A robust utility for PII anonymization, local de-anonymization, and deterministic local detector checks for various text and file formats.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install privacy-protector
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install privacy-protector using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Privacy Protector?
The Privacy Protector skill is a powerful addition to the Openclaw Skills library, designed to handle PII (Personally Identifiable Information) with precision. It allows developers to run anonymization workflows locally or via API, ensuring sensitive data like emails or phone numbers are redacted and can be safely restored using saved maps. This skill is essential for maintaining privacy compliance when providing data to AI agents or external language models.
Built for versatility, it supports a wide range of formats from standard text and Markdown to complex documents like DOCX and PDF. By utilizing the Openclaw Skills framework, Privacy Protector provides a deterministic approach to data privacy, offering different levels of protection—ranging from fully local lite mode for offline processing to advanced cross-border compliance checks.
Privacy Protector Use Cases
- Redacting sensitive user data before sending logs or documents to an LLM.
- Restoring anonymized placeholders in AI-generated responses using local mapping files.
- Tuning PII detection thresholds and managing allowlists for specific organizational terminology.
- Automating privacy-first file processing within automated CI/CD or data engineering pipelines.
How Privacy Protector Works
- The user provides input text or a file path and selects a protection level, such as lite for local-only processing.
- The anonymization script scans the input using deterministic local detectors or optional API-backed intelligence.
- Identified PII is replaced with secure placeholders (e.g., [EMAIL_1]), and a mapping file is generated to link these placeholders to original values.
- When the original data needs to be recovered, the deanonymization script references the saved map to restore the text.
- Developers can refine the process by tuning the local detector with custom allowlists and blocklists to ensure high accuracy and low false-positive rates.
Privacy Protector Setup
Ensure you have python3 installed on your system. To install optional dependencies for specific file types or API features, use the following commands:
pip install requests python-docx PyMuPDF
Navigate to the skill directory to execute commands:
cd privacy-protector
You can override default storage or API settings using environment variables like ANONYMIZE_API_URL or MODEIO_REDACT_MAP_DIR for custom configurations.
Privacy Protector Data Schema & Taxonomy
Privacy Protector organizes data through local mapping files and JSON-structured outputs for seamless integration.
| Component | Description |
|---|---|
| Maps | JSON files stored in ~/.modeio/redact/maps linking placeholders to original PII. |
| Inputs | Supports .txt, .md, .csv, .json, .docx, .pdf, .yaml, and more. |
| Output Envelopes | Standardized JSON output when using the --json flag for machine readability. |
| Tuning Files | External allowlist, blocklist, and threshold JSON files for configuring the local detector. |
Privacy Protector Advanced Features
- Multi-level protection modes including lite (local), dynamic, and strict.
- Cross-border compliance support requiring sender and recipient code parameters.
- Deterministic local detector diagnostics with configurable probability thresholds.
- Extensive file format support including rich text, office documents, and web formats.
- Customizable map storage paths via environment variables for containerized or cloud environments.
SKILL.md
Loading
Related Openclaw Skills
Skill Audit
A deterministic static safety auditor for Openclaw Skills designed to screen third-party repositories for security risks before installation.
Security Live Safety Check
A robust backend-backed safety gate that validates risky AI agent instructions before execution to prevent destructive or unauthorized actions.
SkillBoss
A unified API and CLI tool to access over 50 AI models across image, video, voice, and text providers with intelligent cost-saving routing.
fund-news-daily
An automated news intelligence skill that aggregates and filters mutual fund industry news from five authoritative Chinese financial platforms.
CodeQL Security Audit Skill
A comprehensive pipeline for CodeQL static analysis, vulnerability triaging, and query optimization.
Lightcloud API Integration
A cross-platform skill for managing Qingyun/Lightcloud form documents and authentication via AI coding agents.
Featured*
