A2A-MCP Multi-AI Orchestration Demo 🚀

English

Production-grade FastAPI application implementing the complete Agent-to-Agent (A2A) protocol with Model Context Protocol (MCP) tool integration.

Key Features

• A2A Protocol: Secure signed messages (HMAC), routing, verification between agents.
• MCP Tools: Mock/demo integration for use_mcp_tool & access_mcp_resource (extendable to real BLACKBOXAI MCP servers).
• Multi-Agent Flow: Client → API → Orchestrator → Specialist → MCP → Response.
• Production Ready: Async FastAPI, Pydantic validation, structured logging (loguru), CORS, health checks, OpenAPI docs, 80%+ test coverage.
• Deployment: Docker, docker-compose (with Redis), Kubernetes-ready.
• Real-time: WebSocket /api/v1/ws/a2a for live A2A streaming.

Architecture

graph LR
    Client[Client/App] -->|POST /api/v1/orchestrate| FastAPI[FastAPI API]
    FastAPI -->|A2A Envelope| Orchestrator[Orchestrator Agent]
    Orchestrator -->|A2A Route| Specialist[Specialist Agent]
    Specialist -->|MCP Request| MCP[MCP Server/Tool]
    MCP -->|Result| Specialist
    Specialist -->|Response| Orchestrator --> FastAPI --> Client
    Redis[(Redis Queue)]

Quick Start

1. Dev Env:

   cd a2a-mcp-demo
   python -m venv venv
   venv/Scripts/activate  # Windows
   pip install -r requirements.txt
   cp app/core/.env.example .env
   uvicorn app.main:app --reload
   

   Open http://localhost:8000/docs

2. Test:

   pytest app/tests/
   

3. Docker:

   docker compose up -d
   

API Example

curl -X POST http://localhost:8000/api/v1/orchestrate \
  -H 'Content-Type: application/json' \
  -d '{"task": "analyze market data"}'

Response: A2A trace + MCP mock result.

• Scalierbar & Enterprise-Ready: Async, containerized, observability-ready (Prometheus add-on easy).
• Modern Stack: Python 3.12, FastAPI, Pydantic, Docker.
• Use Case: AI Agent Orchestration for enterprise automation (e.g., data analysis pipelines).
• Deploy to Railway.app for live demo link.
• Screenshots: Add your demo GIF here.