Getting Started
Everything you need to know to protect your data with Prompt Guard.
Installation
Chrome
- 1Visit the Chrome Web Store
- 2Click "Add to Chrome"
- 3The extension icon will appear in your toolbar
Firefox
- 1Visit Firefox Add-ons
- 2Click "Add to Firefox"
- 3Grant permissions when prompted
Edge
- 1Visit the Edge Add-ons Store
- 2Click "Get" to install
- 3Pin to toolbar for easy access
Basic Usage
1. Visit any supported AI platform
Navigate to ChatGPT, Claude, Gemini, or any other supported platform. Prompt Guard activates automatically.
2. Type your prompt
As you type, Prompt Guard scans for sensitive data in real-time. If something is detected, a warning indicator appears.
3. Review detected items
When you try to submit, a popup shows all detected sensitive data. You have three options:
Replace sensitive items with [REDACTED TYPE] placeholders
Submit the original text (use with caution)
Go back and modify your prompt manually
Settings & Configuration
Click the extension icon in your toolbar to access settings:
Detection Types
Toggle PII, API Keys, and Secrets detection independently
Entropy Detection (Pro)
Enable/disable Shannon entropy analysis for unknown secrets
Auto-scan
Enable/disable real-time scanning as you type
Notifications
Show/hide badge counts on the extension icon
Understanding Detection
Prompt Guard uses two methods to detect sensitive data:
Pattern Matching
Regular expressions match known formats like email addresses, phone numbers, and API keys with known prefixes.
Entropy Analysis (Pro)
Information theory identifies high-randomness strings that are likely secrets, even without known patterns.
Adding Custom Patterns
For developers: You can add custom detection patterns by editingextension/utils/detector.js
myNewPattern: {
pattern: /your-regex-here/g,
type: 'PII', // or 'API_KEY', 'SECRET'
label: 'Pattern Name',
severity: 'medium' // critical, high, medium, low
}