Web Security Implementation

Implement defense-in-depth web application security using HTTP headers, input validation, output encoding, and authentication hardening.

Guiding Principle

"Security is not a feature you bolt on — it's a property of every decision in the development lifecycle. The cheapest vulnerability to fix is the one you never wrote."

---

Procedure

Step 1 — Security Audit [HECHO]

1. Run npx audit-ci or npm audit --audit-level=high — fix critical/high CVEs first.
2. Scan HTTP headers with curl -I https://yourdomain.com — check for missing security headers.
3. Test with securityheaders.com — free graded report.
4. Run OWASP ZAP active scan against staging environment.
5. Check for secrets in codebase: npx trufflesecurity/trufflehog --regex --entropy=False .
6. Inspect CSP: look for unsafe-inline, unsafe-eval, * wildcards.
7. Tag all findings: [HECHO] for confirmed issues, [SUPUESTO] for untested assumptions.

Step 2 — Content Security Policy (CSP)

1. Start with report-only mode to learn your baseline:

   Content-Security-Policy-Report-Only: default-src 'self'; report-uri /csp-report
   

2. Collect violations for 1-2 weeks, analyze patterns.
3. Build a strict policy starting from default-src 'none':

   Content-Security-Policy:
     default-src 'none';
     script-src 'self' 'nonce-{random}';
     style-src 'self' 'nonce-{random}';
     img-src 'self' data: https://cdn.yourdomain.com;
     font-src 'self';
     connect-src 'self' https://api.yourdomain.com;
     frame-ancestors 'none';
     base-uri 'self';
     form-action 'self';
   

4. Use nonces (preferred over hashes) for inline scripts when needed:

   // Express: generate nonce per request
   res.locals.nonce = crypto.randomBytes(16).toString('base64');
   res.setHeader('Content-Security-Policy', `script-src 'nonce-${res.locals.nonce}'`);
   

5. Add upgrade-insecure-requests to force HTTPS.
6. Never use unsafe-eval — it enables XSS via eval(). Refactor code that requires it.

Step 3 — CORS Configuration

1. Identify all origins that need cross-origin access — be explicit, no wildcards.
2. Implement CORS at the application layer:

   // Express + cors package
   const corsOptions = {
     origin: (origin, callback) => {
       const allowed = ['https://app.yourdomain.com', 'https://admin.yourdomain.com'];
       if (!origin || allowed.includes(origin)) callback(null, true);
       else callback(new Error('Not allowed by CORS'));
     },
     methods: ['GET', 'POST', 'PUT', 'DELETE'],
     allowedHeaders: ['Content-Type', 'Authorization'],
     credentials: true,
     maxAge: 86400, // Cache preflight for 24 hours
   };
   

3. For credentials (cookies, auth headers): Access-Control-Allow-Credentials: true AND Access-Control-Allow-Origin must be specific domain (not *).
4. Restrict allowed methods to only what the API uses.
5. Never reflect the Origin header directly without validation — this creates a bypass.

Step 4 — XSS Prevention

1. Output encoding: escape all user-controlled data before rendering.
   • HTML context: &, <, >, ", ' → HTML entities
   • JS context: use JSON.stringify() or a dedicated encoder
   • URL context: encodeURIComponent()
2. In React/Vue/Angular: use the framework's built-in escaping. Never use:
   • React: dangerouslySetInnerHTML with user data
   • Vue: v-html with user data
   • Angular: [innerHTML] with user data — use DomSanitizer
3. Implement DOMPurify for legitimate rich-text user content:

   import DOMPurify from 'dompurify';
   const clean = DOMPurify.sanitize(userHTML, { ALLOWED_TAGS: ['b', 'i', 'em'] });
   

4. Set HttpOnly flag on cookies to prevent JS access.
5. CSP script-src 'nonce-...' is the last line of defense — not the first.

