Agent Skills: Audit Skill

Audit Skill

Overview

The audit skill provides comprehensive on-demand security and code quality audits for your codebase. It identifies vulnerabilities, security issues, code smells, outdated dependencies, exposed secrets, and compliance problems across all supported technology stacks.

When to Use:

• Security audits and vulnerability scans
• Pre-deployment security checks
• Compliance verification (GDPR, HIPAA, SOC2)
• Code quality assessment
• Dependency vulnerability scanning
• Secret exposure detection
• Third-party license compliance

Technology Coverage:

• React/TypeScript/JavaScript projects
• Go applications
• Rust projects
• Python codebases
• Full-stack applications
• Monorepos and microservices

Audit Categories

1. Security Vulnerabilities (OWASP Top 10)

What Gets Checked:

• SQL injection vulnerabilities
• Cross-site scripting (XSS)
• Authentication and session management flaws
• Security misconfigurations
• Sensitive data exposure
• XML external entities (XXE)
• Broken access control
• Cross-site request forgery (CSRF)
• Using components with known vulnerabilities
• Insufficient logging and monitoring

Detection Methods:

• Static code analysis
• Pattern matching for common vulnerabilities
• Framework-specific security checks
• API endpoint security validation
• Input validation analysis

2. Dependency Vulnerabilities

What Gets Checked:

• Outdated packages with known CVEs
• Unmaintained dependencies
• License compatibility issues
• Transitive dependency risks
• Version conflicts

Package Managers Supported:

• npm/yarn/pnpm (JavaScript/TypeScript)
• go.mod (Go)
• Cargo.toml (Rust)
• requirements.txt/pyproject.toml (Python)
• Gemfile (Ruby)

Tools Used:

• npm audit / yarn audit / pnpm audit
• go mod verify + vulnerability databases
• cargo audit
• pip-audit / safety

3. Exposed Secrets and Credentials

What Gets Detected:

• Hardcoded API keys and tokens
• Database credentials
• Private keys and certificates
• OAuth tokens and secrets
• AWS/GCP/Azure credentials
• JWT secrets
• Encryption keys

Detection Patterns:

• Regex patterns for common secret formats
• Environment variable misuse
• Configuration file analysis
• Git history scanning (if requested)
• Common secret naming patterns

False Positive Reduction:

• Ignore test fixtures and mocks
• Respect .gitignore patterns
• Check for proper environment variable usage
• Validate against .env.example templates

4. Code Quality Issues

What Gets Analyzed:

• Code complexity (cyclomatic complexity)
• Code duplication
• Dead code and unused exports
• Large functions and files
• Naming convention violations
• Magic numbers and hardcoded values
• Lack of error handling
• Poor separation of concerns

Metrics Calculated:

• Cyclomatic complexity per function
• Duplication percentage
• Lines of code (LOC) per file/function
• Comment-to-code ratio
• Test-to-code ratio

Running Audits

Full Audit

Run all audit categories:

Please run a full security and quality audit of this codebase

The audit will:

1. Scan all source files for vulnerabilities
2. Check dependencies for known CVEs
3. Search for exposed secrets
4. Analyze code quality metrics
5. Generate comprehensive report

Category-Specific Audits

Security Only:

Run a security audit focusing on OWASP top 10 vulnerabilities

Dependencies Only:

Audit all dependencies for vulnerabilities and outdated packages

Secrets Only:

Scan for exposed secrets and credentials

Code Quality Only:

Analyze code quality and identify code smells

Targeted Audits

Specific Directory:

Audit the /src/auth directory for security issues

Specific Files:

Audit UserController.ts and AuthService.ts for vulnerabilities

Pre-Deployment:

Run pre-deployment audit checklist

Audit Report Format

Severity Levels

Reports classify findings by severity:

• CRITICAL: Immediate security risk, exploitable vulnerability
• HIGH: Significant security or quality issue
• MEDIUM: Moderate issue that should be addressed
• LOW: Minor issue or code smell
• INFO: Informational finding, best practice suggestion

Report Structure

# Security and Quality Audit Report

**Generated**: 2026-01-28 14:32:00
**Scope**: Full codebase audit
**Files Scanned**: 247
**Duration**: 8.3 seconds

## Executive Summary

- CRITICAL: 2 findings
- HIGH: 5 findings
- MEDIUM: 12 findings
- LOW: 23 findings
- INFO: 8 findings

