Inference from real examples
The tool analyzes objects, arrays, strings, numbers, booleans, and null to create a starter model. It also detects common string formats like email, URL, UUID, and dates.
Arrays with different objects become combined types where possible. The output is a starting point that should be reviewed before production.
Three outputs for one payload
TypeScript helps with autocomplete and code contracts. Zod adds runtime validation, while JSON Schema works for documentation, validation, and external tooling.
Processing happens in the browser whenever possible. Files and inputs do not need to be uploaded to the site server for the main workflow.
- TypeScript interface
- Zod schema
- JSON Schema draft 2020-12
When to review manually
A JSON example does not reveal missing optional fields, business rules, closed enums, or numeric limits. Adjust generated code against the real API.