JSON to JSON Schema

JSON to JSON Schema Online

JSON Schema is a vocabulary that describes the structure of JSON data: expected types, required fields, format rules and the shape of nested objects. Validators such as Ajv check data against it, and it is used for API validation, code generation and documentation.

Paste a representative JSON object or array and ZeroTool will infer a draft-07 JSON Schema with object properties, required fields, array item types, and nested structures. This is a fast starting point for API validation, documentation, mocks, contract tests, and data pipeline checks.

If you searched for to_json_schema or jinja2schema infer, note the difference: template inference tools inspect Jinja templates, while this generator infers a schema from concrete JSON examples.

Understanding the Output

The generated schema always includes $schema pointing to draft-07. Object properties are inferred from the JSON keys. The required array lists every non-null key found. Array items use an inferred schema based on the elements present.

Next Steps After Generation

A generated schema is a starting point. Common refinements include: adding description fields for documentation, tightening string constraints with minLength or pattern, marking optional properties by removing them from required, and adding enum for fields with a known set of values.

Example

Input:

{
  "orderId": 1001,
  "total": 59.9,
  "coupon": null,
  "items": [{ "sku": "A1", "qty": 2 }],
  "tags": []
}

Output:

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "orderId": { "type": "integer" },
    "total": { "type": "number" },
    "coupon": { "type": "null" },
    "items": {
      "type": "array",
      "items": {
        "type": "object",
        "properties": {
          "sku": { "type": "string" },
          "qty": { "type": "integer" }
        },
        "required": ["sku", "qty"]
      }
    },
    "tags": { "type": "array", "items": {} }
  },
  "required": ["orderId", "total", "items", "tags"]
}

• A whole number becomes integer and a decimal becomes number. If qty can be 1.5, change it to number.
• coupon is null in the sample, so its type is null and it is not in required. If it can also be a string, change the type to ["string", "null"].
• An empty array gives "items": {}, which accepts any item.

Example: Success and Error Responses

An API that returns {"ok": true, "data": …} on success and {"ok": false, "error": …} on failure needs both shapes in the sample. Put one of each in an array:

[{"ok": true, "data": {"id": 7}}, {"ok": false, "error": "not found"}]

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "array",
  "items": {
    "type": "object",
    "properties": {
      "ok": { "type": "boolean" },
      "data": { "type": "object", "properties": { "id": { "type": "integer" } }, "required": ["id"] },
      "error": { "type": "string" }
    },
    "required": ["ok"]
  }
}

ok appears in both responses, so it is the only required key; data and error are optional. Use the items schema to validate a single response. A sample with only the success response would make data required and reject every error response.

Using the Schema with Ajv

Ajv 8 validates draft-07 with its default Ajv class. The format keyword needs the ajv-formats package; without it, compile() throws unknown format "email":

import Ajv from 'ajv';
import addFormats from 'ajv-formats';

const ajv = new Ajv();
addFormats(ajv);

const validate = ajv.compile({
  type: 'object',
  properties: {
    email: { type: 'string', format: 'email' },
    coupon: { type: ['string', 'null'] },
  },
  required: ['email'],
  additionalProperties: false,
});

validate({ email: 'a@example.com', extra: 1 }); // false: must NOT have additional properties

The generator does not add format, minimum, maxLength, pattern, enum, or additionalProperties: false. Add them for the rules your data must follow.

Limits

• Array items are merged. For [{"id": 1, "name": "Alice"}, {"id": 2, "email": "b@example.com"}], items lists id, name, and email, and only id, which every item has, is required. Integers and decimals in one array become "number". The schema always accepts the sample it came from; GenSON merges samples the same way.
• Required comes from the sample. In a single object every non-null key is required. If a field is optional in your data but present in the sample, remove it from required yourself.
• Mixed types give a type list. An array such as [1, "x", {"a": 1}] gives "type": ["integer", "string", "object"] with the object’s properties and required next to it. Those keywords apply only to the object items.
• Draft-07 only. To use draft 2020-12, change $schema and use Ajv’s Ajv2020 class.