Multi-Protocol RAG Server (MCP + A2A)

May 21, 2025 ยท View on GitHub

Advanced implementation of a Retrieval Augmented Generation (RAG) server that supports both Model Context Protocol (MCP) and Google A2A protocols. This dual-protocol architecture enables flexible integration with different AI systems while maintaining a unified RAG backend.
A2AJava is swiss knife of agentic applications in Java and Kotlin. It supports multiple protocols including Google A2A, MCP, and others. This project demonstrates how to build a powerful RAG search application that integrates Google's Agent-to-Agent (A2A) protocol with Spring Boot and MongoDB Atlas.

Architecture Overview

The server implements a unique dual-protocol approach:

  1. MCP (Model Context Protocol) - Enables direct communication with MCP-compatible AI models
  2. A2A (Agent-to-Agent Java) - Provides Java-based agent communication capabilities

This architecture allows for:

  • Seamless integration with multiple AI protocols
  • Unified RAG backend serving different client types
  • Flexible deployment options
  • Protocol-specific optimizations

Screenshots and Features

A2A Protocol Integration

Agent Chat Interface

A2A Chat Interface A2A protocol enabling agent-to-agent communication

Agent Events

A2A Events Event handling in the A2A protocol implementation

Agent Addition

Add Agent Interface for adding new agents to the system

MCP Protocol Integration

Claude MCP Agent

Claude MCP Agent MCP protocol integration with Claude AI

MCP Conversation Flow

Claude MCP Talk Claude MCP Talk Extended Demonstration of MCP-based conversation flow

RAG Retrieval Process

Claude Retrieve RAG retrieval process using the MCP protocol

Key Components

  • A2ARagService: Handles A2A protocol-specific RAG operations
  • RagRetrivalController: Manages RAG retrievals across both protocols
  • YardShareA2ACardController: Controls A2A-specific card interactions
  • RAGConfiguration: Configures RAG settings for both protocols

Building and Running

To build the project:

mvn clean install

To run the server:

java -jar target/mcp-rag.jar

Configuration

The server can be configured through:

  • application.yml: Main application configuration
  • tools4ai.properties: AI-specific settings

Benefits of Dual-Protocol Architecture

  1. Protocol Flexibility: Support for both MCP and A2A allows integration with a wider range of AI systems
  2. Unified RAG Backend: Single source of truth for knowledge retrieval
  3. Scalable Design: Easy to add support for additional protocols
  4. Optimized Performance: Protocol-specific optimizations while maintaining a common core

License

See LICENSE file for details.