Model Context Protocol (MCP)

Language Server Protocol (LSP) revolutionized how programming languages integrate with developer tools. Model Context Protocol (MCP) aims to do the same for a new generation of AI tools.

But before we lay our scene of two protocols (both alike in dignity), let’s take a moment to motivate the problem a bit more.

→ nshipster.com/model-context-protocol/

The Model Context Protocol (MCP) is an emerging open-source standard from Anthropic, designed to standardise AI integration with external tools, data sources, and services, addressing the M × N problem of custom integrations by transforming it into an M + N solution, similar to the Language Server Protocol (LSP).

It enables AI applications, like Claude or ChatGPT, to access resources such as databases, calendars, or APIs via a client-server architecture using JSON-RPC 2.0 over Stdio or HTTP with Server-Sent Events, supporting prompts, resources, and tools for enhanced AI capabilities.

Notable implementations include iMCP, a macOS app connecting Apple user data to AI, and the mcp-swift-sdk for building MCP servers and clients in Swift, alongside tools like hype for Python and emcee for generating MCP servers from OpenAPI specs.

Recent updates, backed by OpenAI and Microsoft, include OAuth 2.1 authorisation, streamable HTTP transport, and JSON-RPC batching, enhancing security and real-time communication as of March 26, 2025. These developments promise simplified, scalable AI integration, with the Swift community encouraged to contribute via the Swift Forums and GitHub repositories like mcp-swift-sdk.