Security Scan SKILL

🎯 Purpose

セキュリティ脆弱性の検出: Identify security vulnerabilities and ensure secure coding practices. Automatically invoked during codex-review for security-sensitive code.

📋 When to Use

Automatic Triggers (via codex-review)

• Authentication/Authorization code
• Database queries (SQL injection risk)
• User input handling (XSS/injection risk)
• File operations (path traversal risk)
• Cryptography usage
• API endpoints
• Session management

Manual Invocation

• User explicitly requests security scan
• Before deploying security-critical features
• After dependency updates

🔐 Security Focus Areas

1. Input Validation

Risks:

• SQL Injection
• XSS (Cross-Site Scripting)
• Command Injection
• Path Traversal
• LDAP Injection
• XML/XXE Injection

Checks:

// ❌ VULNERABLE
const query = `SELECT * FROM users WHERE id = ${userId}`;

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

2. Authentication & Authorization

Risks:

• Weak password policies
• Insecure session management
• Missing authentication
• Broken access control
• Privilege escalation

Checks:

// ❌ VULNERABLE
if (user.role === 'admin') {
  // No verification of user identity
}

// ✅ SECURE
if (authenticatedUser.id === user.id && user.role === 'admin') {
  // Verify both identity and role
}

3. Data Protection

Risks:

• Sensitive data exposure
• Insecure cryptography
• Weak encryption
• Plaintext credentials
• Insufficient SSL/TLS

Checks:

// ❌ VULNERABLE
const password = req.body.password; // Plaintext
localStorage.setItem('token', token); // Insecure storage

// ✅ SECURE
const hashedPassword = await bcrypt.hash(password, 10);
// Use httpOnly, secure cookies instead

4. Dependencies

Risks:

• Known vulnerabilities (CVEs)
• Outdated packages
• Supply chain attacks
• License compliance issues

Checks:

• npm audit / go mod verify / pip check
• Dependency version analysis
• Vulnerability database lookup

🔍 Security Scan Process

Step 1: Identify Security-Sensitive Code

Scan for patterns:

- Database queries: (SELECT|INSERT|UPDATE|DELETE|query|exec)
- User input: (req\.body|params|query|input|form)
- Authentication: (auth|login|password|token|session)
- File operations: (readFile|writeFile|fs\.|path\.)
- Crypto: (crypto|encrypt|decrypt|hash|sign)
- Dangerous functions: (eval|exec|system|shell)

Step 2: Vulnerability Detection

Check against OWASP Top 10:

1. A01: Broken Access Control
2. A02: Cryptographic Failures
3. A03: Injection
4. A04: Insecure Design
5. A05: Security Misconfiguration
6. A06: Vulnerable and Outdated Components
7. A07: Identification and Authentication Failures
8. A08: Software and Data Integrity Failures
9. A09: Security Logging and Monitoring Failures
10. A10: Server-Side Request Forgery (SSRF)

Step 3: Risk Assessment

Severity Levels:

• Critical: Immediate exploitable vulnerability
• High: Significant security risk
• Medium: Potential security concern
• Low: Best practice violation
• Info: Security improvement opportunity

Step 4: Generate Security Report

Output with remediation guidance.

📊 Output Format to User

## セキュリティスキャン結果

### 概要
- **スキャンファイル**: 5ファイル
- **検出された問題**: 3件
  - Critical: 1件
  - High: 1件
  - Medium: 1件

### 🚨 Critical (即座に修正が必要)

#### 1. SQL Injection 脆弱性
- **ファイル**: `src/api/users.ts:45-48`
- **問題**: ユーザー入力を直接SQLクエリに埋め込んでいます
- **リスク**: データベース全体が侵害される可能性
- **OWASP**: A03:2021 - Injection

