commercetools introduced AI Plugins that combine three components: a Knowledge MCP (Model Context Protocol) providing live documentation and schema lookup, a Commerce MCP enabling direct API interaction, and Skills that encode battle-tested implementation patterns (commercetools Blog). The plugins install into Claude, Cursor, VS Code Copilot, and OpenAI Codex, allowing coding agents to build with commercetools knowledge wherever developers work rather than only within documentation sites.
The gap the plugins address is critical for commerce practitioners: knowing APIs is not the same as knowing how to assemble them correctly. The Knowledge MCP stops hallucinations and grounds facts in live docs and schemas, while Skills encode the patterns that experienced architects use implicitly—storefront structure, cart and checkout flows, and proper composition (commercetools Blog). In a demonstration, a Claude-powered agent using these plugins built and deployed a production-ready B2C storefront with product search, faceted listing, cart, checkout with payment flow, authentication, and wishlist functionality from a single prompt, then deployed it to Vercel.
The plugins are free and open to start, with no API key required for the Knowledge MCP; connecting the Commerce MCP to a project requires credentials. This approach moves beyond the industry pattern of bolting on an MCP for better facts, instead pairing retrieval with encoded judgment to solve both correctness and architectural soundness in a single install.