Zod

Zod: TypeScript-First Schema Validation

Overview

Zod is a TypeScript-first validation library that enables developers to define schemas for validating data at runtime while automatically inferring static TypeScript types. With zero dependencies and a compact core (~5kb gzipped; ~1.9kb for zod/mini), Zod provides immutable, composable validation with comprehensive error handling.

Installation

bun add zod
# or
npm install zod
# or
pnpm add zod
# or
yarn add zod

Requirements:

• TypeScript v5.5+ with "strict": true in tsconfig.json
• Zod 4.x (4.4.x recommended; z.codec() requires 4.1+)

Important: This skill documents Zod 4.x features. The following APIs require Zod 4 and are NOT available in Zod 3.x:

• z.codec() - Bidirectional transformations
• z.iso.date(), z.iso.time(), z.iso.datetime(), z.iso.duration() - ISO format validators
• z.toJSONSchema() - JSON Schema generation
• z.treeifyError(), z.prettifyError(), z.flattenError() - New error formatting helpers
• .meta() - Enhanced metadata (Zod 3.x only has .describe())
• Unified error parameter - Replaces message, invalid_type_error, required_error, errorMap
• .check() - Low-level multi-issue validation (composes check factories like z.minLength(3))
• z.file(), z.json(), z.stringbool(), z.xor(), z.templateLiteral() - New schema types
• z.exactOptional(), .prefault(), .safeExtend(), z.fromJSONSchema() - New utilities

For Zod 3.x compatibility or migration guidance, see https://zod.dev

Import paths (all ship in the zod package):

• import { z } from "zod" — standard (v4 since 4.0)
• import { z } from "zod/v4" — pin for libraries supporting both v3 (3.25+) and v4 users
• import { z } from "zod/mini" — functional, tree-shakable API (~1.9kb; checks via .check(z.minLength(3)))
• import * as core from "zod/v4/core" — low-level $-prefixed internals for library authors

Migrating from Zod v3 to v4

Load references/migration-guide.md for complete v3 to v4 migration documentation.

Quick Summary

Zod v4 introduces breaking changes for better performance:

• Error customization: Use unified error parameter (replaces message, invalid_type_error, required_error)
• Number validation: Stricter - rejects Infinity and unsafe integers
• String formats: Prefer top-level functions (z.email() instead of z.string().email(); the method forms still work but are deprecated)
• Object defaults: Applied even in optional fields
• Deprecated APIs: Use .extend() (not .merge()), z.treeifyError() (not error.format())
• Function validation: Use .implement() method
• UUID validation: Stricter RFC 9562/4122 compliance

→ Load references/migration-guide.md for: Complete breaking changes, migration checklist, gradual migration strategy, rollback instructions

Core Concepts

Basic Usage Pattern

import { z } from "zod";

// Define schema
const UserSchema = z.object({
  username: z.string(),
  age: z.number().int().positive(),
  email: z.email(),
});

// Infer TypeScript type
type User = z.infer<typeof UserSchema>;

// Validate data (throws on error)
const user = UserSchema.parse(data);

// Validate data (returns result object)
const result = UserSchema.safeParse(data);
if (result.success) {
  console.log(result.data); // Typed!
} else {
  console.error(result.error); // ZodError
}

Parsing Methods

Use the appropriate parsing method based on error handling needs:

• .parse(data) - Throws ZodError on invalid input; returns strongly-typed data on success
• .safeParse(data) - Returns { success: true, data } or { success: false, error } (no exceptions)
• .parseAsync(data) - For schemas with async refinements/transforms
• .safeParseAsync(data) - Async version that doesn't throw

Best Practice: Use .safeParse() to avoid try-catch blocks and leverage discriminated unions.

