PII

Detects personally identifiable information (PII) in text using a local NLP engine. Supports automatic masking (replace PII with placeholders) or blocking, across 40+ entity types spanning 10+ countries.

Engine: Local — no external API call, no tokens consumed.

Advanced security features:

Unicode normalization: Prevents bypasses using fullwidth characters (e.g. ＠ → @) or zero-width spaces
Encoded PII detection: Optionally detects PII hidden in Base64, URL-encoded, or hex strings
URL context awareness: Detects emails in query parameters (e.g. GET /api?user=john@example.com)
Extended entity support: CVV, BIC_SWIFT, and more beyond standard entity types

Policy configuration

Field	Type	Default	Description
`name`	string	—	Must be `"PII"` (registry name).
`suppress_enforcement`	bool	`true`	If `true`, a detection does not raise `GuardrailEnforcementTriggered`; if `false`, blocking behaviour can stop the pipeline when `block` is `true`.
`config.entities`	list[str]	all entity types	Presidio entity identifiers to scan (e.g. `US_SSN`, `EMAIL_ADDRESS`). An empty list disables checks in the SDK.
`config.block`	bool	`false`	`true`: signal enforcement when PII is found. `false`: mask only (pre-flight / input masking path).
`config.detect_encoded_pii`	bool	`false`	Also scan Base64-, URL-, and hex-encoded segments.
`config.allow_list`	list[str]	`[]`	Extra spans (exact match) allowed in addition to built-in defaults.

{
  "version": 1,
  "pre_flight": {
    "version": 1,
    "guardrails": [
      {
        "name": "PII",
        "suppress_enforcement": true,
        "config": {
          "entities": ["US_SSN", "EMAIL_ADDRESS", "PHONE_NUMBER"],
          "block": false,
          "detect_encoded_pii": true,
          "allow_list": []
        }
      }
    ]
  }
}

Supported stages

Stage	`block=false` (masking)	`block=true` (blocking)	Notes
`input`	✅	✅	Masking must use this stage. The masked text is forwarded to the LLM instead of the original
`output`	—	✅	Block LLM responses that contain PII before they reach the user

Stage usage

Stage	`block`	Behaviour
`input`	`false` (default)	Masks PII — replaces with `<ENTITY_TYPE>` before sending to LLM
`input`	`true`	Blocks the request entirely when PII is found
`output`	`true`	Blocks LLM response when it contains PII
`output`	`false`	Not supported — masking on output does not work as expected

Supported entities

Global

Entity	Description
`CREDIT_CARD`	Credit card numbers
`CRYPTO`	Cryptocurrency wallet addresses
`DATE_TIME`	Date and time expressions
`EMAIL_ADDRESS`	Email addresses
`IBAN_CODE`	International Bank Account Numbers
`IP_ADDRESS`	IPv4 and IPv6 addresses
`NRP`	Nationalities, religious and political groups
`LOCATION`	Location names
`PERSON`	Person names
`PHONE_NUMBER`	Phone numbers in various formats
`MEDICAL_LICENSE`	Medical license numbers
`URL`	URLs

Custom (built-in extensions)

Entity	Description
`CVV`	Credit card CVV/CVC security codes
`BIC_SWIFT`	Bank BIC/SWIFT codes

United States

Entity	Description
`US_BANK_NUMBER`	US bank account numbers
`US_DRIVER_LICENSE`	US driver's license numbers
`US_ITIN`	Individual Taxpayer Identification Numbers
`US_PASSPORT`	US passport numbers
`US_SSN`	Social Security Numbers

United Kingdom

Entity	Description
`UK_NHS`	UK National Health Service numbers
`UK_NINO`	UK National Insurance numbers

Spain

Entity	Description
`ES_NIF`	Spanish tax identification numbers
`ES_NIE`	Spanish foreign identity numbers

Italy

Entity	Description
`IT_FISCAL_CODE`	Italian fiscal codes
`IT_DRIVER_LICENSE`	Italian driver's license numbers
`IT_VAT_CODE`	Italian VAT codes
`IT_PASSPORT`	Italian passport numbers
`IT_IDENTITY_CARD`	Italian identity card numbers

Poland

Entity	Description
`PL_PESEL`	Polish national identification numbers

Singapore

Entity	Description
`SG_NRIC_FIN`	Singapore NRIC/FIN numbers
`SG_UEN`	Singapore Unique Entity Numbers

Australia

Entity	Description
`AU_ABN`	Australian Business Numbers
`AU_ACN`	Australian Company Numbers
`AU_TFN`	Australian Tax File Numbers
`AU_MEDICARE`	Australian Medicare card numbers

India

Entity	Description
`IN_PAN`	Indian Permanent Account Numbers
`IN_AADHAAR`	Indian Aadhaar numbers
`IN_VEHICLE_REGISTRATION`	Indian vehicle registration numbers
`IN_VOTER`	Indian voter ID numbers
`IN_PASSPORT`	Indian passport numbers

Finland

Entity	Description
`FI_PERSONAL_IDENTITY_CODE`	Finnish personal identity codes

South Korea

Entity	Description
`KR_RRN`	Korean Resident Registration Numbers

What it returns

{
    "guardrail_name": "PII",
    "detected_entities": {
        "EMAIL_ADDRESS": ["user@example.com"],
        "US_SSN": ["457-55-5462"]
    },
    "entity_types_checked": ["EMAIL_ADDRESS", "US_SSN", "CREDIT_CARD"],
    "checked_text": "Contact me at <EMAIL_ADDRESS>, SSN: <US_SSN>",
    "block_mode": false,
    "pii_detected": true,
    "detect_encoded_pii": false
}

Field	Description
`detected_entities`	Map of entity type → list of matched values
`entity_types_checked`	Entity types configured for this run
`checked_text`	Input text with PII replaced by `<ENTITY_TYPE>` placeholders
`block_mode`	Whether blocking mode was active
`pii_detected`	`true` if any PII was found
`detect_encoded_pii`	Whether encoded PII scanning was enabled

With encoded PII detection enabled

When detect_encoded_pii: true, encoded PII is masked with <ENTITY_TYPE_ENCODED>:

{
    "detected_entities": {
        "EMAIL_ADDRESS": [
            "user@example.com",
            "am9obkBleGFtcGxlLmNvbQ==",
            "%6a%6f%65%40domain.com"
        ]
    },
    "checked_text": "Contact <EMAIL_ADDRESS> or <EMAIL_ADDRESS_ENCODED> or <EMAIL_ADDRESS_ENCODED>"
}