pii-sanitizer-js
v1.0.2
Published
A lightweight utility to sanitize PII from strings and prompts.
Maintainers
zenzzjReadme
🛡️ PII Sanitizer JS
A lightweight, zero-dependency, high-performance utility for detecting and anonymizing Personally Identifiable Information (PII) in text.
Ideal for: Sanitizing prompts before sending them to LLMs (OpenAI, Claude), cleaning log files, and ensuring GDPR/data privacy compliance in backend services.
🚀 Key Features
- Zero Dependencies: No bloated node_modules. Native, fast, and secure.
- High Performance: Optimized regex patterns for real-time sanitization.
- Bun-First: Built with Bun for maximum speed, fully compatible with Node.js.
- AI-Ready: Designed to prevent data leakage in AI-integrated applications.
📊 Supported Data Types
| Type | Tag Replacement | Example Detection |
| :--- | :--- | :--- |
| Email | [HIDDEN_EMAIL] | [email protected] |
| Phone | [HIDDEN_PHONE] | +1 555-0199, (123) 456-7890 |
| Credit Card | [HIDDEN_CARD] | 4532 XXXX XXXX XXXX |
| IPv4 Address | [HIDDEN_IP] | 192.168.1.1 |
📦 Installation
# Using npm
npm install pii-sanitizer-js
# Using bun
bun add pii-sanitizer-js🛠️ Usage
As a Library Integrate it easily into your middleware or data processing pipeline.
import { sanitize } from 'pii-sanitizer-js';
const input = "Send the invoice to [email protected] or call 555-012-345. My IP is 10.0.0.5";
const cleanText = sanitize(input);
console.log(cleanText);
// Output: "Send the invoice to [HIDDEN_EMAIL] or call [HIDDEN_PHONE]. My IP is [HIDDEN_IP]"🔒 Privacy First
All sanitization is performed locally. No data is ever sent to external servers. This is a pure string-manipulation utility designed to be safe for production environments.
🛠️ Motivation
When working with LLMs (like OpenAI or Claude), it's easy to accidentally leak sensitive data in prompts. I built this to have a lightweight, zero-dependency way to scrub text before it leaves my infrastructure.
🤝 Contributing
Contributions are welcome! If you find a bug or want to support a new PII type (like IPv6 or specific ID formats), feel free to:
- Fork the project.
- Create your feature branch.
- Open a Pull Request.