Security By Design

Security by Design Skill

🔴 AI FIRST Quality Principle

Apply the AI FIRST principle: never accept first-pass quality. Minimum 2 iterations. Read all output, improve every section. No shortcuts.

Purpose

Apply security by design principles to ensure security is integrated from the earliest stages of development, not bolted on as an afterthought.

Core Principles

1. Secure by Default

• Principle: Systems should be secure in their default configuration
• Application:
  • Default to HTTPS, never HTTP
  • Default to least privilege access
  • Default to encrypted communications
  • Default to secure password policies
  • Disable unnecessary features by default

2. Defense in Depth

• Principle: Multiple layers of security controls protect against single point of failure
• Application:
  • Network security (TLS, firewalls, DDoS protection)
  • Application security (input validation, output encoding)
  • Access control (authentication, authorization, MFA)
  • Data security (encryption at rest and in transit)
  • Monitoring and detection (logging, SIEM, alerts)

3. Least Privilege

• Principle: Grant minimum necessary permissions to users, processes, and systems
• Application:
  • GitHub Actions: minimal required permissions
  • User roles: grant only needed access
  • Service accounts: scoped credentials
  • API tokens: limited scope and expiration
  • File permissions: restrictive by default

4. Fail Securely

• Principle: Failures should not compromise security
• Application:
  • Error messages don't leak sensitive info
  • Failed authentication locks out, doesn't grant access
  • Exception handling preserves security state
  • Graceful degradation maintains protection
  • Logs capture security-relevant failures

5. Don't Trust User Input

• Principle: All external input is untrusted and must be validated
• Application:
  • Validate all input formats
  • Sanitize before processing
  • Use parameterized queries
  • Encode output appropriately
  • Whitelist over blacklist

6. Keep Security Simple

• Principle: Complexity is the enemy of security
• Application:
  • Use well-tested libraries over custom code
  • Avoid complex cryptographic implementations
  • Clear, reviewable security logic
  • Minimize attack surface
  • Remove unused features and dependencies

7. Separation of Duties

• Principle: Critical operations require multiple parties
• Application:
  • Code review required before merge
  • Separate dev/prod environments
  • Different roles for different functions
  • No single person can compromise system
  • Audit trail for accountability

8. Economy of Mechanism

• Principle: Keep security mechanisms as simple as possible
• Application:
  • Avoid over-engineering security
  • Use standard protocols (TLS, OAuth)
  • Leverage platform security features
  • Document security architecture clearly
  • Regular security architecture reviews

Static Website Security by Design

Eliminate Server-Side Attack Surface

✅ Benefits of static HTML/CSS:
- No SQL injection (no database)
- No XSS from server-side rendering
- No CSRF tokens needed
- No session management vulnerabilities
- No server-side code execution risks

Transport Layer Security

✅ Always enforce:
- HTTPS-only (TLS 1.3)
- HSTS headers (Strict-Transport-Security)
- Secure cookie flags (if any cookies used)
- Certificate pinning where appropriate

Content Security

✅ Implement CSP headers:
Content-Security-Policy: default-src 'self'; 
  script-src 'self'; 
  style-src 'self' 'unsafe-inline' fonts.googleapis.com;
  font-src 'self' fonts.gstatic.com;
  img-src 'self' data:;
  connect-src 'self';

✅ Additional headers:
X-Content-Type-Options: nosniff
X-Frame-Options: DENY
X-XSS-Protection: 1; mode=block
Referrer-Policy: strict-origin-when-cross-origin

Dependency Security

✅ Minimize dependencies:
- Avoid JavaScript frameworks (for static sites)
- Use trusted CDNs (Google Fonts, etc.)
- Enable Subresource Integrity (SRI) for CDN assets
- Regular dependency scanning (Dependabot)

CI/CD Security by Design

Workflow Hardening

# ✅ Security by design in GitHub Actions
permissions:
  contents: read  # Least privilege
  pull-requests: write  # Only if needed

jobs:
  secure-job:
    runs-on: ubuntu-latest
    
    steps:
      # 1. Harden the runner
      - uses: step-security/harden-runner@SHA
        with:
          egress-policy: audit
      
