2025-09-04
Zod + OpenAPI + AWS Lambda: How Documentation Drift Led Me to Schema-First Development
How a 'simple' API change broke an enterprise client integration overnight, why documentation drift causes real problems, and a practical system that generates OpenAPI specs from Zod schemas automatically.
A painful lesson about API documentation drift: adding an optional field to a user API without updating the OpenAPI spec broke an enterprise client’s integration overnight. Their code generation pipeline produced TypeScript interfaces expecting the old schema, resulting in hundreds of failed user registrations and significant revenue loss.
This incident highlighted that API documentation isn’t just nice-to-have - it’s critical business infrastructure. This approach involves rebuilding systems to generate OpenAPI specs automatically from Zod schemas, enabling safer API evolution.
The Critical Lesson: Why Documentation Drift Kills Businesses
Before our incident, we had the classic serverless API development problem: four different sources of truth:
// 1. TypeScript interfaces (what developers think the API does)
interface CreateUserRequest {
email: string;
username: string;
age?: number;
// I added this field...
company?: string;
}
// 2. OpenAPI spec (what clients generate code from)
const openApiSpec = {
paths: {
'/users': {
post: {
requestBody: {
// But forgot to update this
schema: {
type: 'object',
properties: {
email: { type: 'string' },
username: { type: 'string' },
age: { type: 'number' }
// Missing: company field
}
}
}
}
}
}
};
// 3. Lambda validation (inconsistent and incomplete)
if (!event.body.email || typeof event.body.email !== 'string') {
throw new Error('Invalid email');
}
// No validation for company field
// 4. Database schema (yet another source of truth)
CREATE TABLE users (
id UUID PRIMARY KEY,
email VARCHAR(255) NOT NULL,
username VARCHAR(50) NOT NULL,
age INTEGER,
company_name VARCHAR(100) -- Different field name!
);
Four different definitions. Four opportunities for bugs. One angry enterprise client.
The pain wasn’t just the immediate cost - it was the follow-up meetings, the trust damage, and the realization that our development approach needed fundamental changes.
The Solution: Single Source of Truth
Working from the principle of Single Source of Truth, this approach uses Zod schemas as the definitive API contract, automatically generating:
- Compile-time TypeScript types (no more interface drift)
- OpenAPI 3.0 specifications (always in sync)
- Runtime validation (catch bad data before it hits your database)
- Structured error responses (clients know exactly what went wrong)
- Database migrations (with custom tooling)
Benefits observed in production:
- Reduced integration failures from schema drift
- Less time spent on manual spec maintenance
- Smoother client onboarding with accurate, auto-generated SDKs
- Clearer error messages leading to fewer support questions
Here’s the architecture:
The Foundation That Prevents Schema Drift
Here’s the setup that supports safe API evolution:
# Core dependencies
npm install zod @anatine/zod-openapi @asteasolutions/zod-to-openapi
npm install uuid @aws-sdk/client-dynamodb @aws-sdk/lib-dynamodb
npm install --save-dev @types/aws-lambda @types/uuid
# For production monitoring
npm install @aws-lambda-powertools/logger @aws-lambda-powertools/tracer @aws-lambda-powertools/metrics
The Type System That Catches Bugs at Compile Time
Here’s a foundation that helps prevent schema drift:
// lib/api/types.ts - Core schema definitions
import { z } from 'zod';
import { extendZodWithOpenApi } from '@anatine/zod-openapi';
import {
APIGatewayProxyEventV2,
APIGatewayProxyResultV2,
Context
} from 'aws-lambda';
// Extend Zod with OpenAPI capabilities
extendZodWithOpenApi(z);
// Error schema for structured error responses
export const ErrorResponseSchema = z.object({
error: z.string().openapi({
example: 'Validation failed',
description: 'Human-readable error message'
}),
details: z.array(z.object({
path: z.string().openapi({
example: 'email',
description: 'The field that failed validation'
}),
message: z.string().openapi({
example: 'Invalid email format',
description: 'Specific validation error'
}),
code: z.string().openapi({
example: 'INVALID_EMAIL',
description: 'Machine-readable error code'
})
})).optional(),
