Static text files

Connect an AI editor or coding agent to current Adobe Commerce Storefront documentation so it can answer questions about storefront APIs, drop-in versions, and Edge Delivery Services instead of relying only on general model training.

On the boilerplate with a capable agent?

This page is a fallback for AI tools that cannot install an MCP server or agent skills. If your agent supports both, see Install the integrations instead — it gives you source-backed drop-in facts, documentation routing, and storefront conventions together. Adding this page’s llms.txt link on top of that setup is redundant.

Static text files

These plain-text files are static copies of the documentation, regenerated each time the site publishes so AI tools always have the latest content. llms.txt is the index. Start there. Topic files under _llms-txt/ cover one subject area each, for tools that shouldn’t load the entire site.

File	Purpose
llms.txt	Index with an overview and links to every file.
llms-full.txt	Complete documentation in one file.
llms-small.txt	Same as llms-full.txt but without the long release changelog. Use when your AI tool has a context size limit.
_llms-txt/*.txt	Topic files, such as drop-ins reference, tutorials, and blocks. llms.txt lists every topic file URL.

These files can be large. When possible, paste a specific page URL into your AI tool instead so it fetches the live page directly, including recent edits and diagrams the static files lack. Use static files as a fallback when your tool can’t fetch a URL.

Publish timing

These files update each time the site publishes, so edits made since the last publish won’t appear until the next one. Because the URLs in your AGENTS.md or CLAUDE.md always point to the same files, your AI tool fetches the latest published version automatically each time it reads them.

Prerequisites

• An AI editor or coding agent that supports project context files, documentation URLs, or pasted context, such as Cursor, Claude Code, or GitHub Copilot Chat.
• A storefront project open in that editor, started from the Adobe Commerce boilerplate or with Commerce blocks and drop-ins added.
• For agent-based tools, an AGENTS.md, CLAUDE.md, or equivalent context file at the project root.

Add context files

For agent-based tools, add the llms.txt URL to a project context file so the agent can fetch the index when storefront documentation is relevant. For other tools, paste a bundle URL or file content directly.

Cursor

Add this section to AGENTS.md at your project root, or to a file under .cursor/rules/:

## Adobe Commerce Storefront Documentation

See https://experienceleague.adobe.com/developer/commerce/storefront/llms.txt for Adobe Commerce Storefront documentation.

Claude Code

Add this line to CLAUDE.md at your project root:

See https://experienceleague.adobe.com/developer/commerce/storefront/llms.txt for Adobe Commerce Storefront documentation.

To confirm the setup, start a new chat and ask a question this documentation answers, such as one about a drop-in. If the reply cites or matches those topics, the tool can read the documentation.

Other AI tools

Tools that implement the llms.txt spec follow the index automatically when you provide this URL:

https://experienceleague.adobe.com/developer/commerce/storefront/llms.txt

For tools without URL-based context, paste the content of llms-small.txt or a relevant _llms-txt/*.txt topic file directly into the chat or system prompt.

Example prompt

After you add the documentation context, validate it with a specific question, such as:

Using the Storefront docs, tell me how to show a "Delivery estimate" message in the product details drop-in, below the existing Add to Cart button.

A successful setup returns a response that references storefront topics, drop-in APIs, or paths aligned with this documentation. Adapt drop-in names or UI labels for your project, then validate the response against the live documentation.

Keep context focused

• Paths and examples in this documentation follow the folder layout of the Adobe Commerce boilerplate . Keep your storefront project open so suggested paths match your workspace and tell the agent in chat if your repository layout differs. For example, if you merged boilerplate code into an existing site.
• To scope one reply to a single page, paste that page’s URL into the chat. Your context file connection stays in place, and the URL narrows only that exchange.