Langgraph_A2A_MCP

# A2A_MCP Built Multi Agent Travel Chat Service System Using Langgraph, A2A, MCP Frameworks ## Overview This project demonstrates a modular, agent-based architecture for travel planning and information retrieval. It uses multiple agents (Planner, Orchestrator, Tool, Reflector) that communicate via the A2A protocol and interact with MCP servers for tool execution. ### Key Components - **Planner Agent**: Decomposes user requests into structured travel tasks. - **Orchestrator Agent**: Manages task execution and dependencies. - **Tool Agent**: Selects and calls the appropriate MCP server/tool (e.g., flight, bus, sightseeing). - **Reflector Agent**: Summarizes tool results into a user-friendly answer. - **MCP Servers**: Provide endpoints for transport, sightseeing, and employee information. ## Directory Structure - `agents/`: Agent implementations. - `mcp_servers/`: Example MCP servers (transport, sightseeing, employee). - `agent_executor.py`: A2A-compliant agent executors. - `main.py`: CLI to launch agent servers. - `test_client.py`, `testclientPrevious.py`: Example clients for end-to-end testing. ## Getting Started 1. **Install dependencies:** ```sh pip install -r requirements.txt ``` 2. **Set environment variables:** Edit `.env` with your API keys and model settings. 3. **Start MCP servers:** ```sh python mcp_servers/transport_server.py python mcp_servers/sightseeing_server.py python mcp_servers/employye.py ``` 4. **Start all agents:** ```sh python main.py all_agents ``` 5. **Run a test client:** ```sh python test_client.py ``` ## Example Usage - **User Input:** `Plan a trip from Paris to Rome with sightseeing.` - **System Output:** The system plans the trip, orchestrates tool calls, and summarizes the results for the user. ## Requirements - Python 3.9+ - See [requirements.txt](requirements.txt) for dependencies. --- For more details, see the code and docstrings in each agent and server file.