      # 2. Pin all actions to SHA
      - uses: actions/checkout@SHA
      
      # 3. Use secrets securely
      - env:
          TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          # Never echo secrets
          # Use secrets only where needed

Supply Chain Security

✅ Required controls:
- Pin all GitHub Actions to SHA commits
- Enable Dependabot security updates
- Use dependency review action
- Scan with CodeQL or Semgrep
- Enable secret scanning
- Require signed commits (GPG)

Secure Development Lifecycle

Phase 1: Design

• Threat model (STRIDE analysis)
• Security requirements defined
• Compliance requirements identified
• Security architecture designed
• Risk assessment documented

Phase 2: Development

• Secure coding standards followed
• Input validation implemented
• Output encoding applied
• Error handling secure
• Logging captures security events

Phase 3: Testing

• SAST scanning (CodeQL)
• Dependency scanning (Dependabot)
• Secret scanning enabled
• Manual security review
• Penetration testing (if applicable)

Phase 4: Deployment

• Secure CI/CD pipeline
• Infrastructure hardening
• Security monitoring enabled
• Incident response ready
• Rollback plan documented

Phase 5: Operations

• Continuous monitoring
• Log analysis
• Vulnerability management
• Patch management
• Periodic security reviews

Security Design Patterns

Pattern: Static Site Security

Architecture: Static HTML/CSS on GitHub Pages

Security Controls:
✅ Attack Surface: Minimal (no server-side code)
✅ Transport: TLS 1.3 / HTTPS-only
✅ Headers: CSP, HSTS, X-Frame-Options
✅ Access: GitHub MFA, SSH keys, GPG signing
✅ Monitoring: Dependabot, CodeQL, secret scanning
✅ Recovery: Git rollback, rapid re-deploy

Pattern: Defense in Depth Layers

Layer 1 (Network): TLS 1.3, CDN, DDoS protection
Layer 2 (Application): Static content, no dynamic processing
Layer 3 (Access Control): GitHub auth, MFA, SSH, GPG
Layer 4 (Data): Git encryption at rest, HTTPS in transit
Layer 5 (Monitoring): Logs, alerts, security scanning
Layer 6 (Response): Incident procedures, rollback capability

Pattern: Least Privilege GitHub Actions

permissions:
  # Grant only what's needed
  contents: read
  pull-requests: write  # Only if creating/updating PRs
  issues: write         # Only if managing issues
  # Deny everything else by default

Security Anti-Patterns to Avoid

❌ Security Theater

• Don't implement controls that look secure but aren't effective
• Don't use outdated or weak crypto (MD5, SHA1, RC4)
• Don't rely on obscurity (hiding, not securing)

❌ Bolt-On Security

• Don't add security as afterthought
• Don't patch over insecure design
• Don't bypass security for convenience

❌ Over-Engineering

• Don't create complex custom security solutions
• Don't implement crypto yourself
• Don't ignore well-tested libraries

❌ Incomplete Implementation

• Don't secure one layer and ignore others
• Don't validate input but forget output encoding
• Don't encrypt data in transit but not at rest

Verification Checklist

Before deploying, verify:

• Threat Model: STRIDE analysis complete
• Secure Defaults: All defaults are secure
• Defense in Depth: Multiple security layers
• Least Privilege: Minimal permissions granted
• Input Validation: All inputs validated
• Output Encoding: All outputs properly encoded
• Error Handling: Fails securely, no info leakage
• Logging: Security events captured
• Access Control: Authentication + authorization
• Encryption: Data protected in transit and at rest
• Dependencies: Scanned and up to date
• Testing: Security tests passing
• Monitoring: Alerts configured
• Documentation: Security architecture documented
• Incident Response: Procedures documented and tested

Remember

• Design First: Security requirements before code
• Simple is Secure: Complexity breeds vulnerabilities
• Layers Matter: Defense in depth protects when one layer fails
• Least Privilege Always: Grant minimum necessary access
• Fail Securely: Errors should not compromise security
• Trust Nothing: Validate all external input
• Document Everything: Security decisions need justification

References

• OWASP Security by Design Principles
• NIST SP 800-160 Vol. 1
• Microsoft SDL
• STRIDE Threat Modeling