sqb init

Creates a new SQLBuild project with a minimal directory structure and configuration files.

Usage

sqb init

No flags. Run in the directory where you want to create the project.

Project layout

sqb init creates the configuration, empty resource directories needed for a standalone project, and the SQLBuild agent skill (see sqb skills):

my-project/
  sqlbuild_project.toml
  .agents/skills/sqlbuild/
  .claude/skills/sqlbuild/
  models/
    staging/
    marts/
  schemas/
  sources/
  seeds/
  python/
  hooks/
    sql/
    python/
  tests/
    unit/
    scenarios/
  functions/
    sql/
    python/
  macros/
  audits/
    generic/
    singular/

Empty directories contain .gitkeep files so the scaffold can be committed. Reusable attached audit definitions belong in audits/generic/; cross-resource singular audits belong in audits/singular/. Add reusable SQL lifecycle hooks to hooks/sql/ and decorated Python lifecycle hooks to hooks/python/; see Hooks.

These top-level roles are project-wide. Once an audit, schema, hook, or macro is used only under one folder, such as models/marts/, SQLBuild requires it to move into that folder’s _sqlbuild/ directory; see Where to Put Declarations.

The generated project uses DuckDB, creates a named developer connection shared by dev and prod, defaults models to table materialization, and makes audits warn by default. Its configuration follows this shape:

name = "my_project"
adapter = "duckdb"
default_target = "dev"

[connections.developer]
database = "my_project.duckdb"

[settings]
default_audit_severity = "warn"

[defaults]
materialized = "table"

[targets.prod]
connection = "developer"
schema = "prod"

[targets.dev]
connection = "developer"
schema = "dev"

The project name is derived from the current directory name, with hyphens converted to underscores.