clawvet

Skill vetting & supply chain security for OpenClaw.

ClawVet scans OpenClaw SKILL.md files for prompt injection, credential theft, remote code execution, typosquatting, and social engineering — before they reach your agent.

Install

npm install -g clawvet

Usage

Scan a local skill

clawvet scan ./my-skill/
clawvet scan ./my-skill/SKILL.md

JSON output (for CI/CD)

clawvet scan ./my-skill/ --format json

Fail on severity threshold

clawvet scan ./my-skill/ --fail-on high
# exits 1 if any high or critical findings

Fetch and scan from ClawHub

clawvet scan weather-forecast --remote

Audit all installed skills

clawvet audit

Watch for new skill installs

clawvet watch --threshold 50

What it detects

ClawVet runs a 6-pass analysis on every skill:

| Pass | What it checks | |------|---------------| | Skill Parser | Extracts YAML frontmatter, code blocks, URLs, IPs, domains | | Static Analysis | 54 regex patterns: RCE, reverse shells, credential theft, obfuscation, DNS exfil, privilege escalation | | Metadata Validator | Undeclared binaries, env vars, missing descriptions, invalid semver | | Dependency Checker | npx -y auto-install, global npm install, risky packages | | Typosquat Detector | Levenshtein distance against popular skills, suspicious naming patterns | | Semantic Analysis | AI-powered detection of social engineering & prompt injection (optional) |

Risk Scoring

| Score | Grade | Action | |-------|-------|--------| | 0-10 | A | Approve | | 11-25 | B | Approve | | 26-50 | C | Warn | | 51-75 | D | Warn | | 76-100 | F | Block |

CI/CD Integration

# GitHub Actions example
- name: Vet skill
  run: npx clawvet scan ./my-skill --format json --fail-on high

License

MIT