requestId: z.string().uuid().openapi({
example: '123e4567-e89b-12d3-a456-426614174000',
description: 'Unique request identifier for debugging'
}),
timestamp: z.string().datetime().openapi({
example: '2023-01-15T10:30:45.123Z',
description: 'When the error occurred'
})
}).openapi('ErrorResponse');
// Response wrapper that provides consistent client experience
export const ApiResponseSchema = <T extends z.ZodType>(dataSchema: T) =>
z.object({
success: z.boolean().openapi({
example: true,
description: 'Whether the request succeeded'
}),
data: dataSchema.optional(),
error: ErrorResponseSchema.optional(),
metadata: z.object({
timestamp: z.string().datetime().openapi({
example: '2023-01-15T10:30:45.123Z'
}),
version: z.string().openapi({
example: '1.2.3',
description: 'API version that processed this request'
}),
requestId: z.string().uuid(),
// Performance metrics for debugging
executionTime: z.number().openapi({
example: 245,
description: 'Request processing time in milliseconds'
})
})
});
// Handler configuration for type-safe API routes
export interface HandlerConfig<
TBody extends z.ZodType,
TQuery extends z.ZodType,
TPath extends z.ZodType,
TResponse extends z.ZodType
> {
body?: TBody;
query?: TQuery;
path?: TPath;
response: TResponse;
headers?: z.ZodType;
// Added after learning from production issues
auth?: {
required: boolean;
roles?: string[];
};
rateLimit?: {
requestsPerMinute: number;
burstLimit?: number;
};
// OpenAPI metadata
openapi?: {
summary: string;
description: string;
tags: string[];
deprecated?: boolean;
};
}
// Type inference that provides full IntelliSense
export type InferredHandler<T extends HandlerConfig<any, any, any, any>> = {
body: T['body'] extends z.ZodType ? z.infer<T['body']> : never;
query: T['query'] extends z.ZodType ? z.infer<T['query']> : never;
path: T['path'] extends z.ZodType ? z.infer<T['path']> : never;
response: z.infer<T['response']>;
};
The Handler Wrapper That Validates API Contracts
This wrapper provides comprehensive request validation and error handling:
// lib/api/handler.ts - Type-safe request/response wrapper
import { APIGatewayProxyEventV2, APIGatewayProxyResultV2, Context } from 'aws-lambda';
import { z, ZodError } from 'zod';
import { v4 as uuidv4 } from 'uuid';
import { Logger } from '@aws-lambda-powertools/logger';
import { Tracer } from '@aws-lambda-powertools/tracer';
import { Metrics } from '@aws-lambda-powertools/metrics';
const logger = new Logger();
const tracer = new Tracer();
const metrics = new Metrics();
export function createHandler<T extends HandlerConfig<any, any, any, any>>(
config: T,
handler: (
event: {
body: T['body'] extends z.ZodType ? z.infer<T['body']> : undefined;
query: T['query'] extends z.ZodType ? z.infer<T['query']> : undefined;
path: T['path'] extends z.ZodType ? z.infer<T['path']> : undefined;
headers: T['headers'] extends z.ZodType ? z.infer<T['headers']> : Record<string, string>;
raw: APIGatewayProxyEventV2;
userId?: string; // Added after auth integration
},
context: Context
) => Promise<z.infer<T['response']>>
): (event: APIGatewayProxyEventV2, context: Context) => Promise<APIGatewayProxyResultV2> {
return async (event: APIGatewayProxyEventV2, context: Context): Promise<APIGatewayProxyResultV2> => {
const requestId = context.requestId || uuidv4();
const startTime = Date.now();
// Add trace metadata
tracer.putAnnotation('requestId', requestId);
tracer.putAnnotation('httpMethod', event.requestContext.http.method);
tracer.putAnnotation('path', event.requestContext.http.path);
logger.info('Request received', {
requestId,
method: event.requestContext.http.method,
path: event.requestContext.http.path,
userAgent: event.headers['user-agent'],
sourceIp: event.requestContext.http.sourceIp,
});
try {
// Parse and validate request components with detailed error tracking
let parsedBody: any;
try {
if (config.body && event.body) {
const bodyJson = JSON.parse(event.body);
parsedBody = config.body.parse(bodyJson);
logger.debug('Body validation successful', { requestId });
}
} catch (error) {
if (error instanceof SyntaxError) {
metrics.addMetric('ValidationErrors', 'Count', 1);
throw new Error('Invalid JSON in request body');
}
throw error;
}
const parsedQuery = config.query && event.queryStringParameters
? config.query.parse(event.queryStringParameters)
: undefined;
const parsedPath = config.path && event.pathParameters
