Manage trigger subscriptions

After you set up a trigger, check its subscriptions in the dashboard to see whether events are flowing, which accounts need attention and how often polling runs. This page covers subscription statuses, intervals, removing a subscription and changing delivery endpoints.

Open AgentKit > Connections, select the connection, then open the Triggers tab. A summary line above the list shows the number of triggers and subscriptions, and how many subscriptions need attention, or all active.

Each trigger row shows its number of subscriptions. Select the count to open the trigger’s subscriptions. Filter them by All, Needs attention or Active, or search by the account’s identifier. Subscriptions that need attention are listed first.

Status	Means	What to do
Active	The subscription receives events.	Nothing.
Throttled	A polling subscription receives events, but Scalekit checks less often than you asked, because the environment has many polling subscriptions.	Nothing, if the slower interval is acceptable. Otherwise, remove subscriptions you don’t use, and subscribe with longer intervals where speed doesn’t matter.
Credential issue — paused	Polling stopped because the account’s access no longer works, for example because the user revoked it.	Ask the user to reconnect the account. Polling resumes on its own, and the first check after it resumes covers the time it was paused.
Paused	The subscription is stopped and receives no events.	Remove it and subscribe again to receive events.
Needs attention	A status this page doesn’t list.	Contact support with the subscription’s trigger and account.

Active and Throttled subscriptions receive events; the others don’t. The summary and the Needs attention filter include every subscription that isn’t Active, so a Throttled subscription shows there too, even though its events still arrive.

Each polling subscription shows Checking every and the interval Scalekit actually uses. That’s the interval you set when you subscribed, or the trigger’s default, unless the subscription is Throttled: then it’s longer.

The interval sets how long a change can take to reach you: an event arrives up to one interval after the change, plus a few seconds. Each trigger has a minimum interval, shown under Cadence in the trigger’s details.

A subscription can be Active while its connected account can’t receive the trigger. The subscription then shows a warning tag, and it won’t fire until the user reconnects the account through your app’s connect flow:

Tag	Means
Needs re-authentication	The account’s credential no longer works. Send the user a new authorization link.
Disconnected	The account was disconnected. Reconnect it with a new authorization link.
Authorized under a different app	The user authorized with different app credentials than the connection uses now, for example before you switched to your own app. Ask the user to reconnect.
Awaiting confirmation	The account works, but Scalekit hasn’t confirmed which workspace or user at the app it belongs to, so it can’t route this trigger’s events to it. Ask the user to reconnect.

Authorize a user shows how to send a new authorization link. To get told when an account’s status changes, listen for account events.

Removing a subscription stops its events right away. It can’t be undone: to receive the events again, subscribe again.

1. On the connection’s Triggers tab, select the trigger’s subscription count.

2. Select Remove next to the subscription.

3. Check the confirmation, which names the account or the whole connection and the subscription’s filters, and confirm.

Open Developers > Trigger Delivery to see the environment’s delivery endpoints, which connections each one receives events from and whether it’s enabled.

• Edit an endpoint to change its URL, description or the connections it receives events from.
• Delete an endpoint to stop sending events to it. It stops right away, deliveries still in progress to it are lost, and it can’t be undone.
• Reveal and copy an endpoint’s signing secret when you set up a new server.

Under Recent delivery attempts, each attempt shows when it ran, the endpoint and your server’s response code, or No response. A failed attempt is retried, so a newer attempt for the same event can still succeed.