Production Patterns

Use one client factory

Create a single shared client factory and apply environment-aware config.

export function createCommerceClient() {
  return createSazitoClient({
    domain: process.env.NEXT_PUBLIC_STORE_DOMAIN!,
    timeout: 20000,
    retry: { enabled: true, retries: 2, retryDelay: 700 },
    debug: process.env.NODE_ENV !== 'production',
  });
}

Normalize error handling

Implement one utility that maps SazitoResponse.error to your UI or API error schema.

// src/lib/sazito-error.ts
import type { SazitoResponse } from '@sazito/client-sdk';

export type AppError = {
  code: 'VALIDATION_ERROR' | 'NETWORK_ERROR' | 'API_ERROR' | 'UNKNOWN_ERROR';
  message: string;
  status: number;
  retryable: boolean;
  details?: unknown;
};

export function mapSazitoError(error: SazitoResponse<unknown>['error']): AppError {
  if (!error) {
    return {
      code: 'UNKNOWN_ERROR',
      message: 'Unknown error',
      status: 500,
      retryable: false,
    };
  }

  if (error.type === 'validation') {
    return {
      code: 'VALIDATION_ERROR',
      message: error.message,
      status: error.status ?? 400,
      retryable: false,
      details: error.details,
    };
  }

  if (error.type === 'network') {
    return {
      code: 'NETWORK_ERROR',
      message: error.message,
      status: error.status ?? 503,
      retryable: true,
      details: error.details,
    };
  }

  return {
    code: 'API_ERROR',
    message: error.message,
    status: error.status ?? 500,
    retryable: (error.status ?? 500) >= 500,
    details: error.details,
  };
}

// UI usage
const res = await client.products.list({ page: 1, pageSize: 12 });
if (res.error) {
  const appError = mapSazitoError(res.error);
  setError(appError.message);
  return;
}

// API route usage
const res = await client.orders.get(123);
if (res.error) {
  const appError = mapSazitoError(res.error);

  return NextResponse.json(
    { error: appError.code, message: appError.message, details: appError.details },
    { status: appError.status }
  );
}

Use request-level controls for critical paths

Set lower timeout for page-critical requests.
Disable cache for real-time cart and payment state.
Attach request IDs through headers for tracing.

Separate public and authenticated clients

For SSR or edge middleware, isolate token injection strategy by runtime context.

Keep checkout state clean

Call credential cleanup APIs after completed/abandoned checkout to prevent stale state leakage.