phi-redaction

Redact PHI fields from FHIR resources during read operations.

27|9|Updated Mar 31, 2025
One-click install
npx skills add https://github.com/aks129/HealthClawGuardrails --skill phi-redaction
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: phi-redaction
Source: https://github.com/aks129/HealthClawGuardrails/tree/main/skills/phi-redaction
Command: npx skills add https://github.com/aks129/HealthClawGuardrails --skill phi-redaction

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

PHI redaction patterns for FHIR resources following HIPAA Safe Harbor. Use when: (1) Redacting patient health information from FHIR resources before AI agent access, (2) Implementing de-identification for clinical data pipelines, (3) Understanding what fields are stripped, masked, or truncated in FHIR resources, (4) Building read paths that automatically protect patient privacy.

Core Features & Use Cases

All redaction is applied at read time, not at storage time. Redacts PHI fields in FHIR resources when read through MCP or data pipelines.

  • Names: Given names truncated to initials; Family names kept; Text narratives removed
  • Identifiers: Values masked to last 4 characters; System and type kept
  • Addresses: Lines/text removed; City/State/Country kept
  • Birth Date: Truncated to year only
  • Telecommunication: Values redacted; System/Use kept
  • Photos: Removed entirely
  • Notes/Comments: Replaced with [Redacted]
  • Contained resources: Redacted recursively
  • Read paths: Applied to direct reads, searches, and context envelopes

Quick Start

Integrate this redaction profile into your FHIR MCP guardrail stack to automatically redact PHI on all read paths.

Frequently Asked Questions about phi-redaction

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically redact PHI from FHIR resources for HIPAA compliance?

To redact PHI from FHIR resources, you apply field-level Safe Harbor rules on read paths. This process masks identifiers, truncates birth dates to year only, and removes addresses and photos to ensure HIPAA-compliant data sharing.

What fields are stripped or masked during FHIR de-identification?

FHIR de-identification masks identifier values to the last 4 characters, truncates given names to initials, removes address lines, replaces notes with '[Redacted]', and recursively redacts contained resources.

Can I apply HIPAA Safe Harbor redaction on read paths without altering stored clinical data?

Yes, HIPAA Safe Harbor redaction can be applied strictly at read time. The utility deeply copies FHIR resources before redaction, ensuring original stored clinical data remains unaltered during AI agent access or analytics workflows.

Does FHIR redaction keep any location or identifier metadata for analytics pipelines?

FHIR redaction keeps structural metadata while removing direct identifiers. It retains family names, identifier systems, and city/state/country values, allowing analytics pipelines to process de-identified clinical data without exposing PHI.

What is the best way to protect patient privacy before AI agent access in clinical data pipelines?

The best way to protect patient privacy before AI agent access is integrating a redaction profile into FHIR MCP guardrails. This automatically truncates birth dates, removes photos, and redacts telecom values across direct reads and searches.