web-forms-zod-validation（Web 表单 Zod 验证）

Zod Schema Validation Patterns

Quick Guide: Use Zod for runtime validation at trust boundaries (API responses, form inputs, config, URL params). Define schemas once, derive types with z.infer. Use safeParse for error handling, refine/superRefine for custom validation, transform for data conversion. Named constants for all validation limits. Version Note: Zod v4 is now the stable release (v4.1+). It brings 14.7x faster string parsing, 57% smaller bundle, and new top-level APIs (z.email(), z.url(), z.iso.*). The v3 method-chain equivalents (z.string().email()) still work but are deprecated. For migration details, see reference.md.

<critical_requirements>

CRITICAL: Before Using This Skill

All code must follow project conventions in CLAUDE.md (kebab-case, named exports, import ordering, import type, named constants)

(You MUST use safeParse instead of parse for user-facing validation - prevents unhandled exceptions)

(You MUST use z.infer<typeof schema> to derive types - never duplicate schema as separate interface)

(You MUST validate at trust boundaries - API responses, form inputs, config files, URL params)

(You MUST use named constants for validation limits - NO magic numbers in .min(), .max(), .length())

</critical_requirements>

Auto-detection: Zod schemas, z.object, z.string, z.number, z.infer, safeParse, refine, superRefine, transform, discriminatedUnion, z.coerce, z.pipe, z.catch, z.brand, z.lazy, z.email, z.url, z.iso

When to use:

• Validating API responses before using data
• Parsing form input data with type safety
• Validating configuration files or environment variables
• Defining contracts between systems (frontend/backend shared schemas)
• Runtime type checking for data from untrusted sources

When NOT to use:

• Internal function parameters (TypeScript is sufficient for trusted data)
• Simple boolean checks that don't need schema definition
• Performance-critical hot paths where validation overhead matters

Philosophy

TypeScript provides compile-time type safety for code you control. Zod provides runtime validation for data you don't control - API responses, user input, configuration files, URL parameters. Use TypeScript for internal contracts; use Zod at trust boundaries where external data enters your system.

Key principle: Define the schema once, derive the type. Never maintain parallel type definitions and validation logic - they will drift apart.

// Schema is the source of truth
const UserSchema = z.object({
  name: z.string(),
  email: z.string().email(),
});

// Type is derived, always in sync
type User = z.infer<typeof UserSchema>;

Core Patterns

Pattern 1: Schema Definition with Named Constants

Define schemas with named constants for all validation limits. Custom error messages for user-facing fields.

const MIN_USERNAME_LENGTH = 3;
const MAX_USERNAME_LENGTH = 50;

const UserSchema = z.object({
  username: z
    .string()
    .min(
      MIN_USERNAME_LENGTH,
      `Username must be at least ${MIN_USERNAME_LENGTH} characters`,
    )
    .max(
      MAX_USERNAME_LENGTH,
      `Username cannot exceed ${MAX_USERNAME_LENGTH} characters`,
    ),
  email: z.string().email("Invalid email format"),
});

type User = z.infer<typeof UserSchema>; // Always derived, never manual interface

Why good: named constants make limits discoverable, custom error messages improve UX, type derived from schema

See examples/core.md for complete schema examples with reusable sub-schemas and CRUD composition patterns.

Pattern 2: Safe Parsing for Error Handling

Use safeParse for user input and API responses. Reserve parse for config/internal data where invalid = programming error.

const result = UserSchema.safeParse(data);

if (!result.success) {
  const errors = result.error.issues.reduce(
    (acc, err) => {
      const field = err.path.join(".");
      acc[field] = err.message;
      return acc;
    },
    {} as Record<string, string>,
  );
  return { success: false, errors };
}

return { success: true, user: result.data };

Why good: safeParse never throws, validation errors handled explicitly, error formatting provides useful field-level feedback

See examples/core.md for form validation and API response validation patterns.

Pattern 3: Refinements and Cross-Field Validation

Use refine for custom validation logic. Use superRefine when you need cross-field validation with specific error paths.

const MIN_PASSWORD_LENGTH = 8;

const PasswordFormSchema = z
  .object({
    password: z.string().min(MIN_PASSWORD_LENGTH),
    confirmPassword: z.string(),
  })
  .superRefine((data, ctx) => {
    if (data.password !== data.confirmPassword) {
      ctx.addIssue({
        code: z.ZodIssueCode.custom,
        message: "Passwords do not match",
        path: ["confirmPassword"],
      });
    }
  });

Why good: superRefine enables cross-field validation with specific error paths, keeps all validation in the schema

See examples/core.md for password refinement chains and conditional validation patterns.

Pattern 4: Transforms and Type Conversion

Use transform to convert data during validation. Use z.input and z.output when transforms change the type.

const DateSchema = z
  .string()
  .datetime()
  .transform((str) => new Date(str));

type DateInput = z.input<typeof DateSchema>; // string
type DateOutput = z.output<typeof DateSchema>; // Date

Gotcha: z.infer returns the output type. When a function accepts pre-validation input, use z.input for the parameter type.

See examples/transforms.md for coercion patterns (URL params, form data) and transform pipelines.

