mcpField
P
pii-sanitizer

pii-sanitizer

github.com/jorgealvesSecurity2258 installsUpdated 2d ago
Repository

About pii-sanitizer

A tool for identifying and masking personally identifiable information in datasets, logs, or communications. It removes sensitive data rather than encrypting it, and is intended for local processing within a secure perimeter.

Key features

  1. 01Identify and mask PII
  2. 02Process datasets, logs, and communications
  3. 03Do not save input to disk
  4. 04Recommended for local secure environments

Use cases

  1. 01Scrub logs before centralized collection
  2. 02Sanitize production data for staging or training
  3. 03Mask details in support tickets shared with engineering

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/jorgealves/agent_skills --skill pii-sanitizer

After installing, type/pii-sanitizerin Claude Code to confirm it loaded.

Bundled filesskill.yaml

Details

From the project README

PII Sanitizer

Purpose and Intent

The pii-sanitizer is a data protection tool designed to identify and mask Personally Identifiable Information (PII) from datasets, logs, or communications to comply with privacy regulations like GDPR and CCPA.

When to Use

  • Log Scrubbing: Clean application logs before sending them to centralized logging platforms (e.g., ELK, Datadog).
  • Dataset Preparation: Sanitize production data before using it in staging or training environments.
  • Customer Support: Mask sensitive info in support tickets before sharing them with engineering teams.

When NOT to Use

  • Encryption: This is a redaction tool, not an encryption tool. It is for removing data, not securing it for later retrieval.
  • Structured Database Migration: While it handles some structure, specialized ETL tools are better for massive DB sanitization.

Error Conditions and Edge Cases

  • False Positives: Strings that resemble PII (like internal serial numbers) might be accidentally redacted.
  • Ambiguous Context: "Rose" could be a name (PII) or a flower; the tool may err on the side of caution.
  • Encoding Issues: Ensure input text is UTF-8 to avoid detection failures on special characters.

Security and Data-Handling Considerations

  • Zero Retention: Input data must never be saved to disk.
  • Local Processing: Highly recommended to run this within a secure perimeter so sensitive raw data never leaves the local environment.

Comments

Markdown supported