GDPR Anonymizer Skill

This skill acts as a pre-processing layer to ensure "Data Minimisation" before content is sent to analysis agents.

Workflow

1. •

   PII Identification:

   • •Scan input text for:
     • •Names (e.g., "John Smith")
     • •Email Addresses (e.g., "john@example.com")
     • •Phone Numbers
     • •Specific Locations (home addresses)
     • •Financial Account Numbers
     • •Protected Characteristics (if irrelevant to the audit): Age, Gender, Religion.
2. •

   Masking / Redaction:

   • •Replace identified entities with generic tokens:
     • •"John Smith" -> [PERSON_1]
     • •"john@example.com" -> [EMAIL_1]
     • •"London Office" -> [LOCATION_OFFICE]
   • •Maintain consistency (if "John Smith" appears twice, use [PERSON_1] both times).
3. •

   Verification:

   • •Ensure no raw PII remains in the "Cleaned Text".
4. •

   Handoff:

   • •Pass the [Cleaned Text] to the bias_detective or noise_judge.
   • •Note: The fact_checker may need the original text only if verifying specific public claims, but internal PII must remain masked.

Usage Rule

• •Mandatory Step: This skill must be applied immediately after Ingestion and before any Multi-Agent Analysis.