What is Model Context Protocol (MCP)?

Home > Glossary > Model Context Protocol (MCP)

Model Context Protocol (MCP) is an open protocol that standardizes how AI applications connect to external tools, data sources, prompts, and services. Instead of building a one-off integration for every model and backend, an MCP server exposes capabilities in a consistent way that an MCP-compatible client can discover and use.

In commerce, MCP can sit between an AI agent and systems such as the product catalog, ERP, PIM, CRM, OMS, search engine, or support platform. The protocol does not define ecommerce business rules by itself; it provides the interface through which an agent can retrieve context or invoke approved tools. That makes permission design and tool behavior more important than simply “connecting the model.”

A commerce-oriented MCP implementation should define:

• Which resources are read-only context and which tools can create or change business data.

• Authentication and authorization rules for customer, employee, and service identities.

• Clear input and output schemas so agents receive structured, predictable results.

• Logging, rate limits, validation, and failure behavior for tool calls that affect live commerce.

MCP is useful because it reduces integration fragmentation, but it is not a substitute for an integration architecture. An MCP server still needs reliable APIs and authoritative data behind it. For enterprise ecommerce, the safest pattern is to expose narrowly scoped capabilities while keeping pricing, inventory, payment, and order validation inside deterministic commerce and back-office services.

Example: An MCP server might expose read-only product search and inventory tools to an AI assistant while keeping order creation behind a separate, higher-trust tool that requires authenticated account context.