Hi Spring AI team,

I've built a production-ready MCP server using Spring AI that demonstrates both MCP server and client patterns in one application. It provides AI assistants with Maven Central dependency intelligence.

What it does

• 8 dependency analysis tools (version lookup, age analysis, health scoring, release patterns)
• Direct maven-metadata.xml parsing from Maven Central for accuracy
• Works with any JVM build tool (Maven, Gradle, SBT, Mill)
• Dual MCP architecture: acts as both MCP server (Maven tools) AND MCP client (Context7 docs)
• Async webflux client with streamable-http transport

Spring AI Integration Patterns

• MCP Server: @Tool annotations, type-safe responses, Caffeine caching
• MCP Client: Async client, tool callbacks, graceful degradation
• Performance: GraalVM native images, <100ms cached responses, 90% cache hit rate

Links

• GitHub: https://github.com/arvindand/maven-tools-mcp
• Docker Hub: https://hub.docker.com/r/arvindand/maven-tools-mcp

If there's interest, this could serve as a sample for dual MCP server + client usage in the Spring AI MCP starters. :)