Website content is increasingly aimed at two target audiences: people who visit a website and AI assistants that use content from the web to answer questions.
The llms.txt convention offers a way to help AI systems find important content on a website and better understand its meaning. The approach follows /robots.txt and /sitemap.xml, using a standard filename at the site root.
Mercury now offers built-in support for creating and maintaining llms.txt files directly from a website’s content.
Editors decide for themselves which content to highlight and how to describe it. Mercury handles the technical implementation and generates the website’s llms.txt file from this information. As a result, editors do not need to know the technical format or maintain a separate text file.
Descriptions can be entered manually or generated automatically from published page content using a configured LLM service.