Skill Safety Auditor

Overview

The skill-safety-auditor skill performs comprehensive security audits on skills to identify vulnerabilities, unsafe code patterns, permission issues, and compliance violations. It helps ensure skills meet security standards before release or deployment.

When to Use

• - During skill development for early issue detection
• Before releasing a new skill to production
• As part of the release approval process
• When the user asks to "audit" or "security check" a skill
• During periodic security reviews
• When reviewing third-party skills

Core Concepts

Audit Types

Vulnerability Categories

Severity Levels

• - Critical: Immediate security risk, must fix before release
• High: Significant risk, strongly recommended to fix
• Medium: Moderate risk, should address
• Low: Minor issue, consider fixing
• Info: Informational, optional to address

Input

Accepts:

• - Path to skill directory
• Audit type specification
• Output format preference
• Vulnerability type filters

Output

Produces:

• - Console output with findings
• JSON audit reports
• Summary statistics
• Pass/fail status

Workflow

Quick Security Check

1. 1. Run quick-scan on target skill
2. Review any immediate issues
3. Address critical/high findings

Comprehensive Audit

1. 1. Run full audit with verbose output
2. Review all findings by severity
3. Generate JSON report for records
4. Address findings in priority order
5. Re-run audit to verify fixes

Pre-Release Audit

1. 1. Audit skill with all checks enabled
2. Generate formal report
3. Review with team
4. Fix all critical/high issues
5. Document accepted low-risk items

Commands

Quick Scan

Full Audit

Generate Report

Filter by Type

List Past Audits

Output Format

Console Output

JSON Report

Audit Checks

Secrets Detection

• - API keys and tokens
• Database passwords
• Private keys
• Access credentials
• Environment variable patterns

Code Safety

• - eval() usage
• Function constructor
• setTimeout/setInterval with strings
• Child process execution
• Dynamic code execution

File Permissions

• - World-writable files
• Executable permissions on data files
• Sensitive file accessibility

Dependencies

• - Known CVE vulnerabilities
• Deprecated packages
• Potentially dangerous packages

Network Security

• - HTTP vs HTTPS URLs
• Insecure API endpoints
• Missing certificate validation

Quality Rules

• - Always run before releasing skills
• Fix all critical/high findings
• Document accepted risks
• Re-audit after fixes
• Keep audit history

Good Trigger Examples

• - "Audit this skill for security issues"
• "Security check before release"
• "Scan for hardcoded secrets"
• "Check for injection vulnerabilities"
• "Run security audit on skill X"

Limitations

• - Static analysis only; cannot detect runtime vulnerabilities
• May produce false positives for legitimate patterns
• Cannot verify external service security
• Does not test actual runtime behavior
• Limited to known vulnerability patterns

Resources

scripts/

• - audit-skill.sh - Full security audit
• INLINECODE10 - Fast security check
• INLINECODE11 - View audit history
• INLINECODE12 - Test skill functionality

references/

• - Security best practices
• Common vulnerability patterns
• OWASP guidelines reference