When does the generator actually use oneOf?

Only when array items are genuinely different value types, not just differently-shaped objects, a mix of strings, numbers, and objects, for example. Differently-shaped objects get merged into one combined schema instead; oneOf is reserved for cases a single object schema truly can't represent.

Why did an empty array produce {"type":"array","items":{}} instead of something more specific?

There's nothing in an empty array to infer a shape from, so items is left as an empty schema, which matches anything, rather than guessing a type it has no evidence for. Fill in a real example and regenerate once you have at least one item to base it on.

Why did a field come out typed "integer" when I know it can sometimes hold a decimal?

The generator infers each field's type from the one example value in your pasted JSON; if that value happens to be whole, like 5, it's typed integer. A schema built from a single sample only reflects what that sample looked like, so it's worth loosening a type by hand afterward for any field you know can vary in ways your example didn't show.

JSON Schema Generator: Auto-Infer Draft-07 & 2020-12

JSON Inputpaste or type · or drop a .json file

Browse

JSON Schema

Required

Readyjson-schema

What is JSON Schema?

JSON Schema is a vocabulary for annotating and validating JSON data. A schema document describes the expected structure of a JSON value: which keys are required, what types are allowed, and how nested objects and arrays are shaped.

It is supported by validators in every major language (Ajv for JavaScript, jsonschema for Python, networknt for Go) and used in OpenAPI specs, VS Code settings, GitHub Actions, and many API platforms. Draft-07 is the most widely supported version; Draft 2020-12 is the latest stable specification.

How the inference works

The generator traverses your JSON recursively. Arrays are typed by their elements: if all items share the same structure, a single schema is inferred; mixed arrays produce {"oneOf":[...]}.

string{"type":"string"}

integer{"type":"integer"}

float{"type":"number"}

boolean{"type":"boolean"}

null{"type":"null"}

object{"type":"object","properties":{...}} + required[]

After generation: refine and validate

The generated schema is a structural starting point. To validate your JSON against it, paste both into an online validator or use a library like Ajv (npm install ajv). For API schemas, consider generating from your OpenAPI spec with tools like openapi-typescript rather than from example JSON.

minLengthmaxLength — bounds for strings

minimummaximum — bounds for numbers

patternregex-validated strings

formatemails, dates, and other known shapes

enumfixed set of allowed values

additionalPropertiesfalse — reject unknown keys

Frequently asked questions

No, it merges them into a single object schema instead of using oneOf. A key present in every item is marked required; a key missing from at least one item is still included in properties but left out of required, so the merged schema accepts the full range of shapes your array actually contains.

JSON → Schema

What is JSON Schema?

How the inference works

After generation: refine and validate

Frequently asked questions