Shelfglance

Agent endpoint

What an agent sees at bikes.com

Searching bikes.com for bikes the way a shopping agent would returned 10 products. This is what it received.

ProductPriceCategorySKU
Reaper 27.5 $2,029.00 set na B0237OS3
Element Alloy 10 $1,349.00 set na B0213SM93
Instinct Carbon 70 $4,549.00 set na B0215SM94
Instinct Carbon 90 $9,799.00 set na B0214XS74
Instinct Alloy 50 $4,599.00 set na B0322XS2
Altitude Powerplay Alloy 30 $6,099.00 set na B0420SM2
Element Carbon 70 $5,599.00 set na B0311SM2
Element Alloy 50 $4,099.00 set na B0315XS2
Reaper 20 $2,699.00 set na B0501OS1
Altitude Carbon 70 $4,874.00 set na B0224MD94

Price, title, description and URL are Shopify’s doing and are filled in for you. Category and SKU are yours, set per product in your admin. Here 10 of 10 carried a category and 10 of 10 carried a SKU.

Read another one

Leave the search term blank and we use the store name. Any Shopify store, not just your own.

What you are looking at

Your store answers AI shopping agents at /api/ucp/mcp, speaking the Universal Commerce Protocol. Shopify switched it on; it is not in your admin and nothing in your theme controls it. We sent it the same search_catalog request an agent would and put the answer in a table. The long version, with what we found across sixty stores, explains why the two columns on the right are the only ones you can change.

This is not a score and nothing here is a fault. If you want the graded version — whether your pages can be read, not just your endpoint — the scan covers that.