Validate Structure
And the Math Inside It.
LLMs generate JSON that looks right but contains structural errors and arithmetic mistakes. QWED's Schema Verifier performs deterministic JSON/type validation with embedded math verification for computed fields.
"items": [
{"name": "Widget A", "qty": 3, "price": 29.99},
{"name": "Widget B", "qty": 2, "price": 49.99}
],
"subtotal": 199.95,
"tax": 18.00,
"total": 210.95
}
Structure + Math = Complete Validation
JSON Schema validation checks types and required fields. But it can't verify that a "total" field actually equals the sum of line items. QWED bridges this gap.
Schema Compliance
Validate against JSON Schema, OpenAPI specs, or custom type definitions. Catch missing fields, wrong types, and structural violations.
Math Delegation
Computed fields (totals, averages, percentages) are automatically delegated to the Math Verifier engine for symbolic validation.
Nested Validation
Recursively validate deeply nested JSON structures with cross-field constraints and referential integrity checks.
Engine Capabilities
The Complete Output Validator
When your AI generates structured data—invoices, reports, API responses, configuration files—QWED ensures both the structure and the numbers are correct.
Math + Schema Synergy
QWED's unique advantage is combining structural validation with mathematical verification. A JSON invoice that passes schema checks but has wrong totals is caught by the Math Verifier integration.