🧠 Study Helper — VS Code Extension for Learning Programming

Study Helper is a VS Code extension that helps students learn programming more effectively.
It detects when you’re stuck or making repeated errors and gently provides hints, documentation, or debugging tips — right inside your editor.

System Architecture

flowchart LR
    A[Signals] --> B[ML Predictive Model]
    B --> C[API Call]
    C --> D[Documentation MCP]
    D --> E[Frontend]

Loading

Project Structure

.
├── backend
│   ├── data
│   │   └── synthetic
│   │       └── training_data.csv
│   ├── database.py
│   ├── main.py
│   ├── models
│   │   └── stuck_predictor_v1.pkl
│   ├── requirements.txt
│   ├── test_api.py
│   └── test_ml.py
├── data
│   ├── feedback.jsonl
│   └── synthetic
│       └── training_data.csv
├── extension
│   ├── out
│   │   ├── extension.js
│   │   ├── extension.js.map
│   │   ├── stuckDetector.js
│   │   └── stuckDetector.js.map
│   ├── package.json
│   ├── src
│   │   ├── extension.ts
│   │   └── stuckDetector.ts
│   └── tsconfig.json
├── ml
│   ├── blackbox.py
│   ├── data
│   │   └── synthetic
│   │       ├── test_data.csv
│   │       └── training_data.csv
│   ├── features.py
│   ├── generation_script.py
│   ├── models.py
│   ├── retrain_model.py
│   └── synthetic_data_generation.py
├── models
│   └── stuck_predictor_v1.pkl
├── README.md
└── requirements.txt

Stuck Signals

• Idle time: user stops typing for a while
• Repetitive editing: repeated backspacing + retyping
• Cursor thrashing: jumping around with minimal progress
• Frequent and similar error messages
• No code execution attempts
• Tab switching (if trackable)
• Specific comments: very long comments, terms such as "todo" "fix" etc

ML integration of stuck signals

• Synthetically generated simulated data, training XGBoost model
• "Stuck?" button that logs user metrics; with each automatic popup, user confirmation on if was an accurate prediction
• Model retrained every 100 new data points

API integration

• API calls linked to MCP with documentation
• Hint hierarchy with progressively more informative hints

Frontend

• Popup with "were you stuck?" confirmation button
• Manual "Stuck?" button
• "Get answer" button with cooldown to minimize abuse
• Professor dashboard to see where studnets are stuck
• Student dashboard
• Area to input class ID (or maybe track github classroom) and upload directory for documentation for LLM to pull from

---

🚀 Features

• ⏱️ Detects inactivity
• 💡 Suggests hints and relevant documentation
• 🧭 Simple, non-intrusive popups for help
• 🔍 Customizable triggers for learning assistance
• 🌐 Integration with AI models and Model Context Protocol (MCP)

---

🧩 Example Use Case

1. You’re writing Python code and stop typing for 20 seconds.
2. The extension notices inactivity and suggests:

   “Need help? Check Python indentation syntax here.”

3. You can see a pop-up to read more or get AI-generated explanations.

---

🛠️ Installation

Now available on the VSCode Extension Marketplace!