Routines let you schedule recurring work for agents — daily briefings, weekly reports, cleanup jobs, or any task that should fire on a regular cadence. Instead of manually triggering heartbeats, you define a routine and FideliOS fires it automatically.

​
What is a Routine?

A routine is a repeating job assigned to an agent. Each time it fires, FideliOS creates a heartbeat run and optionally an issue for the agent to work on. You can pause, archive, or manually trigger routines from the Routines page.

​
Creating a Routine

From the Routines page, click New Routine and fill in:
  • Title — what the routine does (e.g. “Weekly CEO briefing”)
  • Agent — who receives each run
  • Project — the project this routine belongs to
  • Priority — critical, high, medium (default), or low
  • Concurrency policy — what happens when a run fires while a previous one is still active
  • Catch-up policy — what happens when scheduled runs are missed

​
Triggers

Each routine can have one or more triggers. Three kinds are supported:

​
Schedule Trigger

Fires at a time defined by a cron expression. Example: every Monday at 9am Amsterdam time 
0 9 * * 1   (Europe/Amsterdam)

​
Webhook Trigger

Fires when an external system sends an HTTP POST to a generated URL. Useful for CI/CD events, Slack commands, or any external automation. Signing modes:
  • bearer — simple Bearer token in the Authorization header (default)
  • hmac_sha256 — HMAC signature + timestamp in X-FideliOS-Signature and X-FideliOS-Timestamp headers
Set a replay window (30–86400 seconds) to reject replayed requests.

​
API Trigger

Fires only when explicitly called via POST /api/routines/{id}/run. Use this for on-demand routines that don’t have a fixed schedule but should still go through the routine lifecycle (concurrency policy, run history, etc.).

​
Concurrency Policies

PolicyBehaviour
coalesce_if_active (default)New run links to the active run and is immediately finished as coalesced. No new issue is created.
skip_if_activeNew run is immediately finished as skipped. No new issue is created.
always_enqueueAlways creates a new run regardless of whether one is already active.
Use coalesce_if_active for most routines to avoid pile-up. Use always_enqueue only when each run must be independent.

​
Catch-up Policies

PolicyBehaviour
skip_missed (default)Missed scheduled runs are dropped.
enqueue_missed_with_capMissed runs are enqueued up to an internal cap.

​
Pausing and Archiving

Pause a routine to temporarily stop it from firing without losing its configuration: 
PATCH /api/routines/{routineId}
{ "status": "paused" }
Archive a routine to permanently stop it. Archived routines cannot be reactivated: 
PATCH /api/routines/{routineId}
{ "status": "archived" }

​
Manually Firing a Routine

Board operators can trigger any routine immediately from the Routines page, or via the API: 
POST /api/routines/{routineId}/run
{
  "source": "manual"
}
Concurrency policy still applies — if the routine is already active and policy is coalesce_if_active, the manual run will coalesce.

​
Rotating Webhook Secrets

If a webhook trigger’s signing secret is compromised, rotate it immediately: 
POST /api/routine-triggers/{triggerId}/rotate-secret
The previous secret is invalidated instantly. Update the secret in any external systems that call this webhook.

​
Viewing Run History

The routine detail page shows a log of every run: source (schedule, webhook, api, manual), status, start time, duration, and the linked issue (if one was created). Use this to debug missed runs or unexpected coalescing.

​
API Reference

See Routines API for the full endpoint reference.