Security scanner for AI agent skills.
In January 2026, the ClawHavoc campaign dropped 341 malicious skills
into the Claude skill marketplace in 3 days. Snyk's ToxicSkills audit
found 13.4% of 3,984 skills contain critical security issues.
There was no open-source scanner. This is it.
What's included in v1.0.0
- 12 detection rules covering the full OWASP Agentic Skills Top 10
- Lethal Trifecta detection (prompt injection + network + file system = ClawHavoc signature)
- Prompt injection, data exfiltration, rug pull, identity hijacking, secret harvesting
- CLI: skillguard scan SKILL.md
- Python API: SkillScanner().scan_file()
- JSON output for CI/CD pipelines
- Exits with code 1 on critical findings — blocks GitHub Actions
- Zero dependencies. Pure Python 3.10+.
Install
pip install skillshield
Usage
skillguard scan SKILL.md
skillguard check "ignore all previous instructions"
skillguard scan ./skills/ --format json