A5 PII Anonymizer v.0.0.1 Release (Mac, Windows, Linux)

A5 PII Anonymizer v0.0.1

A5 PII Anonymizer is an Electron desktop application for locally anonymizing documents before sending them to advanced Large Language Models (LLMs). This first release provides:

• Context-Aware PII Detection
  Uses a locally downloaded ONNX model (not included in the repo due to size) to identify names, addresses, phone numbers, and other sensitive entities far more accurately than traditional regex.

• Multiple File Formats
  Supports .txt, .docx, .xlsx, .csv, .pdf, and more. Anonymizes text by replacing personal data with consistent pseudonyms (e.g., NAME_1, PHONE_NUMBER_2).

• Daily 100-File Limit
  The free version enforces a 100-documents-per-day limit. For heavier usage, a “Pro” mode is available, which also provides a JSON mapping file for re-identification.

• Cross-Platform
  This release primarily targets macOS (Intel & Apple Silicon). Windows packaging is planned soon.

Getting Started

1. Download & Install:
   • [DMG link here if hosting on GitHub Release assets]
2. Acquire the Model:
   • Place the ONNX model in ./models/protectai/lakshyakh93-deberta_finetuned_pii-onnx/.
3. Run the App:
   • Drag & drop or select files/folders, pick an output directory, and click Anonymize Files.

Notes

• Beta Quality: v0.0.1 is an early release. Some entity edge cases may not be perfectly anonymized.
• Open Source: MIT-licensed. Feel free to modify or remove the daily limit if you prefer.
• Contributions Welcome: If you encounter bugs or have improvements, please open an issue or PR.

Enjoy anonymizing your data safely before leveraging powerful LLMs!