JSON to TypeScript

Q: Why is a property typed null or unknown[]?

The sample held only null, or only an empty list, at that place, so the real type cannot be known. The tool lists these paths below the result. Paste a sample with real values there, or edit the type by hand. Turn on Use any where the type is unknown if you prefer any.

Q: Can it generate a Zod schema?

Yes. Choose Zod schema to get z.object(…) schemas with .optional(), .nullable(), unions, arrays and records, each followed by its z.infer type. Nested schemas are declared before the ones that use them, and the code runs on both Zod 3 and Zod 4.

Q: How are the types named?

From the property key, in PascalCase: shipping_address becomes ShippingAddress. Items of a list use the singular (orders → Order). If two different structures would get the same name, the parent’s name is added, as in OfficeAddress.

Paste an API response and get TypeScript interfaces, type aliases or Zod schemas as you type. Optional properties, unions and nested types are worked out from the data, and nothing is uploaded.

Runs in your browser Free · no sign-up Works on any device

Optionsinterfaces from “Root”, nested types named, exported

Root type name

Nested objects

Each gets its own name, taken from its key: address → Address, orders → Order.

Add exportPut export in front of every declaration.Make every property optionalname?: string — for partial updates and drafts.Readonly propertiesreadonly id: number — for data that is never changed in place.Use any where the type is unknownFor values that are always null and for empty lists. Off: null and unknown[].

JSON

TypeScript

The TypeScript appears here as you type. Paste an API response, or try the Sample to see optional properties, nested types and unions.

At a glance

Turns JSON into TypeScript interfaces, type aliases or Zod schemas with inferred types, instantly as you type.
Every object in a list is combined into one type: a property missing from some objects becomes optional (coupon?: string), and a value seen with several types becomes a union (string | number).
Nested objects get their own named types taken from their keys — address → Address, orders → Order[] — or are written inline. Places with the same name and structure share one type.
Objects whose keys are IDs or dates, such as {"1042": {…}, "1043": {…}}, are typed as Record<string, …> instead of one property per ID.
What the sample cannot show is reported instead of guessed: a value that is always null is typed null, and an empty list unknown[], each listed with its path.
Zod output declares every schema before it is used and adds a z.infer type for each one.
Accepts JSON with comments, trailing commas and single quotes, and points to the exact line and column of real errors.
Runs in your browser: the API responses you paste are never uploaded.

Step by step

How to convert JSON to TypeScript

1
Add your JSON
Paste a JSON object or array — an API response, a config, a database record — or click Open file.
2
Pick the output
Choose Interface, Type or Zod schema, and set the root type name under Options.
3
Read the notes
Below the result, check which properties were made optional and which values had no type to infer.
4
Copy or download
Copy the code into your project or download it as a .ts file.

Features

Everything you need, nothing you don’t

Three output styles

interface, type or a Zod schema with its inferred type. Switch between them without pasting again.

Optional properties found for you

Paste a list of records and every property that is missing from some of them is marked with ?, with the paths listed below the result.

Readable type names

Types are named from their keys and list items in the singular: categories gives Category[]. Name clashes are resolved, and names that would hide built-ins like Date are avoided.

Honest about gaps

A null or an empty list tells nothing about the real type. Those places are listed so you can fill them in, rather than hidden behind any.

Your conventions

Choose the root name, export, readonly, all-optional properties, inline or named nested types, and 2 spaces, 4 spaces or tabs.

Private

Responses with tokens, customer data or internal fields stay on your device. Nothing is sent to a server.

How JSON values become types

JSON	TypeScript	Zod
`"name": "Asha"`	`name: string`	`z.string()`
`"id": 1042`	`id: number`	`z.number()`
`"active": true`	`active: boolean`	`z.boolean()`
`"roles": ["admin"]`	`roles: string[]`	`z.array(z.string())`
`"address": {…}`	`address: Address`	`addressSchema`
in some objects only	`coupon?: string`	`z.string().optional()`
text in one object, `null` in another	`note: string \| null`	`z.string().nullable()`
a number in one object, text in another	`value: string \| number`	`z.union([z.string(), z.number()])`
`"avatar": null` every time	`avatar: null`	`z.null()`
`"tags": []`	`tags: unknown[]`	`z.array(z.unknown())`
`"extra": {}`	`extra: Record<string, unknown>`	`z.record(z.string(), z.unknown())`

Getting better types

Paste a list of several records rather than one. Optional properties and unions can only be found by comparing objects.
Include a record where nullable fields have a value, so avatar: null becomes avatar: string | null.
Types come from the sample, not from the API’s rules. A property present in every sample is typed as required even if the API may omit it — use Make every property optional for partial data.
Dates, emails and URLs are strings in JSON, so they are typed string. Add stricter checks in your Zod schema where you need them.
Property names that are not valid identifiers, such as last-login, are written in quotes: 'last-login': string.

Interface, type or Zod?

An interface and a type describe the same object shape; use whichever your codebase prefers. Both exist only while compiling and add nothing to the code that runs.

A Zod schema also checks data at run time. Parse an API response with it and you get an error when the response does not match, plus the TypeScript type from z.infer — so the type and the check never drift apart. The generated code needs the zod package in your project.

FAQ

Frequently asked questions

Is my JSON uploaded anywhere?

No. The types are generated in your browser, so API responses and internal data never leave your device.

How does it decide a property is optional?

When a list contains several objects, they are combined into one type. A property that some of them do not have is marked optional with ?. With a single object there is nothing to compare, so every property is required.

Why is a property typed null or unknown[]?

The sample held only null, or only an empty list, at that place, so the real type cannot be known. The tool lists these paths below the result. Paste a sample with real values there, or edit the type by hand. Turn on Use any where the type is unknown if you prefer any.

Can it generate a Zod schema?

Yes. Choose Zod schema to get z.object(…) schemas with .optional(), .nullable(), unions, arrays and records, each followed by its z.infer type. Nested schemas are declared before the ones that use them, and the code runs on both Zod 3 and Zod 4.

How are the types named?

From the property key, in PascalCase: shipping_address becomes ShippingAddress. Items of a list use the singular (orders → Order). If two different structures would get the same name, the parent’s name is added, as in OfficeAddress.

What about an object with IDs as keys?

An object whose keys are all numbers, UUIDs, long hex IDs or dates is treated as a lookup table and typed as Record<string, …>, with the values combined into one type.

Can I paste JSON with comments or trailing commas?

Yes. Comments, trailing commas, single quotes and unquoted keys are repaired and listed. Real mistakes are reported with their line and column.

Keep going

New

JSON Formatter & Validator

Format, validate, minify and repair JSON online. Get the exact line and column of every error, a collapsible tree view and sorted keys — private, in-browser.

Developer ToolsOpen

New

JSON to YAML Converter

Convert JSON to YAML and YAML to JSON online. Quotes yes, no and 0755 so every parser reads them the same, keeps numbers exact and never uploads your data.

Developer ToolsOpen

New

JSON to CSV Converter

Convert JSON to CSV for Excel or Google Sheets: nested objects become columns, lists are kept and big numbers stay exact. Also CSV to JSON. No upload.

Developer ToolsOpen

New

JWT Debugger

Decode any JWT and verify its signature in your browser: header, claims and expiry explained; HS256, RS256, ES256 and EdDSA with PEM or JWK keys. No upload.

Developer ToolsOpen

New