JSON Schema Validator

Draft-07, 2019-09, and 2020-12. Pick the draft with the selector above the schema editor to match your schema's own "$schema" URI (e.g. a schema declaring "https://json-schema.org/draft/2020-12/schema" should be validated against 2020-12) — a schema written for an older draft can compile or behave differently under a newer draft's meta-schema, especially if it uses draft-specific keywords, so the selector should track the schema rather than default to the newest option.

Yes. Both the schema and the data you paste are validated locally using the Ajv validation engine running in your browser tab. Nothing is uploaded to a server, logged, or stored — closing the tab clears everything (unless you explicitly use Share, which encodes your input into the URL itself).

Common formats from the "format" keyword are validated via ajv-formats, including email, uri, uri-reference, date, date-time, time, uuid, hostname, ipv4, and ipv6. Format checks only run when your schema actually declares a "format" on a string field.

An unresolved $ref is reported as a schema problem (a compile error), not silently ignored and not treated as a data validation failure. Make sure every $ref points at a definition that actually exists in the same schema document — this validator does not fetch remote $ref URLs.

JSON to JSON Schema (/json/to-schema/) goes one direction: give it example data and it infers a starting schema. This tool goes the other direction: give it a schema you already have (hand-written or generated) plus a document, and it tells you exactly which parts of the document violate which rules. Use the generator first if you don't have a schema yet, then validate real data against it here.

The validator collects every rule violation in one pass (Ajv's "allErrors" mode) so you can fix a document in one editing pass instead of a slow fix-one-error-see-the-next loop.