+420 728 089 029jsme@webotvurci.czCZ
Let's set sail
+420 728 089 029CZ

Shoptet MCP: connect AI to your store's data

· ⏱ 4 min read

Shoptet MCP: connect AI to your store's dataAI at the helm

TL;DR: We built an open-source MCP server for Shoptet Premium that connects an AI assistant (Claude, Cursor or Claude Code) straight to your online store data. Instead of exporting to Excel, you simply ask in plain language – "what was last week's revenue day by day?" – and the AI answers from live data. The server is deliberately read-only, so AI can only read and analyze data, never change it.

AI and Shoptet are our daily work – we build add-ons, integrations and AI agents for real online stores. We keep hitting the same barrier: the store owner wants to "use AI", but the AI has no way to reach their data. This MCP server removes that barrier. Let's go through what MCP is, what the server can do and how you deploy it.

What MCP is and why it matters

MCP (Model Context Protocol) is an open standard from Anthropic that works as a universal "cable" between an AI model and your systems. Without MCP, AI only knows what you type into the chat. With an MCP server, AI can connect to a specific data source on its own, find out which tools it has available, and use them.

For an online store that's a major shift. Instead of pulling data out of your Shoptet admin by hand, exporting it and pasting it into ChatGPT, the AI pulls the data itself – current, complete and in the right format. Your only job is to ask the question.

What the Shoptet MCP server can do

The server connects to the Shoptet Premium API and exposes a set of tools to the AI, split into three groups:

Summaries and analytics – precomputed overviews for the questions you ask most often:

  • orders_summary, revenue_trend – revenue and orders over time
  • top_products, top_customers – best-selling products and top customers
  • unpaid_orders – unpaid orders
  • inventory_overview, stock_movements_summary – stock levels and stock movements
  • reviews_summary, refunds_summary – reviews and returns

Lists and details – access to individual records: orders, products, customers, invoices, reviews, credit notes, complaints, discount coupons, articles and pages (list_orders/get_order, list_products/get_product and more).

Code lists and metadata – order statuses, payment and shipping methods, categories, brands, store information. For advanced users there's also shoptet_raw_get as an "escape hatch" to any API endpoint.

A few real questions you can ask the AI this way:

  • "What was last week's revenue day by day?"
  • "Show the 20 best-selling products by revenue over the last 3 months."
  • "Which orders have been unpaid for more than a month?"
  • "Compare sales by sales channel over the last month."

Why the server is read-only (and why that's good)

The biggest worry about connecting AI to live data is this: what if the AI deletes or changes something? This MCP server answers it at the architecture level – the HTTP client only knows get(). There is no post, put, patch or delete in it. Even if the AI "decided" to change an order, it technically has no way to. It can only read and analyze data.

For a store, that means you can point MCP at production data without any risk. It fits exactly what AI is best at: reporting, analysis and decision-making, not writing.

How to deploy it

The server is written in TypeScript, runs on Node.js 20+ and is free under the MIT license. The setup is short:

  1. Get an API token in your Shoptet admin: Connections → Shoptet API (a private token in the format 123456-p-789012-…).
  2. Clone and build the repository:
    git clone https://github.com/Webotvurci-s-r-o/shoptet-premium-readonly-mcp.git
cd shoptet-premium-readonly-mcp
npm install && npm run build
  3. Connect it to Claude Desktop (or Cursor / Claude Code) through claude_desktop_config.json, where you add the path to dist/index.js and your token in the SHOPTET_PRIVATE_API_TOKEN variable.

You'll find detailed configuration for each platform in the README on GitHub. Because the project is open source, you can modify it, extend it with your own tools, or have us deploy it for you.

Who it's for

You'll get the most out of the MCP server if you sell on Shoptet Premium and want to:

  • stop exporting data to Excel for every single report,
  • ask about your numbers in plain language instead of clicking through the admin,
  • give AI the context for smarter decisions about stock, marketing or pricing.

You don't need to know how to code – set it up once and then just ask. And if you get stuck with the setup or want to go further (your own AI agents, automations, connections to other systems), that's exactly what we teach at the Shoptet AI training and what we build as custom AI solutions.

MCP is only just getting started and we want to be part of it – that's why this server is public and free. Try it out, send us your ideas, or let us connect AI to your online store for you.