Model Context Protocol (MCP) is an open protocol that standardizes how applications provide tools and context to LLMs. LangChain agents can use tools defined on MCP servers using the langchain-mcp-adapters library.
The protocol itself is a standardized set of rules (built on JSON-RPC) that dictates exactly how an AI application and a data source should talk to each other. It ensures that no matter what LLM you are using, it can understand and use external data in a uniform way.
An MCP Server is a lightweight, specialized application that exposes specific data or tools through the Model Context Protocol. It acts as the "translator" or "bridge" between a specific technology (like a PostgreSQL database, Slack, or GitHub) and the AI. Because the server speaks the standard MCP language, any AI application that supports MCP can connect to it instantly.
Before MCP, if you wanted an AI model to read your local files, browse the web, or query a database, developers had to write custom, fragmented integration code for every single tool and every single LLM. MCP standardizes this connection.