AgentSkillsCN

security-auditor

针对 OWASP Top 10、常见漏洞和不安全模式的持续安全漏洞扫描。在审查代码、部署前或文件更改时使用。扫描 SQL 注入、XSS、秘密暴露、认证问题。在文件更改、安全提及、部署准备时触发。

SKILL.md
--- frontmatter
name: security-auditor
description: Continuous security vulnerability scanning for OWASP Top 10, common vulnerabilities, and insecure patterns. Use when reviewing code, before deployments, or on file changes. Scans for SQL injection, XSS, secrets exposure, auth issues. Triggers on file changes, security mentions, deployment prep.
allowed-tools: Read, Grep, Bash

Security Auditor Skill

Automatic security vulnerability detection.

When I Activate

  • •✅ Code files modified (especially auth, API, database)
  • •✅ User mentions security or vulnerabilities
  • •✅ Before deployments or commits
  • •✅ Dependency changes
  • •✅ Configuration file changes

What I Scan For

OWASP Top 10 Patterns

1. SQL Injection

javascript
// CRITICAL: SQL injection
const query = `SELECT * FROM users WHERE id = ${userId}`;

// SECURE: Parameterized query
const query = 'SELECT * FROM users WHERE id = ?';
db.query(query, [userId]);

2. XSS (Cross-Site Scripting)

javascript
// CRITICAL: XSS vulnerability
element.innerHTML = userInput;

// SECURE: Use textContent or sanitize
element.textContent = userInput;
// or
element.innerHTML = DOMPurify.sanitize(userInput);

3. Authentication Issues

javascript
// CRITICAL: Weak JWT secret
const token = jwt.sign(payload, 'secret123');

// SECURE: Strong secret from environment
const token = jwt.sign(payload, process.env.JWT_SECRET);

4. Sensitive Data Exposure

python
# CRITICAL: Exposed password
password = "admin123"

# SECURE: Environment variable
password = os.getenv("DB_PASSWORD")

5. Broken Access Control

javascript
// CRITICAL: No authorization check
app.delete('/api/users/:id', (req, res) => {
  User.delete(req.params.id);
});

// SECURE: Authorization check
app.delete('/api/users/:id', auth, checkOwnership, (req, res) => {
  User.delete(req.params.id);
});

Additional Security Checks

  • •Insecure Deserialization
  • •Security Misconfiguration
  • •Insufficient Logging
  • •CSRF Protection Missing
  • •CORS Misconfiguration

Alert Format

code
🚨 CRITICAL: [Vulnerability type]
📍 Location: file.js:42
🔧 Fix: [Specific remediation]
📖 Reference: [OWASP/CWE link]

Severity Levels

  • •🚨 CRITICAL: Must fix immediately (exploitable vulnerabilities)
  • •⚠️ HIGH: Should fix soon (security weaknesses)
  • •📋 MEDIUM: Consider fixing (potential issues)
  • •💡 LOW: Best practice improvements

Real-World Examples

SQL Injection Detection

javascript
// You write:
app.get('/users', (req, res) => {
  const sql = `SELECT * FROM users WHERE name = '${req.query.name}'`;
  db.query(sql, (err, results) => res.json(results));
});

// I alert:
🚨 CRITICAL: SQL injection vulnerability (line 2)
📍 File: routes/users.js, Line 2
🔧 Fix: Use parameterized queries
  const sql = 'SELECT * FROM users WHERE name = ?';
  db.query(sql, [req.query.name], ...);
📖 https://owasp.org/www-community/attacks/SQL_Injection

Password Storage

python
# You write:
def create_user(username, password):
    user = User(username=username, password=password)
    user.save()

# I alert:
🚨 CRITICAL: Storing plain text password (line 2)
📍 File: models.py, Line 2
🔧 Fix: Hash passwords before storing
  from bcrypt import hashpw, gensalt
  hashed = hashpw(password.encode(), gensalt())
  user = User(username=username, password=hashed)
📖 Use bcrypt, scrypt, or argon2 for password hashing

API Key Exposure

javascript
// You write:
const stripe = require('stripe')('sk_live_abc123...');

// I alert:
🚨 CRITICAL: Hardcoded API key detected (line 1)
📍 File: payment.js, Line 1
🔧 Fix: Use environment variables
  const stripe = require('stripe')(process.env.STRIPE_SECRET_KEY);
📖 Never commit API keys to version control

Dependency Scanning

I can run security audits on dependencies:

bash
# Node.js
npm audit

# Python
pip-audit

# Results flagged with severity

Relationship with @code-reviewer Sub-Agent

Me (Skill): Quick vulnerability pattern detection @code-reviewer (Sub-Agent): Deep security audit with threat modeling

Workflow

  1. •I detect vulnerability pattern
  2. •I flag: "🚨 SQL injection detected"
  3. •You want full analysis → Invoke @code-reviewer sub-agent
  4. •Sub-agent provides comprehensive security audit

Common Vulnerability Patterns

Authentication

  • •Weak password policies
  • •Missing MFA
  • •Session fixation
  • •Insecure password storage

Authorization

  • •Missing access control
  • •Privilege escalation
  • •IDOR (Insecure Direct Object Reference)

Data Protection

  • •Unencrypted sensitive data
  • •Weak encryption algorithms
  • •Missing HTTPS
  • •Insecure cookies

Input Validation

  • •SQL injection
  • •Command injection
  • •XSS
  • •Path traversal

Sandboxing Compatibility

Works without sandboxing: ✅ Yes Works with sandboxing: ✅ Yes

Optional: For dependency scanning

json
{
  "network": {
    "allowedDomains": [
      "registry.npmjs.org",
      "pypi.org",
      "api.github.com"
    ]
  }
}

Integration with Tools

With secret-scanner Skill

code
security-auditor: Checks code patterns
secret-scanner: Checks for exposed secrets
Together: Comprehensive security coverage

With /review Command

bash
/review --scope staged --checks security

# Workflow:
# 1. My automatic security findings
# 2. @code-reviewer sub-agent deep audit
# 3. Comprehensive security report

Customization

Add company-specific security patterns:

bash
cp -r ~/.claude/skills/security/security-auditor \
      ~/.claude/skills/security/company-security-auditor

# Edit SKILL.md to add:
# - Internal API patterns
# - Company security policies
# - Custom vulnerability checks

Learn More