Server2mcp Spring Boot Starter

🚀 Server2MCP Spring Boot Starter

A Spring Boot starter for automating the integration of MCP (Model Control Protocol) services.

🚀 Quick Start

Since this library has not been pushed to the central repository yet, you can download the source code, execute mvn clean install, and then reference it as a dependency.

Add Dependency

<dependency>
    <groupId>com.ai.plug</groupId>
    <artifactId>server2mcp-spring-boot-starter</artifactId>
    <version>0.0.1-SNAPSHOT</version>
</dependency>

Then add the following configuration to your configuration file:

plugin.mcp.enabled=true

If you are using YAML, the configuration should be:

plugin:
  mcp:
    enabled: true

The above is the basic configuration for starting the project, which includes all the native configurations such as spring.ai.mcp.server-side. By default, it will register all controllers under your startup classpath as MCP interfaces. If an interface method or class is annotated with @Deprecated, it will not be registered.

✨ Features

• Automatically configures the MCP service. Similar to the relationship between MyBatis Plus and MyBatis, it is non-intrusive and purely enhancing.
• Supports all native functions of MCP in the Java SDK, providing tool registration and callback mechanisms.
• Supports custom tool parsing.
• Users can customize the Parser without relying on the responsibility chain to implement attribute parsing for unique interface annotations.

👀 Incomplete Points

There are still many parsing extension points to be integrated in Spring AI. For example, currently, only the Javadoc version of annotations has been completed, but the parsing architecture uses the responsibility chain and templates, which is extremely easy to expand. In the future, various mainstream methods will be integrated to describe annotation parsing.

📚 Documentation

It can be understood as exposing interfaces to AI. These interfaces are the same as ordinary interfaces, but can be called by AI. Relevant knowledge documents: Model Context Protocol (MCP) :: Spring AI Reference, Introduction - Model Context Protocol, and Introduction - Model Context Protocol.

💕 Best Practices

With this framework, you no longer need to rebuild an MCP service application from scratch, nor do you need to add the @Tool annotation to highly coupled copied code or add MCP functionality to the source code. You can easily register MCP interfaces by simply adding a custom @ToolScan annotation based on the configuration class. What if the MCP SDK is revised? Don't worry, I maintain the core content, and the usage remains the same.

1. You can easily build a multi-agent application by using multiple custom AI dialogue interfaces and then simply call the corresponding MCP interfaces on the client side.
2. It can quickly enable your management system to access AI dialogue calls with a high degree of customization. You don't need to pay attention to any details in the AI field. Just focus on your favorite backend and frontend areas, and you can achieve cool effects. In contrast, applications like DB-GPT may seem cumbersome and difficult to expand.
3. It can be easily used with a simple MCP client like Cursor for interface debugging.

🔔 Summary

This framework is actually very simple. There may be many bugs and deficiencies in the code. Please bear with them.

📄 License

The code is released under the Apache 2.0 License.