? config.path.parse(event.pathParameters)
: undefined;
const parsedHeaders = config.headers && event.headers
? config.headers.parse(event.headers)
: event.headers;
// Auth validation (learned from security incidents)
let userId: string | undefined;
if (config.auth?.required) {
const authHeader = event.headers.authorization;
if (!authHeader) {
metrics.addMetric('AuthenticationErrors', 'Count', 1);
throw new Error('Authorization header required');
}
// Extract user ID from JWT or other auth mechanism
userId = await validateAuthToken(authHeader);
tracer.putAnnotation('userId', userId);
}
// Execute handler with validated inputs
const result = await handler({
body: parsedBody,
query: parsedQuery,
path: parsedPath,
headers: parsedHeaders,
raw: event,
userId
}, context);
// Validate response against schema
const validatedResponse = config.response.parse(result);
const executionTime = Date.now() - startTime;
// Record success metrics
metrics.addMetric('SuccessfulRequests', 'Count', 1);
metrics.addMetric('ExecutionTime', 'Milliseconds', executionTime);
logger.info('Request completed successfully', {
requestId,
executionTime,
responseSize: JSON.stringify(validatedResponse).length
});
return {
statusCode: 200,
headers: {
'Content-Type': 'application/json',
'X-Request-Id': requestId,
'X-API-Version': process.env.API_VERSION || '1.0.0',
// Security headers learned from production
'X-Content-Type-Options': 'nosniff',
'X-Frame-Options': 'DENY',
'X-XSS-Protection': '1; mode=block'
},
body: JSON.stringify({
success: true,
data: validatedResponse,
metadata: {
timestamp: new Date().toISOString(),
version: process.env.API_VERSION || '1.0.0',
requestId,
executionTime: Date.now() - startTime
}
})
};
} catch (error) {
const executionTime = Date.now() - startTime;
// Track error metrics
metrics.addMetric('ErrorRequests', 'Count', 1);
metrics.addMetric('ErrorExecutionTime', 'Milliseconds', executionTime);
// Handle validation errors (these save support tickets)
if (error instanceof ZodError) {
logger.warn('Validation error', {
requestId,
errors: error.errors,
path: event.requestContext.http.path,
method: event.requestContext.http.method
});
metrics.addMetric('ValidationErrors', 'Count', 1);
return {
statusCode: 400,
headers: {
'Content-Type': 'application/json',
'X-Request-Id': requestId,
'X-API-Version': process.env.API_VERSION || '1.0.0'
},
body: JSON.stringify({
success: false,
error: {
error: 'Validation failed',
details: error.errors.map(e => ({
path: e.path.join('.'),
message: e.message,
code: `INVALID_${e.path.join('_').toUpperCase()}`,
received: e.input
})),
requestId,
timestamp: new Date().toISOString()
},
metadata: {
timestamp: new Date().toISOString(),
version: process.env.API_VERSION || '1.0.0',
requestId,
executionTime
}
})
};
}
// Handle authentication errors
if (error.message === 'Authorization header required' ||
error.message.includes('Invalid token')) {
return {
statusCode: 401,
headers: {
'Content-Type': 'application/json',
'X-Request-Id': requestId
},
body: JSON.stringify({
success: false,
error: {
error: 'Unauthorized',
requestId,
timestamp: new Date().toISOString()
},
metadata: {
timestamp: new Date().toISOString(),
version: process.env.API_VERSION || '1.0.0',
requestId,
executionTime
}
})
};
}
// Handle other errors (with proper logging for debugging)
logger.error('Handler error', {
requestId,
error: error.message,
stack: error.stack,
path: event.requestContext.http.path,
method: event.requestContext.http.method,
userId: userId || 'anonymous'
});
metrics.addMetric('InternalErrors', 'Count', 1);
return {
statusCode: 500,
headers: {
'Content-Type': 'application/json',
'X-Request-Id': requestId,
'X-API-Version': process.env.API_VERSION || '1.0.0'
},
body: JSON.stringify({
success: false,
error: {
error: 'Internal server error',
requestId,
timestamp: new Date().toISOString(),
// In development, show actual error
...(process.env.NODE_ENV === 'development' && {
details: error.message,
stack: error.stack
})
},
metadata: {
timestamp: new Date().toISOString(),
version: process.env.API_VERSION || '1.0.0',
requestId,
executionTime
}
})
};
}
};
}
// Helper function for auth validation
async function validateAuthToken(authHeader: string): Promise<string> {
// Implementation depends on your auth provider
// Could be JWT validation, Cognito, etc.
