Tether

Real-time AI learning intervention system that observes behavior, detects risk patterns, and intervenes live with context-aware guidance.

The Problem

Procrastination is a huge unsolved problem that affects many high school and college students today. Most learning tools are passive and only react after performance drops.

Tether is built to intervene in the moment.

What Tether Does

Tether monitors live work behavior across coding, studying, writing, and browsing contexts, then detects:

• distraction / inactivity
• procrastination patterns

When risk is detected, it triggers actionable interventions in real time.

Why This Is Different

Tether is not a passive tutor chatbot.

It is a behavior-first intervention system that:

• senses what is happening now
• explains why risk is increasing
• proposes the best next action
• tracks whether the action improved outcomes

Core Features

• Context-aware behavior tracking:
  • typing speed
  • pause duration
  • idle duration
  • repeated actions / edit churn
  • tab switches
  • scroll behavior
  • time on task
• Strict inactivity detection:
  • 60s first inactivity trigger
  • ~150s follow-up inactivity threshold (2nd+ trigger)
  • "Distraction / Inactivity" intervention
  • 3-minute follow-up reminder when ignored
• Live intervention actions:
  • Lock In (2 min focus)
  • Resume Task
  • Ignore
  • Break into Steps
  • Try New Approach
• Grades & Risk module:
  • baseline grade + 6 course-grade slots
  • risk scoring and trend
  • outcome prediction
  • recovery actions
• Browser notification controls:
  • notification mode (Normal / Focus / Chill)
  • reminder log
  • browser permission status
• Session dashboard:
  • intervention history
  • timeline
  • focus/risk improvement metrics

Tech Stack

• Frontend: React + Vite + Recharts
• Backend: Node.js + Express + OpenAI SDK
• Extension: Chrome Extension (Manifest V3)
• Runtime storage: in-memory/local storage (hackathon optimized)

Project Structure

Tether/
  backend/
    src/
      server.js
      detectionEngine.js
      aiService.js
      knowledgeGraph.js
      sessionStore.js
  frontend/
    src/
      pages/
        WorkspacePage.jsx
        DashboardPage.jsx
      lib/api.js
      styles.css
    vercel.json
  extension/
    manifest.json
    icons/
      nudge-128.png
    src/
      background.js
      content.js
      popup.html
      popup.js
  .env.example
  package.json

Quick Start

1. Install dependencies

npm install

2. Configure environment

cp .env.example .env

Set at least:

• OPENAI_API_KEY (optional, recommended)
• OPENAI_MODEL (default gpt-4.1-mini)
• PORT (default 8787)
• CLIENT_ORIGIN (default http://localhost:5173)
• VITE_API_BASE (default http://localhost:8787)

3. Run frontend + backend

npm run dev

• Frontend: http://localhost:5173
• Backend: http://localhost:8787

Demo Flow

1. Start a session from the homepage.
2. Work normally for 20–30 seconds.
3. Stop all activity for ~60 seconds.
4. Observe "Distraction / Inactivity" intervention + notification.
5. Click Lock In (2 min focus).
6. Watch risk/focus improve in timeline and dashboard.

Backend API Snapshot

• GET /api/health
• GET /api/context/profiles
• POST /api/session/start
• POST /api/session/:sessionId/metrics
• POST /api/session/:sessionId/intervention-action
• POST /api/session/:sessionId/end
• GET /api/session/:sessionId/summary

Chrome Extension MVP

The extension in extension/ can:

• monitor behavior directly on websites,
• run live detection in a background worker,
• show interventions in an on-page floating widget,
• show metrics/timeline in popup view.

Load Extension Locally

1. Open chrome://extensions
2. Turn on Developer mode
3. Click Load unpacked
4. Select the extension folder
5. Reload any target tab

Notification Notes

• Browser notifications are real when permission is granted.
• Reminder sequence supports 1st alert immediately, then follow-ups at 3 and 8 minutes.