Shopify ยท agentic commerce

See your store the way an AI shopper does

A prompt you paste into an AI assistant. It reads the Shopify developer docs, queries your live catalog the way a shopper would, and hands back a scorecard of what to fix.

Before you start

This prompt only works if your AI assistant can read the Shopify developer docs. Install the Shopify AI Toolkit first. It gives tools like Claude Code, Cursor, Codex, and VS Code direct access to shopify.dev, which is what lets the assistant cite a real doc page for every claim instead of guessing.

One-line install for Claude Code:

claude plugin install shopify-ai-toolkit@claude-plugins-official

Other tools and full setup: shopify.dev/docs/apps/build/ai-toolkit

How to use it

  1. Fill in the brackets. Your store domain, what you sell, your best sellers, and five shopper-style questions. Highlighted in amber below.
  2. Paste it into an AI assistant that can browse the web and call your store's Storefront MCP endpoint, such as Claude Code or Cursor. The prompt is read-only. It does not create, edit, or delete anything.
  3. Read the scorecard. Fixes come ranked by how much they change what an AI agent tells a shopper, each with a link to the Shopify doc that explains it.

What comes back

Part 1

The surfaces

What AI agents actually query, what you need to turn on in admin, and which product fields they can see.

Part 2

Live shopper queries

Five real-world searches run against your catalog, with the raw response and the plain-English answer a shopper would hear.

Part 3

Scorecard

Intent match per product, thin or empty fields, price and inventory checks, and missing products with the likely cause.

The prompt

Replace every [bracketed] field before you run it. 
I run a Shopify store at [your-store].myshopify.com and I want to see how my products show up when a shopper asks an AI assistant (ChatGPT, Perplexity, Copilot, etc.) about products like mine. Use the Shopify developer docs to guide this. Do not answer from memory. Cite the shopify.dev page for every claim about how the agentic surfaces work.

Store context:
- We sell [category, e.g. GPS units and outdoor navigation gear]
- Our best sellers are [3 to 5 product names]
- Shoppers usually search by [use case, brand, spec, price range, compatibility]


Part 1. Confirm the surfaces
Using the docs, tell me:
- What AI agents actually query when they look up a Shopify merchant's products (Storefront MCP on my shop domain, Shopify Catalog / agentic storefront, anything else)
- Which of those I need to enable in admin, and where
- Which product fields those surfaces read (title, description, variants, media, metafields, category, availability, price) so I know what the agent can see

Part 2. Query my catalog the way a shopper would
Read-only. Do not create, edit, or delete anything in my store.
Call my store's Storefront MCP endpoint and run these shopper-style requests:
1. "[natural language query a real customer would ask, e.g. best handheld GPS for hiking under $300]"
2. "[query by brand or product name]"
3. "[query by compatibility or spec]"
4. "[vague or misspelled query]"
5. "[a query for something we do NOT sell, to see what comes back]"
For each one, show me the raw response and then the plain-English version of what the AI would tell the shopper. Then pull full top result of query 1.

Part 3. Scorecard
For each product that surfaced, tell me:
- Did it match the intent of the query? Yes / partly / no
- What fields came back empty or thin (description, images, variants, category, price)
- Whether inventory and price looked correct
- Any product I expected to see that did not appear, and the likely reason
Rank the fixes by impact on how an AI agent would present my products. For each fix, cite the doc that explains the field or setting.

If anything cannot be verified from the docs or from a live response, say so plainly rather than guessing.