const token = authHeader.replace('Bearer ', '');
// This is where you'd validate the token
// For example purposes, we'll just extract a fake user ID
try {
// Your token validation logic here
return 'user-123'; // Return actual user ID
} catch (error) {
throw new Error('Invalid token');
}
}
A Complete User Management API Example
Here’s a comprehensive user management API that demonstrates these patterns:
// lib/api/schemas/user.ts
import { z } from 'zod';
import { extendZodWithOpenApi } from '@anatine/zod-openapi';
extendZodWithOpenApi(z);
// Shared schemas
export const UserIdSchema = z.string().uuid().openapi({
example: '123e4567-e89b-12d3-a456-426614174000'
});
export const EmailSchema = z.string().email().openapi({
example: '[email protected]',
description: 'Valid email address'
});
// User entity
export const UserSchema = z.object({
id: UserIdSchema,
email: EmailSchema,
username: z.string().min(3).max(30).regex(/^[a-zA-Z0-9_-]+$/).openapi({
example: 'john_doe',
description: 'Alphanumeric username with underscores and hyphens'
}),
fullName: z.string().min(1).max(100).openapi({
example: 'John Doe'
}),
age: z.number().int().min(13).max(120).optional().openapi({
example: 25,
description: 'User age (13-120)'
}),
role: z.enum(['user', 'admin', 'moderator']).default('user').openapi({
example: 'user'
}),
metadata: z.object({
createdAt: z.string().datetime(),
updatedAt: z.string().datetime(),
lastLoginAt: z.string().datetime().optional()
}),
preferences: z.object({
notifications: z.boolean().default(true),
theme: z.enum(['light', 'dark', 'system']).default('system'),
language: z.string().default('en')
}).optional()
}).openapi('User');
// Request schemas
export const CreateUserRequestSchema = UserSchema
.pick({ email: true, username: true, fullName: true, age: true })
.openapi('CreateUserRequest');
export const UpdateUserRequestSchema = UserSchema
.pick({ fullName: true, age: true, preferences: true })
.partial()
.openapi('UpdateUserRequest');
export const ListUsersQuerySchema = z.object({
limit: z.string().regex(/^\d+$/).transform(Number).pipe(
z.number().int().min(1).max(100)
).default('20').openapi({
example: '20',
description: 'Number of users to return (1-100)'
}),
offset: z.string().regex(/^\d+$/).transform(Number).pipe(
z.number().int().min(0)
).default('0').openapi({
example: '0'
}),
role: z.enum(['user', 'admin', 'moderator']).optional(),
sortBy: z.enum(['createdAt', 'username', 'email']).default('createdAt'),
sortOrder: z.enum(['asc', 'desc']).default('desc')
}).openapi('ListUsersQuery');
// Response schemas
export const UserResponseSchema = UserSchema.openapi('UserResponse');
export const UsersListResponseSchema = z.object({
users: z.array(UserResponseSchema),
pagination: z.object({
total: z.number().int(),
limit: z.number().int(),
offset: z.number().int(),
hasMore: z.boolean()
})
}).openapi('UsersListResponse');
Lambda Handlers
Now let’s implement the Lambda functions using our type-safe wrapper:
// lambda/handlers/createUser.ts
import { createHandler } from '../../lib/api/handler';
import { CreateUserRequestSchema, UserResponseSchema } from '../../lib/api/schemas/user';
import { DynamoDBClient } from '@aws-sdk/client-dynamodb';
import { DynamoDBDocumentClient, PutCommand } from '@aws-sdk/lib-dynamodb';
import { v4 as uuidv4 } from 'uuid';
const dynamodb = DynamoDBDocumentClient.from(new DynamoDBClient({}));
const TABLE_NAME = process.env.USERS_TABLE!;
export const handler = createHandler({
body: CreateUserRequestSchema,
response: UserResponseSchema
}, async ({ body }) => {
const userId = uuidv4();
const now = new Date().toISOString();
const user = {
id: userId,
...body,
role: 'user' as const,
metadata: {
createdAt: now,
updatedAt: now
}
};
await dynamodb.send(new PutCommand({
TableName: TABLE_NAME,
Item: {
PK: `USER#${userId}`,
SK: `USER#${userId}`,
...user
},
ConditionExpression: 'attribute_not_exists(PK)'
}));
return user;
});
// lambda/handlers/listUsers.ts
import { createHandler } from '../../lib/api/handler';
import { ListUsersQuerySchema, UsersListResponseSchema } from '../../lib/api/schemas/user';
import { DynamoDBClient } from '@aws-sdk/client-dynamodb';
import { DynamoDBDocumentClient, QueryCommand } from '@aws-sdk/lib-dynamodb';
const dynamodb = DynamoDBDocumentClient.from(new DynamoDBClient({}));
const TABLE_NAME = process.env.USERS_TABLE!;
export const handler = createHandler({
query: ListUsersQuerySchema,
response: UsersListResponseSchema
}, async ({ query }) => {
const { limit, offset, role, sortBy, sortOrder } = query;
// In production, implement proper pagination with DynamoDB
const result = await dynamodb.send(new QueryCommand({
TableName: TABLE_NAME,
KeyConditionExpression: 'begins_with(PK, :pk)',
ExpressionAttributeValues: {
':pk': 'USER#'
},
Limit: limit + 1, // Fetch one extra to check hasMore
ScanIndexForward: sortOrder === 'asc'
}));
const items = result.Items || [];
const hasMore = items.length > limit;
