Presidio Node for n8n

This custom n8n node integrates Microsoft Presidio for PII (Personally Identifiable Information) detection and anonymization.

Features

The Presidio node provides three operations:

1. Analyze - Detect PII entities in text
2. Anonymize - Anonymize text using previously detected entities
3. Analyze and Anonymize - Detect and anonymize in one operation (recommended)

Installation & Setup

1. Install Dependencies

npm install

2. Build the Node

npm run build

3. Start Presidio Services

You need to have Presidio Analyzer and Anonymizer services running. The easiest way is using Docker:

# Start Presidio Analyzer
docker run -d -p 5002:5002 --name presidio-analyzer mcr.microsoft.com/presidio-analyzer

# Start Presidio Anonymizer  
docker run -d -p 5001:5001 --name presidio-anonymizer mcr.microsoft.com/presidio-anonymizer

Verify the services are running:

curl http://localhost:5002/health
curl http://localhost:5001/health

4. Start n8n Development Server

npm run dev

This will start n8n with your custom Presidio node loaded. Access n8n at http://localhost:5678

Configuration

Setting up Credentials

1. In n8n, go to Credentials > New Credential
2. Search for "Presidio API"
3. Configure the endpoints:
   • Analyzer Endpoint: http://localhost:5002 (or your Presidio Analyzer URL)
   • Anonymizer Endpoint: http://localhost:5001 (or your Presidio Anonymizer URL)
4. Save the credentials

Usage Examples

Example 1: Analyze and Anonymize (Simple)

1. Add a Manual Trigger node
2. Add the Presidio node
3. Select operation: Analyze and Anonymize
4. Set the text parameter:

   Hi Joe, how are you doing? Is +1-2345 6789 still your number? And you live in 6th Ave?

5. Select your Presidio credentials
6. Execute the workflow

Expected output:

{
  "analyzerResults": [
    {
      "entity_type": "PERSON",
      "start": 3,
      "end": 7,
      "score": 0.85
    },
    {
      "entity_type": "PHONE_NUMBER",
      "start": 32,
      "end": 45,
      "score": 0.75
    },
    {
      "entity_type": "LOCATION",
      "start": 70,
      "end": 78,
      "score": 0.85
    }
  ],
  "anonymizedText": {
    "text": "Hi <PERSON>, how are you doing? Is <PHONE_NUMBER> still your number? And you live in <LOCATION>?",
    "items": [...]
  }
}

Example 2: Analyze Only

Use this when you want to inspect what PII entities are detected before anonymizing:

1. Set operation to Analyze
2. Optionally specify which entity types to detect (e.g., PERSON,EMAIL_ADDRESS,PHONE_NUMBER)
3. Set a score threshold (e.g., 0.7) to filter low-confidence detections
4. Execute to see the detected entities

Example 3: Two-Step Process (Analyze then Anonymize)

This gives you more control to filter or modify the analyzer results before anonymization:

1. First node: Presidio with Analyze operation
2. (Optional) Process/filter the results using a Code node
3. Second node: Presidio with Anonymize operation
   • Set text to the original text
   • Set analyzer results to: {{ $json.analyzerResults }}

Advanced Configuration

Custom Anonymization

You can specify custom anonymization strategies using the Anonymizers field:

{
  "PERSON": {
    "type": "replace",
    "new_value": "[REDACTED NAME]"
  },
  "PHONE_NUMBER": {
    "type": "mask",
    "masking_char": "*",
    "chars_to_mask": 12,
    "from_end": false
  }
}

Filtering Entities

In the Analyze operation, you can limit detection to specific entity types:

• Set Entities to: PERSON,EMAIL_ADDRESS,PHONE_NUMBER,LOCATION

Score Threshold

Set a minimum confidence score (0-1) for entity detection to reduce false positives:

• Set Score Threshold to: 0.7

Comparison with HTTP Request Workflow

The attached workflow Anonymize Text.json used HTTP Request nodes to call Presidio APIs directly. The custom Presidio node provides several advantages:

Benefits:

• ✅ Cleaner workflow - One node instead of two HTTP Request nodes
• ✅ Better UX - Proper field labels, descriptions, and validation
• ✅ Type safety - Proper handling of JSON parameters
• ✅ Reusability - Centralized credential management
• ✅ Error handling - Better error messages and handling
• ✅ Tool integration - Can be used as an AI tool with usableAsTool: true

HTTP Request Workflow Issues:

• ❌ Manual JSON construction in body parameters
• ❌ No validation of entity types or score thresholds
• ❌ Endpoint URLs hardcoded in each node
• ❌ Complex expression syntax for passing data between nodes
• ❌ No built-in documentation

Development

Running Lint

npm run lint

Auto-fix Lint Issues

npm run lint:fix

Building for Production

npm run build

Troubleshooting

"Cannot connect to Presidio" errors

• Verify Presidio services are running: docker ps
• Check health endpoints: curl http://localhost:5002/health
• Ensure correct ports in credentials (5002 for analyzer, 5001 for anonymizer)

"n8n exited with code 1" during dev

This can happen on first run while dependencies are being installed. Common solutions:

• Wait a few minutes for dependency installation
• Check that package.json has valid values (name, author, etc.)
• Try running npm run build before npm run dev
• Clear npm cache: npm cache clean --force

Node doesn't appear in n8n

• Verify the build was successful: npm run build
• Check that the node is registered in package.json under n8n.nodes
• Restart the n8n dev server

License

MIT

Resources

• Presidio Documentation
• n8n Custom Node Development
• n8n Community Nodes