CLI Subcommands & Flags

Reference manual for compare, schema, and fingerprint operations.

drift compare <left> <right>

Compares two snapshots row by row using one or more key columns.

Flag Argument Behavior
-k, --keycol[,col...]Primary identifier fields (repeatable for composite keys)
--ignorecol[,col...]Skip volatile columns such as ingestion timestamps
--epsilonfloatAbsolute numeric tolerance before flagging float divergence
--formattext | json | ndjsonSelect human-readable report or machine-readable diff events
--strict-schemanoneFail with exit code 2 if column order or presence differs

drift schema <file>

Infers column types (integer, decimal, boolean, iso-date, text), nullability rates, and cardinality hints from a sample window.

$ drift schema monthly_ledger.jsonl --sample 50000
COLUMNS 6
  entry_id    text     non-null  unique=100.0%
  amount_eur  decimal  non-null  min=-420.00 max=19850.50
  booked_on   iso-date non-null  span=2026-01-01..2026-10-31

drift fingerprint <file>

Computes a canonical order-independent BLAKE3 digest across key-sorted records so identical exports in different physical row orders yield identical checksums.