28 prompts for Shopify's new agentic guidelines (an operator playbook)

Shopify's Spring ‘26 Editions release auto-indexed every eligible store into the agent network. ChatGPT, Copilot, and Gemini are now reading your catalog and returning results without you doing anything - and most operators haven't acted on what that actually means for their products.

Ranking happens across two layers. Shopify scores your products first using Listing Quality, five signals the merchant directly controls. Each AI channel then re-ranks on top of that using its own logic, which is mostly invisible from the admin side.

We went through both layers in detail in the , including what Shopify infers about your products when your metafields are empty - .

This is about what to do next. I put together 28 prompts that cover the full stack in Claude with the Shopify MCP - from diagnosing where your catalog stands right now to running a weekly ops review of how your products are performing across AI channels.

Six sections

Section 01 - Diagnose

Five prompts to see where your store actually stands before changing anything - a Listing Quality audit per collection, a check for descriptions where compliance text sits below the 6,000-character cutoff, a map of which fields Shopify is currently guessing on, an agents.md fetch, and a direct checkout eligibility audit by exclusion category.

Section 02 - Enrich

These prompts read your existing product descriptions and extract what the missing fields should be - material, style, occasion, technical specs, top features, and what makes the product unique - then return the suggested values for you to review before anything is written. You look at what Claude proposes, confirm, and it fills. There's also a bulk version that runs across a full collection once you've approved the candidates.

Section 03 - Descriptions

Five prompts to fix the content itself. Compliance disclosures get moved to the first 6,000 characters so they reach the agent in direct checkout contexts. Thin descriptions get rewritten to the Listing Quality spec using the five signals as a structure. Variant names that use abbreviations or codes - BLK, OS, M-14A - get flagged with plain-language alternatives, and missing image alt text gets surfaced across the collection.

Section 04 - Discovery

This is where you run a natural-language search against your own catalog the way an agent would - something like "linen shirt for a hot weather wedding under $200" - and see what actually comes back. Which products surface, which ones don't, and what's missing on the ones that should but didn't? There's also a prompt that checks for products sitting in a curated collection that don't show up when you search for what that collection is called - a gap that's more common than it sounds.

Section 05 - Operate

Five prompts to run on a regular cadence once the catalog work is in place. A daily orders briefing in three sentences via ShopifyQL, a Klaviyo re-engagement cohort you can pull and export directly, a restock list that ranks urgency by how fast a product is selling and how strong its Listing Quality is - because a weak listing about to run out is a different call from a strong one - discounts set up through Claude with collection, customer segment, and time window handled in one prompt, and a weekly Monday review that shows Listing Quality trend, fill gap vs the prior week, and revenue per agentic channel.

Section 06 - Non-US playbook

A US/Canada shipping audit for Shopify Catalog eligibility, an agents.md.liquid template that pulls from your live store data and stays dynamic, and a Catalog Mapping walkthrough for stores using custom metafields.

Let's connect

If you run something like this and want to compare notes on what is actually working, connect with me on LinkedIn.

Or drop an email at ankit@atomz.ai

Share this with someone who can find it useful: The Agentic Operator

- Ankit