• CLEANED
  • STRUCTURED
  • GUARDED
  • HUMAN

How to redact sensitive PII with your AI

Connect Pangea once to strip sensitive data from your inputs. Your agent handles the scrubbing so you don't leak private info to models or logs.

Ask AI about this page

Short answer

How do I redact PII from my AI prompts?

Use Pangea to automatically identify and mask personally identifiable information in text or JSON. This ensures your data stays private before it reaches an LLM or a logging service.

The results.

Where your data lands.

The AI uses Pangea to clean your data based on the format you provide.

CLEANED

Safe text strings

The AI replaces names and emails with placeholders. You get a sanitized version of your original text.

STRUCTURED

Clean JSON objects

Sensitive keys and values in JSON are scrubbed. This keeps your data structures intact but private.

GUARDED

Prompt protection

The AI applies safety rules to your prompts. This prevents sensitive info from leaking through conversational context.

HUMAN

Manual review

If the AI flags something it can't confidently scrub, it leaves the original data for you to check.

The workflow

What your AI does when the data arrives.

The agent intercepts your data and runs it through Pangea's security tools.

  1. Identify PII

    The agent scans your raw text to find names, addresses, or phone numbers.

    redact_text

  2. Scrub JSON

    For structured data, the agent targets specific fields to remove sensitive values.

    redact_structured

  3. Apply safety rules

    The agent uses specific instructions to ensure the prompt stays within your privacy bounds.

    ai_guard_prompt

  4. Verify output

    The agent checks the redacted version against the original to ensure no leaks occurred.

    redact_text

Try it

Copy these to start.

Use these prompts to trigger the redaction process.

Starting points

Swap out the bracketed text with your actual data or file names.

Pangea Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_6PGcCGWSsRTH8ftWAASrBfIUfCx2NlDoXWkeA96h/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Pangea capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "pangea-security-apis-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_6PGcCGWSsRTH8ftWAASrBfIUfCx2NlDoXWkeA96h/mcp"
    }
  }
}
Copy into chat04

  • Redact all PII from this customer support email: [paste email text]

  • Scrub the sensitive fields in this JSON object: [paste JSON]

  • Use the guard prompt to clean this prompt before I send it: [paste prompt]

  • Check this log file for any leaked email addresses: [paste log content]

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Pangea once, then ask.

Just link your Pangea account. Your credentials stay encrypted, and you can immediately start scrubbing data in your favorite AI client.

Connect Pangea to your AI

FAQ

How this task behaves.

  • 01

    Does the AI delete my original data?

    No. The tools only create a redacted copy of your data. Your original input remains unchanged.

  • 02

    Can the AI redact images?

    No. These tools only work on text and JSON objects.

  • 03

    Will the AI automatically fix my logs?

    No. You must explicitly ask the agent to redact the text or JSON from your logs.

  • 04

    Can the AI redact data in a database?

    No. This MCP only redacts the text or objects you provide in the chat context.

  • 05

    Does it work with any AI client?

    Yes, it works with any MCP-compatible client like Claude, Cursor, or Windsurf.

  • More questions about Pangea? See everything the Pangea Connector can do

Connect Pangea to Claude, Cursor, ChatGPT & more