AgentXray Blog Pricing

One list across every Amazon client account you manage

An agency running fifty Amazon accounts does not have a dashboard problem, it has a which-of-my-fifty-clients-needs-me-today problem, and no per-account view answers that. AgentXray reads every connected account on a schedule and returns one list ranked across the whole book, worst first, with each row naming the account it belongs to and the source its number was measured from: search-term history, inventory age, the listing's own health issues in Amazon's wording, settlement line items, and the account health tier Amazon itself reports.

Your AI drafts the change; a person approves it

Your own Claude or ChatGPT connects over the Model Context Protocol, names which client it is acting for, and files a bid cut, negative keyword, budget or listing change as a draft. The draft waits in an approval queue. The connector token that wrote it cannot authenticate against the approval endpoint, so a leaked connector can propose anything and change nothing.

Naming the client is required, not optional

Twenty clients on amazon.de is twenty accounts and one marketplace. Where more than one connected account could match, the system refuses to guess and asks which client you meant. A resolver that quietly picks the oldest match is how a change drafted for one client lands on another, with Amazon accepting it and nothing reporting an error.

What we do not claim

Agency pricing · What the platform measures · How seller data is kept apart