Shopify open-sources AI Toolkit with MCP plugins for Claude Code and Cursor

Shopify has open-sourced its AI Toolkit, a set of MCP plugins that connect AI coding assistants directly to Shopify APIs. The project is available on GitHub under the MIT license, and the first supported tools include Claude Code, Cursor, and Gemini CLI.

The release matters because it turns Shopify’s developer surface into something AI assistants can use without custom glue code from the community. Instead of asking developers to switch between an AI tool and a separate API client, the plugins let the assistant call Shopify functions through the Model Context Protocol, or MCP, which is a standard for giving AI systems access to external tools and data sources.

For Shopify, this is a practical way to make its platform easier to work with inside the growing ecosystem of AI coding tools. For developers, it means the assistant can understand and act on Shopify-specific tasks in a more structured way, using official integrations rather than unofficial wrappers that may break when APIs change.

The toolkit was published on GitHub at github.com/Shopify/ai-toolkit, where Shopify describes it as an open-source project for building and using MCP-based integrations. By shipping the plugins itself, Shopify is doing more than releasing sample code. It is providing a maintained path for AI clients to reach its APIs in the format those clients increasingly expect.

Claude Code, Cursor, and Gemini CLI all sit in the category of AI development tools that can read context, generate code, and invoke connected services. With MCP support, they can go beyond text generation and reach into real systems, such as a commerce platform, through a standardized interface. That matters because it reduces the amount of one-off integration work each tool vendor, developer, or internal platform team would otherwise need to do.

Shopify’s move also highlights a shift in how major software companies are thinking about developer access. In the past, platform companies mostly exposed REST or GraphQL APIs and expected developers to build their own clients. Now, some are packaging those APIs for AI agents as well, with manifests, tool definitions, and permissions designed for machine-to-machine use.

That is one reason the release stands out. Official MCP support is still relatively new, and most integrations have come from community projects or individual developers building adapters around popular apps. A large platform company publishing its own MCP tooling signals that this is becoming part of the standard developer experience, not just an experiment at the edges.

The AI Toolkit is also useful as a reference for how a platform company can structure tools for AI consumption. The plugin layer, the manifest format, and the way functions are described all matter when an AI assistant has to decide what it can do, what it needs permission for, and how to call the right endpoint.

Because the project is open source, anyone can inspect the implementation, adapt it, or use it as a starting point for their own MCP-based integrations. For teams building internal automation or custom AI assistants, that makes Shopify’s release less like a standalone feature and more like a published example of how an API-first company can expose its services to AI clients in a controlled way.