skll-scan（骨骼扫描）

Skill Scan - Security Scanner for OpenClaw Skills

Purpose

Skill Scan is a security tool that analyzes OpenClaw Skills for potentially malicious code and threat intelligence indicators before installation or during audits.

Features

• 🔍 Static Code Analysis: Detects dangerous patterns (exec, network, filesystem, sensitive data access)
• 🌐 Domain Extraction: Identifies all domains referenced in Skill code
• 🛡️ Threat Intelligence Check: Validates domains against abuse databases (Abuse.ch, AbuseIPDB, VirusTotal)
• 📊 Risk Scoring: Generates risk levels (low/medium/high) with detailed reports
• 💾 JSON Export: Saves scan results for automation and CI/CD integration

Usage

Basic Scan

python3 ~/.openclaw/skills/skill-scan/scripts/skill-scan.py <skill-path>

Examples

# Scan a specific Skill
python3 ~/.openclaw/skills/skill-scan/scripts/skill-scan.py ~/.openclaw/extensions/mem9

# Scan all installed Skills
for skill in ~/.openclaw/extensions/*/; do
  echo "Scanning: $skill"
  python3 ~/.openclaw/skills/skill-scan/scripts/skill-scan.py "$skill"
done

# Scan before installation
tar -xzf new-skill.tgz -C /tmp/skill-check/
python3 ~/.openclaw/skills/skill-scan/scripts/skill-scan.py /tmp/skill-check/

Output

Risk Levels

Report Format

============================================================
📊 Skill Security Scan Report
============================================================
Skill Path: /path/to/skill
Risk Level: low
Total Findings: 2
Domains Checked: 1

📋 Findings by Category:
  - network: 2

📝 Details:
  [network] /path/to/file.ts:30
    const resp = await fetch(this.baseUrl + "/v1alpha1/mem9s", {
============================================================

Detection Categories

1. Exec (System Command Execution)

• exec(), execSync(), spawn()
• child_process, subprocess.*
• os.system(), shell_exec()

2. Network (Network Requests)

• fetch(), axios.*
• http.get, https.get
• requests.*, urllib.request
• XMLHttpRequest

3. Filesystem (File Operations)

• fs.writeFile, fs.readFile, fs.unlink
• open(..., 'w')
• shutil.(copy|move|remove)

4. Sensitive (Sensitive Data Access)

• process.env, process.argv
• os.environ
• Hardcoded secrets: secret=, password=, token=, api_key=

Threat Intelligence Sources

Built-in (Free APIs)

Optional (API Key Required)

Configuration

To enable additional threat intelligence APIs, edit the script and add your API keys:

THREAT_INTEL_APIS = {
    "virustotal": {
        "url": "https://www.virustotal.com/api/v3/domains/",
        "key_param": "x-apikey",
        "api_key": "YOUR_API_KEY"  # Add your key here
    }
}

Integration

Pre-installation Hook

Add to your CI/CD pipeline:

#!/bin/bash
# Pre-installation security check

SKILL_PATH=$1
REPORT=$(python3 ~/.openclaw/skills/skill-scan/scripts/skill-scan.py "$SKILL_PATH")

if echo "$REPORT" | grep -q "Risk Level: high"; then
  echo "❌ Security check failed: High risk detected"
  exit 1
fi

echo "✅ Security check passed"

Periodic Audit

Create a cron job for weekly audits:

# /etc/cron.d/skill-scan
0 2 * * 1 niuqun python3 ~/.openclaw/skills/skill-scan/scripts/skill-scan.py ~/.openclaw/extensions/* >> /var/log/skill-scan.log

Security Best Practices

1. Always scan before installing - Never install Skills from unknown sources without scanning
2. Review medium/high risks - Don't ignore warnings
3. Keep threat intel updated - Regularly update local blacklists
4. Report false positives - Help improve the tool
5. Contribute signatures - Add new malicious patterns you discover

Troubleshooting

Issue: Script fails to run

Solution: Ensure Python 3 is installed and script has execute permission

chmod +x ~/.openclaw/skills/skill-scan/scripts/skill-scan.py

Issue: Threat intelligence API timeout

Solution: Check network connection or API key validity

curl -I https://urlhaus-api.abuse.ch/

Issue: Too many false positives

Solution: Adjust detection patterns in the script or add domains to whitelist

Limitations

• Static analysis only - Cannot detect runtime behavior
• API rate limits - Free APIs have request limits
• Evasion techniques - Obfuscated code may bypass detection
• No sandbox - Does not execute code in isolation

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Add new detection patterns
3. Integrate additional threat intelligence sources
4. Submit pull requests

License

MIT License - See LICENSE file for details

Support

• Issues: https://github.com/yourusername/skill-scan/issues
• Documentation: https://github.com/yourusername/skill-scan/wiki
• Threat Intel: Report malicious domains to abuse@yourdomain.com

Remember: This tool is a first line of defense. Always combine with manual code review and other security measures for critical systems.