const users = items.slice(0, limit).map(item => ({
id: item.id,
email: item.email,
username: item.username,
fullName: item.fullName,
age: item.age,
role: item.role,
metadata: item.metadata,
preferences: item.preferences
}));
return {
users,
pagination: {
total: result.Count || 0,
limit,
offset,
hasMore
}
};
});
// lambda/handlers/getUser.ts
import { createHandler } from '../../lib/api/handler';
import { UserIdSchema, UserResponseSchema } from '../../lib/api/schemas/user';
import { DynamoDBClient } from '@aws-sdk/client-dynamodb';
import { DynamoDBDocumentClient, GetCommand } from '@aws-sdk/lib-dynamodb';
import { z } from 'zod';
const dynamodb = DynamoDBDocumentClient.from(new DynamoDBClient({}));
const TABLE_NAME = process.env.USERS_TABLE!;
export const handler = createHandler({
path: z.object({ userId: UserIdSchema }),
response: UserResponseSchema
}, async ({ path }) => {
const result = await dynamodb.send(new GetCommand({
TableName: TABLE_NAME,
Key: {
PK: `USER#${path.userId}`,
SK: `USER#${path.userId}`
}
}));
if (!result.Item) {
throw new Error('User not found');
}
return {
id: result.Item.id,
email: result.Item.email,
username: result.Item.username,
fullName: result.Item.fullName,
age: result.Item.age,
role: result.Item.role,
metadata: result.Item.metadata,
preferences: result.Item.preferences
};
});
Generating OpenAPI Specifications
The real magic happens when we automatically generate OpenAPI specs from our Zod schemas:
// lib/api/openapi.ts
import { OpenAPIRegistry, OpenApiGeneratorV3 } from '@asteasolutions/zod-to-openapi';
import {
CreateUserRequestSchema,
UpdateUserRequestSchema,
ListUsersQuerySchema,
UserResponseSchema,
UsersListResponseSchema,
UserIdSchema
} from './schemas/user';
import { ErrorResponseSchema, ApiResponseSchema } from './types';
import { z } from 'zod';
export function generateOpenApiSpec() {
const registry = new OpenAPIRegistry();
// Register all schemas
registry.register('User', UserResponseSchema);
registry.register('CreateUserRequest', CreateUserRequestSchema);
registry.register('UpdateUserRequest', UpdateUserRequestSchema);
registry.register('ErrorResponse', ErrorResponseSchema);
// Define paths
registry.registerPath({
method: 'post',
path: '/users',
summary: 'Create a new user',
tags: ['Users'],
request: {
body: {
content: {
'application/json': {
schema: CreateUserRequestSchema
}
}
}
},
responses: {
200: {
description: 'User created successfully',
content: {
'application/json': {
schema: ApiResponseSchema(UserResponseSchema)
}
}
},
400: {
description: 'Validation error',
content: {
'application/json': {
schema: ApiResponseSchema(z.never()).extend({
error: ErrorResponseSchema
})
}
}
}
}
});
registry.registerPath({
method: 'get',
path: '/users',
summary: 'List users',
tags: ['Users'],
request: {
query: ListUsersQuerySchema
},
responses: {
200: {
description: 'Users retrieved successfully',
content: {
'application/json': {
schema: ApiResponseSchema(UsersListResponseSchema)
}
}
}
}
});
registry.registerPath({
method: 'get',
path: '/users/{userId}',
summary: 'Get user by ID',
tags: ['Users'],
request: {
params: z.object({
userId: UserIdSchema
})
},
responses: {
200: {
description: 'User retrieved successfully',
content: {
'application/json': {
schema: ApiResponseSchema(UserResponseSchema)
}
}
},
404: {
description: 'User not found',
content: {
'application/json': {
schema: ApiResponseSchema(z.never()).extend({
error: ErrorResponseSchema
})
}
}
}
}
});
// Generate OpenAPI document
const generator = new OpenApiGeneratorV3(registry.definitions);
return generator.generateDocument({
openapi: '3.0.0',
info: {
version: '1.0.0',
title: 'User Management API',
description: 'Type-safe serverless API with automatic OpenAPI generation'
},
servers: [
{
url: 'https://api.example.com',
description: 'Production'
}
]
});
}
// Build script to generate spec file
if (require.main === module) {
const fs = require('fs');
const path = require('path');
const spec = generateOpenApiSpec();
const outputPath = path.join(__dirname, '../../openapi.json');
fs.writeFileSync(outputPath, JSON.stringify(spec, null, 2));
console.log(`OpenAPI spec generated at: ${outputPath}`);
}
CDK Infrastructure
Now let’s wire everything together with AWS CDK:
// lib/stacks/api-stack.ts
import * as cdk from 'aws-cdk-lib';
import { Construct } from 'constructs';
import * as lambda from 'aws-cdk-lib/aws-lambda';
import * as apigateway from 'aws-cdk-lib/aws-apigateway';
import * as dynamodb from 'aws-cdk-lib/aws-dynamodb';
import { NodejsFunction } from 'aws-cdk-lib/aws-lambda-nodejs';
import * as path from 'path';
import { generateOpenApiSpec } from '../api/openapi';
export class ApiStack extends cdk.Stack {
constructor(scope: Construct, id: string, props?: cdk.StackProps) {
super(scope, id, props);
// DynamoDB table
const usersTable = new dynamodb.Table(this, 'UsersTable', {
partitionKey: { name: 'PK', type: dynamodb.AttributeType.STRING },
sortKey: { name: 'SK', type: dynamodb.AttributeType.STRING },
