skill-vetter

name	skill-vetter
description	Security review and audit tool for AI agent skills. Use when asked to "vet a skill", "check skill security", "audit a skill", or when reviewing skills before installation to detect malicious code, unauthorized access, and suspicious patterns.

Skill Vetter

Security review and audit tool for AI agent skills. Checks for potential security risks before skill installation.

When to Use

• Before installing a new skill from an untrusted source
• User asks to "vet" or "audit" a skill
• Reviewing skill code for security vulnerabilities
• Checking for malicious patterns in skill files

Security Review Process

1. File Analysis

Scan all files in the skill directory:

• SKILL.md - Main instruction file
• scripts/ - Helper scripts
• references/ - Reference documents
• Any other included files

2. Check Categories

🔴 Critical Risks

• Shell injection: Commands that could execute arbitrary code
• File system access: Unauthorized read/write to sensitive paths
• Network exfiltration: Sending data to external servers
• Credential theft: Accessing API keys, tokens, passwords
• Privilege escalation: Commands requiring elevated permissions

🟡 Medium Risks

• Broad file permissions: Overly permissive file access patterns
• External dependencies: Downloading scripts from URLs at runtime
• Environment variable access: Reading sensitive env vars
• Hidden obfuscation: Base64 encoded strings, eval statements

🔵 Low Risks

• Overly broad instructions: Instructions that could lead to unintended actions
• Missing input validation: No checks on user-provided data
• Unclear scope: Ambiguous instructions that could be misinterpreted

3. Output Format

## Skill Security Report: [skill-name]

### Risk Level: [LOW/MEDIUM/HIGH/CRITICAL]

### Findings
| # | Severity | File | Description |
|---|----------|------|-------------|
| 1 | 🔴 CRITICAL | script.py:15 | Shell injection via user input |
| 2 | 🟡 MEDIUM | SKILL.md:42 | Broad file access pattern |

### Recommendation
[SAFE TO INSTALL / INSTALL WITH CAUTION / DO NOT INSTALL]

### Details
[Detailed explanation of each finding]

Best Practices

• Always vet skills from unknown sources
• Review scripts line by line for suspicious patterns
• Check for obfuscated code
• Verify network access is justified
• Ensure file access is scoped appropriately