Paste any JSON response and get production-ready types in six dialects at once: TypeScript interfaces, a Zod schema, Python dataclasses, Go structs, C# classes and a JSON Schema document. Nested objects are detected, de-duplicated and named; optional, nullable and literal-union fields are inferred from real data.
Every object is fingerprinted by its sorted key list, so the same shape appearing in three places becomes
one shared type instead of three copies. A property is optional when it is missing from some instances of
that shape, and nullable when null appears among its values. With literal-union inference on,
a string property whose values repeat from a small closed set (for example paid,
shipped, refunded) is emitted as a union of literals rather than a bare string.
Numbers are all treated as 64-bit floats because JSON does not distinguish integers from decimals, except
in Python and Go where a whole-number property becomes int / int64.