billingMode: dynamodb.BillingMode.PAY_PER_REQUEST,
pointInTimeRecovery: true
});
// Common Lambda environment
const environment = {
USERS_TABLE: usersTable.tableName,
API_VERSION: '1.0.0',
NODE_OPTIONS: '--enable-source-maps'
};
// Lambda functions
const createUserFn = new NodejsFunction(this, 'CreateUserFunction', {
entry: path.join(__dirname, '../../lambda/handlers/createUser.ts'),
runtime: lambda.Runtime.NODEJS_20_X,
architecture: lambda.Architecture.ARM_64,
environment,
bundling: {
minify: true,
sourceMap: true,
sourcesContent: false,
target: 'es2022',
format: 'esm'
}
});
const listUsersFn = new NodejsFunction(this, 'ListUsersFunction', {
entry: path.join(__dirname, '../../lambda/handlers/listUsers.ts'),
runtime: lambda.Runtime.NODEJS_20_X,
architecture: lambda.Architecture.ARM_64,
environment
});
const getUserFn = new NodejsFunction(this, 'GetUserFunction', {
entry: path.join(__dirname, '../../lambda/handlers/getUser.ts'),
runtime: lambda.Runtime.NODEJS_20_X,
architecture: lambda.Architecture.ARM_64,
environment
});
// Grant permissions
usersTable.grantReadWriteData(createUserFn);
usersTable.grantReadData(listUsersFn);
usersTable.grantReadData(getUserFn);
// API Gateway
const api = new apigateway.RestApi(this, 'UserApi', {
restApiName: 'User Management API',
description: 'Type-safe API with Zod and OpenAPI',
deployOptions: {
stageName: 'prod',
tracingEnabled: true,
loggingLevel: apigateway.MethodLoggingLevel.INFO,
dataTraceEnabled: true
},
defaultCorsPreflightOptions: {
allowOrigins: apigateway.Cors.ALL_ORIGINS,
allowMethods: apigateway.Cors.ALL_METHODS
}
});
// API resources
const users = api.root.addResource('users');
const user = users.addResource('{userId}');
// Wire up endpoints
users.addMethod('POST', new apigateway.LambdaIntegration(createUserFn));
users.addMethod('GET', new apigateway.LambdaIntegration(listUsersFn));
user.addMethod('GET', new apigateway.LambdaIntegration(getUserFn));
// Generate and export OpenAPI spec
const openApiSpec = generateOpenApiSpec();
new cdk.CfnOutput(this, 'OpenApiSpec', {
value: JSON.stringify(openApiSpec),
description: 'OpenAPI specification for the API'
});
// Export API URL
new cdk.CfnOutput(this, 'ApiUrl', {
value: api.url,
description: 'API Gateway URL'
});
}
}
Advanced Patterns
Middleware System
Create reusable middleware for common concerns:
// lib/api/middleware.ts
export interface MiddlewareContext<T> {
event: T;
context: Context;
next: () => Promise<any>;
}
export type Middleware<T = any> = (
ctx: MiddlewareContext<T>
) => Promise<void>;
export function compose<T>(...middlewares: Middleware<T>[]): Middleware<T> {
return async (ctx: MiddlewareContext<T>) => {
let index = -1;
async function dispatch(i: number): Promise<void> {
if (i <= index) throw new Error('next() called multiple times');
index = i;
const middleware = middlewares[i];
if (!middleware) return;
await middleware({
...ctx,
next: () => dispatch(i + 1)
});
}
await dispatch(0);
};
}
// Auth middleware
export const authMiddleware: Middleware = async (ctx) => {
const token = ctx.event.headers?.authorization?.replace('Bearer ', '');
if (!token) {
throw new Error('Unauthorized');
}
// Verify token (example with Cognito)
const payload = await verifyToken(token);
ctx.event.user = payload;
await ctx.next();
};
// Rate limiting middleware
export const rateLimitMiddleware: Middleware = async (ctx) => {
const ip = ctx.event.requestContext.identity.sourceIp;
const key = `rate-limit:${ip}`;
// Check rate limit (Redis/DynamoDB)
const count = await incrementCounter(key);
if (count > 100) {
throw new Error('Rate limit exceeded');
}
await ctx.next();
};
Schema Composition
Build complex schemas from reusable parts:
// lib/api/schemas/common.ts
import { z } from 'zod';
// Pagination mixin
export const PaginationSchema = z.object({
page: z.number().int().min(1).default(1),
pageSize: z.number().int().min(1).max(100).default(20),
sortBy: z.string().optional(),
sortOrder: z.enum(['asc', 'desc']).default('asc')
});
// Timestamps mixin
export const TimestampsSchema = z.object({
createdAt: z.string().datetime(),
updatedAt: z.string().datetime(),
deletedAt: z.string().datetime().nullable().optional()
});
// Audit fields mixin
export const AuditFieldsSchema = z.object({
createdBy: z.string().uuid(),
updatedBy: z.string().uuid(),
version: z.number().int().min(0)
});
// Resource schema factory
export function createResourceSchema<T extends z.ZodRawShape>(
name: string,
shape: T
) {
return z.object({
id: z.string().uuid(),
type: z.literal(name.toLowerCase()),
attributes: z.object(shape),
metadata: TimestampsSchema.merge(AuditFieldsSchema)
}).openapi(name);
}
// Usage
export const ProductResourceSchema = createResourceSchema('Product', {
name: z.string().min(1).max(255),
description: z.string().optional(),
price: z.number().positive(),
currency: z.enum(['USD', 'EUR', 'GBP']),
inventory: z.object({
quantity: z.number().int().min(0),
