Configuration Reference

DBSprout is configured via a dbsprout.toml file in your project root. All settings are optional — DBSprout works with sensible defaults.

Full Example

[database]
url = "postgresql://user:pass@localhost:5432/myapp"

[generation]
default_rows = 1000
engine = "heuristic"
seed = 42
batch_size = 1000

[generation.tables.users]
rows = 500

[generation.tables.orders]
rows = 10000

[output]
format = "sql"
directory = "./seeds/"

[llm]
provider = "embedded"
model = "qwen2.5-1.5b"
temperature = 0.1
max_tokens = 4096

[privacy]
tier = "local"
audit = true

[train]
backend = "unsloth"
epochs = 3
batch_size = 4
learning_rate = 2e-4

Sections

[database]

Key	Type	Description	Default
url	string	Database connection string	—

[generation]

Key	Type	Description	Default
default_rows	integer	Rows per table	100
engine	string	Generation engine	"heuristic"
seed	integer	Random seed	random
batch_size	integer	Rows per batch	1000

[generation.tables.<name>]

Per-table overrides:

Key	Type	Description
rows	integer	Row count for this table

[output]

Key	Type	Description	Default
format	string	Output format	"sql"
directory	string	Output directory	"./dbsprout_output/"

[llm]

Key	Type	Description	Default
provider	string	LLM provider	"embedded"
model	string	Model name	"qwen2.5-1.5b"
temperature	float	Generation temperature	0.1
max_tokens	integer	Max output tokens	4096

[privacy]

Key	Type	Description	Default
tier	string	Privacy tier (local, redacted, cloud, training)	"local"
audit	boolean	Enable audit logging	true

[train]

Key	Type	Description	Default
backend	string	Training backend (unsloth, mlx)	"unsloth"
epochs	integer	Training epochs	3
batch_size	integer	Training batch size	4
learning_rate	float	Learning rate	2e-4