Skip to content

part2-study-buddy.md

Latest commit

 

History

History
173 lines (127 loc) · 7.56 KB

part2-study-buddy.md

File metadata and controls

173 lines (127 loc) · 7.56 KB

Part 2: Using MCP Servers - Building Python Study Buddy App

Overview

In this section, you'll learn how to use MCP servers by building an interactive Python learning companion that helps developers at different skill levels master Python concepts. This hands-on tutorial will teach you how MCP servers work from a consumer perspective before you build your own advanced server.

study-buddy-app

Step-by-Step Walkthrough

2.1 Project Setup

If you haven't already set up your python environment:

2.1.1. Visual Studio Code

  • Download and install VS Code
  • Essential for MCP development and integration

2.1.2. Python 3.12+

  • Install Python 3.12 or later from Python.org
  • Verify installation: python --version or python3 --version
  • Ensure pip is installed: pip --version or pip3 --version

2.1.3. Python Extension for VS Code

  • Install the Python extension
  • Provides comprehensive Python development support
  • Includes IntelliSense, debugging, and virtual environment management

2.1.4. Install UV

To install UV, run the following command in the terminal:

pip install uv

2.1.5. Create Virtual Environment

# Using venv (recommended)
python -m venv mcp-env

# Activate on macOS/Linux
source mcp-env/bin/activate

# Activate on Windows
mcp-env\Scripts\activate

2.1.6. Install packages

uv sync

2.2 MCP Server Configuration and Initial Setup

Configure Python Learning MCP

  • Install and configure a Python Learning MCP server in VS Code
  • to get the path to uv, run which uv

Open the .vscode folder and make sure you see a file named mcp.json with the following content:

"learnpython-mcp": {
         "command": "{path-to-uv}",
         "args": [
            "--directory",
            ".",
            "run",
            "server_part_2.py"
         ]
      }

2.2 Learning about Prompts

  • To get started we will learn how to use MCP prompts. To do this, edit the server in the mcp.json file to be prompts_server.py instead of server_part_2.py.
  • Next run the play button to start the server.
  • Examine the prompts_server.py file and note the @mcp.prompts decorator. This is the prompt we will be looking for in copilot.
  • In Copilot chat type \ and look for the prompt that you saw in the previous step.
  • Add the beginner variable to generate a beginner list of topics and press enter.
  • Once the prompt shows up in your terminal press enter and note the result!

Key Learning Points:

  • Prompts are great for storing information
  • Prompts can be dynamic and take variables
  • Prompts need to explicilty be called by the user unless sampling is done.

2.3 Learning about Tools and Sampling

  • To learn about MCP tools, edit the server in the mcp.json file to be tools_server.py instead of prompts_server.py.
  • Next run the play button to start the server.
  • Examine the tools_server.py file and note the @mcp.tool() decorator. These are the tools we will be using in Copilot.
  • In Copilot chat, you can now use the tools to generate exercises. Try asking: "Use the generate_and_create_exercises tool to create 5 beginner Python exercises about variables"
  • Notice how this tool uses sampling - it calls the LLM to generate the content and then processes the result
  • You can also use the list_exercises tool to see all created exercises
  • Try creating exercises for different levels (beginner, intermediate, advanced) and topics

Key Learning Points:

  • Tools extend what LLMs can do by calling functions
  • Sampling allows tools to use the LLM's capabilities within the tool execution
  • Tools can work together to create more complex workflows

2.4 Learning about Resources

  • To learn about MCP resources, edit the server in the mcp.json file to be resources_server.py instead of tools_server.py.
  • Next run the play button to start the server.
  • Examine the resources_server.py file and note the @mcp.resource() decorator. These are file-like data sources that can be read by clients.
  • Resources in this server provide access to:
    • user://study-progress/{username} - Get study progress for a specific user
    • user://exercises/{level} - List exercises available for a specific level
  • You can also use the get_users_progress tool that combines resources with sampling to provide intelligent analysis
  • Try asking Copilot to "Get the study progress for Marlene" and see how it accesses the resource
  • Resources are different from tools - they provide data that can be read, while tools perform actions

Key Learning Points:

  • Resources provide structured data access to LLMs
  • Resources can be dynamically accessed with parameters (like username or level)
  • Resources enable LLMs to read current state and context information

2.5 Running the Study Buddy App

Now let's use the full Study Buddy application with all MCP features integrated:

  • Edit the server in the mcp.json file back to server_part_2.py (this is the complete server with all features)
  • Run the play button to start the server
  • The server_part_2.py includes:
    • Prompts: python_topics for generating topic lists
    • Tools: generate_exercises, create_exercise, track_study_progress, start_study_buddy
    • Resources: Access to exercises and progress data
  • Open the .Github folder and read the copilot-instructions.md file. This will give copilot instructions on the workflow! This makes the experience more predictable.

To run the complete Study Buddy experience:

  1. Start Study Session: Tell Copilot in the chat that you want to learn Python and then follow the instructions.

What you'll experience:

  • 🐍 Welcome screen with your personalized progress
  • 📚 Menu of available exercises at your level
  • 💡 Hints and solutions for each exercise
  • ✅ Progress tracking as you complete exercises
  • 🎯 Achievement system to motivate continued learning

When the final command is run by copilot in the terminal you should observe the following in the terminal:

Sample Study Session Output:

🐍 Python Study Buddy

Welcome, CodeLearner! 
Ready to practice Python at the beginner level?

Current Streak: 🔥 0 days
Exercises Completed: ✅ 0

Available Beginner Exercises
┌────┬─────────────────┬────────────┬──────────────┐
│ No.│ Title           │ Difficulty │ Status       │
├────┼─────────────────┼────────────┼──────────────┤
│ 1  │ Hello World     │ ⭐         │ 📝 Not started │
│ 2  │ Variables       │ ⭐⭐       │ 📝 Not started │
└────┴─────────────────┴────────────┴──────────────┘

Choose an exercise (number): 1

The Study Buddy app demonstrates the full power of MCP by combining prompts, tools, resources, and sampling to create a personalized, interactive learning experience!

Next Step: Now that you've built a learning application with MCP, let's create an advanced MCP server for AI research discovery!

Continue to: Part 3 - Building Your AI Research Learning Hub →