CLI and daemon

The CLI manages the same schedule store the desktop and server read, and it can run the scheduler loop itself. You can run it on a laptop, a utility host, or inside a container.

Commands

flow schedule list                                  # show every schedule
flow schedule add my-flow --frequency daily         # daily at "now" (UTC)
flow schedule add my-flow --frequency cron --cron "0 9 * * 1-5"
flow schedule add my-flow --frequency every_n_minutes --every-minutes 15
flow schedule rm my-flow                            # remove a schedule
flow schedule daemon                                # run the scheduler loop here

flow schedule add flags

Flag	Values	Default
--frequency	hourly / daily / weekly / monthly / yearly / cron / every_n_minutes / once	required
--start	RFC3339 anchor instant	now
--timezone	IANA timezone name	UTC
--cron	5/6-field cron expression	with --frequency cron
--every-minutes	interval in minutes	with --frequency every_n_minutes
--catchup	skip / run-once / run-all	run-once

Schedules attach to saved-template slugs, so save the flow as a template first (flow template save, or from the desktop).

The daemon

flow schedule daemon runs the same scheduler loop the desktop and server run in-process:

• Polls on the settings interval for due schedules.
• Fires each due flow with the destructive-confirm gate off (headless).
• Records every run to the shared history with the scheduled trigger.
• Respects catch-up policies for windows missed while the daemon was down.

Because the store is shared, a schedule created on the desktop can be served by a daemon on the same data directory. Runs fired by the daemon appear in the desktop’s History and Schedules views.

Deployment patterns

• Utility host. Install the CLI, point FLOW_STUDIO_DIR at the shared data directory, and run flow schedule daemon under your service manager. Schedules keep firing with no desktop session.
• Flow Server instance. You do not need a daemon here. The instance runs the scheduler loop itself, so browser-managed schedules fire server-side.
• Local model note. A scheduled flow with local ai nodes auto-starts the managed model server on the daemon’s host. You can disable this with --no-auto-llm semantics on runs. Size the host accordingly, or keep scheduled flows deterministic.

Related

• Scheduling overview
• CI integration covers exit codes and isolation for unattended runs.
• Tutorial: scheduled fleet scan