MCP

Your store, operable by AI agents.

Every Shopretailor store includes a native MCP (Model Context Protocol) server. Connect an AI agent, such as Claude, and it works on your store through purpose-built tools.

  • 490+purpose-built agent tools
  • 12tool categories, from commerce to page building
  • 3key roles: admin, support and writer
  • Everycall recorded in the audit log, with key and IP

What an agent can do

The same jobs your team does. On tap.

Everything below is a tool the agent calls, not a screen it has to click through.

  • Catalogue

    Create products and combinations, upload images, put products on sale and manage redirects.

  • Pages and content

    Build pages and blog posts from page builder blocks, and manage menus and sliders.

  • Email

    Write, compile and save email templates, campaigns and automations, with no browser involved.

  • Support

    Read tickets and prepare reply drafts for your team to send.

  • Diagnostics

    Read orders, the error board, site health, checkout health and cache state to find out what went wrong.

  • Search and SEO

    Reindex search, update SEO titles and descriptions, and fix broken links.

Safety first

An agent with the keys needs limits.

So the limits live in the server itself, where a clever prompt cannot argue with them.

  • Every response states which environment it acted on: development, staging or production.
  • The first change to a production store needs an explicit confirmation.
  • Many tools can preview a change as a dry run before anything is written.
  • Every call is recorded in an audit log with the key and IP address, kept for a period you choose.
  • Built on the stateless revision of the Model Context Protocol, with origin validation.
  • Admin

    The full toolset, for the agent you trust with the shop.

  • Support agent

    Read only. It can look things up for your support team and change nothing.

  • Blog writer

    Drafts only. It can write content and cannot publish it.

Ready when you are

Hand it the keys.

See an agent build a page, create a product and trace a fault on a live demo store. Then ask it something awkward.

  1. Tell us about your shop
  2. We set up your store on its own private server
  3. You build, launch and sell