Primitive Types

Strings

z.string()                    // Basic string
z.string().min(5)            // Minimum length
z.string().max(100)          // Maximum length
z.string().length(10)        // Exact length
z.email()                    // Email validation (top-level in v4)
z.url()                      // URL validation (top-level in v4)
z.uuid()                     // UUID (RFC 9562; use z.guid() for permissive)
z.uuidv4()                   // Version-specific UUIDs (also z.uuidv7)
z.httpUrl()                  // http/https URLs only (rejects ftp:// etc.)
z.e164()                     // Phone numbers (E.164 format)
z.string().regex(/^\d+$/)    // Custom pattern
z.string().startsWith("pre") // Prefix check
z.string().endsWith("suf")   // Suffix check
z.string().trim()            // Auto-trim whitespace
z.string().toLowerCase()     // Auto-lowercase
z.string().toUpperCase()     // Auto-uppercase

// ISO formats (Zod 4+)
z.iso.date()                 // YYYY-MM-DD
z.iso.time()                 // HH:MM:SS
z.iso.datetime()             // ISO 8601 datetime
z.iso.duration()             // ISO 8601 duration

// Network formats
z.ipv4()                     // IPv4 address
z.ipv6()                     // IPv6 address
z.cidrv4()                   // IPv4 CIDR notation
z.cidrv6()                   // IPv6 CIDR notation

// Other formats
z.jwt()                      // JWT token
z.nanoid()                   // Nanoid
z.cuid()                     // CUID
z.cuid2()                    // CUID2
z.ulid()                     // ULID
z.base64()                   // Base64 encoded
z.hex()                      // Hexadecimal
z.hash("sha256")             // Hash of algorithm-specific length
z.guid()                     // Permissive UUID/GUID (non-RFC)
z.hostname()                 // Hostname
z.mac()                      // MAC address
z.uuid({ version: "v4" })    // UUID with version constraint
z.email({ pattern: z.regexes.email }) // Email with explicit pattern

Numbers

z.number()                   // Basic number
z.number().int()             // Integer only
z.number().positive()        // > 0
z.number().nonnegative()     // >= 0
z.number().negative()        // < 0
z.number().nonpositive()     // <= 0
z.number().min(0)            // Minimum value
z.number().max(100)          // Maximum value
z.number().gt(0)             // Greater than
z.number().gte(0)            // Greater than or equal
z.number().lt(100)           // Less than
z.number().lte(100)          // Less than or equal
z.number().multipleOf(5)     // Must be multiple of 5
z.int()                      // Shorthand for z.number().int()
z.int32()                    // 32-bit integer (also z.uint32, z.int64, z.uint64)
z.float32()                  // 32-bit float (also z.float64)
z.nan()                      // NaN value

Coercion (Type Conversion)

z.coerce.string()            // Convert to string
z.coerce.number()            // Convert to number
z.coerce.boolean()           // Convert to boolean
z.coerce.bigint()            // Convert to bigint
z.coerce.date()              // Convert to Date

// Example: Parse query parameters
const QuerySchema = z.object({
  page: z.coerce.number().int().positive(),
  limit: z.coerce.number().int().max(100).default(10),
});

// "?page=5&limit=20" -> { page: 5, limit: 20 }

Other Primitives

z.boolean()                  // Boolean
z.date()                     // Date object
z.date().min(new Date("2020-01-01"))
z.date().max(new Date("2030-12-31"))
z.bigint()                   // BigInt
z.symbol()                   // Symbol
z.null()                     // Null
z.undefined()                // Undefined
z.void()                     // Void (undefined)

// File validation (v4) — sizes are bytes
z.file().min(1).max(5 * 1024 * 1024).mime(["image/png"])

// JSON-representable values (v4) — validates the VALUE is JSON-serializable
// (recursive union of string | number | boolean | null | array | record)
z.json()

// Env-style boolean coercion: "true"/"yes"/"1" → true (v4)
z.stringbool({ truthy: ["true", "yes", "1"], falsy: ["false", "no", "0"] })

Complex Types

Objects

const PersonSchema = z.object({
  name: z.string(),
  age: z.number(),
  address: z.object({
    street: z.string(),
    city: z.string(),
    country: z.string(),
  }),
});

type Person = z.infer<typeof PersonSchema>;

// Object methods
PersonSchema.shape                 // Access shape
PersonSchema.keyof()              // Get union of keys
PersonSchema.extend({ role: z.string() })  // Add fields
PersonSchema.safeExtend({ role: z.string() }) // Type-safe extend (inherits refinements)
PersonSchema.pick({ name: true }) // Pick specific fields
PersonSchema.omit({ age: true })  // Omit fields
PersonSchema.partial()            // Make all fields optional
PersonSchema.partial({ age: true }) // Make specific fields optional
PersonSchema.required()           // Make all fields required