Step 5 — CSRF Protection

1. Same-Site cookies (recommended, works for most cases):

   Set-Cookie: session=abc123; SameSite=Strict; Secure; HttpOnly; Path=/
   

   • SameSite=Strict: cookie not sent on cross-site requests (most secure).
   • SameSite=Lax: cookie sent on top-level GET navigations (good default).
2. CSRF tokens (for APIs or when SameSite isn't sufficient):

   // Generate token per session
   const csrfToken = crypto.randomBytes(32).toString('hex');
   // Include in forms and validate on POST
   if (req.body.csrf !== req.session.csrf) throw new Error('CSRF validation failed');
   

3. Double Submit Cookie pattern for stateless APIs:
   • Set a random cookie value, require same value in custom request header.
   • Attacker cannot read the cookie from a different origin.
4. Custom request header check: Simple X-Requested-With: XMLHttpRequest prevents form-based CSRF.
5. Verify Origin or Referer header matches expected domain for sensitive mutations.

Step 6 — Security Headers

Implement all of the following headers:

# Prevent clickjacking
X-Frame-Options: DENY

# Enable XSS filter in older browsers
X-XSS-Protection: 1; mode=block

# Prevent MIME sniffing
X-Content-Type-Options: nosniff

# Force HTTPS for 2 years
Strict-Transport-Security: max-age=63072000; includeSubDomains; preload

# Control referrer information
Referrer-Policy: strict-origin-when-cross-origin

# Restrict browser features
Permissions-Policy: camera=(), microphone=(), geolocation=(), payment=()

# CSP (configured in Step 2)
Content-Security-Policy: ...

Step 7 — Input Validation and Injection Prevention

1. Validate on the server, always — client-side validation is UX, not security.
2. Use schema validation: Zod (TypeScript), Joi (Node.js), Pydantic (Python).
3. SQL injection prevention — use parameterized queries:

   // Wrong
   db.query(`SELECT * FROM users WHERE id = ${userId}`);
   // Right
   db.query('SELECT * FROM users WHERE id = $1', [userId]);
   

4. Never pass user input to eval(), Function(), child_process.exec(), vm.runInContext().
5. Validate file uploads: check MIME type (by magic bytes, not extension), limit size, scan with ClamAV.
6. Rate limit authentication endpoints: express-rate-limit or API gateway throttling.

Step 8 — Verification

1. Re-run securityheaders.com — target A+ grade.
2. Run npm audit — zero critical/high vulnerabilities.
3. Test CSP with csp-evaluator.withgoogle.com.
4. Verify CORS: curl -H "Origin: https://evil.com" -I https://yourdomain.com/api — should not return Access-Control-Allow-Origin: https://evil.com.
5. Test CSRF: attempt cross-origin POST without token — must be rejected.
6. Test XSS: inject <script>alert(1)</script> in all user inputs — must be encoded or blocked by CSP.

---

Decision Framework

CSP Strictness Level

Environment	Approach
New app (full control)	Nonce-based strict CSP, no unsafe- directives
App with inline scripts (legacy)	Start report-only, migrate to nonces incrementally
Third-party heavy (GTM, analytics)	Allowlist specific domains, consider trusted types
Marketing site with CMS	Hash-based CSP for CMS-generated scripts

Cookie Security

Scenario	SameSite	HttpOnly	Secure
Session cookie	Strict	Yes	Yes
Auth token	Lax	Yes	Yes
Preferences (non-sensitive)	Lax	No	Yes
Cross-site OAuth flow	None	Yes	Yes (required when None)

---

Anti-Patterns

1. Access-Control-Allow-Origin: * with credentials — Browsers block this combination, but some misconfigured proxies allow it. Always use specific origins.
2. Storing JWT in localStorage — Accessible to XSS. Use HttpOnly cookies. If localStorage is required, accept the XSS risk and mitigate via CSP.
3. unsafe-inline in CSP — Defeats 90% of CSP's XSS protection value. Migrate to nonces or hashes.
4. Reflecting Origin without validation — Some CORS middleware reflects Origin directly. Always validate against an allowlist.
5. Security through obscurity — Hiding admin routes, using non-standard ports, or obfuscating JS does not replace proper authentication.
6. Missing Secure flag on cookies in production — Without Secure, session cookies transmit over HTTP, enabling MITM theft.
7. Broad Permissions-Policy — Default allow on geolocation, camera, payment opens unnecessary attack surface. Default-deny all browser features.

---

Output Artifacts

• security-audit-{date}.md — Findings with CVE references and severity
• middleware/security-headers.ts — Security headers middleware
• middleware/cors.ts — CORS configuration
• middleware/csrf.ts — CSRF protection middleware
• docs/CSP-policy.md — Documented CSP policy with rationale
• ADR-SEC-001.md — Security architecture decisions