| name | security-auditor |
| description | Comprehensive security analysis and vulnerability detection |
| trigger | security|vulnerability|CVE|OWASP|audit|pentest|harden|compliance|secret|exploit |
| priority | 1 |
Security Auditor Skill
Activation Conditions
This skill activates when:
- Keywords detected: security, vulnerability, CVE, OWASP, audit, pentest, harden, compliance, secret, exploit
- File patterns:
.env*, *secret*, *credential*, Dockerfile, *.tf, *security*
- User requests security review, audit, or hardening
Workflow
Phase 1: Reconnaissance
Identify the security surface and project characteristics.
Actions:
1. Detect project type and languages
2. Identify sensitive file patterns
3. Check for existing security configurations
4. Review .gitignore for security patterns
Phase 2: Static Analysis (SAST)
Run static code analysis for vulnerabilities.
Tools:
- Semgrep: Multi-language SAST
- Bandit: Python-specific security
- ESLint security plugin: JavaScript/TypeScript
- Gosec: Go security
Rulesets:
- p/security-audit
- p/owasp-top-ten
- p/secrets
- p/sql-injection
- p/xss
Phase 3: Dependency Analysis (SCA)
Check for vulnerable dependencies.
Tools:
- Trivy: Universal vulnerability scanner
- npm audit: Node.js dependencies
- pip-audit: Python dependencies
- OSV-Scanner: Cross-language CVE database
Focus:
- Known CVEs (Critical/High priority)
- Outdated packages
- License compliance
Phase 4: Secret Detection
Find exposed credentials and sensitive data.
Tools:
- Gitleaks: Fast secret detection
- TruffleHog: Verified secret scanning
Patterns:
- AWS credentials (AKIA...)
- API keys
- Private keys
- Database connection strings
- JWT tokens
Phase 5: Configuration Review
Analyze security configurations.
Checks:
- Security headers (HSTS, CSP, X-Frame-Options)
- CORS configuration
- Cookie security flags
- TLS configuration
- Authentication settings
Phase 6: Reporting
Generate prioritized findings report.
Format:
- SARIF for tool integration
- Markdown for human review
- CVSS scoring
- OWASP Top 10 mapping
- CWE correlation
OWASP Top 10 (2021) Coverage
| Category | Checks |
|---|
| A01 Broken Access Control | Authorization checks, IDOR, CORS |
| A02 Cryptographic Failures | TLS, encryption, key management |
| A03 Injection | SQLi, XSS, Command injection |
| A04 Insecure Design | Threat modeling, secure patterns |
| A05 Security Misconfiguration | Headers, defaults, errors |
| A06 Vulnerable Components | Dependencies, CVEs |
| A07 Auth Failures | Session, passwords, MFA |
| A08 Data Integrity | Signatures, CI/CD security |
| A09 Logging Failures | Audit logs, sensitive data |
| A10 SSRF | URL validation, network segmentation |
Finding Severity Matrix
| Severity | CVSS | Response Time | Examples |
|---|
| CRITICAL | 9.0+ | Immediate | RCE, Active exploit, Secrets |
| HIGH | 7.0-8.9 | 24 hours | SQLi, XSS, Auth bypass |
| MEDIUM | 4.0-6.9 | 7 days | Info disclosure, CSRF |
| LOW | 0.1-3.9 | 30 days | Minor misconfig |
Output Templates
Finding Report
{
"finding_id": "SEC-001",
"severity": "CRITICAL",
"cvss": 9.8,
"category": "A03:2021-Injection",
"cwe": "CWE-89",
"title": "SQL Injection",
"location": {
"file": "src/db/queries.ts",
"line": 45,
"column": 12
},
"description": "User input directly interpolated in SQL query",
"evidence": "const query = `SELECT * FROM users WHERE id = ${userId}`",
"remediation": "Use parameterized queries with prepared statements",
"references": [
Summary Report
# Security Audit Summary
## Risk Assessment: HIGH
## Findings Overview
| Severity | Count | Categories |
|----------|-------|------------|
| Critical | 2 | A03, A06 |
| High | 5 | A01, A02, A07 |
| Medium | 8 | A05, A09 |
| Low | 3 | A05 |
## Critical Issues (Immediate Action)
1. SQL Injection in user input handling
2. Hardcoded AWS credentials
## Recommendations
1. Fix critical issues within 24 hours
2. Update vulnerable dependencies
3. Add security headers
4. Implement proper secret management
Agent Coordination
Dispatches To
@dependency-analyzer: Deep SCA analysis
@code-reviewer: Security-focused code review
@devops-engineer: Infrastructure security
Escalates To
- Human security team for critical findings
- Legal/compliance for regulatory issues
Reports To
@master-orchestrator: Status updates
- Session memory: Store patterns and decisions
Self-Learning Integration
After completing security audits:
- Save new vulnerability patterns to memory
- Record false positive patterns for future filtering
- Document remediation approaches that worked
- Update baseline configurations
/mem-save pattern "SQL injection pattern in TypeScript template literals"
/mem-save solution "Remediated XSS by switching to React's built-in escaping"
/mem-save decision "Adopted parameterized queries project-wide"