CloakIt

CloakIt

CloakIt scans your messages on Gen AI chat platforms and blocks sensitive data before it leaves your device. Everything runs locally — no data is ever sent to external servers.

Supported Platforms

CloakIt works on these sites:

Platform URL
ChatGPT chatgpt.com
ChatGPT (legacy) chat.openai.com
Claude claude.ai
Gemini gemini.google.com
DeepSeek chat.deepseek.com

How It Works

  1. You type a message into the chat composer on any supported platform.
  2. When you press Enter or click Send, CloakIt scans the message for sensitive data.
  3. If sensitive data is found, a modal appears showing what was detected. You then choose:
    • Cancel — return to the chat to edit your message.
    • Send Redacted — replaces detected data with [REDACTED_TYPE] placeholders and sends the sanitised message.
    • Send Anyway — sends the original message unchanged (bypasses redaction).

The modal is draggable — click and drag the header bar to reposition it.

Detection modal


Click the CloakIt icon in your browser toolbar to open the popup. It has five tabs:


1. Features Tab

Lists all built-in detection patterns with their risk levels:

Severity Detected Patterns
High Risk (red) SSN, credit card numbers, API keys (OpenAI, AWS, generic), GitHub tokens, JWT tokens, passwords
Medium Risk (yellow) Email addresses, phone numbers, dates of birth

Also summarises the three on-send actions (Cancel, Send Redacted, Send Anyway).

Features tab


2. Platform Tab

Toggle protection on or off for each supported platform individually.

This is useful if you trust a particular platform or need to send known-safe data without interruption.

Platform tab


3. Custom Tab

Add your own detection rules beyond the built-in patterns. Two types are supported:

Custom tab

Custom Text Patterns (Regex)

Define regex patterns for locale-specific or organisation-specific sensitive data. Examples:

Use Case Pattern
Indian PIN code \b\d{6}\b
Japanese phone number 0\d{1,4}-\d{1,4}-\d{4}
Employee ID EMP-\d{5}

To add a pattern:

  1. Click + Add under “Custom Text Patterns”.
  2. Enter a Label (e.g. “Indian PIN”).
  3. (Optional) Enter a Sample (e.g. “123456”) and click Generate — CloakIt will auto-create a regex from your sample.
  4. The Generated regex field can be edited manually if needed.
  5. Click Save.

Pattern Generator behaviour:

Specific Text

Add exact words or phrases to block — such as names, project codenames, or internal jargon you do not want shared with AI models.

To add a text:

  1. Click + Add under “Specific Text”.
  2. Type the text (e.g. “John Smith” or “Project Phoenix”).
  3. Click Save.

Matched items appear as BLOCKED_TEXT with High Risk severity in the detection modal.

Import / Export

Tier Limits

The free tier limits you to 1 custom pattern and 1 specific text. Items beyond the limit are stored but dimmed with an “(inactive)” label and are not scanned. Delete existing items to make room for new ones.


4. Test Tab

Preview how CloakIt detects and redacts sensitive data — offline, without sending anything.

  1. Paste any text into the text area.
  2. Click Preview Redaction.
  3. The Detected list shows every match with its type, value, and risk level.
  4. The Redacted Message pane shows the sanitised output with [REDACTED_TYPE] placeholders.

This mirrors the live scanning engine exactly. Use it to verify custom patterns or check what a message looks like after redaction.

Test tab


5. About Tab

Displays:

About tab


Detection Modal Keyboard Shortcuts

Key Action
Escape Close the modal (same as Cancel)
Tab / Shift+Tab Cycle focus between buttons and editor

Privacy

Pricing

Free

Install CloakIt

Firefox | Chrome