openclaw-master-skills/skills/ka88-agent-shield

ka88-agent-shield

Professional security audit skill for AI agents.

Description

ka88-agent-shield is a skill for AI agents providing comprehensive protection against:

• Prompt Injection
• SSRF Attacks
• Credential Exfiltration
• Malicious JavaScript
• Phishing Patterns
• Obfuscation (hidden code)

Features

🔍 4-Phase Audit System

1. Pre-Visit Scan — Check URL before visiting
2. Content Analysis — Analyze content for threats
3. Command Safety — Validate commands before execution
4. Self-Audit — Periodic self-monitoring

📊 216 Detection Patterns

Complete pattern set for threat detection, based on ClawGuard and OWASP Agentic AI Top 10.

🔧 Tools

Script	Description	Requirements
quick-scan.sh	Fast scan without LLM	bash/grep only
scan-skill-scanner.sh	Full scan with LLM	skill-scanner + LM Studio

Installation

Via OpenSkills (recommended)

git clone <repo-url> ka88-agent-shield
cd ka88-agent-shield
openskills install ./ --global
openskills sync --yes

Manual

git clone <repo-url> ka88-agent-shield
mkdir -p ~/.claude/skills
ln -s $(pwd)/ka88-agent-shield ~/.claude/skills/ka88-agent-shield

Usage

Activation

Skill activates automatically when agent:

• Visits websites
• Analyzes URL content
• Executes commands (curl, wget, pip, npm)
• Processes HTML/JS/CSS

Quick Scan

./scripts/quick-scan.sh <path> [--dry-run] [--verbose] [--help]

Full Scan

./scripts/scan-skill-scanner.sh <path> [--install] [--force] [--help]

Project Structure

ka88-agent-shield/
├── SKILL.md
├── LICENSE
├── README.md
├── config/
├── scripts/
├── procedures/
└── templates/

License

MIT

Version

1.0.0

Author

Danilka88