mcp-edtech-agent

🧠 Sunbird Ed AI Assistant (MCP Protocol Integration)

This project is a prototype AI assistant integrated with the Sunbird Ed learning platform using Model Context Protocol (MCP) concepts. The goal is to enable contextual, conversational support for learners and admins, powered by LLaMA 3 (via the Groq API) and Sunbird APIs.

✅ Successfully demonstrates API tooling, agent reasoning, and installation-level context using mocked data.

For architecture workflow, please refer to: architecture.md

🚀 Project Overview

Sunbird Ed is a modular Digital Public Good (DPG) that supports personalized learning and skilling at scale. This project introduces an AI assistant that can:

• Answer user queries about course metadata, enrollments, and progress.
• Simulate Sunbird Ed behavior using mock API responses.
• Provide intelligent responses using an LLM (LLaMA 3 via the Groq API) with tool-calling logic.
• Lay the foundation for personalized learning support using MCP.

🛠 Tech Stack

• Language: Python
• Model: LLaMA 3 (via Groq API)
• MCP Protocol: Integrated conceptually, leveraging Groq and LLaMA's contextual AI capabilities. (Note: No official Python MCP Agent SDK used yet)
• API Simulation: Local mock APIs (Mockoon)
• Env Management: dotenv
• CLI Interface: Command-line interface
• IDE: Pycharm

🚧 Updates & Next Steps

• MCP SDK: MCP SDK is conceptualized and integrated manually via the tool schemas, leveraging Groq's LLaMA 3 model. Future versions may integrate a more formal MCP SDK when available.

⚙️ Setup Instructions

1. Install Requirements

   pip install -r requirements.txt
   

2. Set Environment Variables

   Create a .env file in the root directory:

   PYTHONPATH=/path/to/sunbird-ai-assistant   # Replace with your actual path
   MCP_MODEL=llama3-8b-8192             # Replace with any other model available in groq
   MCP_API_BASE=https://api.groq.com/openai/v1
   MCP_API_KEY=your_groq_api_key_here    # Paste your API key here
   

🧪 Testing Instructions (LLaMA 3 Required)

To run the project using LLaMA 3:

1. Start the Assistant: To run the assistant using Groq's hosted LLaMA 3 or other model:

   Sign up and get an API key from Groq Console Add the API key to your .env file as shown above Start the assistant:

   python cli/main.py
   

Note: PYTHONPATH is automatically appended to sys.path during runtime.

✅ Current Features

• Fetch course listings via course/v1/search (mocked)
• Retrieve user enrollment data via user/enrollment/list (mocked)
• Display user profile and progress via user/v1/profile (mocked)
• Run contextual CLI chat using LLaMA 3 with tool calling logic

🌟 Future Improvements

• 🔄 Replace mocked endpoints with real Sunbird API integration
• 🎯 Add user-level personalization and intelligent guidance
• 🧠 Incorporate official Python MCP Agent SDK (once available)
• 🌐 Launch a web-based interface (Phase 2)

🔐 Requesting API Access

Currently using mock APIs. Access to official Sunbird APIs (course/v1/search, user/v1/profile, user/enrollment/list) is requested to enable real-world deployment and validation.

📄 License

This project is part of Code4GovTech and is open-sourced under the MIT License.

👥 Contributors

• Ganasekhar Kalla – Developer & MCP Agent Integration
• Sunbird Community – API & Documentation Support