Fill in these first
- Schema: [paste CREATE TABLE statements, an ORM model file or a table list with columns and types]
- Database and tooling: [e.g. PostgreSQL with Prisma, MySQL with Laravel seeders, SQLite with pytest]
- Output format: [SQL inserts, JSON fixtures, CSV, or seeder code]
- Rows per table: [e.g. 10 users, 50 orders, 200 order lines]
- Locale for names, addresses and phone numbers: [e.g. India, UK, mixed]
- Edge cases tests need: [e.g. user with no orders, cancelled order, very long name, zero-price item]
Task
Create seed data that respects every constraint in the schema and covers the edge cases, then write the code to load it.
Data rules
- Insert parent tables before child tables, and every foreign key points to a row that exists.
- Respect unique, not null, check constraints, enum values and column lengths.
- Dates are consistent: an order is never shipped before it was placed.
- Totals match their lines where the schema stores both.
- Each edge case appears at least once and is labelled in a comment.
- Fixed IDs or a fixed random seed so tests get the same data on every run.
Output format
- | Table | Rows | Edge cases covered |
- The data in the format requested, one block per table.
- The loading script, with a reset step that clears tables in reverse order.
- Assumptions about any column whose meaning was unclear.
Rules
- No real people, real email domains or real phone numbers; use example.com addresses and clearly fake numbers.
- Never include real passwords; store a placeholder hash and note it.
- Do not change the schema; list suggested schema fixes separately.






