dlt
Create, inspect, and deploy dlt data-load pipelines
TLDR
SYNOPSIS
dlt [options] init|pipeline|schema|deploy|dashboard|telemetry [args]
DESCRIPTION
dlt (data load tool) is the Python CLI for dltHub pipelines. It scaffolds a project with init, then inspects pipeline state, traces, load packages, and destination schemas after you run pipeline scripts.Global flags belong on dlt, before the subcommand. `dlt --debug pipeline` enables debug traces; `dlt pipeline --debug` does not.pipeline abort-packages marks the oldest normalized package aborted, records failed jobs, deletes other pending packages, and restores local state from the destination snapshot. drop-pending-packages is deprecated in favor of that command. pipeline show launches a marimo dashboard and needs marimo installed.Pipeline scripts stay ordinary Python. The CLI does not run extracts by itself; it manages the working directory those scripts create.
PARAMETERS
-v, --verbose
Increase verbosity. Repeat for more (-v, -vv, -vvv).--non-interactive
Use prompt defaults; fail if a prompt has none. Implied when stdin is not a tty.-y, --yes
Auto-accept confirmations.--debug
Show full stack traces on exceptions.--version
Print the dlt version and exit.--disable-telemetry / --enable-telemetry
Turn telemetry off or on before the command runs.init source destination
Create a pipeline in the current folder from a verified source or a template. --list-sources and --list-destinations list the known names.pipeline [--list-pipelines] [pipelinename] subcommand_
Inspect a pipeline working directory. Subcommands include info, show, failed-jobs, abort-packages, sync, trace, schema, drop, load-package, and mcp.schema [--format json|yaml|dbml|dot|mermaid] file
Load, validate, and print a schema from a yaml or json file.deploy
Create a deployment package for a selected pipeline script.dashboard
Show the dltHub workspace dashboard.telemetry
Show whether telemetry is on.
CAVEATS
Needs Python 3 and a dlt install in the environment (`pip install dlt`). Run pipeline inspection from the same folder as the pipeline script so destination credentials in `.dlt/secrets.toml` resolve. ai moved to `dlthub ai` (install `dlt[hub]`). Telemetry is on unless disabled.
