ah-mcp, developed by Mrserzhan, is an MCP server that connects Albert Heijn services to AI assistants to enable conversational grocery workflows. The tool lets assistants query the supermarket catalog, manage shopping lists and baskets, access past orders, and surface Bonus discounts through OAuth2-authenticated sessions. It runs on Node.js and supports Stdio and Server-Sent Events transports, aimed at tech-savvy Albert Heijn customers who use MCP hosts to automate meal planning and shopping tasks.

What tasks can you actually use it for?

The tool performs catalog queries and account-backed shopping actions, exposing product search results with detailed product information, personal shopping-list creation and modification, basket add/remove operations, access to order history, Bonus offer retrieval, and favorites management. These outputs arrive as structured responses suitable for an assistant to present or act on. The developer implemented direct API calls to Albert Heijn so responses reflect data pulled from the supermarket systems.

How accurate and timely are the shopping outputs?

Real-time accuracy depends on direct API access rather than scraped snapshots, because the server queries Albert Heijn’s systems directly. That design produces current prices, stock hints, and live Bonus offers when the supermarket API returns them. Accuracy therefore tracks the upstream service state; the tool does not add independent fact-checking. Users should treat assistant suggestions as derived from the retailer’s live responses.

What inputs and environment are required to run it?

Running the server requires a modern JavaScript runtime and an MCP host. The project runs on Node.js and needs a Model Context Protocol client such as Claude Desktop to interface with the assistant. Authentication uses an OAuth2 flow for secure login and token management. Basket manipulation and list edits can be performed through the assistant, while final checkout commonly requires confirmation in the retailer’s official channel.

Does it fit into an assistant workflow without heavy rework?

Integration suits power users who operate MCP hosts and automate workflows. The server supports both Stdio and Server-Sent Events transports, giving flexibility for local or remote deployments and scripted interactions. The implementation is an independent, community-oriented project, which makes source inspection and extension possible for teams that prefer to audit or customise behaviour before production use.

Best for technical users who want assistant-driven shopping automation

The tool is a practical choice for technically proficient users who accept a moderate setup and want programmatic access to their supermarket data inside an assistant. Its community-maintained codebase makes auditing and customisation straightforward, so it suits automation projects and workflows that require tracked, reproducible actions rather than off-the-shelf consumer convenience.