privateGPT MCP Server

# PrivateGPT ChatBot Agent ## Overview The **PrivateGPT ChatBot Agent** is a crucial component of the Fujitsu PrivateGPT multi-agent system. It acts as a client that communicates with a private GPT server to respond to queries and provide information. This agent leverages both local knowledge databases and server-based responses, employing **FIPA ACL (Agent Communication Language)** to standardize communications within the system. --- ## Features - **Multi-Agent Communication via FIPA ACL:** - Handles structured messages with standardized fields such as `performative`, `sender`, `receiver`, `ontology`, and `content`. - Sends FIPA ACL failure messages if a connection to the MCP server cannot be established. - **Production WSGI Servers: Waitress & Gunicorn:** - Manages RESTful endpoints (e.g., `/ask`, `/logs`, `/status`) using Flask. - Uses Waitress for serving requests in Windows environments and Gunicorn for Unix-based systems, ensuring readiness for production environments. - Implements Cross-Origin Resource Sharing (CORS) to allow requests from any domain (`origins: "*"`) for broad compatibility. - **Authentication:** - All endpoints, except for `OPTIONS` and `/status`, require an API key sent via the `X-API-KEY` header. - **MCP Server Connectivity:** - Ensures connectivity to the MCP server (as defined in `config.json`) is validated before processing requests. - **Logging:** - Maintains detailed logs for both the agent operations (`agent.log`) and the Flask server (`flask.log`). ## Prerequisites - **Python:** Version 3.8 or higher - **Dependencies:** Install the required packages listed in the corresponding `requirements.txt` which includes Flask, waitress (for Windows), gunicorn (for Unix-based systems), flask-cors, paho-mqtt, paramiko, etc. --- ## Setup 1. **Clone the Repository:** ```bash git clone https://github.com/pgpt-dev/MCP-Server-for-MAS-Developments.git cd MCP-Server-for-MAS-Developments ``` 2. **(Optional) Create and Activate a Virtual Environment:** - **Windows:** ```bash python -m venv venv .\venv\Scripts\activate ``` - **Unix/MacOS:** ```bash python -m venv venv source venv/bin/activate ``` 3. **Install Dependencies:** ```bash pip install -r agents/ChatBotAgent/requirements.txt ``` 4. **Configure the Agent:** Copy the example configuration file and adjust it according to your environment: ```bash cp agents/ChatBotAgent/config.json.example agents/ChatBotAgent/config.json ``` **Example `config.json`:** ```json { "email": "<YOUR EMAIL>", "password": "<YOUR PASSWORD>", "api_ip": "0.0.0.0", "api_port": 5001, "api_key": "<YOUR_API_KEY>", "mcp_server": { "host": "172.24.123.123", "port": 5000 }, "language": "en", "groups": ["<Your Group>"] } ``` **Note:** All sensitive parameters, such as `email` and `password`, should be securely stored and managed. ## Running the Agent - **Using Waitress (Windows):** To start the ChatBot Agent, ensure you are in the repository's root directory and execute: ```bash python -m agents.ChatBotAgent.Python.chatbot_agent ``` - **Using Gunicorn (Unix-based systems):** To start the application using Gunicorn, run the following command: ```bash gunicorn -w 4 -b 0.0.0.0:5001 agents.ChatBotAgent.Python.chatbot_agent:app ``` This command launches the Flask API server using Gunicorn on the configured `api_ip` and `api_port` and uses multiple workers to handle requests efficiently. The `fcntl` package is needed to run this under Linux, use `pip` to install it. --- ## API Endpoints ### `/ask` (POST) - **Purpose:** Accepts questions for the ChatBot Agent. This endpoint supports both a **FIPA ACL** structured message and a legacy JSON format. - **Authentication:** Requires the `X-API-KEY` header with the correct API key. - **FIPA ACL Request Example:** ```json { "performative": "request", "sender": "IoT_MQTT_Agent", "receiver": "Chatbot_Agent", "ontology": "fujitsu-iot-ontology", "content": { "question": "What is the system status?", "usePublic": false, "groups": ["group1"], "language": "en" } } ``` - **Legacy JSON Request Example:** ```json { "question": "What is the system status?", "usePublic": false, "groups": ["group1"], "language": "en" } ``` - **Response:** Returns a JSON object containing the answer generated by the PrivateGPT server. If the MCP server connection fails, a FIPA ACL failure message is returned. ### `/logs` (GET) - **Purpose:** Provides access to the Flask server's log file (`flask.log`) for debugging and monitoring purposes. ### `/status` (GET) - **Purpose:** Outputs a simple JSON message confirming that the agent is operational. This endpoint does **not** require authentication. --- ## Multi-Agent Communication & FIPA ACL The ChatBot Agent utilizes **FIPA ACL** to structure its communications with other agents, ensuring reliable interactions within a multi-agent system. It provides standardized message fields for consistency and effective communication. --- ## Logging & Debugging - **Agent Logs:** Key events, errors, and status updates are logged in `agent.log`. - **Flask Logs:** Logs specific to the Flask server are stored in `flask.log`. These logs are critical for troubleshooting and provide insights into the agent's operation. --- ## Example Conversation ```plaintext 🎉 Welcome to the PrivateGPT ChatBot Agent. You: What is your name? Agent: I am the PrivateGPT ChatBot Agent. You: Provide system status. Agent: The system is running normally. ``` ---