wb04307201/spring-ai-loom-agent

What it solves

Spring AI LoomAgent provides an out-of-the-box framework for building AI agents within the Spring Boot ecosystem. It eliminates the need to build core agentic infrastructure from scratch by providing integrated capabilities for conversation memory, knowledge retrieval (RAG), tool execution, and user management.

How it works

The project acts as a "batteries-included" starter for Spring Boot applications. It integrates with Spring AI to connect to various LLMs and implements a modular architecture where tools are defined by interfaces and default implementations. It leverages the Model Context Protocol (MCP) for tool integration, allowing the agent to interact with external systems. The system uses a database-driven approach to manage "Skills" (prompt templates) and "Knowledge Bases" (vectorized documents), which can be shared via internal markets or assigned via Role-Based Access Control (RBAC).

Who it’s for

Java and Spring Boot developers who want to quickly deploy a fully featured AI agent with a chat interface, admin console, and RAG capabilities without implementing the underlying plumbing.

Highlights

• Integrated RAG: Built-in knowledge base management with Tika parsing and JVector local storage, with support for swapping in other Spring AI vector stores.
• MCP Tooling: Support for Model Context Protocol (MCP) servers, including built-in tools for file system operations, Git, Maven, and automated deployment pipelines.
• Skill & Knowledge Markets: A system for creating, sharing, and pulling prompt templates and knowledge bases between users.
• Agentic Task Management: Ability to delegate work to synchronous sub-models and create scheduled tasks that persist across restarts.
• Enterprise-Ready: Includes a built-in admin console for user/role management, RBAC for gating tool access, and comprehensive logging.