// NOTE: .deepPartial() was REMOVED in v4 — build recursive partials manually
// (top-level z.deepPartial() landed post-4.4.3 and is not yet stable)

// Strict vs loose objects
z.strictObject({ ... })           // No extra keys allowed (throws)
z.object({ ... })                 // Strips extra keys (default)
z.looseObject({ ... })            // Allows extra keys
z.object({ ... }).catchall(z.number()) // Validate (and keep) unknown keys

Arrays

z.array(z.string())              // String array
z.array(z.number()).min(1)       // At least 1 element
z.array(z.number()).max(10)      // At most 10 elements
z.array(z.number()).length(5)    // Exactly 5 elements
z.array(z.number()).nonempty()   // At least 1 element

// Nested arrays
z.array(z.array(z.number()))     // number[][]

Tuples

z.tuple([z.string(), z.number()]) // [string, number]
z.tuple([z.string(), z.number()]).rest(z.boolean()) // [string, number, ...boolean[]]

Enums and Literals

// Enum
const RoleEnum = z.enum(["admin", "user", "guest"]);
type Role = z.infer<typeof RoleEnum>; // "admin" | "user" | "guest"

// Literal values
z.literal("exact_value")
z.literal(42)
z.literal(true)

// Native TypeScript enum
enum Fruits {
  Apple,
  Banana,
}
z.nativeEnum(Fruits)

// Enum methods
RoleEnum.enum.admin              // "admin"
RoleEnum.exclude(["guest"])      // Exclude values
RoleEnum.extract(["admin", "user"]) // Include only

Unions

// Basic union
z.union([z.string(), z.number()])

// Discriminated union (better performance & type inference)
const ResponseSchema = z.discriminatedUnion("status", [
  z.object({ status: z.literal("success"), data: z.any() }),
  z.object({ status: z.literal("error"), message: z.string() }),
]);

type Response = z.infer<typeof ResponseSchema>;
// { status: "success", data: any } | { status: "error", message: string }

// Exclusive union (v4): must match exactly ONE branch
z.xor([z.object({ a: z.string() }), z.object({ b: z.number() })])

Intersections

const BaseSchema = z.object({ id: z.string() });
const ExtendedSchema = z.object({ name: z.string() });

const Combined = z.intersection(BaseSchema, ExtendedSchema);
// Equivalent to: z.object({ id: z.string(), name: z.string() })

Records and Maps

// Record: object with typed keys and values (BOTH args required in v4)
z.record(z.string(), z.number())  // { [key: string]: number }
z.record(z.enum(["a", "b"]), z.string()) // Exhaustive: both keys REQUIRED

// Partial record (enum keys optional)
z.partialRecord(z.enum(["a", "b"]), z.string())

// Map
z.map(z.string(), z.number())    // Map<string, number>
z.set(z.string())                // Set<string>

Advanced Patterns

Load references/advanced-patterns.md for complete advanced validation and transformation patterns.

Quick Reference

Refinements (custom validation):

z.string().refine((val) => val.length >= 8, "Too short");
z.object({ password, confirmPassword }).superRefine((data, ctx) => { /* ... */ });

Transformations (modify data):

z.string().transform((val) => val.trim());
z.string().pipe(z.coerce.number());
z.preprocess((v) => (typeof v === "string" ? v.trim() : v), z.string()); // Transform BEFORE parsing
z.custom<string>((v) => typeof v === "string", { error: "Not a string" }); // Fully custom schema

Multi-issue validation (.check(), v4):

z.string().check(z.minLength(3), z.startsWith("a")); // Compose check factories
z.string().check((payload) => {
  if (payload.value === "forbidden") {
    payload.issues.push({ code: "custom", message: "Not allowed", input: payload.value });
  }
});

Template literals (v4):

const UserRole = z.templateLiteral(["role-", z.string()]); // "role-admin", "role-editor", ...

