spring-ai-mcp

Implement MCP tools with Spring AI by exposing them via infrastructure layer annotations and delegating execution to application layer Use Cases, ensuring domain models remain cleanly separated. This structure maintains strict DDD boundaries throughout your LLM tooling.

The Model Context Protocol in LLM tooling is used to expose structured tools and domain models to AI providers like Grok and Ollama, allowing LLMs to directly invoke application use cases through a standardized interface.

Spring AI works with Grok and Ollama for tool invocation by integrating the providers in development and production contexts, where the infrastructure layer delegates tool calls to use cases that the configured LLM invokes.

You need Spring Boot and a Java-based tool implementation to build MCP-based tools for LLMs using this approach, along with Spring AI starters to configure the LLM interactions and infrastructure annotations.

The best way to structure LLM tools in a DDD context is to place tool annotations in the infrastructure layer, business logic in application use cases, and data models in the domain layer, ensuring clean separation of concerns.

Configure Grok or Ollama as the AI provider for Spring AI tools by setting up the Spring Boot module, running the app in development mode, and following the documentation to invoke sample tools from an LLM-based interface.