Keywords​

AI suggested code, quizzes, educational, leetcode, practice

Project Abstract​

AutoSuggestion Quiz is a web-based educational application that helps students develop critical thinking skills when working with AI-generated code commonly found in modern IDEs. Teachers create and upload quizzes, and students can write and run code directly in the platform while answering by selecting from multiple-choice code suggestions. After submission, teachers can score responses and provide feedback. This format encourages students to question AI-generated code and builds confidence in tackling complex programming problems.

High Level Requirement​

Describe the requirements – i.e., what the product does and how it does it from a user point of view – at a high level.

Conceptual Design​

The frontend is built with React (Create React App), providing a Monaco editor-based coding environment where students write and run Python code directly in the browser using Pyodide. The backend is implemented in Python using FastAPI, which handles teacher authentication, problem management, AI suggestion generation, and submission storage. The database is PostgreSQL hosted on Supabase. AI code suggestions are generated via the OpenAI API (GPT-4o mini). OTP-based teacher authentication is handled through Supabase's email auth service.

Background​

AutoSuggestion Quiz (ASQ) was created to address the problem of students accepting AI-suggested code without fully understanding it. Many students rely on these suggestions because they are unsure how to approach LeetCode-style questions, which can often be challenging.

ASQ helps solve this issue by presenting students with options for the next line of auto-suggested code. This gives them direction while encouraging them to think critically about how to solve the problem and evaluate AI-generated suggestions. After a code option is selected, an explanation is provided to clarify why the AI suggested that code.

ASQ also supports teachers by allowing them to upload problems and give students opportunities to practice solving challenging coding tasks. In this way, ASQ helps students build confidence, improve problem-solving skills, and develop a deeper understanding of the code they use.

Required Resources​

• Windows 11 / macOS
• Internet access
• Node.js v16+
• Python 3.10+
• A free Supabase account (PostgreSQL database and OTP auth)
• An OpenAI API key for AI code suggestions

Running Locally​

For full setup instructions including Supabase configuration, schema SQL, environment variables, and how to run all tests, see LOCAL_SETUP.md.

Backend​

cd backend
python3 -m venv venv
source venv/bin/activate        # macOS/Linux
# venv\Scripts\activate         # Windows
pip install -r requirements.txt

Environment variables

Copy the template and fill in your values:

cp backend/.env.example backend/.env

Then edit backend/.env:

Start the server

uvicorn main:app --reload

The API runs at http://localhost:8000. The database is created and seeded automatically on first startup — no manual migration needed. A seed teacher (seed@autoquiz.dev) and a sample problem (access code 123456) are added when the database is empty.

Self-Hosted Compiler With Docker​

The backend uses a local Judge0 CE instance for code execution. This keeps the existing /code/execute backend endpoint the same while moving compilation and execution onto your machine.

Docker Desktop must already be installed and running before you start Judge0. The helper scripts manage the local Judge0 stack, but they do not install Docker for you.

Start Judge0 locally:

./scripts/start_judge0.sh

Then set JUDGE0_URL=http://localhost:2358 in backend/.env and start the backend normally.

Stop Judge0 when you are done:

./scripts/stop_judge0.sh

If Judge0 starts successfully but code execution still fails, reset the local Judge0 volumes and start again:

./scripts/stop_judge0.sh
cd .judge0/judge0-v1.13.1 && docker compose down -v
cd ../..
./scripts/start_judge0.sh

Mac​

For macOS users setting up Judge0: You may need to change a Docker settings file if Docker and Judge0 are incompatible on your machine.

To start, close docker and every instance of it currently running.

You will need to locate the settings-store.json for docker in order to use the deprecatedCgroupv1.

In Terminal:

open -a TextEdit "/Users/[Your Computer Name]/Library/Group Containers/group.com.docker/settings-store.json"

Add 'deprecatedCgroupv1: true' at the bottom of the json. It should look like this:

{
  "AutoStart": false,
  "DisplayedOnboarding": true,
  "DockerAppLaunchPath": "/Applications/Docker.app",
  "EnableDockerAI": true,
  "LastContainerdSnapshotterEnable": 1773866500,
  "LicenseTermsVersion": 2,
  "SettingsVersion": 43,
  "ShowInstallScreen": false,
  "UseContainerdSnapshotter": true,
  "deprecatedCgroupv1": true
}

Reopen Docker, and the app should be able to reach Judge0.

Windows​

For Windows user, trying to run the script will prompt a choice between Git Windows and VS Studio Code. Pick Git Windows, and do not click "Always run this program".

Frontend​

cd frontend
npm install

Optional: If your backend is not on the default port, create frontend/.env and set REACT_APP_API_URL=http://localhost:<your_port>. Otherwise no .env is needed.

Start the app

npm start

The app opens at http://localhost:3000.

Supabase / OTP Login​

All Supabase communication happens on the backend — the frontend does not need any Supabase credentials. To enable OTP email login:

1. Create a free account at supabase.com.
2. Create a new project.
3. Go to Project Settings → API.
4. Copy the Project URL → paste as SUPABASE_URL in backend/.env.
5. Copy the anon / public key → paste as SUPABASE_SERVICE_KEY in backend/.env.

Dev bypass — no Supabase account needed

On the login page, enter dev as the email and click Send OTP. This skips Supabase entirely and signs you in as the seed teacher (seed@autoquiz.dev) without sending any email. This only works when DEBUG=True is set in backend/.env.

Testing​

Backend​

Make sure the virtual environment is activated, then run:

cd backend
source venv/bin/activate        # macOS/Linux
# venv\Scripts\activate         # Windows
pytest backUnitTest/

Frontend​

cd frontend
npm test

Press a to run all tests.

First Release​

Live app: https://autosuggestions.onrender.com/

Features​

• Teachers can create an account by entering their email and receiving a one-time password.
• Teachers can access their dashboard and create problems by providing a problem prompt and boilerplate code for the student to start with.
• Teachers will receive a one-time password after creating a problem. Save this password to distribute to students.
• Students can access their problems by using the one-time password for the assignment (one-time password provided by the teacher).
• Students can attempt a problem and will be prompted with AI suggestions whenever they are inactive for a short period of time.
• Python is the only language available currently.

Collaborators​

Bwosley Temi Raymond Yong Huang Ethan Friedman Henry Le Aidan McCammitt