a2ajava implements both sides of Google's Agent2Agent protocol in Java — server and client — for integration into Spring Framework and Spring Boot applications. A2A standardises how independent agents advertise capabilities and exchange tasks, and this library covers sending and retrieving tasks between agents built with different tools by different authors. The MCP relevance is the automatic dual exposure: an agent defined with a2ajava's annotations is published as an A2A agent and as an MCP tool at the same time, from the same declaration, so the same Java method is reachable by an MCP client and by another agent without writing two integrations. The library also supports building plain MCP servers in Java through those annotations, which puts it in the same role as an MCP server framework for the JVM. It is multi-language across the JVM, working from Java, Kotlin and Groovy, and multi-platform on the model side, documented against Gemini, OpenAI, Claude and Grok. For Java teams the significant point is that agent capabilities can be added to an existing Spring service as annotated methods, reusing the dependency injection, security and configuration already in place, rather than standing up a separate Python or Node process. The audience is enterprise Java developers exposing existing business logic to agents.
https://github.com/vishalmysore/a2ajava
The MCP Java SDK is the official Java SDK enabling Java applications to interact with AI models and tools through the Model Context Protocol, supporting both synchronous and asynchronous communication and integrating with Spring AI for enhanced functionality.
Quarkus MCP Server is a Quarkus extension that enables developers to implement Model Context Protocol server features for seamless integration between LLM applications and external data sources and tools.
Clojure library for building Model Context Protocol servers and clients natively on the JVM, implementing the stdio transport of the 2024-11-05 spec without using the Java SDK.