Codecs (bidirectional transforms - NEW in v4.1):

const DateCodec = z.codec(
  z.iso.datetime(),
  z.date(),
  {
    decode: (str) => new Date(str),
    encode: (date) => date.toISOString(),
  }
);

Recursive Types:

const CategorySchema: z.ZodType<Category> = z.lazy(() =>
  z.object({ name: z.string(), subcategories: z.array(CategorySchema) })
);

Optional/Nullable:

z.string().optional()            // string | undefined
z.string().nullable()            // string | null
z.string().default("default")    // Provides default if undefined

Readonly & Brand:

z.object({ ... }).readonly()     // Readonly properties
z.string().brand<"UserId">()     // Nominal typing

→ Load references/advanced-patterns.md for: Complete refinement patterns, async validation, codec examples, composable schemas, conditional validation, performance optimization

Error Handling

Load references/error-handling.md for complete error formatting and customization guide.

Quick Reference

Error Formatting Methods:

// For forms
const { fieldErrors } = z.flattenError(error);

// For nested data
const tree = z.treeifyError(error);
const nameError = tree.properties?.user?.properties?.name?.errors?.[0];

// For debugging
console.log(z.prettifyError(error));

Custom Error Messages (three levels):

// 1. Schema-level (highest priority)
z.string({ error: "Custom message" });
z.string().min(5, "Too short");

// 2. Per-parse level
schema.parse(data, { error: (issue) => "..." });

// 3. Global level
z.config({ customError: (issue) => "..." });

Localization (40+ languages):

z.config(z.locales.es());  // Spanish
z.config(z.locales.fr());  // French

→ Load references/error-handling.md for: Complete error formatting examples, custom error patterns, localization setup, error code reference

Type Inference

Load references/type-inference.md for complete type inference and metadata documentation.

Quick Reference

Basic Type Inference:

const UserSchema = z.object({ name: z.string() });
type User = z.infer<typeof UserSchema>; // { name: string }

Input vs Output (for transforms):

const TransformSchema = z.string().transform((s) => s.length);
type Input = z.input<typeof TransformSchema>;   // string
type Output = z.output<typeof TransformSchema>; // number

JSON Schema Conversion:

const jsonSchema = z.toJSONSchema(UserSchema, {
  target: "openapi-3.0",
  // .meta() data from the global registry is included by default
});

Metadata:

// Add metadata
const EmailSchema = z.email().meta({
  title: "Email Address",
  description: "User's email address",
});

// Create custom registry
const formRegistry = z.registry<FormFieldMeta>();

→ Load references/type-inference.md for: Complete type inference patterns, JSON Schema options, metadata system, custom registries, brand types

Functions

Validate function inputs and outputs with the v4 factory signature:

const AddFunction = z.function({
  input: [z.number(), z.number()], // Arguments
  output: z.number(),              // Return type
});

// Implement typed function
const add = AddFunction.implement((a, b) => {
  return a + b; // Type-checked!
});

// Async functions
const FetchFunction = z.function({
  input: [z.string()],
  output: z.object({ data: z.any() }),
});

const fetchJson = FetchFunction.implementAsync(async (url) => {
  const response = await fetch(url);
  return response.json();
});

Note: the v3 chaining style (z.function().args(...).returns(...)) was removed in v4.

Common Patterns

Environment Variables

const EnvSchema = z.object({
  NODE_ENV: z.enum(["development", "production", "test"]),
  DATABASE_URL: z.url(),
  PORT: z.coerce.number().int().positive().default(3000),
  API_KEY: z.string().min(32),
});

// Validate on startup
const env = EnvSchema.parse(process.env);

// Now use typed env
console.log(env.PORT); // number

API Request Validation

const CreateUserRequest = z.object({
  username: z.string().min(3).max(20),
  email: z.email(),
  password: z.string().min(8),
  age: z.number().int().positive().optional(),
});

// Express example
app.post("/users", async (req, res) => {
  const result = CreateUserRequest.safeParse(req.body);

  if (!result.success) {
    return res.status(400).json({
      errors: z.flattenError(result.error).fieldErrors,
    });
  }

  const user = await createUser(result.data);
  res.json(user);
});