**脆弱なコード:**
```typescript
const query = `SELECT * FROM users WHERE email = '${email}'`;
db.query(query);

修正案:

const query = 'SELECT * FROM users WHERE email = ?';
db.query(query, [email]);
// または ORMを使用
const user = await User.findOne({ where: { email } });

影響: 攻撃者が任意のSQLを実行可能 修正優先度: 最高

---

⚠️ High (早急に修正を推奨)

2. 認証トークンの不適切な保存

• ファイル: src/auth/session.ts:23
• 問題: JWTトークンがlocalStorageに保存されています
• リスク: XSS攻撃でトークンが盗まれる可能性
• OWASP: A07:2021 - Identification and Authentication Failures

脆弱なコード:

localStorage.setItem('authToken', token);

修正案:

// httpOnly, secure cookieを使用
res.cookie('authToken', token, {
  httpOnly: true,
  secure: true,
  sameSite: 'strict',
  maxAge: 3600000
});

---

📋 Medium

3. パスワードハッシュの不十分な強度

• ファイル: src/auth/password.ts:12
• 問題: bcryptのsalt roundsが10未満
• 推奨: 12以上のsalt roundsを使用

---

✅ セキュリティベストプラクティス

検出された良好な実装:

• ✅ CORS設定が適切に構成されています
• ✅ CSRFトークン検証が実装されています
• ✅ HTTPSが強制されています

---

📚 推奨される追加対策

1. セキュリティヘッダーの追加

app.use(helmet({
  contentSecurityPolicy: true,
  hsts: true,
  noSniff: true
}));

2. レート制限の実装

const limiter = rateLimit({
  windowMs: 15 * 60 * 1000,
  max: 100
});
app.use('/api/', limiter);

3. 入力バリデーションライブラリの使用

import { z } from 'zod';

const userSchema = z.object({
  email: z.string().email(),
  password: z.string().min(8)
});

---

🔗 参考リソース

• OWASP Top 10 2021
• CWE/SANS Top 25
• Node.js Security Best Practices

これらの問題を修正してから進めますか?

## 🛡️ Language-Specific Checks

### JavaScript/TypeScript
```javascript
// Dangerous patterns
- eval() usage
- new Function() with user input
- innerHTML with unescaped data
- document.write()
- setTimeout/setInterval with string argument

// Security libraries
- helmet (security headers)
- express-rate-limit
- joi/zod (validation)
- bcrypt (password hashing)

Go

// Dangerous patterns
- SQL string concatenation
- exec.Command() with user input
- filepath.Join() without validation
- crypto/md5, crypto/sha1 (weak hashing)

// Security libraries
- golang.org/x/crypto/bcrypt
- github.com/go-playground/validator
- database/sql with prepared statements

Python

# Dangerous patterns
- eval(), exec() with user input
- pickle.loads() on untrusted data
- SQL string formatting
- os.system() with user input

# Security libraries
- bcrypt
- SQLAlchemy (ORM)
- bleach (XSS prevention)
- cryptography

🔗 Integration with codex-review

Security scan runs automatically during codex-review:

codex-review triggers security-scan when detecting:
├─ SQL queries → Check for injection
├─ User input handling → Check for XSS/injection
├─ Auth code → Check authentication/authorization
├─ File operations → Check path traversal
├─ Crypto usage → Check weak algorithms
└─ Dependencies → Check known vulnerabilities

Security findings are included in codex-review output as blocking issues.

🔧 Configuration Parameters

Parameter	Default	Description
severity_threshold	medium	Minimum severity to report
include_dependencies	true	Check dependency vulnerabilities
owasp_checks	all	OWASP categories to check
auto_fix_suggestions	true	Provide code fix examples

⚠️ Important Reminders

1. Security is non-negotiable - All Critical/High issues are blocking
2. Output in Japanese for user-facing text
3. Provide specific code examples for fixes
4. Link to OWASP/CWE for educational value
5. Integrate with codex-review for automatic scanning
6. Check dependencies for known CVEs
7. Fail fast - Don't allow insecure code to proceed

📝 Security Checklist

Before marking code as secure:

• All inputs validated and sanitized
• SQL queries use prepared statements/ORM
• Authentication/authorization properly implemented
• Sensitive data encrypted at rest and in transit
• Security headers configured
• Dependencies up to date and vulnerability-free
• Error messages don't leak sensitive information
• Logging doesn't include sensitive data
• Rate limiting implemented for APIs
• CSRF protection enabled