wtf is kotlin-sdk?

The MCP Kotlin SDK lets Kotlin developers connect their applications to AI models in a standardized way. Instead of writing custom integrations for every large language model or AI assistant your app might talk to, this toolkit provides a common interface so your app can expose its own data, prompts, and tools, or connect to other services that do the same. It's an official collaboration between the maintainers of the Model Context Protocol and JetBrains, the company behind Kotlin. At its core, the SDK helps you build either a "client" (an app that connects to an external AI service to ask questions) or a "server" (an app that exposes its own capabilities to an AI). A server can share three main things: prompts (templates for common questions or tasks), resources (data sources like files or API responses that give the AI context), and tools (actions the AI can ask your app to perform). The SDK handles the underlying communication, supporting several common network methods so the client and server can reliably pass messages back and forth. A developer building a customer support chatbot, for instance, could use this SDK to create a server that exposes a tool for looking up a user's recent orders. When the AI assistant receives a question about a delivery, it can invoke that tool to fetch the real order data, then use that context to give an accurate answer. A developer could also build a client that connects to an existing MCP server to pull in external context before sending a prompt to a model. One notable aspect of this project is that it supports Kotlin Multiplatform. This means a development team can write their MCP integration once and run it across multiple environments, including standard Java-based servers, JavaScript for the web, and native applications. The README notes that it relies on a framework called Ktor for handling network connections, which is a common choice for Kotlin projects.