How does this skill help with GDPR compliance?

It provides standardized patterns to ensure PII is masked or redacted before being logged or transmitted to third-party observability tools, helping satisfy data protection requirements.

Can I integrate this with Langfuse?

Yes, the skill includes specific implementation patterns for the Langfuse mask callback, allowing you to sanitize traces before they leave your environment.

Does it support enterprise-grade PII detection?

Absolutely. It features patterns for Microsoft Presidio, which uses NLP and recognizers to detect names, locations, and other complex entities.

Is the masking reversible?

The skill includes patterns for LLM Guard's Vault, which enables reversible anonymization (deanonymization) when you need to recover original values securely.

PII Masking & Privacy Patterns

Name: PII Masking & Privacy Patterns
Author: yonatangross

byyonatangross

•

Security & Testing

Protects sensitive user data by implementing automated PII detection and redaction across LLM observability pipelines and logs.

This skill provides production-ready patterns for identifying and masking Personally Identifiable Information (PII) within AI applications to ensure data privacy and regulatory compliance. It enables developers to integrate advanced security measures like Microsoft Presidio and LLM Guard into their workflows, facilitating safe handling of sensitive information. By offering specific implementations for Langfuse tracing, structured logging, and regex-based redaction, it helps maintain high observability standards while preventing the leakage of sensitive data to external monitors or log files.

Key Features

01Pre-logging redaction patterns for structlog and loguru frameworks

02Input and output sanitization with LLM Guard and Vault storage

0369 GitHub stars

04Secure Langfuse integration via custom mask callbacks

05Lightweight regex-based masking for emails, SSNs, and credit cards

06Automated PII detection using Microsoft Presidio and NLP models

Use Cases

01Implementing GDPR-compliant logging for enterprise-grade AI applications

02Redacting sensitive user data before sending traces to LLM observability platforms

03Sanitizing prompts and responses in production pipelines to prevent data leakage

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit pii-masking-patterns

For use in Claude.ai and ChatGPT

Download Skill