JobRight AI - Chrome Extension

AI-powered auto-fill extension for job applications. Automatically detects form fields, matches them to your profile data, and suggests answers with confidence scores.

Features

• Smart Form Detection: Automatically detects form fields and extracts question text from labels, placeholders, and surrounding context
• AI-Powered Matching: Matches form questions to your profile data using intelligent pattern matching and snippet retrieval
• Multiple Profiles: Create and manage multiple profiles for different job types or career stages
• Confidence Scoring: Each suggestion includes a confidence score and explanation
• Safety First: Blocks auto-fill for sensitive fields (SSN, passwords, etc.)
• Local-First: All data stored locally in your browser by default
• Privacy-Focused: Encrypted storage for sensitive profile data

Installation

Development Setup

1. Install dependencies:

   npm install

2. Build the extension:

   npm run build

3. Load in Chrome:

   • Open Chrome and navigate to chrome://extensions/
   • Enable "Developer mode" (toggle in top right)
   • Click "Load unpacked"
   • Select the extension for job application directory

4. Add Icons:

   • Create an icons folder in the project root
   • Add three PNG files: icon16.png, icon48.png, and icon128.png
   • Or use placeholder icons (16x16, 48x48, 128x128 pixels)

Usage

First Time Setup

1. Click the JobRight AI icon in your Chrome toolbar
2. Click "Manage Profiles" to create your first profile
3. Fill in your information:
   • Name, email, phone, location
   • Custom fields (e.g., current_title, experience_years, expected_salary)
   • Cover letter snippets for common questions
4. Select your profile as the active profile

Using on Job Applications

1. Navigate to a job application page
2. The extension will automatically scan for forms
3. A panel will appear on the right side showing suggestions
4. Review each suggestion:
   • Green background: High confidence (>85%), safe to auto-fill
   • Yellow background: Sensitive field, requires review
   • Gray background: Lower confidence, review before accepting
5. Click "Accept" for individual fields or "Accept All Safe Fields" for bulk fill
6. Edit any field before accepting if needed

Project Structure

.
├── src/
│   ├── types.ts              # TypeScript type definitions
│   ├── content.ts            # Content script (runs on pages)
│   ├── background.ts         # Service worker (background tasks)
│   ├── utils/
│   │   ├── storage.ts        # IndexedDB storage utilities
│   │   ├── encryption.ts     # WebCrypto encryption
│   │   └── formDetector.ts   # Form detection logic
│   └── services/
│       └── aiMatcher.ts      # AI matching service
├── popup.html/js             # Extension popup UI
├── options.html/js           # Settings page
├── manifest.json             # Extension manifest
├── tsconfig.json             # TypeScript configuration
└── package.json              # Dependencies and scripts

Development

Build Commands

• npm run build - Build TypeScript to JavaScript
• npm run watch - Watch mode for development
• npm run copy-assets - Copy assets to dist folder

Architecture

• Content Script (content.ts): Runs on every page, detects forms, shows overlay UI
• Background Service Worker (background.ts): Handles storage, message routing, settings
• Storage Layer (storage.ts): IndexedDB operations for profiles, mappings, logs
• AI Matcher (aiMatcher.ts): Matching logic (currently local algorithm, ready for LLM integration)

Adding LLM Integration

The extension is designed to work with LLM APIs. To integrate:

1. Update src/services/aiMatcher.ts:

   • Replace localMatch() with an API call to your LLM provider
   • Use the buildUserPrompt() function to format requests
   • Parse structured JSON responses

2. Add API key management:

   • Store API keys securely in chrome.storage.local
   • Add settings UI for API key configuration

3. Example LLM call:

   const response = await fetch('https://api.openai.com/v1/chat/completions', {
     method: 'POST',
     headers: {
       'Authorization': `Bearer ${apiKey}`,
       'Content-Type': 'application/json'
     },
     body: JSON.stringify({
       model: 'gpt-4',
       messages: [
         { role: 'system', content: SYSTEM_PROMPT },
         { role: 'user', content: userPrompt }
       ],
       response_format: { type: 'json_object' }
     })
   });

Claude Code CLI Provider (no API key)

The extension can use your locally installed Claude Code CLI instead of a paid API, via Chrome native messaging:

1. Load the extension and copy its ID from chrome://extensions (Developer mode on)
2. Run: ./native-host/install.sh <extension-id>
3. Restart Chrome, then pick Claude Code CLI as the AI provider in Options → AI Settings

Requests run through claude -p on your machine — covered by your Claude subscription, no per-token billing.

Privacy & Security

• Local Storage: All profile data stored locally in IndexedDB
• Encryption: Sensitive fields encrypted using WebCrypto API
• No Auto-Submit by Default: Requires explicit user confirmation
• Sensitive Field Detection: Automatically blocks auto-fill for SSN, passwords, etc.
• No Data Collection: Extension does not send data to external servers (unless you configure LLM API)

Limitations & Future Work

Current MVP Features

• ✅ Local pattern matching
• ✅ Basic form detection
• ✅ Profile management
• ✅ Overlay UI with suggestions
• ✅ Safety checks for sensitive fields

Implemented Enhancements ✅

• LLM integration - Structure in place with OpenAI API support (configure in Advanced settings)
• Embedding-based snippet retrieval - Enhanced local matching + optional embedding API support
• Form mapping persistence - Save mappings per domain, auto-apply on return visits
• Application history and audit logs - Track all filled applications with history UI
• Multi-page form support - State persistence across navigation, SPA support
• Shadow DOM support - Detects form fields in Shadow DOM and web components
• File upload handling - Detects resume fields and provides user guidance
• Cloud sync structure - Optional encrypted cloud sync framework (ready for backend)

New Features Added

• Save Mapping button on each suggestion to remember field patterns
• History tab in options page to view application logs
• Advanced settings for LLM and embedding API configuration
• Enhanced snippet matching with better similarity scoring
• Form state persistence across page navigation
• Shadow DOM traversal for modern web apps

License

MIT License - feel free to use and modify for your needs.

Contributing

This is a prototype/MVP. Contributions welcome! Areas for improvement:

• Better form detection heuristics
• LLM integration examples
• UI/UX enhancements
• Test coverage
• Documentation

Support

For issues or questions, please open an issue in the repository.