How a scan works

The free scan fetches the homepage, robots.txt, a sitemap, llms.txt, a UCP well-known URL if it exists, policy pages linked from the footer, and one product page. A paid scan can read up to 50 product pages. Requests stay at or under two per second, time out at 15 seconds, and stop at 5 MB. Private and internal addresses are refused.

The scanner identifies itself as CartReadyScanner and follows robots.txt. If a shopping crawler is disallowed, we record that and do not pretend to be a browser to get around it.

The score

Each check is a pass, a partial, a fail, or not applicable. High-impact checks count three, medium two, and low one. A pillar score is the weighted result of its checks. The total gives Access and discovery 25%, product data 35%, trust and policies 20%, and channel connections 20%.

0 to 44 is hard for agents to read. 45 to 69 is readable, with gaps. 70 to 89 is mostly ready. 90 to 100 is agent-ready. The same inputs and the same rule version produce the same score.

What paid delivery adds

The Full Report widens the product sample, keeps every finding with evidence, and exports a PDF. The Fix Pack adds JSON-LD templates, an OpenAI-style product CSV, a Google Merchant table, a Stripe catalog CSV, llms.txt, a robots.txt diff, and install notes. Missing GTINs are listed. They are never invented.

No one can guarantee that an AI assistant will show your products. Each service decides what it displays. CartReady checks published requirements and whether a store is easy for those services to read.