The skill
Install once. Attach a SKU and promo export and get a per-product policy, agent-eligible verdict, margin floor, promos to suppress, and checkout gaps.
What's inside
Three files, one job: set your terms for agent buyers
Built for ecommerce and growth teams, not data scientists.
Install guide
A two-minute setup, plus the exact columns to export so the margin math is real instead of unknown.
Plan B prompt
No skills access? Paste the included prompt into any Claude chat with your export. Same rules, same ranking, zero setup.
How it works
Three steps, from a SKU export to a policy you can configure.
Step 1Export your SKUs and promos
Pull your products with unit cost, price, active promotions and whether each stacks, inventory, and revenue per SKU.
Step 2Hand it to Claude
The skill computes margin, sets an agent-eligible verdict against your floor, decides which promos to suppress from agent channels, and flags checkout blockers.
Step 3Apply the policy
Back comes an action list ranked by revenue times the size of the problem, so you fix the products leaking the most margin first.
No connection to your store. You bring the export, the skill writes the policy. It recommends the guardrails, you configure them in your own store settings.
How to protect your margin from AI shopping agents
Agent checkout went self-serve this month, and AI shopping agents now compare your exact price and terms and stack every discount they can find. A promo that lifts human order value can quietly leak margin to an agent. This free Claude skill turns your SKU and promo export into a policy that decides what you expose to agents and on what terms.
Why an agent buyer is different
An AI agent does not respond to brand, social proof, or urgency. It reads structured price and terms, picks the cheapest qualifying option, stacks eligible promos, and abandons the moment checkout needs a human to click. The levers built for human buyers can lose money against an agent, so the terms you sell on need to be set deliberately, not left at the default.
How the skill builds your policy
It computes margin per SKU, then sets an agent-eligible verdict against your margin floor. Thin-margin and loss-leader items get excluded from agent channels, configured and bundle items go to review for a human rule, and healthy-margin products stay eligible. It checks each promo's effect on margin net of the discount, not gross, so a stackable promo that breaks the floor is flagged for suppression.
Why the action list is ranked by revenue
A poorly governed SKU that carries heavy revenue is where the margin leak is largest. The skill ranks fixes by revenue times the size of the problem, so the products that sell a lot and leak the most rise to the top. Each item names the specific guardrail, the promo to suppress, the floor breach, or the checkout step to fix.
Why now
Shopify made its Universal Commerce Protocol self-serve and now auto-syndicates eligible products to ChatGPT, Copilot, Google AI Mode, and Gemini, while Visa and OpenAI turned on tokenized agent checkout. Your store is exposed to agent buyers by default. The full background is in the source article, Shopify Just Opened Your Store to AI Buyers. Set the Terms Before They Set Themselves.
Who it is for
Ecommerce and growth marketers, Shopify and marketplace merchants, and retail and revenue teams that need to keep margin control as agent-driven buying scales.