CLI Reference

Global Options

Flag	Description
--help	Show help for any command
--version	Show DBSprout version
--verbose	Enable verbose output
--quiet	Suppress non-error output

dbsprout init

Initialize a DBSprout project by reading a database schema.

dbsprout init --db <connection-string>
dbsprout init --file <schema-file>

Option	Description
--db	Database connection string (SQLAlchemy format)
--file	Path to schema file (SQL, DBML, Prisma, Mermaid, PlantUML)
--parser	Force a specific parser (auto-detected by default)
--output	Output path for the schema snapshot (default: .dbsprout/schema.json)

dbsprout generate

Generate seed data from the current schema.

dbsprout generate [OPTIONS]

Option	Description	Default
--rows	Number of rows per table	100
--engine	Generation engine (heuristic, spec, vectorized)	heuristic
--output-format	Output format (sql, csv, json, parquet, direct, upsert)	sql
--output-dir	Output directory	./dbsprout_output/
--db	Target database for direct insertion	—
--seed	Random seed for deterministic output	—
--llm-provider	LLM provider for spec engine	embedded
--no-cache	Ignore cached specs	false
--tables	Generate only specific tables (comma-separated)	all

dbsprout validate

Validate generated data against schema constraints.

dbsprout validate [OPTIONS]

dbsprout diff

Detect schema changes and generate incremental seed data.

dbsprout diff --db <connection-string>

dbsprout models

Manage embedded AI models.

dbsprout models download   # Download the default model
dbsprout models list       # List available models
dbsprout models remove     # Remove a downloaded model

dbsprout audit

View the audit log of all generation runs.

dbsprout audit [--last N]

dbsprout train

Fine-tune a model on your data (requires dbsprout[train-cuda] or dbsprout[train-mlx]).

dbsprout train --data ./training-data/ --output ./model/