Codex_Projects

Student Task Portal MVP

https://student-task-portal-wru2.onrender.com

A presentation-ready hackathon MVP that helps students turn a broad assignment prompt into a focused study dashboard.

---

1) Elevator pitch

Student Task Portal is a lightweight web app where a student enters a task (for example, “Write an essay on the New Deal”), optionally uploads class documents, and receives a structured dashboard with:

• Best matches
• Course materials
• Past assignments
• Reputable online sources
• Suggested study plan

The product goal is to reduce time spent searching and increase time spent learning.

---

2) Problem this solves

Students often start with an unclear prompt and scattered resources. They need to quickly answer:

1. What should I review first?
2. Which class materials are relevant?
3. Which external sources are trustworthy?

This MVP solves that by centralizing retrieval and showing organized, source-labeled results in one place.

---

3) Demo flow (what to show in a presentation)

1. Open the homepage.
2. Paste a task prompt.
3. Upload course notes or prior assignments.
4. Click Generate Study Dashboard.
5. Walk through each result section.
6. Highlight the academic-integrity warning (guidance only, no copying).

---

4) Architecture at a glance

Frontend (Vanilla HTML/CSS/JS)

• Collects task input + files
• Sends multipart request to backend
• Renders categorized result cards from JSON

Backend (Node.js + Express)

• Accepts POST /api/process-task
• Stores uploaded files via Multer
• Runs retrieval orchestration service
• Returns structured JSON payload to UI

Retrieval service

• Mock mode for offline/demo reliability
• OpenAI mode scaffold using Responses API + web search tool
• Normalizes online source links to be task-specific (avoids generic homepage links)

---

5) Codebase walkthrough

Hackathon-ready MVP web app that helps students organize what to review for an assignment or quiz.

What it does

A student enters a task such as:

• “Write an essay on the New Deal”
• “Study for a chemistry quiz on gas laws”

The app returns:

1. Best matches
2. Course materials
3. Past assignments
4. Reputable online sources
5. Suggested study plan

Tech stack

• Frontend: HTML, CSS, vanilla JavaScript
• Backend: Node.js + Express
• Upload handling: Multer
• Optional orchestration: OpenAI Responses API

Folder structure

.
├── public/
│   ├── app.js
│   ├── index.html
│   └── styles.css
├── routes/
│   └── taskRoutes.js
├── services/
│   └── retrievalService.js
├── uploads/
├── .env.example
├── package.json
├── README.md
└── server.js

Setup

1. Install dependencies:

   npm install
   

2. Create env file:

   cp .env.example .env
   

3. Start server:

   npm start
   

4. Open:

   http://localhost:3000
   

API

POST /api/process-task

Multipart form fields:

• task (required string)
• documents (optional file array)

Returns structured JSON:

• task
• mode (mock or openai)
• warning
• sections object:
  • bestMatches
  • courseMaterials
  • pastAssignments
  • onlineSources
  • suggestedStudyPlan
• adapters status object

OpenAI wiring notes

• If OPENAI_API_KEY is missing or USE_MOCK_DATA=true, the app uses mock retrieval data.
• services/retrievalService.js includes TODO markers for full OpenAI file search wiring (vector store + file indexing).
• Web search tool is scaffolded via Responses API when mock mode is disabled.

MVP guardrail

The UI includes a warning that source suggestions are for guidance only and should not be copied as final coursework.