Vendooly
// Model Context Protocol for Amazon

An Amazon MCP server built for controlled operations.

Vendooly sits between an AI assistant and the Amazon APIs you authorize. It exposes defined tools for Seller Central and Amazon Ads, keeps each request scoped to the right seller account and marketplace, and waits for explicit approval before writes.

AI assistantCalls a defined Vendooly tool with the authorized scope.
Vendooly MCPQueries the relevant Amazon API and returns the minimum operating context.
Human approvalReviews the proposed write and explicitly confirms or rejects it.
No unrestricted seller credentials in the chat
What the MCP layer changes

A tool boundary between the model and Amazon.

An AI model can explain Amazon concepts on its own. Operational work needs fresh, authorized data and a clear boundary around what may be read or changed.

Defined capabilities

The assistant calls named tools for catalog, listings, inventory, orders, pricing, account health and Sponsored Products, depending on the granted scopes.

Tenant and context isolation

Requests stay attached to the selected org, seller account and marketplace. Agencies can work across authorized contexts without combining them.

Approval before writes

Read operations and proposed changes are different jobs. Listing, feed, price, bid, budget and campaign writes require a visible confirmation.

Operational coverage

Investigate first. Prepare the next action second.

Seller API

Catalog, inventory and orders

  • Catalog and listing quality checks
  • Inventory and FBA availability
  • Orders, pricing and Buy Box context
  • Account-health signals within authorized scopes
Amazon Ads API

Sponsored Products

  • Campaign and ad-group performance
  • Keywords, targets, bids and budgets
  • ACoS, ROAS and spend investigation
  • Stock-aware operating review
Action layer

Controlled execution

  • Prepare listing and feed changes
  • Prepare price and catalog updates
  • Prepare bid, target and budget changes
  • Record the approval and outcome
A practical MCP flow

Three boundaries stay visible.

01 · Authorization

Amazon access is granted through OAuth

The seller authorizes the relevant Seller API or Ads API scopes. Vendooly does not ask the operator to paste Amazon credentials into an AI conversation.

02 · Tool call

The assistant requests a bounded operation

The MCP tool validates the org, seller account, marketplace and parameters before retrieving or preparing data.

03 · Confirmation

The responsible person decides

For a write, the proposed scope remains visible and nothing is executed until the user confirms.

Try Amazon MCP on one seller account.

Start with a bounded question about listings, stock, pricing or Sponsored Products, then inspect exactly what the tool used.

Start free