Detection Engine

Smart Detection

Industry-leading detection of 50+ sensitive data types, powered by advanced pattern matching and validation algorithms.

Detection Categories

Secrets & Credentials

API Keys & Tokens

  • • OpenAI API Keys
  • • Anthropic API Keys
  • • AWS Access Keys
  • • GitHub Tokens
  • • Stripe Keys
  • • Slack Tokens
  • • JWT Tokens

Authentication

  • • Bearer Tokens
  • • OAuth Tokens
  • • RSA Private Keys
  • • SSH Private Keys
  • • Database Connection Strings
  • • Password Hashes

Personal Identifiable Information (PII)

Identity Documents

  • • Social Security Numbers (SSN)
  • • National Insurance Numbers (NIN)
  • • Passport Numbers
  • • Driver's License Numbers
  • • Tax IDs

Personal Data

  • • Full Names
  • • Date of Birth
  • • IP Addresses (IPv4/IPv6)
  • • MAC Addresses
  • • Physical Addresses

Financial Data

Payment Cards

  • • Credit Card Numbers (Visa, Mastercard, Amex)
  • • CVV Codes
  • • Card Expiry Dates
  • • Luhn Algorithm Validation

Banking

  • • IBAN (International Bank Account Numbers)
  • • SWIFT/BIC Codes
  • • Routing Numbers
  • • Account Numbers

Healthcare Data

Medical Identifiers

  • • Medical Record Numbers (MRN)
  • • DEA Numbers
  • • NPI (National Provider Identifier)

Health Information

  • • Diagnosis Codes (ICD-10)
  • • Prescription Numbers
  • • Insurance Member IDs

Contact Information

Digital Contact

  • • Email Addresses
  • • Phone Numbers (50+ country formats)
  • • URLs

Physical Contact

  • • Postal Codes
  • • Street Addresses
  • • Geographic Coordinates

How Detection Works

1

Pattern Matching

Advanced regex patterns scan your document for potential sensitive data. Each pattern is optimized for accuracy with minimal false positives.

2

Validation

Matches are validated using domain-specific algorithms:

  • Luhn Algorithm for credit card numbers
  • Checksum Verification for IBANs and SSNs
  • Format Validation for emails and phone numbers
  • Entropy Analysis for API keys and tokens
3

Contextual Analysis

Surrounding text is analyzed to reduce false positives. For example, "credit card: 4532 ..." is more likely to be real than a random 16-digit number.

4

Whitelist Checking

Known safe values (like example.com or 555-0100) are whitelisted to prevent unnecessary redactions.

Detection Accuracy

97.38%
F1 Score
Harmonic mean of precision & recall
97.70%
Precision
Correctly identifies sensitive data
97.06%
Recall
Catches all sensitive data

Validated across 200,000 samples. See full benchmark results →

🔒

100% Local Processing

All detection happens on your machine. The Redactorr Engine runs entirely in your browser—no data is ever sent to external servers for analysis. Your sensitive data never leaves your control. Open DevTools → Network tab to verify.

Learn More

Explore related documentation