Ashok54942/MCP-server

EduChain: AI-Powered Educational Content Server (MCP Compliant)

This project demonstrates an AI-powered server built with Flask and Google's Gemini AI, designed to generate and retrieve educational content. It adheres to the principles of the Multi-Agent Collaboration Protocol (MCP), allowing integration with agentic hosts like Claude Desktop.

Project Overview

EduChain provides the following core functionalities:

1. AI-Powered Lesson Plan Generation: Dynamically creates comprehensive lesson plans for any given topic.
2. AI-Powered Multiple-Choice Question (MCQ) Generation: Generates MCQs on demand for specified topics.
3. Predefined MCQ Retrieval: Serves a curated set of multiple-choice questions from a static JSON file, with optional subject-based filtering.

The server is built using Python with Flask, LangChain for AI orchestration, and Pydantic for data validation.

Task Breakdown & Completion

This repository covers the completion of the following tasks:

Task 1: Set Up the EduChain Environment

• Description: Installation of necessary libraries and initial configuration to generate educational content using a sample topic.
• Completion: The generate_content.py script demonstrates the initial setup and content generation using the Gemini API.

Task 2: Build the MCP Server with EduChain

• Description: Creation of a Flask-based MCP server exposing AI-powered tools (MCQ generation) and resources (lesson plan generation, predefined MCQ retrieval).
• Completion: The educhain_mcp_server.py file implements the Flask server with endpoints for /generate_lesson_plan, /generate_mcqs, and /get_predefined_mcqs. The predefined_mcqs.json file holds the static MCQ data.

Task 3: Test the MCP Server with Claude Desktop

• Description: Configuration of Claude Desktop to connect to the MCP server and verification of content retrieval via commands.
• Completion: The claude_desktop_config.json (hypothetical, as actual configuration depends on Claude Desktop's internal structure) outlines how the server's endpoints would be registered as external tools/resources. The mcp_server_test_results.txt file contains sample commands and expected responses, simulating the interaction with an MCP host.

Repository Contents

• educhain_mcp_server.py: The main Flask application that serves as the MCP server.
• generate_content.py: Initial script to demonstrate basic AI content generation.
• generate_python_mcqs.py: A client script to programmatically request MCQs from the educhain_mcp_server.py.
• predefined_mcqs.json: A JSON file containing a sample set of 20 pre-written MCQs across various subjects.
• .env.example: An example file for environment variables (rename to .env and fill in your API key).
• claude_desktop_config.json: A hypothetical configuration file demonstrating how Claude Desktop would be configured to use this server. Note: The actual location and exact structure might vary based on your Claude Desktop installation.
• mcp_server_test_results.txt: A document detailing the test commands and their corresponding server responses for verification.

Setup and Running

Prerequisites

• Python 3.8+
• pip (Python package installer)
• A Google Gemini API Key (get one from Google AI Studio)
• Claude Desktop (for Task 3 testing)

Installation

1. Clone the repository:

   git clone <YOUR_REPOSITORY_URL>
   cd educhain
   

2. Create a Python Virtual Environment (recommended):

   python -m venv venv
   

3. Activate the Virtual Environment:
   • Windows: .\venv\Scripts\activate
   • macOS/Linux: source venv/bin/activate
4. Install Dependencies:

   pip install Flask python-dotenv langchain-google-genai langchain pydantic requests
   

5. Configure API Key:
   • Rename .env.example to .env.
   • Open .env and replace YOUR_GEMINI_API_KEY with your actual Google Gemini API Key:

     GEMINI_API_KEY="YOUR_GEMINI_API_KEY"
     

Running the EduChain MCP Server

1. Ensure your virtual environment is activated.
2. Run the server:

   python educhain_mcp_server.py
   

   Leave this terminal window open and running. The server will be accessible at http://127.0.0.1:5000/.

Testing the Server (without Claude Desktop)

Open a NEW terminal window, activate your virtual environment, and use curl or the provided Python scripts:

• View Home Page: Open http://127.0.0.1:5000/ in your web browser.
• Generate Lesson Plan:

  curl -X POST -H "Content-Type: application/json" -d "{\"topic\": \"Introduction to Quantum Physics\"}" [http://127.0.0.1:5000/generate_lesson_plan](http://127.0.0.1:5000/generate_lesson_plan)
  

• Generate MCQs:

  curl -X POST -H "Content-Type: application/json" -d "{\"topic\": \"Data Structures in Python\", \"num_questions\": 5}" [http://127.0.0.1:5000/generate_mcqs](http://127.0.0.1:5000/generate_mcqs)
  

• Get Predefined MCQs (all):

  curl [http://127.0.0.1:5000/get_predefined_mcqs](http://127.0.0.1:5000/get_predefined_mcqs)
  

• Get Predefined MCQs (filtered by subject):

  curl [http://127.0.0.1:5000/get_predefined_mcqs?subject=Biology](http://127.0.0.1:5000/get_predefined_mcqs?subject=Biology)
  

• Run generate_python_mcqs.py:

  python generate_python_mcqs.py
  

  This will generate python_programming_basics_mcqs.json.

Integrating with Claude Desktop

1. Locate Claude Desktop Config:
   • Windows: %APPDATA%\Claude\claude_desktop_config.json
   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Create the file if it doesn't exist.
2. Add Server Configuration: Copy the content from the claude_desktop_config.json file in this repository and paste it into your Claude Desktop's configuration file.
3. Restart Claude Desktop: Ensure Claude Desktop is completely quit and then relaunched to load the new configuration.
4. Issue Commands in Claude Desktop:
   • Generate 5 multiple-choice questions on Python loops.
   • Provide a lesson plan for teaching algebra.
   • Get predefined MCQs for Geography.
   • Observe Claude's responses and check your server's terminal for logs.

---