CLI Reference

Durable execution for AI agents.

Usage

kitaru COMMAND

Flag	Description
`--help`, `-h`	Display help and exit
`--version`, `-V`	Display the installed version and exit

Most agent-facing commands support --output json (or -o json) in addition to the default text output.

Text output is designed for people reading the terminal directly.
JSON output is designed for agents and scripts that need a stable structure.
Single-item commands emit {command, item}.
List commands emit {command, items, count}.
kitaru executions logs --follow --output json is the special case: it emits one JSON event per line while following the stream.

Command	Description
`executions`	Inspect and manage flow executions.
`info`	Show detailed environment information for the current setup.
`init`	Initialize a Kitaru project in the current directory.
`log-store`	Manage global runtime log-store settings.
`login`	Connect to a remote server, or start and connect to a local server.
`logout`	Log out from the current Kitaru server and clear stored auth state.
`model`	Manage local model aliases for kitaru.llm().
`secrets`	Manage centralized runtime secrets.
`stack`	Inspect, create, delete, and switch stacks.
`status`	Show the current connection state and active stack context.