Covers the keywords you actually meet day to day (types, required, enum, const, numeric and string bounds, items, uniqueItems, anyOf/oneOf/allOf/not, $ref to local pointers). Exotic draft corners like dynamic refs and content encoding are not implemented.
Paste a payload and a schema and get a precise list of what fails, each error naming the exact path inside the document and the keyword that rejected it — the two things you need to fix it. It pairs with the schema generator: infer a schema from one payload, then test it against the awkward ones.
Every failure names the instance path and the keyword that rejected it — the two things you need.
Types, required, enum, const, bounds, items, uniqueItems, combinators, patternProperties and local $ref.
See everything that is wrong at once, or stop at the first failure while iterating.
Enforce email, uri, uuid and date formats, or treat them as annotations like the spec does.
Types, required, enum, const, minimum/maximum and their exclusive forms, multipleOf, minLength/maxLength/pattern, format, items and prefixItems, minItems/maxItems/uniqueItems/contains, properties, patternProperties, additionalProperties, minProperties/maxProperties, dependentRequired, allOf/anyOf/oneOf/not, and $ref to local pointers.
No, and it says so. Dynamic references, content encoding and the more exotic corners of the specification are not implemented. For everyday API schemas it will tell you what is wrong.
Format checking is on by default. The email, uri, uuid, date-time, date, time, ipv4 and hostname formats are checked against patterns; switch off "Enforce formats" to treat them as annotations, which is what the specification does by default.