Structured Output Validator
Validate AI outputs against JSON Schema — test OpenAI, Anthropic, Gemini structured responses
Supported Schema Formats
OpenAI Function Calling
{ "name": "fn", "parameters": { "type": "object", ... } }Anthropic Tool Use
{ "name": "fn", "input_schema": { "type": "object", ... } }OpenAI Structured Output
{ "schema": { "type": "object", ... } }Raw JSON Schema
{ "type": "object", "properties": { ... } }Supported Validation Keywords
- type, const, enum
- required, properties, additionalProperties
- minimum, maximum, exclusiveMinimum, exclusiveMaximum, multipleOf
- minLength, maxLength, pattern
- items, prefixItems, minItems, maxItems, uniqueItems
- minProperties, maxProperties
- allOf, anyOf, oneOf, not
- $ref (local references)
You’re using structured output from an AI model — OpenAI’s JSON mode, Anthropic’s tool use, or Gemini’s response schemas — and need to verify the output matches your expected schema before processing it in your application.
What This Tool Does
Paste an AI model’s structured output and a JSON Schema, and instantly see whether the output conforms. The validator highlights schema violations, missing required fields, type mismatches, and unexpected properties. Test your prompts’ structured output reliability.
Everything runs in your browser. Your data never leaves your device.
When to Use This
- Prompt development — verify that your prompt consistently produces schema-compliant output
- Schema design — iterate on your JSON Schema definition until it captures the right structure
- Integration testing — validate AI outputs before building downstream processing logic
- Debugging — diagnose why your application fails to parse an AI model’s response
FAQ
Does this send my data to a server?
No. All validation happens entirely in your browser.
Which schema drafts are supported?
JSON Schema Draft 4, Draft 7, and Draft 2020-12 are all supported.