JSON to TypeScript Interface Converter

Each JSON value has a natural TypeScript counterpart, and the generator applies these rules to every key in your sample:

Mixed arrays become unions, so [1, "a"] is typed (string | number)[]. Objects keyed by IDs, such as {"101": {...}, "102": {...}}, become Record<string, T> instead of an interface with one property per ID.

Nested objects are named after their key, and array items get the singular form: an orders array produces an Order interface, categories produces Category. Objects with identical shapes share one type instead of producing duplicates.

For object shapes like API responses the two are interchangeable, and the TypeScript handbook suggests using an interface until you need a feature only type aliases have.

Pick one and stay consistent. Choose type if your codebase already composes types with unions or utility types such as Pick and Omit, and Readonly if API data should never be mutated after it arrives.

A single object can only show what one response looked like. Paste an array of several responses instead and the generator merges them: a key missing from some items becomes optional (?), and a key that is null in any item becomes nullable. These mean different things, and the difference matters:

interface User {
  nickname?: string;        // the key may be absent
  avatarUrl: string | null; // the key is always there; the value may be null
}

If your code treats both cases the same way, set Null values to T?. With the exactOptionalPropertyTypes compiler option enabled, an optional property can't be assigned undefined explicitly, so the distinction becomes a compile-time check.

TypeScript types vanish when your code compiles. If an API returns "price": "12.50" where your interface says number, nothing stops it, and the bug surfaces far from its cause. Types describe what you expect; they don't check it.

For data you don't control, validate at the boundary. A schema library such as Zod checks the data and gives you the static type from the same definition, so the two can't drift apart. The JSON to Zod generator builds that schema from the same sample:

const user = UserSchema.parse(await res.json()); // throws on a mismatch

• Dates arrive as strings, because JSON has no date type and JSON.parse doesn't revive them. ISO 8601 fields are typed string with a comment; convert them with new Date(value) where you use them.
• Integers above 2^53 are typed number with a warning comment, because JSON.parse has already rounded them. Ask the API for strings if exact IDs matter.
• Keys that aren't identifiers, such as first name or 2fa, are quoted: "first name": string. Reserved words like class and default are fine as property names.
• Fields that were only ever null are typed null with a comment, so you remember to widen them once you know the real type.
• Empty arrays become unknown[], or any[] with Use any, since there is nothing to infer from.