reserved: z.number().int().min(0).default(0),
available: z.number().int().min(0)
})
});
Performance Optimization
Optimize cold starts with lazy loading:
// lib/api/lazy.ts
export class LazyContainer<T> {
private instance?: T;
private initializer: () => T;
constructor(initializer: () => T) {
this.initializer = initializer;
}
get(): T {
if (!this.instance) {
this.instance = this.initializer();
}
return this.instance;
}
}
// Usage in Lambda
const dynamoClient = new LazyContainer(
() => DynamoDBDocumentClient.from(new DynamoDBClient({}))
);
export const handler = createHandler({
// ... config
}, async ({ body }) => {
const client = dynamoClient.get();
// Use client
});
Testing Strategies
Unit Testing Schemas
// tests/schemas/user.test.ts
import { CreateUserRequestSchema } from '../../lib/api/schemas/user';
describe('CreateUserRequestSchema', () => {
it('validates correct input', () => {
const result = CreateUserRequestSchema.safeParse({
email: '[email protected]',
username: 'test_user',
fullName: 'Test User',
age: 25
});
expect(result.success).toBe(true);
});
it('rejects invalid email', () => {
const result = CreateUserRequestSchema.safeParse({
email: 'not-an-email',
username: 'test_user',
fullName: 'Test User'
});
expect(result.success).toBe(false);
expect(result.error?.issues[0].path).toEqual(['email']);
});
it('enforces username constraints', () => {
const result = CreateUserRequestSchema.safeParse({
email: '[email protected]',
username: 'a', // Too short
fullName: 'Test User'
});
expect(result.success).toBe(false);
expect(result.error?.issues[0].message).toContain('at least 3');
});
});
Integration Testing
// tests/integration/createUser.test.ts
import { handler } from '../../lambda/handlers/createUser';
import { APIGatewayProxyEventV2, Context } from 'aws-lambda';
describe('Create User Handler', () => {
it('creates user with valid input', async () => {
const event: Partial<APIGatewayProxyEventV2> = {
body: JSON.stringify({
email: '[email protected]',
username: 'test_user',
fullName: 'Test User'
}),
headers: {
'content-type': 'application/json'
}
};
const context: Partial<Context> = {
requestId: '123',
functionName: 'createUser'
};
const response = await handler(
event as APIGatewayProxyEventV2,
context as Context
);
expect(response.statusCode).toBe(200);
const body = JSON.parse(response.body!);
expect(body.success).toBe(true);
expect(body.data.email).toBe('[email protected]');
expect(body.data.id).toBeDefined();
});
it('returns validation error for invalid input', async () => {
const event: Partial<APIGatewayProxyEventV2> = {
body: JSON.stringify({
email: 'invalid-email',
username: 'test_user'
})
};
const response = await handler(
event as APIGatewayProxyEventV2,
{} as Context
);
expect(response.statusCode).toBe(400);
const body = JSON.parse(response.body!);
expect(body.success).toBe(false);
expect(body.error.details).toBeDefined();
});
});
Deployment and CI/CD
GitHub Actions Workflow
# .github/workflows/deploy.yml
name: Deploy API
on:
push:
branches: [main]
pull_request:
branches: [main]
jobs:
test:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Setup Node.js
uses: actions/setup-node@v3
with:
node-version: '20'
cache: 'npm'
- name: Install dependencies
run: npm ci
- name: Run tests
run: npm test
- name: Type check
run: npm run typecheck
- name: Generate OpenAPI spec
run: npm run generate:openapi
- name: Upload OpenAPI spec
uses: actions/upload-artifact@v3
with:
name: openapi-spec
path: openapi.json
deploy:
needs: test
if: github.ref == 'refs/heads/main'
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- name: Configure AWS credentials
uses: aws-actions/configure-aws-credentials@v2
with:
aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
aws-region: us-east-1
- name: Setup Node.js
uses: actions/setup-node@v3
with:
node-version: '20'
cache: 'npm'
- name: Install dependencies
run: npm ci
- name: Deploy CDK
run: npx cdk deploy --require-approval never
Monitoring and Observability
Add structured logging and tracing:
// lib/api/observability.ts
import { Tracer } from '@aws-lambda-powertools/tracer';
import { Logger } from '@aws-lambda-powertools/logger';
import { Metrics } from '@aws-lambda-powertools/metrics';
const tracer = new Tracer();
const logger = new Logger();
const metrics = new Metrics();
export function createObservableHandler<T extends HandlerConfig<any, any, any, any>>(
config: T,
handler: HandlerFunction<T>
) {
const baseHandler = createHandler(config, handler);
return tracer.captureLambdaHandler(
logger.injectLambdaContext(
metrics.logMetrics(baseHandler)
)
);
}
// Usage
export const handler = createObservableHandler({
body: CreateUserRequestSchema,
response: UserResponseSchema
}, async ({ body }, context) => {
logger.info('Creating user', { email: body.email });
// Add custom metric
metrics.addMetric('UserCreated', 'Count', 1);
// Add trace annotation
tracer.putAnnotation('userEmail', body.email);
// Business logic...
