Spring Boot

Connects your existing Java services to AI agents — Claude, GitHub Copilot, Cursor — via SSE (IDE clients) and stdio (CLI clients). Drop in your own @Tool methods and your Java backend becomes AI-callable.

---

What it does

Exposes ready-to-use MCP tools that any MCP-compatible AI agent can call:

Tool	What it does	Try asking Claude
queryDatabase	Read-only SQL with SELECT guard	"Show me the 5 most recent orders"
listFiles	Browse sandboxed workspace files	"What files are in the reports folder?"
readFile	Read file contents from workspace	"Read the contents of config.json"
getCustomer	Fetch customer from CRM REST API	"Get details for customer ID abc-123"
searchOrders	Filter orders by status	"Show me all PENDING orders"

---

Quick start

git clone https://github.com/anirbandashfx-commits/spring-boot-mcp-server.git
cd spring-boot-mcp-server
cp application.yml.example src/main/resources/application.yml
# Edit application.yml with your DB credentials
mvn spring-boot:run

Or with Docker:

docker compose up

---

Connect to Claude Desktop (stdio)

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "spring-boot-mcp": {
      "command": "java",
      "args": ["-jar", "/absolute/path/to/target/spring-boot-mcp-server.jar"]
    }
  }
}

Restart Claude Desktop. Look for the 🔨 hammer icon in the chat input — your tools will be listed there.

---

Connect to VS Code / Cursor (SSE)

Start the app first, then add to .vscode/mcp.json:

{
  "servers": {
    "spring-boot-mcp": {
      "type": "sse",
      "url": "http://localhost:8080/sse"
    }
  }
}

---

The transport rule (saves you hours of debugging)

Client	Transport	Use this
Claude Desktop, Claude Code CLI	stdio	command: java -jar ...
VS Code, Cursor, Windsurf	SSE	url: http://localhost:8080/sse

Using the wrong transport causes silent connection failures with no useful error message.

---

Add your own tools

@Service
public class MyCustomTools {

    @Tool(description = "Describe what this tool does in plain English")
    public String myTool(
        @ToolParam(description = "What this parameter expects") String input
    ) {
        return result;
    }
}

Spring AI auto-registers any @Service with @Tool methods. No extra config needed.

---

Built with

• Java 21 + Spring Boot 3.4
• Spring AI 1.0 MCP Server
• Tested: Claude Desktop (stdio), VS Code Agent mode (SSE)

---

Need a custom MCP server for your Java team?

I build these for teams integrating AI agents into existing Spring Boot applications. → Connect on LinkedIn