Form Validation

const FormSchema = z.object({
  firstName: z.string().min(1, "First name required"),
  lastName: z.string().min(1, "Last name required"),
  email: z.email({ error: "Invalid email" }),
  age: z.coerce.number().int().min(18, "Must be 18+"),
  agreeToTerms: z.literal(true, {
    error: () => "Must accept terms",
  }),
});

type FormData = z.infer<typeof FormSchema>;

Partial Updates

const UserSchema = z.object({
  id: z.string(),
  name: z.string(),
  email: z.email(),
});

// For PATCH requests: make everything optional except id
const UpdateUserSchema = UserSchema.partial().required({ id: true });

type UpdateUser = z.infer<typeof UpdateUserSchema>;
// { id: string; name?: string; email?: string }

Composable Schemas

// Base schemas
const TimestampSchema = z.object({
  createdAt: z.date(),
  updatedAt: z.date(),
});

const AuthorSchema = z.object({
  authorId: z.string(),
  authorName: z.string(),
});

// Compose into larger schemas (.merge() is deprecated — use .extend with .shape)
const PostSchema = z.object({
  id: z.string(),
  title: z.string(),
  content: z.string(),
}).extend(TimestampSchema.shape).extend(AuthorSchema.shape);

Ecosystem Integration

Load references/ecosystem-integrations.md for complete framework and tooling integration guide.

Quick Reference

ESLint Plugins:

• eslint-plugin-zod-x - Enforces best practices
• eslint-plugin-import-zod - Enforces import style

Framework Integrations:

• tRPC - End-to-end typesafe APIs
• React Hook Form - Form validation (see react-hook-form-zod skill)
• Prisma - Generate Zod from database models
• NestJS - DTOs and validation pipes

Code Generation:

• orval - OpenAPI → Zod
• Hey API - OpenAPI to TypeScript + Zod
• kubb - API toolkit with codegen

→ Load references/ecosystem-integrations.md for: Setup instructions, integration examples, Hono middleware, Drizzle ORM patterns

Troubleshooting

Load references/troubleshooting.md for complete troubleshooting guide, performance tips, and best practices.

Quick Reference

Common Issues:

1. TypeScript strict mode required → Enable in tsconfig.json
2. Large bundle size → Use z.lazy() for code splitting
3. Slow async refinements → Cache or debounce
4. Circular dependencies → Use z.lazy()
5. Slow unions → Use z.discriminatedUnion()
6. Transform vs refine confusion → Use .refine() for validation, .transform() for modification

Performance Tips:

• Use .discriminatedUnion() (5-10x faster than .union())
• Cache schema instances
• Use .safeParse() (avoids try-catch overhead)
• Lazy load large schemas

Best Practices:

• Define schemas at module level
• Use type inference (z.infer)
• Add custom error messages
• Validate at system boundaries
• Compose small schemas
• Document with .meta()

→ Load references/troubleshooting.md for: Detailed solutions, performance optimization, best practices, testing patterns

Quick Reference

// Primitives
z.string(), z.number(), z.boolean(), z.date(), z.bigint()

// Collections
z.array(), z.tuple(), z.object(), z.record(), z.map(), z.set()

// Special types
z.enum(), z.union(), z.discriminatedUnion(), z.intersection()
z.literal(), z.any(), z.unknown(), z.never()

// Modifiers
.optional(), .nullable(), .nullish(), .default(), .prefault()
.catch(), .readonly(), .brand(), .exactOptional()

// Validation
.min(), .max(), .length(), .regex()
.refine(), .superRefine(), .check(z.minLength(3))

// Top-level formats (v4)
z.email(), z.uuid(), z.url(), z.iso.datetime(), z.file(), z.json()

// Transformation
.transform(), .pipe(), .codec(), z.preprocess(), z.custom()

// Composition helpers
z.strictObject(), z.looseObject(), z.xor(), z.templateLiteral()

// Parsing
.parse(), .safeParse(), .parseAsync(), .safeParseAsync()

// Type inference
z.infer<typeof Schema>, z.input<typeof Schema>, z.output<typeof Schema>

// Error handling
z.flattenError(), z.treeifyError(), z.prettifyError()

// JSON Schema
z.toJSONSchema(schema, options)

