MCP_Host_Client_Server_SSE

This MCP server implements a host-client-server architecture over the MCP protocol with Server-Sent Events (SSE) for real-time communication. The host component integrates OpenAI to translate natural language questions into SQL queries executed against a SQLite database, returning results and natural language explanations back to the user. The server acts as an intermediary between the client and the database, orchestrating API calls to OpenAI (or compatible interfaces) and executing SQL against SQLite while streaming updates via SSE. You can run the MCP server and then connect a client to issue natural language prompts, receive incremental responses, and see the resulting SQL-backed data in real time.

To use it, start the MCP server (mcp_server.py) and run the host/client application (mcp_host.py). The client sends natural language queries, which the server forwards through the MCP framework to the host, which queries OpenAI to generate SQL, then executes the SQL against the SQLite database and streams the results back to the client via SSE. The system also logs interactions for traceability and debugging. Ensure you have a valid OpenAI API key configured to enable natural language processing and SQL generation.

Prerequisites:

• Python 3.8+
• pip (Python package manager)
• OpenAI API key

Installation steps:

1. Clone the repository and navigate into it:

git clone https://github.com/HanShan006/MCP_Host_Client_Server_SSE.git
cd MCP_Host_Client_Server_SSE

2. Create and activate a Python environment (optional but recommended):

python -m venv venv
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

3. Install dependencies:

pip install openai mcp-framework starlette uvicorn

4. Configure API key (OpenAI) by creating config.ini as described in the project or set environment variable:

# config.ini
[secrets]
API_KEY=your_api_key_here

Or set in the environment:

export API_KEY=your_api_key_here

5. Prepare the database (as included in the repo):

python create_database.py

6. Run the MCP server and host client:

python mcp_server.py

python mcp_host.py

Notes and tips:

• The configuration expects an API key for OpenAI access. If the key is not provided, requests to the OpenAI API will fail.
• The system uses SQLite; ensure the database file path (database.db by default) is writable and present.
• SSE streams will deliver incremental results; ensure your client supports EventSource or equivalent.
• If you modify config.ini, you may need to restart the MCP server to pick up changes.
• When debugging, enable logging to capture request/response traces for OpenAI and database queries.