Official Model Context Protocol Java SDK released, giving Java apps a standard interface to AI toolchains

Maintained by the Spring ecosystem team, the SDK offers a standardized interface for Java apps to interact with AI models and tools, supporting both synchronous and asynchronous communication. Core modules cover STDIO, JDK HttpClient client transport, and Jakarta Servlet server transport; the public API is based on Reactive Streams, implemented internally with Project Reactor, and offers a synchronous facade for blocking scenarios. JSON serialization defaults to Jackson but is replaceable via an abstraction layer in mcp-core; logging uses SLF4J, and observability data propagates through Reactor Context for easy Micrometer or OpenTelemetry integration. Modules include mcp-bom, mcp-core, mcp-json-jackson2, mcp-json-jackson3, mcp, and mcp-test. Spring integration (WebClient, WebFlux, WebMVC) is merged into Spring AI 2.0+. For Java developers building enterprise AI applications with Spring Boot, this means no longer assembling an MCP communication layer yourself—you can plug in via the mcp dependency or Spring AI's starter. Teams with their own JSON libraries or authorization systems can keep their existing stack through the JSON binding abstraction and pluggable authorization hooks. For teams needing long connections and streaming, the Reactive Streams public interface with Reactor internals handles concurrent requests, notifications, and SSE/Streamable HTTP transport, while simple request-response projects can use the synchronous facade to lower the learning curve. Teams exposing MCP endpoints with Spring WebFlux or WebMVC can depend directly on the corresponding transport modules in Spring AI 2.0+.