Notifications

Notifications — the conductor.* lifecycle source

Conductor emits its own operational events as a built-in source, so alerting is an ordinary trigger — the retired notify: block auto-migrates to these (see Migration). Every attention event is written to the audit log regardless of configuration (status / report see escalations with no trigger at all); a trigger selects which events also deliver externally. Nothing here is ever posted as a PR/issue comment — it is private, addressed to you alone.

event	fires when
conductor.dispatch	work started on a target
conductor.escalate	a target gave up after retries
conductor.needs_input	a workflow handed a PR to a live agent and is waiting on you
conductor.complete	a run finished
conductor.failed	a run errored
conductor.updated	conductor self-updated (fires on the first boot of the new release)
conductor.update_available	a newer release was detected under update: { apply: workflow }

Context: {{.message}} (the composed line), {{.event}}, {{.ref}} (repo#number), {{.repo}}, {{.number}}, {{.origin_kind}} (the originating work's kind), {{.title}}; the update events add {{.version}}.

connectors:
  alerts: { use: ntfy, topic: your-topic }
  pager:  { use: pushover, token: ${PUSHOVER_TOKEN}, user: ${PUSHOVER_USER} }

triggers:
  - name: act-now
    on: [ conductor.escalate, conductor.failed, conductor.needs_input ]
    steps:
      - { uses: alerts.publish, options: { title: conductor, message: "{{.message}}" } }
  - name: page-on-giveup
    on: conductor.escalate
    steps:
      - { uses: pager.notify, options: { message: "conductor {{.message}}" } }
  - name: daily-digest
    on: conductor.complete
    group: { key: '"digest"', window: 24h }
    steps:
      - { uses: alerts.publish, options: { title: conductor, message: "[digest] {{ len .group.events }} completed run(s) today" } }

Routing, filtering, per-event fan-out (on: lists), quiet hours (policy:), and digests (group: { window: … }) all come from the normal trigger grammar — there is no separate notification subsystem.

Loop guard. Events emitted by a conductor-lifecycle trigger's own run are never re-fed into conductor.* — a notification workflow's own dispatch/complete/failed cannot storm. Its escalations still land in the audit log.

Acting on conductor. The same connector also exposes verbs (conductor.update / pause / resume / restart / reload / run {name, inputs}, plus gh.sweep for the github catch-up) — see Configuration for the gated-update workflow (update: { apply: workflow } → a trigger on conductor.update_available drains, updates, and smoke-tests with hooks around each step).

Related: Configuration · Workflows · Migration