Which JSON Schema draft does this target?

The output uses the draft 2020-12 dialect, declared with the $schema keyword set to the official 2020-12 meta-schema URI. This is the current widely-supported draft used by OpenAPI 3.1 and most modern validators.

Is the generated schema actually valid?

Yes. Every document is a well-formed object schema with a type, a properties map, and optionally a required array whose entries all exist in properties. It parses and validates as a real JSON Schema, so you can feed it straight into a validator.

What property types are produced?

Properties are drawn from string (with formats like email, date-time, and uri, plus minLength and maxLength), integer and number (with minimum and maximum), boolean, array (with an items subschema), and nested object types when nesting is enabled.

Why generate random schemas?

Random schemas exercise validation libraries, schema-to-form generators, fake-data producers, and documentation renderers across many shapes without hand-writing each one. They are useful for fuzzing and regression-testing tools that consume schemas.

Does this send data anywhere?

No. The schema is built locally in JavaScript and serialised with JSON.stringify, so nothing is uploaded and it works offline.

Random JSON Schema Generator

This tool generates valid JSON Schema documents in the draft 2020-12 dialect, complete with randomly chosen property types, formats, and constraints. It is built for testing the tools that consume schemas — validators, form generators, mock-data producers, and documentation renderers — without hand-writing each example.

How it works

The generator builds an object schema and fills its properties map with a configurable number of fields. Each field is assigned a random type and type-appropriate constraints:

string  -> format: email | date-time | uri (sometimes), minLength, maxLength
integer -> minimum, maximum
number  -> minimum, maximum
boolean -> (no extra constraints)
array   -> items: { type: <random scalar> }
object  -> nested properties (when nesting is enabled)

If you enable required fields, a random subset of the property names is copied into a required array — and because those names are taken from the actual properties, the schema stays internally consistent. The additionalProperties flag and $schema declaration are added so the document is a complete, standalone schema.

Tips and notes

Enable nested objects to test how your tooling handles recursive subschemas, and toggle additionalProperties to check both the open and closed object cases. Use a high property count to stress-test renderers and form generators. The output is deterministic only within a single generation — regenerate for a new shape, or copy and save a schema you want to reuse in a fixed test.