specific robots rules.
A minimum viable file
# Your Product
> One-sentence description.
We are X, not Y.
## Key pages
- https://example.com/
- https://example.com/pricing
Host it at /llms.txt with Content-Type: text/plain. Keep it short. Update it when the product changes.
robots.txt still wins
If Cloudflare (or your WAF) disallows GPTBot and Google-Extended, those systems will never see your beautiful llms.txt. Check robots.txt for User-agent: GPTBot / Google-Extended blocks before you obsess over the briefing file.
Check yours
Our llms.txt checker fetches the root file as part of the live audit, alongside the AEO / GEO checker for crawler access and FAQ markup.