a2a-adk-mcp-example

# AI Agent Project with Google ADK, A2A Protocol and MCP Integration This project represents an implementation of an AI agent with integration of Google ADK (Agent Development Kit), supporting A2A (Agent-to-Agent) protocol for interaction with other agents and extensible through MCP (Model Context Protocol) servers. The project includes a ready-made infrastructure for developing, testing, and deploying AI agents with monitoring and tracing. Detailed documentation on the A2A protocol: [documentation](./documentation-a2a-russian.md) and [Postman collection](./postman-collection-a2a-spec.json) ## 🎯 Features - **Google ADK Integration** - Utilization of an advanced SDK for agent development - **MCP Tools Support** - Integration with Model Context Protocol for extensible tools - **LiteLLM** - Support for various LLM models through a unified interface - **Phoenix Monitoring** - Optional monitoring and tracing of execution - **Docker Ready** - Full containerization with automatic setup - **A2A Protocol** - Agent-to-Agent communication - **Flexible Configuration** - Configuration through environment variables ## 📁 Project Structure ``` a2a-adk-mcp-example/ ├── agent/ # Agent directory │ ├── app/ # Main application code │ │ ├── __main__.py # Entry point with Click CLI │ │ ├── agent.py # Class AgentEvolution │ │ └── agent_executor.py # Agent executor │ ├── .env.example # Example configuration file │ ├── pyproject.toml # Project configuration and dependencies │ └── README.md # Agent documentation ├── mcp-weather/ # Weather MCP server │ ├── server.py # Main server │ ├── Dockerfile # Docker configuration │ └── README.md # MCP server documentation ├── nginx/ # Nginx configuration ├── docker-compose.phoenix.yml # Configuration with Phoenix ├── docker-compose.yml # Main Docker Compose configuration ├── Makefile # Command automation └── .gitignore # Ignored files ``` ## 🛠 Installation and Quick Start ### Quick Start with Docker (recommended) ```bash # Build images make build # Start main services make up # The agent will be available at http://localhost:10002 # The MCP server will be available at http://localhost:8001 ``` ### Using Makefile ```bash # View all available commands make help # Start main services make up # Start with Phoenix monitoring make phoenix # Development mode make dev # View logs make logs # Stop all services make down ``` ## ⚙️ Configuration ### Creating an Environment File ```bash # Create .env file from example make env # Or manually cp agent/.env.example .env ``` ### Main Environment Variables In the `.env` file, you can configure the following parameters: ```bash # Main agent settings AGENT_NAME=jira_mcp_agent AGENT_DESCRIPTION="Jira MCP agent for managing projects, tasks, sprints, and agile processes" AGENT_VERSION=1.0.0 # Model configuration LLM_MODEL="evolution_inference/model-for-agent-space-test" LLM_API_BASE="https://your-model-api-base-url/v1" # MCP Configuration MCP_URL=http://mcp-weather:8001/sse # Phoenix monitoring (optional) PHOENIX_PROJECT_NAME="ip_agent_adk" PHOENIX_ENDPOINT="http://phoenix:6006/v1/traces" # Server settings HOST="0.0.0.0" PORT="10002" # Monitoring ENABLE_PHOENIX="false" ENABLE_MONITORING="true" ``` ## 🚀 Usage ### Running the Agent ```bash # Start main services make up # The agent will be available at http://localhost:10002 ``` ### Development Mode ```bash # Run in development mode with live reload make dev # Run with Phoenix monitoring for debugging make dev-phoenix ``` ### Checking Status ```bash # Status of all services make status # Health check make health # View logs make logs ``` ## 🧩 Project Components ### Agent (agent/) The main component of the project is the AI agent based on Google ADK with support for MCP tools. The agent implements the A2A Protocol for interaction with other agents. #### API Endpoints - `GET /` - Information about the agent (Agent Card) - `POST /tasks` - Create a new task - `GET /tasks/{task_id}` - Get the status of a task - `GET /tasks/{task_id}/stream` - SSE stream of task execution ### Weather MCP Server (mcp-weather/) The server provides tools for obtaining weather data through the MCP protocol. It uses the free Open-Meteo API. #### Tools - `get_today_weather(city: str)` - Gets the current weather for today for the specified city - `get_weekly_forecast(city: str)` - Gets the weekly weather forecast for the specified city #### Endpoints - **SSE**: `http://localhost:8001/sse` - **Messages**: `http://localhost:8001/messages/` ## 📊 Monitoring ### Phoenix Tracing To enable Phoenix monitoring, set the environment variable: ```bash ENABLE_PHOENIX=true ``` Run with Phoenix: ```bash make phoenix ``` The Phoenix Dashboard will be available at http://localhost:6006 ## 🐳 Docker Commands ```bash # Main commands make build # Build images make up # Start services make down # Stop services make restart # Restart make logs # View logs # Phoenix monitoring make phoenix # Start with monitoring make phoenix-down # Stop Phoenix # Utilities make shell # Enter the agent container make clean # Clean up Docker resources ``` ## 📋 Requirements - **Docker**: For containerization - **Make**: For command automation - **MCP Server**: For tools (optional) ## 📄 License This project is licensed under the MIT License. See the `LICENSE` file for details (if available). ## 🔗 Useful Links - [Google ADK Documentation](https://developers.google.com/adk) - [Model Context Protocol](https://modelcontextprotocol.io/) - [LiteLLM Documentation](https://docs.litellm.ai/) - [Phoenix Tracing](https://phoenix.arize.com/)