Security Patterns

Security Patterns

Purpose: Security best practices (validation, PII protection, crypto, defense against vulnerabilities)

• Keywords: security, validation, auth, password, token, sanitize, xss, sql injection, zod, pii, webhook, signature, verify, encrypt, decrypt, hash, jwt, oauth, csrf, authentication, authorization, stripe

Quick Reference

Input Validation (Zod)

import { z } from "zod"

const orderSchema = z.object({
  name: z.string().min(1).max(100).trim(),
  email: z.string().email().max(255).toLowerCase(),
  amount: z.number().positive().max(1000000),
  phone: z.string().regex(/^\+?[1-9]\d{1,14}$/).optional()
})

type Result<T, E> = { ok: true; value: T } | { ok: false; error: E }

const validate = (input: unknown): Result<Order, z.ZodError> => {
  const result = orderSchema.safeParse(input)
  return result.success
    ? { ok: true, value: result.data }
    : { ok: false, error: result.error }
}

Key: Use .safeParse(), add constraints (.min(), .max(), .trim()), return Result

PII Protection

import { sanitizeForLogging } from "./utils/sanitize"

// ✅ Auto-removes PII
logger.audit('order_created', sanitizeForLogging({
  user_id: "123",
  full_name: "John", // Removed
  email: "john@example.com", // Removed
  amount: 100
}))
// Logged: { user_id: "123", amount: 100 }

Auto-removed: Passwords, tokens, API keys, emails, names, SSN, cards, Bitcoin keys/addresses, phones

See resources/pii-protection.md for implementation.

Secure Random

import { randomInt } from "crypto"

// ✅ Cryptographically secure
const sessionId = randomInt(0, Number.MAX_SAFE_INTEGER).toString(36)
const code = randomInt(100000, 999999).toString()

// ✅ Browser
const array = new Uint32Array(1)
crypto.getRandomValues(array)

// ❌ NEVER for security
const code = Math.random() * 1000000

Secure Client Storage

import { encrypt, decrypt } from "./crypto"

const saveSession = (data: SessionData) => {
  const encrypted = encrypt(JSON.stringify(data))
  localStorage.setItem("session", encrypted)
}

// ❌ NEVER store Bitcoin keys/passwords (even encrypted)

Error Handling

export const createOrder = async (input: unknown) => {
  try {
    const order = await processOrder(input)
    return { ok: true, value: order }
  } catch (error) {
    // ✅ Log full details server-side
    logger.error('Order failed', {
      error,
      stack: error.stack,
      input: sanitizeForLogging(input)
    })

    // ✅ Generic to client
    return {
      ok: false,
      error: new Error("Failed to create order")
    }
  }
}

Webhook Verification

export const handleStripeWebhook = async (req: Request) => {
  const signature = req.headers.get("stripe-signature")
  const body = await req.text()

  // ✅ Verify signature BEFORE processing
  if (!verifySignature(body, signature, WEBHOOK_SECRET)) {
    logger.warn('Invalid signature', { ip: req.headers.get("x-forwarded-for") })
    return new Response("Invalid signature", { status: 401 })
  }

  // ✅ Check timestamp (replay attack)
  const event = JSON.parse(body)
  const eventTime = event.created * 1000
  if (Date.now() - eventTime > 5 * 60 * 1000) {
    return new Response("Timestamp too old", { status: 400 })
  }

  await processWebhookEvent(event)
  return new Response("OK", { status: 200 })
}

Pre-Implementation Checklist

Before implementing, ask:

1. ✅ Can users access data they shouldn't? (Authorization)
2. ✅ Can this be automated/abused? (Rate limiting)
3. ✅ Am I trusting client data? (Validation)
4. ✅ Could this leak PII? (Sanitization)
5. ✅ Can concurrent requests cause issues? (Race conditions)
6. ✅ Can this be spammed? (Rate limiting, CAPTCHA)

If ANY "yes" → redesign before implementing

Rate Limits

See resources/rate-limiting.md

Incident Response

Steps: Assess → Contain → Investigate → Patch → Notify (<72h if PII)

Resources

• resources/input-validation.md - Zod patterns
• resources/pii-protection.md - Sanitization
• resources/crypto-operations.md - Secure random, encryption
• resources/rate-limiting.md - Rate limit implementation