simple

MCP server from prashantkul/simple-mcp-server

Installation

Run this command in your terminal to add the MCP server to Claude Code.

Run in terminal:

Command

claude mcp add --transport stdio prashantkul-simple-mcp-server python -m mcp_server \
  --env NGROK_TOKEN="your-ngrok-token" \
  --env SERVER_PORT="8000"

How to use

This educational MCP server demo provides a Flask-based HTTP server with SSE streaming that implements a simple Customer Management System. It exposes six MCP tools for managing customers: get_customer, list_customers, add_customer, update_customer, disable_customer, and activate_customer. The notebook-based setup demonstrates building the server, connecting to an SQLite database, and testing the tools using MCP protocol flows (JSON-RPC 2.0) alongside Server-Sent Events for streaming responses. The accompanying Google ADK agent notebook shows how to auto-discover and interact with these MCP tools using natural language, enabling conversational data access to the customer store. To try it locally, run the server from the notebook (Cell 11 as described) and use MCP Inspector to interact with the tools via the provided public URL (through ngrok).

How to install

Prerequisites:

Python 3.9+ installed on your machine
pip (Python package installer) available in PATH
SQLite (usually bundled with Python as sqlite3)
ngrok or a public tunneling alternative (for public URL) and an ngrok auth token

Installation steps:

Create a Python virtual environment (optional but recommended): python -m venv venv source venv/bin/activate # Linux/macOS venv\Scripts\activate # Windows
Install required Python packages (Flask, sqlite3 is built-in, PyNgrok for ngrok integration): pip install flask pyngrok
Prepare the MCP server module placeholder (as used in this demo):
- Ensure you have a module named mcp_server.py or adjust the -m module_name in the run command accordingly
- The notebook will provide the actual server implementation and wiring for Flask + SSE
Start the server (as shown in the notebook cells): python -m mcp_server
If using ngrok for a public URL, set your NGROK_TOKEN environment variable and run ngrok: ngrok authtoken <your_token> ngrok http 8000
Open the public URL printed by ngrok and append /mcp to reach the MCP endpoint (e.g., https://<id>.ngrok.io/mcp).

Note: Because this project is demonstrated within notebooks, the exact server module and wiring are provided in the notebooks. Adapt the module name in the -m flag to match your local setup if you implement the server differently.

Additional notes

Tips and common issues:

Ensure Python 3.9+ compatibility and install required packages exactly as listed.
If the MCP Inspector cannot connect, verify the server is running and that the public URL (via ngrok) is accessible from your browser.
For local testing, you can bypass ngrok by using localhost directly and pointing the MCP Inspector to http://localhost:8000/mcp.
The six tools rely on an SQLite database; if you modify schemas, remember to migrate or recreate the test data accordingly.
Environment variables like NGROK_TOKEN and SERVER_PORT can be tuned in your run configuration to fit your environment.
The notebooks also demonstrate proper error handling and validation within the tool implementations.