Generate verified JSON Schema test fixtures
Turn a supported JSON Schema definition into valid examples and intentionally invalid test cases. Inspect the target validation rule, reproduce results with a seed, and export a fixture package. Your schema and generated fixtures are processed in this browser tab.
JSON Schema Test Fixture Studio
Paste a schema, check compatibility, then generate and verify fixture cases.
Use the same schema, seed, count and application version to reproduce the same fixture data. The seed is not cryptographic randomness.
3. Generated test cases
Fixture list
No fixtures yet.
Fixture details
Select a fixture to inspect its JSON and validation result.
Rule coverage and mutation results
Coverage appears after fixture generation.
How the JSON Schema test generator works
The studio reads your JSON Schema, checks that it fits the documented supported subset, creates candidate instances and validates each candidate with the built-in subset validator. Positive fixtures are included only when accepted. Negative mutations are marked verified only when the validator rejects them for the intended rule without additional validation failures.
- Paste or upload: provide a JSON Schema definition as text or a local JSON file.
- Check compatibility: the tool reports unsupported keywords before generation rather than silently ignoring them.
- Generate: a seeded pseudo-random process creates valid instances and targeted negative mutations.
- Verify: every candidate is checked against the same documented validation subset.
- Export: download the package as JSON or one-fixture-per-line JSONL.
Worked example: an age boundary
For a property with "type":"integer", "minimum":18 and "maximum":65, a valid boundary example can use 18. A targeted lower-bound mutation can use 17. The fixture is marked verified only when the validator reports the intended minimum failure.
{
"age": {
"type": "integer",
"minimum": 18,
"maximum": 65
}
}This example explains the boundary rule; actual fixture values are generated from the schema and selected seed.
Supported JSON Schema features and limitations
This release implements a deliberately limited JSON Schema Draft-07 validation and generation subset. It is not a complete Draft-07 validator. The root schema must declare type: "object".
Supported in this release
- Nested object properties and required properties.
- String, number, integer, boolean and null types.
- Enum and const values.
- Numeric minimum, maximum, exclusiveMinimum and exclusiveMaximum.
- String minLength and maxLength.
- Arrays with a single items schema, minItems, maxItems and uniqueItems.
- minProperties, maxProperties and boolean additionalProperties.
- Nested object and array traversal up to a configured depth limit.
Methodology, accuracy and reproducibility
The generator uses a deterministic xoshiro128** pseudo-random stream. The same schema text, seed, requested valid count and application version reproduce the same fixture data within that version. Changing the schema, seed or generator version can change the output. This is reproducibility for software testing, not cryptographic randomness.
Numeric and string boundary mutations are derived from the supported constraints. Array and object mutations are constructed from the existing valid base fixture. The built-in validator checks expected and actual outcomes. A mutation that fails for more than its intended rule is marked unverified. A rule that cannot produce a supported isolated case is reported as unavailable.
Validation follows the documented subset behavior, not the complete JSON Schema specification. Review the exported fixtures and verify them with the validator used by your own application before relying on them in a test pipeline.
Privacy and local processing
The application reads pasted schema text and selected JSON files in the browser. Fixture generation, validation and export are performed locally. This theme does not include an application backend, remote schema fetcher, validator CDN or analytics script, and it does not send schema or fixture contents to a RandomGenerator application server.
The website itself is hosted by Blogger/Google. Blogger, browser features, and any site-level analytics or advertising services configured separately by the publisher may process ordinary visit, request or device information under their own terms. See the Privacy Policy for the deployed configuration.
Frequently asked questions
What does this JSON Schema test fixture generator do?
It creates valid JSON examples and targeted invalid test cases from supported Draft-07 object schemas. It verifies the result against its documented built-in subset validator and exports the fixtures.
Does it support every JSON Schema feature?
No. It supports a limited subset described above. Patterns, formats, references, composition keywords, conditionals and other advanced features are not supported. Compatibility checks report unsupported keywords.
Are my schemas uploaded?
The tool processes schemas and fixtures in your browser and does not send their contents to a RandomGenerator application backend. The site is hosted by Blogger/Google; other services configured on the site may have separate data practices.
Can I test a live API?
No. The tool creates fixture files. It does not send HTTP requests to your API or execute your application's tests.
Is the seed secure for passwords or tokens?
No. The seed is for deterministic test-data reproduction only. Do not use it to create passwords, credentials, tokens or secrets.
Can I import the exported fixtures into Jest or Vitest?
You can import the downloaded JSON or JSONL fixture records into your own test code. This release does not generate or execute Jest, Vitest, Playwright or Postman tests.
Why is a negative fixture marked unverified?
The candidate may fail for an additional rule or may not demonstrate the intended validation keyword in isolation. Review the validator errors and schema constraints before using that case.
What are the generation limits?
The schema input is limited to 64 KB, schema depth to 10, schema nodes to 500, valid examples to 1–20, array generation to 50 items and mutation coverage to 100 rules.