Relational fake datasets
Generate connected test data (users, their orders, the orders' line items) with primary and foreign keys drawn from one seeded stream. Same seed, same locale, same preset: byte-identical files. CSV/JSONL per entity with a manifest, or one SQL bundle.
runs in the faker family worker. Locale data never loads into this page
same seed + locale + preset reproduces every file
- Presets are the registry. No arbitrary user-defined entity graph in this round.
- Values come from the same bounded field set as the bulk test-data builder; no finance fields anywhere.
- The SQL bundle reuses the CSV-to-SQL quoting rules (TEXT columns, '' doubling) and nothing more: no schema design, no execution.
Nothing in yet
Paste, drop, or type to begin. Everything stays on this device.
parents first, children reference parent keys