Triggers

A trigger is a scheduled execution of an agent prompt, a shell command, or an installed skill. Triggers run on the daemon — the daemon’s trigger scheduler goroutine fires them on schedule. State lives in cmdop.db.

Triggers only fire when the daemon is running. If the host is offline, a missed trigger is not automatically replayed — it’s logged as missed and visible in cmdop trigger history.

List / add / remove


cmdop trigger list
cmdop trigger add --name nightly-backup \
  --schedule "0 3 * * *" \
  --machine vps-audi \
  --prompt "rotate logs and tar /var/log to /backup"
cmdop trigger remove nightly-backup

Schedule format: standard 5-field cron (minute hour day-of-month month day-of-week).

Trigger types

Prompt — runs an agent prompt via the local LLM. Useful for “investigate disk usage and post to Slack”.
Shell — runs a shell command. Useful for raw cron replacement.
Skill — runs an installed skill. Useful for parameterised templates.


# Prompt-based
cmdop trigger add --name disk-watch --schedule "*/30 * * * *" \
  --machine prod-server \
  --prompt "if disk usage > 80% on any partition, page on-call"
 
# Shell-based
cmdop trigger add --name log-rotate --schedule "0 4 * * *" \
  --machine vps-audi \
  --shell "logrotate /etc/logrotate.conf"
 
# Skill-based
cmdop trigger add --name health-check --schedule "*/15 * * * *" \
  --skill check-services --machine prod-server

Enable / disable


cmdop trigger disable nightly-backup
cmdop trigger enable  nightly-backup

Disabled triggers stay in the registry but don’t fire. Use this for maintenance windows without losing the configuration.

Health


cmdop trigger health                 # all triggers
cmdop trigger health nightly-backup  # one trigger

Reports last-run time, last-run outcome, next-fire ETA, and consecutive-failure count.


NAME            STATUS    LAST RUN    OUTCOME   NEXT FIRE    FAILS
nightly-backup  enabled   8h ago      ok        in 16h       0
disk-watch      enabled   12m ago     ok        in 18m       0
log-rotate      disabled  3d ago      ok        -            0
health-check    enabled   3m ago      fail      in 12m       3

History


cmdop trigger history nightly-backup
cmdop trigger history nightly-backup --limit 50 --json

Shows past runs with start/end timestamps, exit code, output snippet, and machine.

Failure handling

After 3 consecutive failures, the trigger:

Emits a desktop notification.
Pushes an issue to the Board labelled trigger-failed.
Stays enabled — but you’ll see the failure count climb in cmdop trigger health.

Re-enable the green path by fixing whatever’s broken, then either let it succeed naturally or:


cmdop trigger disable health-check
cmdop trigger enable health-check    # resets the failure counter

Run a trigger manually


cmdop trigger run nightly-backup

Runs the trigger immediately, regardless of schedule. Useful for testing without waiting for the next firing.

Where triggers live

State lives in cmdop.db (SQLite) under the trigger tables. Portable across machines via cmdop session save / restore-style snapshots is not supported for triggers — they’re host-bound (the daemon that owns the schedule is the daemon that fires them).