Model Context Protocol
Authoritative Definition
An open standard protocol (MCP) that unifies and securely standardizes how AI models and applications connect to external data repositories, databases, and tool servers.
Overview & Technical Description
Created by Anthropic, the Model Context Protocol (MCP) defines a universal client-server JSON-RPC messaging architecture. It enables AI applications, IDEs, and agent frameworks to seamlessly discover and interact with external resources, prompt templates, and executable tools. By decoupling the AI model from the underlying data integrations, MCP fosters a modular and highly extensible ecosystem. MCP operates via a standardized transport layer, typically utilizing standard input/output (stdio) for local processes or Server-Sent Events (SSE) for remote HTTP integrations. An MCP server exposes specific capabilities—such as querying a database, searching a filesystem, or invoking a third-party API—while the MCP client (the AI application) dynamically discovers and leverages these capabilities to ground its responses and execute actions. This standardization drastically reduces the integration overhead for developers, who no longer need to write custom glue code for every new data source. MCP provides a secure, sandboxed environment for tool execution and context retrieval, rapidly becoming a foundational layer for modern AI agent infrastructure.
Editorial Notes
Rapidly becoming the universal protocol for AI tool and context integration. Implementations must carefully manage access control and security boundaries, as MCP servers expose powerful execution capabilities directly to the LLM.