**Risk Score**: 7.2/10 (HIGH)

## Critical Findings

### [CRITICAL-001] SQL Injection Vulnerability
**File**: src/database/queries.ts:42
**Severity**: CRITICAL
**Category**: Security - SQL Injection

**Issue**: User input concatenated directly into SQL query without sanitization.

**Code**:
```typescript
const query = `SELECT * FROM users WHERE email = '${email}'`;

Impact: Attacker can execute arbitrary SQL commands.

Recommendation: Use parameterized queries or ORM.

Fix:

const query = db.prepare('SELECT * FROM users WHERE email = ?').bind(email);

[CRITICAL-002] Exposed API Key

File: src/config/api.ts:15 Severity: CRITICAL Category: Security - Exposed Secret

Issue: Hardcoded API key found in source code.

Code:

const STRIPE_SECRET_KEY = "sk_live_abc123xyz789";

Impact: Unauthorized access to Stripe account.

Recommendation: Move to environment variable.

Fix:

const STRIPE_SECRET_KEY = process.env.STRIPE_SECRET_KEY;

High Priority Findings

[HIGH-001] Outdated Dependency with Known CVE

File: package.json:23 Severity: HIGH Category: Dependencies

Issue: lodash@4.17.15 has known vulnerability (CVE-2021-23337)

CVE Details:

• CVE-2021-23337: Command injection in template
• Published: 2021-02-15
• CVSS Score: 7.2

Recommendation: Upgrade to lodash@4.17.21 or higher

Fix:

npm install lodash@latest

[... additional findings ...]

Compliance Checks

OWASP Top 10 Coverage

• [FAIL] A1: Injection - 2 SQL injection vulnerabilities found
• [PASS] A2: Broken Authentication
• [FAIL] A3: Sensitive Data Exposure - 1 exposed secret
• [PASS] A4: XML External Entities
• [PASS] A5: Broken Access Control
• [FAIL] A6: Security Misconfiguration - CORS too permissive
• [PASS] A7: Cross-Site Scripting
• [PASS] A8: Insecure Deserialization
• [FAIL] A9: Using Components with Known Vulnerabilities - 3 outdated deps
• [WARN] A10: Insufficient Logging - Limited audit logging

Overall: 6/10 passing (60%)

Recommendations

1. Immediate Actions (CRITICAL/HIGH):

   • Fix SQL injection in queries.ts
   • Remove hardcoded API key
   • Upgrade vulnerable dependencies

2. Short Term (MEDIUM):

   • Improve error handling consistency
   • Add input validation to all API endpoints
   • Configure stricter CORS policy

3. Long Term (LOW/INFO):

   • Reduce code duplication (12% current)
   • Refactor complex functions (8 functions >15 complexity)
   • Improve test coverage (currently 67%)

## Integration with Dev Plugin

### With Code Analysis Agent

Use code-analysis enrichment before audit:

First enrich the codebase with claudemem, then run security audit

This provides:
- Context-aware vulnerability detection
- Better false positive filtering
- Dependency graph analysis

### With Test Architect Agent

Combine audit with test coverage:

Run audit and identify untested critical code paths

### With Optimize Skill

Security-performance trade-offs:

Audit security implications of performance optimizations

## Best Practices

### 1. Regular Audits

**Recommended Schedule**:
- Daily: Dependency vulnerability scans (CI/CD)
- Weekly: Full security audit
- Pre-deployment: Comprehensive audit
- Post-incident: Targeted security review

### 2. Incremental Audits

For large codebases:

Audit files changed in the last 7 days

This focuses on recent changes and reduces noise.

### 3. Baseline and Track

**First Audit**:

Run full audit and establish security baseline

**Subsequent Audits**:

Run audit and compare against baseline

Track improvements over time.

### 4. Prioritize by Risk

Focus on:
1. User-facing authentication code
2. Payment processing
3. Data storage and retrieval
4. API endpoints with PII
5. Third-party integrations

### 5. Automate in CI/CD

**GitHub Actions Example**:
```yaml
- name: Security Audit
  run: |
    npm audit --audit-level=high
    # Run custom audit script

Pre-commit Hook:

#!/bin/bash
# Scan staged files for secrets
git diff --cached --name-only | xargs grep -E "(api_key|secret|password)"

Examples

Example 1: Pre-Deployment Security Audit

Request:

We're deploying to production tomorrow. Run a comprehensive security audit.

