Model Context Protocol (MCP)

Model Context Protocol (MCP) is an open standard for connecting AI applications to external tools, data sources and services through one shared interface. The application acting as host runs an MCP client for each server it connects to, and the client exchanges JSON-RPC 2.0 messages with an MCP server that offers tools the model can call, resources the application can read, and prompts that users pick as reusable templates. A vendor's function-calling API defines how one model asks for a tool, while MCP standardizes how tools are described, found and reached, so one server works with any compatible client. The model still decides which tool to call. MCP covers agent-to-tool access, and agent-to-agent communication uses separate protocols such as A2A.

Anthropic released MCP in November 2024 and in December 2025 donated it to the Agentic AI Foundation (AAIF), a directed fund under the Linux Foundation co-founded by Anthropic, Block and OpenAI. The specification is versioned by date and runs over two transports, stdio for local servers and Streamable HTTP for remote ones, and the older HTTP+SSE transport is deprecated. The 2026-07-28 revision made the core protocol stateless, deprecated the client-side sampling and roots features, and moved long-running Tasks into an official extension. MCP standardizes the connection and leaves consent and permissions to each implementation, so hosts should treat tool descriptions and returned content as untrusted input and put write operations behind scoped permissions.

Work with us

Ready to put agentic AI to work?

Book a free 30-minute consultation. We'll map one real process worth automating with production-grade AI.