Documentation for AI agents

Optimizely publishes its documentation in formats built for large language models (LLMs) and AI agents. These formats sit alongside the standard web pages at docs.optimizely.com . They give an AI tool a cleaner, more compact view of the content than HTML, which uses fewer tokens and parses more reliably.

You can work with the documentation in two ways:

You do not need an Optimizely account or any setup. Any AI tool that can read a URL can use these formats.

The llms.txt file is a plain-text index of the documentation, following the llms.txt convention . It lists every page, grouped by product area, and each entry links to the Markdown version of that page.

Find the index at the root of the documentation site:

Use the index when you want an AI tool to find relevant pages before it reads them. Point the tool at the URL, or paste the URL into a prompt. The tool reads the index, identifies the pages that match your question, and then fetches only those pages.

The index covers every Optimizely product, so the file is large. Use it to locate the pages you need, then load only those individual pages into the tool rather than the whole index.

Every documentation page has a Markdown version. The Markdown holds the page content without the site navigation, scripts, or styling. This uses fewer tokens than HTML and reduces the chance that a tool misreads the page.

To get the Markdown version of a page, add .md to the end of its URL. For example, the Explorations page in Analytics has these two addresses:

This works for any page on the site. Give an AI tool the .md address when you want it to read a single page accurately with the least overhead.

Use the format that fits the size of the task:

Give an AI tool that can read a URL a prompt like one of these.

To find pages on a topic:

To answer a question from a single page: