Security patterns and OWASP guidelines. Triggers on: security review, OWASP, XSS, SQL injection, CSRF, authentication, authorization, secrets management, input validation, secure coding.
This skill is limited to using the following tools:
references/auth-patterns.mdreferences/crypto-patterns.mdreferences/owasp-detailed.mdreferences/secure-headers.mdscripts/dependency-audit.shscripts/security-scan.shEssential security patterns for web applications.
| Rank | Vulnerability | Prevention |
|---|---|---|
| A01 | Broken Access Control | Check permissions server-side, deny by default |
| A02 | Cryptographic Failures | Use TLS, hash passwords, encrypt sensitive data |
| A03 | Injection | Parameterized queries, validate input |
| A04 | Insecure Design | Threat modeling, secure defaults |
| A05 | Security Misconfiguration | Harden configs, disable unused features |
| A06 | Vulnerable Components | Update dependencies, audit regularly |
| A07 | Auth Failures | MFA, rate limiting, secure session management |
| A08 | Data Integrity Failures | Verify signatures, use trusted sources |
| A09 | Logging Failures | Log security events, protect logs |
| A10 | SSRF | Validate URLs, allowlist destinations |
# WRONG - Trust user input
def search(query):
return db.execute(f"SELECT * FROM users WHERE name = '{query}'")
# CORRECT - Parameterized query
def search(query):
return db.execute("SELECT * FROM users WHERE name = ?", [query])
Always validate:
- Type (string, int, email format)
- Length (min/max bounds)
- Range (numeric bounds)
- Format (regex for patterns)
- Allowlist (known good values)
Never trust:
- URL parameters
- Form data
- HTTP headers
- Cookies
- File uploads
// WRONG - Direct HTML insertion
element.innerHTML = userInput;
// CORRECT - Text content (auto-escapes)
element.textContent = userInput;
// CORRECT - Template with escaping
render(`<div>${escapeHtml(userInput)}</div>`);
| Context | Encoding |
|---|---|
| HTML body | HTML entity encode |
| HTML attribute | Attribute encode + quote |
| JavaScript | JS encode |
| URL parameter | URL encode |
| CSS | CSS encode |
# Password hashing (use bcrypt, argon2, or scrypt)
import bcrypt
def hash_password(password: str) -> bytes:
return bcrypt.hashpw(password.encode(), bcrypt.gensalt(rounds=12))
def verify_password(password: str, hashed: bytes) -> bool:
return bcrypt.checkpw(password.encode(), hashed)
# WRONG - Check only authentication
@login_required
def delete_post(post_id):
post = Post.get(post_id)
post.delete()
# CORRECT - Check authorization
@login_required
def delete_post(post_id):
post = Post.get(post_id)
if post.author_id != current_user.id and not current_user.is_admin:
raise Forbidden("Not authorized to delete this post")
post.delete()
# WRONG - Hardcoded secrets
API_KEY = "sk-1234567890abcdef"
# CORRECT - Environment variables
API_KEY = os.environ["API_KEY"]
# BETTER - Secrets manager
API_KEY = secrets_client.get_secret("api-key")
DO:
- Use environment variables or secrets manager
- Rotate secrets regularly
- Use different secrets per environment
- Audit secret access
DON'T:
- Commit secrets to git
- Log secrets
- Include secrets in error messages
- Share secrets in plain text
Content-Security-Policy: default-src 'self'; script-src 'self'
X-Content-Type-Options: nosniff
X-Frame-Options: DENY
Strict-Transport-Security: max-age=31536000; includeSubDomains
Referrer-Policy: strict-origin-when-cross-origin
Permissions-Policy: geolocation=(), camera=()
# Find hardcoded secrets
rg -i "(password|secret|api_key|token)\s*=\s*['\"][^'\"]+['\"]" --type py
# Find SQL injection risks
rg "execute\(f['\"]|format\(" --type py
# Find eval/exec usage
rg "\b(eval|exec)\s*\(" --type py
# Check for TODO security items
rg -i "TODO.*security|FIXME.*security"
./references/owasp-detailed.md - Full OWASP Top 10 details./references/auth-patterns.md - JWT, OAuth, session management./references/crypto-patterns.md - Encryption, hashing, signatures./references/secure-headers.md - HTTP security headers guide./scripts/security-scan.sh - Quick security grep patterns./scripts/dependency-audit.sh - Check for vulnerable dependencies