CodeMentor AI (Chrome Extension)

CodeMentor AI is a privacy-friendly coding assistant for LeetCode and NeetCode. It provides subtle hints, detailed approaches, and quick code analysis directly on problem pages. When supported, it uses on-device AI (Gemini Nano) for low-latency, offline-friendly help, and falls back gracefully when AI is unavailable.

Features

• Smart hints and detailed solution approaches
• Code analysis of your current editor content (Monaco/CodeMirror)
• Voice commands (hint, explain, analyze, screenshot, open settings)
• Real-time code monitoring (optional)
• Progress tracking: hints, explanations, problems explored
• On-device AI via Gemini Nano when available

Requirements

• Google Chrome (or Chromium-based) with Manifest V3 support
• Optional: Gemini Nano availability in the browser (window.ai.languageModel)

Installation (Developer Mode)

1. Download or clone this repository.
2. Open Chrome and navigate to chrome://extensions/.
3. Enable "Developer mode" (top-right toggle).
4. Click "Load unpacked" and select the extension/ folder.
5. Pin the extension to your toolbar for quick access.

Permissions Used (Why)

• activeTab, tabs, scripting: Inject and communicate with content scripts
• storage: Persist your session and preferences locally
• alarms: Auto-save your session every 5 minutes
• tabCapture: Capture screenshots (for the Smart Capture feature)
• host_permissions: Restricted to leetcode.com and neetcode.io

Quick Start

1. Open a LeetCode or NeetCode problem.
2. Click the CodeMentor AI icon to open the popup.
3. Use one of the main actions:
   • Smart Hint: subtle guidance tailored to the problem
   • Detailed Approach: step-by-step solution strategy
   • Code Analysis: review the code currently in your editor
4. Try the mic button for voice commands if your browser allows microphone access.

Voice Commands (Examples)

• "Give me a hint"
• "Explain the approach"
• "Analyze my code"
• "Take a screenshot"
• "Open settings"

Settings

• Hint Level: subtle, medium, detailed
• Explanation Style: concise, step-by-step, detailed
• Toggle: voice commands, code monitoring
• Clear session data

How It Works

• content-script.js parses the problem page (title, difficulty, description, examples, constraints) and extracts code from Monaco/CodeMirror.
• background.js manages session persistence (via chrome.storage.local), autosave (via chrome.alarms), and orchestrates on-device AI prompts by messaging the content script.
• popup.html, popup.js, styles.css power the UI and interactions.

Gemini Nano Integration

• Checks window.ai.languageModel.capabilities() in page context.
• If available, creates a session to generate hints, explanations, and code feedback.
• If not available, uses robust non-AI fallbacks.

Troubleshooting

• "Please navigate to a LeetCode problem page": open a problem page before using the popup.
• Voice errors:
  • not-allowed: allow microphone access in the browser address bar
  • no-speech / audio-capture: check mic/OS audio settings
• AI unavailable:
  • You’ll see a friendly error; the extension falls back automatically
  • Ensure your Chrome supports on-device AI (Gemini Nano) and it’s enabled
• "No code detected":
  • Ensure code exists in the editor; if live extraction fails, cached code is used when available.

Development

• Load the extension/ folder via "Load unpacked" in Chrome.
• Key files:
  • manifest.json – MV3 manifest and permissions
  • content-script.js – page parsing, code extraction, AI session
  • background.js – session, autosave, messaging, AI orchestration
  • popup.html, popup.js, styles.css – UI and interactions
• Use Chrome DevTools → Applications → Service Workers to view background logs.

Tech Stack

• Languages: JavaScript (ES6+), HTML, CSS
• Platform: Chrome Extension (Manifest V3)
• Browser APIs: chrome.runtime, chrome.tabs, chrome.scripting, chrome.storage, chrome.alarms, chrome.tabCapture
• AI: Gemini Nano Prompt API (window.ai.languageModel)
• Editors: Monaco, CodeMirror (DOM-based extraction)

Roadmap

• Deeper code understanding: test case generation, complexity checks
• Personalized learning paths with adaptive difficulty
• Multi-language support and richer templates
• Optional cloud sync of session data
• Support beyond LeetCode/NeetCode

License

MIT