Pattern 5: Discriminated Unions

Use discriminatedUnion when objects share a common discriminator field. Provides better error messages and TypeScript narrowing than union.

const NotificationSchema = z.discriminatedUnion("type", [
  z.object({
    type: z.literal("email"),
    email: z.string().email(),
    subject: z.string(),
  }),
  z.object({ type: z.literal("sms"), phone: z.string(), message: z.string() }),
  z.object({
    type: z.literal("push"),
    deviceId: z.string(),
    title: z.string(),
  }),
]);

Why good over z.union: discriminatedUnion reports which variant failed (not "Invalid input"), TypeScript narrows type in switch statements

See examples/core.md for payment method union and type narrowing examples.

Pattern 6: Schema Composition

Compose schemas using extend, pick, omit, and partial for CRUD operations.

const BaseEntitySchema = z.object({
  id: z.string().uuid(),
  createdAt: z.string().datetime(),
  updatedAt: z.string().datetime(),
});

const UserSchema = BaseEntitySchema.extend({
  email: z.string().email(),
  name: z.string(),
});
const CreateUserSchema = UserSchema.omit({
  id: true,
  createdAt: true,
  updatedAt: true,
});
const UpdateUserSchema = CreateUserSchema.partial();
const UserSummarySchema = UserSchema.pick({ id: true, name: true });

See examples/core.md for full CRUD schema composition example.

Pattern 7: Coercion for String Inputs

Use z.coerce for URL params and form data that arrive as strings. Simpler than manual parsing.

const DEFAULT_PAGE = 1;
const DEFAULT_LIMIT = 20;
const MAX_LIMIT = 100;

const PaginationSchema = z.object({
  page: z.coerce.number().int().positive().default(DEFAULT_PAGE),
  limit: z.coerce
    .number()
    .int()
    .positive()
    .max(MAX_LIMIT)
    .default(DEFAULT_LIMIT),
});

// "3" -> 3, "50" -> 50, missing -> defaults

Gotcha: z.coerce.boolean() coerces any truthy value to true, including the string "false". Use explicit comparison for string booleans.

See examples/transforms.md for complete pagination and query param patterns.

Pattern 8: Optional, Nullable, and Nullish

const ProfileSchema = z.object({
  name: z.string(), // Required
  bio: z.string().optional(), // string | undefined
  avatar: z.string().url().nullable(), // string | null
  nickname: z.string().nullish(), // string | null | undefined
  theme: z.string().default("light"), // string (always defined)
});

Key distinction: nullable = explicitly set to null (API returns null), optional = may be omitted entirely, nullish = either.

Detailed Resources:

• examples/core.md - Schema definition, safe parsing, error formatting, discriminated unions, composition, nested schemas
• examples/transforms.md - Transforms, coercion, pipe chains
• examples/advanced-patterns.md - Branded types, catch fallbacks, readonly, recursive schemas, ISO validators
• reference.md - Decision frameworks, method reference, anti-patterns, v4 migration guide

<red_flags>

RED FLAGS

High Priority Issues:

• Using parse for user-facing validation - Throws exceptions for expected invalid input, requiring try-catch and losing detailed error info
• Magic numbers in validation limits - .min(3).max(50) is undocumented; use named constants like MIN_USERNAME_LENGTH
• Defining separate TypeScript interfaces - Creates drift between schema and type; always use z.infer<typeof schema>
• Not validating at trust boundaries - API responses, user input, and config should always be validated at entry points
• Async refinements with parse instead of parseAsync - Async refinements silently fail with sync parse methods

Medium Priority Issues:

• Overly strict validation on optional fields - Empty strings should often be treated as undefined for optional fields
• Missing custom error messages - Default "Invalid input" messages are not user-friendly
• Validating internal function parameters with Zod - TypeScript is sufficient for trusted internal code
• Using .passthrough() by default - Allows unexpected fields through; use .strict() when you want to reject extras

Gotchas & Edge Cases:

• z.coerce.boolean(): Coerces any truthy value to true, including string "false" - use explicit string comparison if needed
• Transform order: .transform() runs after all other validations; refinements on transformed values need .pipe() to validate after
• Empty strings: z.string().email() rejects empty strings; use .email().or(z.literal("")) to allow empty
• Extend with refinements: .extend() on a schema with .refine() throws; apply refinements after extending instead
• Date parsing: z.coerce.date() uses new Date() which accepts many formats; use .datetime() for strict ISO format
• z.union vs z.discriminatedUnion: Union tries all schemas and reports combined errors; discriminatedUnion uses discriminator for targeted validation and better errors
• v4: .refine() function second arg removed: z.string().refine(fn, (val) => ({message:...})) no longer works; use superRefine() for dynamic messages
• v4: ctx.path removed in .superRefine(): No longer available for performance reasons; ctx.addIssue() still works
• v4 deprecations: .flatten() deprecated - use z.flattenError() instead; .format() deprecated - use z.treeifyError() instead; .merge() deprecated - use .extend() instead

</red_flags>

<critical_reminders>

CRITICAL REMINDERS

All code must follow project conventions in CLAUDE.md

(You MUST use safeParse instead of parse for user-facing validation - prevents unhandled exceptions)

(You MUST use z.infer<typeof schema> to derive types - never duplicate schema as separate interface)

(You MUST validate at trust boundaries - API responses, form inputs, config files, URL params)

(You MUST use named constants for validation limits - NO magic numbers in .min(), .max(), .length())

Failure to follow these rules will create type mismatches, unhandled exceptions, and unmaintainable validation code.

</critical_reminders>