Why a schema-based test data builder?
The fake data generator gives you ready-made profiles (name, address, tax ID). But sometimes you need your own schema: a product with SKU, price (10 to 500), category (5-value enum), tag array. Or an order with a nested object for the customer.
Here you design the structure field-by-field: type, range, pattern, options. Generate N records matching exactly. Export to JSON, CSV, SQL INSERT or NDJSON to drop into your app or database.
Available types: string, number, integer, boolean, date, email, UUID, enum (pick-list), array, nested object, and "faker" (pulls from our pools - first name, city, company).
How to use it
- Click "Add field" - creates a new field in the schema. Type a name (e.g. "price").
- Pick a type - string, number, date, enum, faker, and so on. Type-specific options appear (range, length).
- For objects you can add fields inside fields (recursive). For arrays pick the item type and min/max count.
- Set record count to generate - 1 to 10 000.
- Pick output format - JSON (most common), CSV (Excel), SQL INSERT (database), NDJSON (one line per record, for streaming).
- Type a seed (optional) for deterministic output. Click "Copy" or "Download".
When this is useful
Real projects that need a custom schema:
- E-commerce app - generate 100 products: SKU (UUID), name (faker), price (number 10-500 with 2 decimals), category (enum: "Electronics", "Clothing", "Home"), tags (array of 1-3 strings). Fill a demo product page in 5 seconds.
- Booking system - reservation record: id (UUID), customer (object with name and email), date range (2024-2025), status (enum: "confirmed", "cancelled", "pending"), guests (array of 1-6 names).
- GraphQL APIs - schema-first design. Define types in the tool, generate mock data, drop into Apollo Server mocked resolvers. Frontend works before backend writes a line.
- Database migration - test a new table schema. Generate 1000 records matching the structure, load with SQL INSERT, check `WHERE` performance on a new index.
- Storybook stories - the `<ProductCard>` React component has 8 props. Generate 5 object variants with different values, each as its own story. Clickable palette instead of hardcoded.
- CSV/Excel samples - a client asks "what would the sales report look like?". Generate a CSV with date, amount, customer, status. Open in Excel, show.
- Validator tests - your `validateOrder(order)` must handle 100 edge-case variants. Generate 100 objects, each with different boundary values.
- Live demos - a "this is how our app works" talk needs a populated DB. Generate 50 records matching the real shape.
Related: ready profiles in fake data generator, themed placeholder text in themed lorem ipsum.
Questions and answers
The other one has fixed fields (name, phone, tax ID). This one has any fields you define. The other generates "a user". This generates whatever you want: a product, order, transaction, Jira ticket, Reddit comment, bank transaction. The other is faster for "give me 100 fake users". This one is more powerful for "give me 100 objects with these specific fields and constraints".