AI-Powered Python Debugging Assistant with Jan AI & Mistral 7B

Introduction

This Python Debugging Assistant is designed to help CS students (ages 7-18) debug their code efficiently. It provides guided step-by-step explanations, encourages critical thinking, and helps students understand Python error messages without simply giving them the answer.

Built using Jan AI and the Mistral 7B Instruct Q4 model, this assistant runs on low-end hardware and can be easily integrated into classrooms or personal study sessions.

Why Jan AI?

Many AI-powered debugging tools rely on cloud-based LLMs that store and analyze user input, which may raise privacy concerns in classrooms. Jan AI is:

✅ Fully open-source – No proprietary lock-in, fully transparent.

✅ Privacy-first – Runs locally, ensuring no student data is sent to external servers.

✅ Highly customizable – Allows full control over assistant behavior.

✅ Works offline – Unlike ChatGPT or Claude, no internet required once set up.

Quick Start Guide (Manual Setup)

Download and Install Jan AI

1. Download the current version of Jan AI for your machine

2. Follow any on-screen installation instructions.

3. Launch Jan AI once the installation is complete.

Import the Model

Option 1: Jan AI's Hugging Face Integration (Recommended)

1. Open Jan AI and go to Model Hub

2. Search for Mistral 7B Instruct Q4 and click Download & Import.

   • Alternative Models (if Mistral does not work)

     • Deepseek Coder 1.3B Instruct Q8 – Powerful lightweight coding model

     • TinyLlama Chat 1.1B Q4 – Lightweight option, ideal for low-end PCs.

   • Don't know if the model will run on your desktop? Check out these VRAM calculators:

     • https://huggingface.co/spaces/Vokturz/can-it-run-llm

     • https://huggingface.co/spaces/NyxKrage/LLM-Model-VRAM-Calculator

3. Wait for the model to be downloaded locally on your machine.

Option 2: Manually Add a Custom GGUF Model

1. Alternative download a custom GGUF model from Hugging Face.

2. Save the .gguf file in a known directory.

3. Open Jan AI, navigate to Model Hub and click Import Model.

4. Select the .gguf file and confirm the import.

Setup Instructions and Prompt Template

1. Open Jan AI and go to the Assistant Settings panel.

2. Copy and paste the following instructions for the assistant:

   You are a friendly Python debugging mentor for students aged 7–18. Use simple language, guide students step by step, and encourage them to analyze error messages. Provide hints first before giving a full solution. Use markdown for code formatting. Reference Python documentation when relevant.
   

How to Use (Testing the Assistant)

Try These Prompts

Once set up, enter the following debugging prompts into Jan AI:

• Example 1: Syntax Error

  • Expected response: The assistant should identify the missing closing parenthesis and suggest adding it.

print("Hello World"

• Example 2: NameError

  • Expected response: The assistant should explain that y is undefined and suggest defining it before printing.

x = 10
print(y)

• Example 3: Logical Error

  • Expected response: The assistant should recognize the incorrect operator and suggest replacing + with *.

def multiply(a, b):
    return a + b

print(multiply(2, 3))

Alternative Debugging Scenarios

• Try debugging JavaScript or C++ by modifying the instructions and prompt template accordingly.

• Experiment with different ways to phrase error descriptions to see how the assistant adapts.

Troubleshooting & FAQ

• Q: Jan AI doesn't recognize my model. What should I do?

  A: Ensure the model is in the correct directory and that Jan AI has access to it. Try restarting Jan AI and re-importing the model.

• Q: The assistant isn't responding as expected.

  A: Check that the instructions and prompt template are correctly configured. If needed, fine-tune responses by adjusting temperature and max token settings.

• Q: Can this be used for other languages?

  A: Yes. You can modify the instructions and prompt template to work with JavaScript, C++, or other languages.

Conclusion

The Python Debugging Assistant helps students debug their code while developing problem-solving skills. It provides structured guidance, avoids over-reliance by encouraging critical thinking, and is easily customizable for different use cases.

For improvements, feedback, or additional debugging templates, feel free to comment on this this creation!