JSON Schema & structured output helpers

JSON Schema Generation, Validation, and LLM Output Guide

JSON Schema powers API contracts, OpenAPI docs, gateway validation, and structured LLM output. Inferring a draft from a real example and then tightening rules is usually safer than starting from an empty file.

FormaX infers type, required, object, and array structures, then validates instances with AJV. You still need to add business rules such as enum, format, minimum, and pattern.

Common Use Cases

• API contract design: generate Schema drafts from real response examples to speed up OpenAPI documentation.
• Gateway validation: intercept non-compliant requests at API Gateway or BFF layers using Schema.
• LLM structured output: create function calling or response_format configs for GPT, Claude, and similar models.
• Frontend-backend alignment: unify request/response format definitions to reduce missing fields and type mismatches.
• Data quality checks: batch-validate logs or imported data against expected structures.

From Example to Usable Schema

1. Copy a real successful response and remove sensitive fields.
2. Generate a draft schema and review whether required fields match the business contract.
3. Add format, pattern, or range constraints for IDs, emails, dates, and amounts.
4. Validate missing fields, wrong types, and edge cases before publishing to OpenAPI or LLM configs.

Good Example JSON

{"id":1001,"email":"user@example.com","plan":"pro","active":true}

Inference Does Not Decide Business Rules

• One example only proves the current value type, not every possible value.
• Empty arrays cannot reveal item type, so include at least one representative element.
• For LLM structured output, add descriptions to reduce wrong field filling.

FAQ

Related Tools

• Formateador JSON
• RAG Chunker
• JWT Decoder