Skip to content

🛡️ Privacy Guardian (privacy-guardian.sh)

"Loose lips sink ships." - Preventing Data Leaks to AI Models.

1. 🏛️ Purpose

Version: v1.0 Description: A pre-commit and pre-upload scanner that checks the generated sod_manifest for sensitive data (PII, API Keys, IPs) to ensure "Sovereign Privacy."

2. 🛡️ Safety Mechanisms

Mechanism Status Description
Zero-Global Pattern ✅ Enforced Local variable scoping.
Exit-on-Error ✅ Active set -e injected.
Heuristic Scan ✅ Active Uses regex for IPv4, Google Keys, Slack Tokens, and Home Paths.

3. ⚙️ Usage

./tools/privacy-guardian.sh

4. 🧠 Logic Flow

  1. Target Verification: Checks if a sod_manifest_YYYY-MM-DD.txt exists.
  2. Regex Scanning: Iterates through an array of dangerous patterns:
    • IPv4 Addresses
    • Emails (Standard Regex)
    • Google API Keys (AIza...)
    • AWS Access Keys (AKIA...)
    • GitHub Tokens (ghp...)
    • OpenAI Keys (sk-...)
    • Slack Tokens (xoxb...)
    • Private Keys (-----BEGIN...)
    • Local User Paths (/home/user/)
  3. Reporting: Exits with 1 if leaks are found, requiring manual remediation.

5. 📝 Extracted Comments

"Scans the generated DSOM reanimation manifest for sensitive information before it is uploaded to an external AI model."

Deep State of Mind (DSOM) For My AI Protocol | Harisfazillah Jamel (LinuxMalaysia) | 2026-07-04 Standard: UK English | DBP-standard Bahasa Melayu Malaysia (Piawai) | GNU General Public License v3.0