Audit Process:

1. Scan all source files for OWASP top 10
2. Check all dependencies for CVEs
3. Search for exposed secrets
4. Verify CORS and authentication config
5. Check logging and monitoring setup

Report Highlights:

• 1 CRITICAL: Exposed database password
• 2 HIGH: Outdated dependencies with CVEs
• 5 MEDIUM: Missing input validation
• Compliance: 8/10 OWASP categories passing

Outcome: Deployment blocked until CRITICAL/HIGH fixed

Example 2: Dependency Vulnerability Scan

Request:

Check all npm dependencies for known vulnerabilities

Process:

1. Run npm audit --json
2. Parse vulnerability reports
3. Check for available fixes
4. Identify breaking changes

Report:

Found 3 vulnerabilities (1 high, 2 moderate)

HIGH: axios@0.21.1 (CVE-2021-3749)
- Fix available: axios@0.21.4
- Breaking: No
- Run: npm install axios@0.21.4

MODERATE: glob-parent@5.1.1 (CVE-2020-28469)
- Fix available: glob-parent@6.0.2
- Breaking: Yes (ESM only)
- Review: Migration guide needed

Example 3: Compliance Check (GDPR)

Request:

Audit for GDPR compliance issues

Checks:

• Data retention policies
• User consent mechanisms
• Right to erasure implementation
• Data export functionality
• Encryption at rest and in transit
• Privacy policy references
• Third-party data sharing

Report:

GDPR Compliance Audit

[PASS] Data encryption at rest
[FAIL] Missing data retention policy
[FAIL] No user data export endpoint
[PASS] Consent management implemented
[WARN] Privacy policy link outdated
[PASS] TLS 1.3 enforced

Example 4: Code Quality Audit

Request:

Identify code quality issues and technical debt

Analysis:

• Complexity: 8 functions exceed threshold
• Duplication: 15% code duplication
• Dead code: 23 unused exports
• Large files: 4 files over 500 lines
• Magic numbers: 47 instances

Top Issues:

1. UserService.ts - Complexity 28 (threshold: 15)
   Recommendation: Extract validation logic

2. utils/helpers.ts - 234 lines duplicated in 3 files
   Recommendation: Create shared utility module

3. api/routes.ts - 847 lines (threshold: 500)
   Recommendation: Split into feature-based modules

Stack-Specific Audit Patterns

React/TypeScript

Security Focus:

• XSS via dangerouslySetInnerHTML
• Insecure refs and DOM manipulation
• Exposed environment variables in client code
• CORS misconfigurations

Quality Focus:

• Prop types and TypeScript coverage
• Component complexity
• Hook dependencies
• Re-render performance

Go

Security Focus:

• SQL injection in database/sql
• Command injection in exec
• Path traversal
• Goroutine leaks

Quality Focus:

• Error handling consistency
• Context propagation
• Resource cleanup (defer)
• Race conditions

Rust

Security Focus:

• Unsafe blocks
• Integer overflow
• Memory leaks in FFI
• Dependency vulnerabilities

Quality Focus:

• Error propagation patterns
• Clippy warnings
• Documentation coverage
• Panic usage

Audit Workflow Integration

[Developer] → Request audit
    ↓
[Audit Skill] → Scan codebase
    ↓
[Generate Report] → Categorize findings
    ↓
[Prioritize] → CRITICAL → HIGH → MEDIUM → LOW
    ↓
[Fix Critical] → Apply fixes
    ↓
[Re-audit] → Verify fixes
    ↓
[Update Baseline] → Track progress

Conclusion

The audit skill provides comprehensive security and quality analysis on-demand. Use it regularly to maintain code health, catch vulnerabilities early, and ensure compliance with industry standards.

Key Takeaways:

• Run audits frequently (daily in CI/CD)
• Prioritize CRITICAL and HIGH findings
• Establish baselines and track progress
• Integrate with other dev skills for comprehensive analysis
• Automate where possible

For performance analysis, see the optimize skill. For test coverage gaps, see the test-coverage skill.