});
Best Practices
1. Schema Versioning
// lib/api/schemas/v1/user.ts
export const UserSchemaV1 = z.object({
// V1 schema
});
// lib/api/schemas/v2/user.ts
export const UserSchemaV2 = UserSchemaV1.extend({
// V2 additions
preferences: PreferencesSchema
});
// Handler with version support
export const handler = createHandler({
headers: z.object({
'api-version': z.enum(['v1', 'v2']).default('v2')
}),
body: z.union([UserSchemaV1, UserSchemaV2]),
response: z.union([UserResponseV1, UserResponseV2])
}, async ({ headers, body }) => {
if (headers['api-version'] === 'v1') {
return handleV1(body);
}
return handleV2(body);
});
2. Error Recovery
export const resilientHandler = createHandler({
// ... config
}, async ({ body }, context) => {
// Circuit breaker pattern
const breaker = new CircuitBreaker(dynamoClient.send, {
timeout: 3000,
errorThresholdPercentage: 50,
resetTimeout: 30000
});
try {
return await breaker.fire(new PutCommand({
TableName: TABLE_NAME,
Item: user
}));
} catch (error) {
// Fallback to SQS
await sqsClient.send(new SendMessageCommand({
QueueUrl: DLQ_URL,
MessageBody: JSON.stringify({ user, error: error.message })
}));
throw new Error('Service temporarily unavailable');
}
});
3. Security Headers
export function addSecurityHeaders(response: APIGatewayProxyResultV2): APIGatewayProxyResultV2 {
return {
...response,
headers: {
...response.headers,
'Strict-Transport-Security': 'max-age=63072000; includeSubDomains; preload',
'X-Content-Type-Options': 'nosniff',
'X-Frame-Options': 'DENY',
'X-XSS-Protection': '1; mode=block',
'Referrer-Policy': 'strict-origin-when-cross-origin',
'Content-Security-Policy': "default-src 'none'; frame-ancestors 'none'"
}
};
}
Conclusion
By combining Zod’s runtime validation with OpenAPI generation, this approach creates a type-safe serverless API that:
- Eliminates manual synchronization between types, validation, and documentation
- Catches errors at compile time with full TypeScript integration
- Validates at runtime with detailed error messages
- Generates accurate documentation automatically
- Scales efficiently with AWS Lambda and CDK
This approach transforms API development from error-prone manual coordination to a streamlined, automated process. Your schemas become the single source of truth, ensuring consistency across every layer of your serverless stack.
Next Steps
- Add authentication with AWS Cognito or custom JWT validation
- Implement caching with API Gateway caching or ElastiCache
- Add WebSocket support for real-time features
- Integrate with AWS X-Ray for distributed tracing
- Set up API versioning with stage variables
- Add contract testing with Pact or similar tools
The foundation we’ve built handles the complexity of modern API development while maintaining the simplicity that makes serverless attractive. Happy building!
Related posts
Setting up a production-grade link shortener with AWS CDK, DynamoDB, and Lambda. Real architecture decisions, initial setup, and lessons learned from building URL shorteners at scale.
Building the redirect engine, analytics collection, and API Gateway configuration. Real performance optimizations and debugging strategies from handling millions of daily redirects.
A comprehensive technical guide to Amazon Cognito's advanced features including custom authentication flows, federation patterns, multi-tenancy architectures, migration strategies, and production-grade security implementation.
Learn how to build a comprehensive testing strategy for AWS Lambda, API Gateway, DynamoDB, and Step Functions with practical patterns for fast feedback and production reliability.
Learn to build automated preview environments using AWS CDK, Lambda, and GitHub Actions for seamless PR testing and review workflows