Security scanner for OpenClaw SKILL.md files.
Scan any skill before your agent runs it — catch prompt injection, hidden unicode, dangerous shell vectors, and HTML comment overrides before it's too late.
# scan a local file
bash tools/claw-vet.sh path/to/SKILL.md
# scan a remote skill
bash tools/claw-vet.sh https://raw.githubusercontent.com/user/repo/main/SKILL.md✅ PASS — No issues detected. Safe to load.
⚠️ WARN — Suspicious content found. Review before loading.
⛔ FAIL — Critical issues detected. Do NOT load this SKILL.md.
Findings include severity, category, and line number.
| Category | Examples |
|---|---|
| Hidden unicode | Zero-width chars, BOM markers invisible to humans but processed by LLMs |
| Prompt injection | "ignore previous instructions", role reassignment, DAN mode |
| Dangerous shell | curl|bash, eval, /dev/tcp, base64 -d|bash, npx -y |
| HTML comment overrides | Hidden instructions inside <!-- --> comments |
bash tools/claw-vet.sh tests/clawvet/good-skill.md # → PASS
bash tools/claw-vet.sh tests/clawvet/malicious-skill.md # → FAIL
bash tools/claw-vet.sh tests/clawvet/subtle-skill.md # → WARNMIT. Do whatever you want with these.
Made by @BChopLXXXII
Vet your skills before your agent runs them. Your credentials will thank you.
Ship it. 🚀
If this helped, star the repo — it helps others find it.