CLI Reference

Synopsis

dfmt [options]                                format all included files
dfmt <file-or-dir> [options]                  format subset of included files
dfmt --dialect <dialect> <file> [options]     yolo mode (no config file)
dfmt --changed-since <ref> [options]          format files changed since git ref
dfmt --model <name> [options]                 format specific dbt model
dfmt --init [dialect]                         create .dfmt.json

Modes

Config mode (default)

Requires a .dfmt.json file, found by walking up from the current directory (or target path). The config provides the dialect, format options, and include/exclude patterns that determine which files to format.

Running dfmt with no arguments formats all files matching the include/exclude patterns.

Running dfmt <path> formats files under <path> that match the patterns.

Yolo mode

When no .dfmt.json exists, --dialect is required:

dfmt --dialect snowflake query.sql
dfmt --dialect bigquery queries/

In yolo mode, all .sql files under the target are formatted.

Options

Core

Flag	Description
--write	Write formatted output back to files (only if changed)
--check	Exit 1 if any file needs formatting. For CI pipelines.
--init [dialect]	Create .dfmt.json in the current directory

File selection

Flag	Description
--dialect <dialect>	Set dialect explicitly. Required in yolo mode.
--model <name>	Format a specific dbt model by name
--changed-since <ref>	Format only .sql files changed since a git ref (branch, tag, or SHA)
--no-jinja	Format as plain SQL even if a dbt project is detected

Shorthand

Flag	Description
--uppercase	Uppercase keywords (equivalent to --keyword-case upper)
--lowercase	Lowercase keywords
--leading-commas	Leading comma style
--indent N	Indent width (spaces)
--width N	Max line width

Format options

All format options from .dfmt.json can also be set via CLI flags. CLI flags override config file values.

Flag	Values	Default
--keyword-case	upper, lower, preserve	lower
--identifier-case	upper, lower, preserve	preserve
--function-case	upper, lower, preserve	lower
--data-type-case	upper, lower, preserve	lower
--max-line-width	integer	80
--indent-width	integer	4
--comma-style	trailing, leading	trailing
--clause-layout	indented, adaptive	indented
--sibling-layout	stacked, consistent, pack	stacked
--short-statements-one-line	boolean	true
--clarify-structure	boolean	false
--blank-line-between-ctes	boolean	false
--align-token	token to align (e.g., AS)	none

Dialects

ansi, bigquery, postgres, duckdb, spark, databricks, snowflake, oracle, db2, tsql, mariadb, presto, trino, redshift, sqlite

Resolution Order

1. Dialect: --dialect flag > .dfmt.json dialect > dbt adapter type > ansi
2. Format options: dialect defaults > .dfmt.json overrides > CLI flag overrides
3. Files: .dfmt.json include/exclude patterns > target path filter > --changed-since filter

Exit Codes

Code	Meaning
0	Success (or --check found no changes needed)
1	Error, or --check found files that need formatting