Question 1

What is a JSON Schema?

Accepted Answer

A JSON Schema is a contract for JSON data: it declares which fields an object must have, what type each value must be, which fields are required, and what formats strings must follow. APIs use it to validate requests before processing them, form builders generate forms from it, and documentation tools turn it into readable specs. Writing one by hand for a large payload is slow and error-prone, which is why generating it from a real sample is the standard way to start.

Question 2

How does the generator infer the schema?

Accepted Answer

It walks your sample value by value. Objects become a properties map with every key marked required, since a single sample shows only what is present. Arrays are sampled across all their elements: if elements are objects, their properties are merged, and a property stays required only if every element has it. Numbers are split into integer and number, and strings are tested for common formats: email addresses, ISO dates and timestamps, and URLs are labeled with the matching format keyword.

Question 3

What should I adjust after generating?

Accepted Answer

Treat the output as an accurate draft, not the final contract. Review the required list first: a sample cannot show which fields are sometimes absent, so loosen fields you know are optional. Tighten strings your API cares about with pattern, minLength or enum values, and add descriptions: future developers, including you in six months, will thank the schema that explains itself. Then validate real payloads against it with the JSON formatter and validator here.

Question 4

Which schema version does it produce?

Accepted Answer

The output targets JSON Schema draft-07, the most widely supported version across validators, form libraries and API tooling. It opens with the $schema declaration so tools detect the dialect automatically. If your stack wants 2019-09 or 2020-12, the generated structure, types, properties, items, required, is compatible; you mainly change the $schema URL and any version-specific keywords you add later.

Question 5

Is my JSON pasted anywhere or stored?

Accepted Answer

No. The inference runs entirely in your browser tab. Sample payloads often contain real production data shapes, tokens in strings, or customer records; that is not something to paste into a stranger's server for a schema you could compute locally, so here you compute it locally.

JSON Schema Generator

Use JSON Schema Generator

How to use the JSON Schema Generator

Paste sample JSON

Read the inferred schema

Copy or download it

Refine and validate

About the JSON Schema Generator