// Metadata
.meta(), .describe()

// Object methods
.extend(), .safeExtend(), .pick(), .omit(), .partial(), .required()
.catchall(), .keyof()  // .merge() is deprecated — use .extend(other.shape)

When to Load References

Load references/migration-guide.md when:

• Upgrading from Zod v3 to v4
• Questions about breaking changes
• Need migration checklist or rollback strategy
• Errors related to deprecated APIs (.merge(), error.format(), etc.)
• Number validation issues with Infinity or unsafe integers

Load references/error-handling.md when:

• Need to format errors for forms or UI
• Implementing custom error messages
• Questions about z.flattenError(), z.treeifyError(), or z.prettifyError()
• Setting up localization for error messages
• Need error code reference or pattern examples

Load references/advanced-patterns.md when:

• Implementing custom refinements or async validation
• Need bidirectional transformations (codecs)
• Working with recursive types or self-referential data
• Questions about .refine(), .transform(), or .codec()
• Need performance optimization patterns
• Implementing conditional validation

Load references/type-inference.md when:

• Questions about TypeScript type inference
• Need to generate JSON Schema for OpenAPI or AI
• Implementing metadata system for forms or documentation
• Need custom registries for type-safe metadata
• Questions about z.infer, z.input, z.output
• Using brand types for ID safety

Load references/ecosystem-integrations.md when:

• Integrating with tRPC, React Hook Form, Prisma, or NestJS
• Setting up ESLint plugins for best practices
• Generating Zod schemas from OpenAPI (orval, Hey API, kubb)
• Questions about Hono middleware or Drizzle ORM
• Need framework-specific integration examples

Load references/troubleshooting.md when:

• Encountering TypeScript strict mode errors
• Bundle size concerns or lazy loading needs
• Performance issues with large unions or async refinements
• Questions about circular dependencies
• Need best practices or testing patterns
• Confusion between .refine() and .transform()

Load references/best-practices.md when:

• Writing production Zod code or reviewing Zod usage
• Deciding strict vs loose objects, optional vs nullable, any vs unknown
• Optimizing hot paths (schema caching, avoiding dynamic schema creation, large arrays)
• Unsure where to validate (boundaries, JSON.parse, double validation)
• Want Incorrect/Correct examples with "when NOT to use" guidance

Additional Resources

• Official Docs: https://zod.dev
• GitHub: https://github.com/colinhacks/zod
• TypeScript Playground: https://zod-playground.vercel.app
• Zod Mini: https://zod.dev/packages/mini
• ESLint Plugin (Best Practices): https://github.com/JoshuaKGoldberg/eslint-plugin-zod-x
• tRPC Integration: https://trpc.io
• Ecosystem: https://zod.dev/ecosystem

Production Notes:

• Package version: 4.4.x (all APIs verified against zod@4.4.3)
• Zero dependencies
• Bundle size: ~5kb core / ~1.9kb zod/mini (gzipped)
• TypeScript 5.5+ required
• Strict mode required
• Last verified: 2026-08-20
• Skill version: 2.1.0 (fidelity audit + best-practices rules)

What's New in 2.1.0:

• 🔍 Fidelity audit: every documented API verified against zod@4.4.3 (typecheck + runtime tests)
• 📜 New references/best-practices.md rulebook (Incorrect/Correct/When-NOT format with impact ratings)
• ➕ New coverage: zod/mini, .check(), catchall, z.preprocess, z.custom, z.templateLiteral, z.file/z.json/z.stringbool/z.xor, z.exactOptional, .safeExtend, z.fromJSONSchema, expanded format validators
• 🐛 Fixed: install commands, z.string().uuid() → z.uuid(), v3 .merge()/z.function() chains, .deepPartial() (removed in v4), single-arg z.record, toJSONSchema metadata option

What's New in 2.0.0:

• ✨ Comprehensive v3 to v4 migration guide with breaking changes
• ✨ Enhanced error customization with three-level system
• ✨ Expanded metadata API with registry system
• ✨ Improved error formatting with practical examples
• ✨ Built-in localization support for 40+ locales
• ✨ Detailed codec documentation with real-world patterns
• ✨ Performance improvements and architectural changes explained