Alerts¶
Skynet can create observations automatically from external astronomical alert streams — gamma-ray bursts, gravitational-wave events, and survey transients. Alerts are a first-class Skynet resource, not something you wire up per telescope: Skynet maintains the connections to the alert brokers (NASA GCN and Fink) centrally, so neither owners nor observers configure a feed or a credential.
This page is the owner-side view — how alert-driven observations reach your telescopes. For the full subscription workflow (sources, filters, response templates, lifecycle), see the observer-facing Alert subscriptions page.
How alert observations reach your queues¶
An alert subscription belongs to an observer or organization. When an alert matches, Skynet builds a normal observation — target filled in from the event — and submits it to the scheduler against the subscriber's observing grants, exactly like an observation created by hand. It competes for time, respects access, and runs through your queues like anything else.
So the levers you already have as an owner apply unchanged:
- Queues and access grants. Alert observations only reach your telescope through a queue access grant the subscriber holds. To carve out time for rapid follow-up without it competing with normal science, issue a dedicated target-of-opportunity (ToO) queue and grant access to the accounts that should be able to trigger on it.
- Prioritization. Give that queue the priority you want urgent events to have relative to your standing programs — see Scheduler prioritization.
- Instruments. A subscription can be restricted to specific instruments; combined with what your queue exposes, that bounds which hardware an alert can actually use.
What you don't manage¶
- Broker connections and credentials — handled centrally by Skynet; there is nothing to install or authorize on your side.
- Filters, response templates, and lifecycle policies — these live on the observer's subscription, not on your telescope.
If you want a telescope to participate in coordinated rapid follow-up, the setup is a queue-and-grant exercise, not an integration.