API calls implement retry logic with exponential backoff

Transient network failures — DNS blips, cloud provider hiccups, brief overloads — are normal in distributed systems. Without retry logic, a single dropped packet causes a user-visible failure on an operation that would succeed 200ms later. ISO 25010 reliability.recoverability requires the application to restore a defined performance level after failure; immediate hard failures with no retry violate that objective. Equally important: retrying non-idempotent POST requests without user confirmation risks duplicate charges, duplicate submissions, or duplicate order creation — a common and costly AI-coding defect.

Implement retry with exponential backoff and jitter. Only auto-retry idempotent requests (GET, safe PUT). For POST/DELETE, surface a retry button to the user instead.

// lib/retry.ts
export async function withRetry<T>(
  fn: () => Promise<T>,
  { maxAttempts = 3, baseDelayMs = 200 } = {}
): Promise<T> {
  for (let attempt = 0; attempt < maxAttempts; attempt++) {
    try {
      return await fn()
    } catch (err) {
      if (attempt === maxAttempts - 1) throw err
      const delay = baseDelayMs * 2 ** attempt + Math.random() * baseDelayMs
      await new Promise((r) => setTimeout(r, delay))
    }
  }
  throw new Error('unreachable')
}

Jitter (the Math.random() term) prevents thundering-herd when multiple clients retry simultaneously after a shared outage.