Features

Define entities and relationships, not just tables. SqlCAD understands how your data connects, allowing it to automatically generating complex JOIN paths so you don't have to write them manually.

Enforces a clean separation of concerns. Each layer has specific rules and optimizations handled by the compiler.

SqlCAD validates your entire graph at compile time. It checks for missing columns, invalid type operations, and broken references before any SQL is run.

DEFINE MART BadMath AS (
  SELECT 
    Amount + 'string' 
  FROM Orders
);
// Error: Cannot apply operator '+' 
// to types DECIMAL and STRING

Every load operation automatically tracks statistics. Define assertions (`not_null`, `distinct`, `expression`) directly on your models.

DEFINE REPORT DailySales {
  tests:
    - not_null: [Date, Revenue]
    - expression: "Revenue >= 0"
} AS ( SELECT ... )

SqlCAD automates `valid_from` and `valid_to` management, generating complex merge logic for you.

DEFINE ENTITY Orders {
  temporal: {
    event_time_col: OrderDate
  }
}
// Compiler generates:
// MERGE ... 
// ON o.id = s.id 
// AND s.valid_to IS NULL...

Define reusable ML features. SqlCAD handles point-in-time correctness to prevent training data leakage.

DEFINE FEATURE_SET CustomerFeatures {
  entity: Customers
  temporal: { point_in_time: true }
  features:
    - recent_order_count
    - lifetime_value
}

Stop writing manual "AS OF" joins. SqlCAD automatically injects the correct temporal logic.

Declaratively manage your database schema. The Diff Engine compares code to specific database state.

> sqlcad schema plan

Diff Engine detected:
- [NEW] Column 'Discount' in 'Orders'
- [MOD] Type 'Amount': INT -> DECIMAL

Generated 'migration_001.sql'

Write logic once, compile to optimized SQL for multiple backends.