JSON Schema
What is JSON Schema?
JSON Schema is the standard vocabulary for describing the shape of JSON data: which properties exist, their types, which are required, and what values are allowed. APIs use it to validate request bodies, config systems use it to guard settings files, and code generators use it as their source of truth. This tool works in both directions: validate a document against an existing schema, or generate a schema from real sample data.
Features
- Draft 2020-12 validation — full structural validation with every error reported (no stop-at-first-failure), each with a JSON Pointer to the exact offending value, capped at 100 errors
- Schema errors vs instance errors — a broken schema (invalid JSON, bad pattern, unresolved
$ref) is reported separately from validation failures, so you always know which side to fix - Schema generation from samples — paste example JSON and get a deterministic draft 2020-12 schema: integer vs number detection, sorted properties and required lists, type unions for mixed content, and
anyOfbranches when objects and primitives mix - Format detection — strings that consistently look like RFC 3339 date-times, UUIDs, emails or URIs get a
formathint in generated schemas - Honest format semantics — during validation,
formatfailures are reported as warnings, not hard errors, matching how most validators treat format as an annotation
Limitations
Validation is pinned to draft 2020-12 and resolves local $refs only (#/... pointers and schemas bundled under $defs/definitions) — remote references are never fetched, and a schema that needs one fails with a clear compile error. Instances and schemas are capped at 1 MB each in validate mode and 5 MB in generate mode.
Privacy
Both validation and generation run locally in your browser — pasted data is not uploaded. (A public API endpoint also exists for programmatic use; anything you send there is, by definition, transmitted.)