Model Context Protocol (MCP) is Anthropic's open standard for connecting AI assistants to external data sources and tools. Released late 2024 and widely adopted across the LLM tooling ecosystem in 2025, MCP defines a JSON-RPC-style spec for how an AI assistant requests data or invokes a tool, and how a server exposes resources to the assistant.
In production deployments, MCP servers expose: filesystem access, database queries, internal API calls, third-party integrations (Slack, Linear, GitHub), and customer-specific data sources. The assistant picks tools dynamically based on the user prompt.
For enterprises, MCP becomes the architectural seam between the LLM and the rest of the stack -- well-scoped MCP servers + RBAC + audit logging make AI-agent workflows operationally manageable.