Task-oriented
how-tos.
Each recipe is a self-contained walkthrough for a real situation — the problem, the exact commands, and what you end up with.
Seed CI test fixtures
Deterministic, fast, reproducible seed data inside a CI pipeline.
Re-seed after a schema migration
Update only what changed instead of regenerating everything.
Match production distributions
Synthetic rows whose distributions and correlations mirror a real sample.
Bulk-load millions of rows
Database-native throughput with COPY and LOAD DATA.
Keep real data private
Control exactly what — if anything — leaves your machine.
FK cycles & self-references
What DBSprout does automatically when foreign keys form a loop.
Fine-tune a local model on your data
Train a private QLoRA adapter and generate from it.
Column-aware values with --engine spec
Let an LLM write a per-column generator spec once, cache it forever, and get semantically correct values for unusual column names.
Review a dataset in Web Studio
Walk the ERD, tune column specs, watch SSE progress, and export — all from a browser.
Keep everything on a local LLM
Pin spec generation to the embedded Qwen2.5-1.5B model or your own Ollama instance — nothing leaves the machine.