Schema & type generation
Choose a count and seed; each instance is checked against your schema.
Schema and instances stay in this browser; nothing is uploaded.
Review the schema, then generate.
JSON Schema Mock Data Generator turns a contract into synthetic instances for tests, demos and API stubs. Start with the constraints your test actually depends on, rather than a production record that happens to pass today.
This seeded JSON generator accepts a defined draft-07 subset. A schema describes each record, while the quantity controls the outer collection. JSON Schema to mock data generation does not infer business relationships from field names.
Save the schema, seed and settings beside the fixture. The same engine version reproduces the same batch, making a regression easier to discuss without copying real customer records.
Boundary sampling favors feasible endpoints for numbers, lengths and collection sizes. Each candidate still passes independent schema validation; an unsupported constraint stops the run rather than becoming a convincing placeholder.
Count starts at 5 and seed at 42. If your schema describes an array, each generated record is itself an array. Use minItems and maxItems inside the schema to constrain those inner arrays. Seed: integer from 0 to 4294967295.
Choose normal or boundary sampling, then Generate or Ctrl/Cmd+Enter. Editing the schema or any option makes previous exports stale. Stop cancels the current run; an older task cannot replace a newer result.
The report lists unsupported schema locations. Fix the contract and regenerate before copying. Example and Clear require a replacement preview, Apply and one-step Undo. Choose JSONL for one complete instance on each line.
Objects and arrays retain their structure. Properties, required fields and additionalProperties guide object generation; a missing optional property is different from a present null, which must be allowed by its schema.
Type unions, enum and const are evaluated together with sibling constraints. A fixed value outside a numeric bound cannot become a valid sample simply because it appears in enum.
String length, item count and property count limits shape the result. uniqueItems applies within each generated array; it does not make every top-level record in a batch different.
JSON always wraps the records in an outer array, including count 1. JSONL test data places an independent JSON value on each line. The input schema validates the records, not the extra JSON export wrapper. JSON uses fixed two-space indentation; JSONL keeps each instance on one line.
A fixed seed initializes deterministic generation. Keep the schema, count, mode and engine version unchanged to repeat a batch. A constant or a one-value enum may produce the same data under every seed.
The tool implements the draft-07 families listed in the rules, with independent validation of every emitted instance. Unsupported keywords stop the whole run; they are never silently removed from your contract.
Choose 1–100 top-level records. Nested collection limits come from the schema and the stated resource budget. Larger constraints may fail even when the requested record count is small.
Yes. Select JSONL for newline-delimited JSON, also commonly called NDJSON. Each line is a full instance and may be an object, array or primitive. It is not a multiline, pretty-printed JSON array.
Yes. JSON Schema Mock Data Generator parses, generates and validates locally after loading. The schema is not sent to a mock API, and generation does not fetch external references.
See a constraint we should support?
Use JSON Schema Mock Data Generator with a small contract first, then compare normal and boundary batches against the expectations of your own tests.
Save both the fixture and the generation settings so another developer can reproduce the same case.