cron-manage

name	cron-manage
description	Create, inspect, enable, disable, and remove scheduled cron jobs on the hotel. Use when the operator or an autonomous workflow needs recurring work at a fixed schedule.
catalog	{"skill_name":"cron.manage","implied_tools":["cron.register","cron.list","cron.enable","cron.disable","cron.remove","session.status"],"validation_state":"validated","skill_markers":["governed","high_agency"],"field_sources":{"required_fields":["schedule","target_role","payload"],"optional_fields":["job_id","guaranteed","paracrine_heartbeat_template"],"repo_skill_path":"skills/cron-manage/SKILL.md","workflow":"cron.list → compose job → cron.register → verify"}}

Cron Management

Use this skill to set up recurring scheduled work on the hotel. Cron jobs fire on a 7-field cron schedule, deliver a JSON payload to a target role's inbox, and survive hotel restarts.

CronJob fields

Field	Required	Notes
schedule	Yes	7-field cron: <sec> <min> <hour> <dom> <month> <dow> <year>. Example: "0 0 8 * * * *" = 08:00 daily.
target_role	Yes	Role name whose inbox receives the trigger payload.
payload	Yes	JSON string delivered to the role. Supports {timestamp}, {iso_timestamp}, {job_id}, {node_id}, {target_role} interpolation.
guaranteed	No	Mesh-coordinated delivery (Slice 2+, ignored in current runtime). Default false.

Workflow

1. Call cron.list to see existing jobs and confirm no duplicate exists for this schedule + role pair.
2. Compose the job:
   • Choose a target_role that is configured and ready to receive the trigger
   • Write the payload JSON — the role's inbox will receive it as a task message
   • Choose the schedule using the 7-field cron format
3. Call cron.register with the full CronJob object (hotel assigns the id).
4. Verify: call cron.list again to confirm the job appears with enabled: true and a next_fire_at in the future.
5. To pause a job: call cron.disable with the job_id.
6. To resume: call cron.enable with the job_id.
7. To permanently remove: call cron.remove with the job_id.

Common schedule patterns

Goal	Schedule
Every 5 minutes	"0 */5 * * * * *"
Daily at 08:00	"0 0 8 * * * *"
Hourly	"0 0 * * * * *"
Every 30 seconds	"*/30 * * * * * *"
Weekdays at 09:00	"0 0 9 * * MON-FRI *"

Payload design

The payload is what the target role receives as its task. Write it as if you are handing work to that role directly:

{
  "action": "scheduled_check",
  "job_id": "{job_id}",
  "fired_at": "{iso_timestamp}",
  "intent": "Review new training samples and flag any requiring correction."
}

The role should be able to act on the payload without additional context injection.

Paracrine heartbeat payloads

For Life Graph heartbeat jobs, prefer the canonical template in ansible_mesh_core::cron::ParacrineHeartbeatTemplate instead of hand-writing the JSON. The template emits a top-level paracrine_signal object, so the hotel ticker dispatches action = "paracrine_signal" and the subscriber role observes the signal without entering the conversational model path.

Minimum shape after template interpolation:

{
  "paracrine_signal": {
    "signal_id": "cron:{job_id}:{timestamp}",
    "signal_type": "open_loop_staleness",
    "scope": "personal",
    "source_hotel": "{node_id}",
    "source_node": "{node_id}",
    "target_role_type": "attention-steward",
    "subject_refs": ["lifegraph:open_loop"],
    "cadence": "daily",
    "priority": "medium",
    "observed_at": "{iso_timestamp}",
    "expires_at": null,
    "payload_summary": "Scan stale open loops and defer unless policy says to record.",
    "policy_tags": ["adhd-support", "re-entry"]
  },
  "payload_summary": "Scan stale open loops and defer unless policy says to record."
}

Use target_role = "attention-steward" for the first observe-only Life Graph subscriber.

Guardrails

• Do not register jobs that fire more frequently than the task actually requires — prefer longer intervals
• Do not register jobs to orchestrator role unless the operator has explicitly authorized recurring orchestrator tasks
• Always call cron.list before cron.register to prevent accidental duplicates
• Jobs targeting roles that are not currently materialized will queue until the role materializes — this is expected behavior, not an error