Reconnect after a credential alert

TL;DR

Follow the alert link to reconnect the named Credential in your Dashboard. Reuse the same name, Provider, and kind so you repair the existing Credential instead of creating another one.

Once a Provider Credential that used to work stops authenticating, your Stack's control plane emails your tenant admin a link to fix it — this page is what to do when that email arrives. See Auth & consent for what a Credential actually is and why re-affirming its health happens by ordinary use, not a separate check-in.

A vendor-policy pause also sends one email per affected Credential. That message explains why subscription credentials are paused and points you to the API-key path; re-enabling the vendor restores the existing Credential without new consent.

Steps

1. Open the link in the email. It's shaped like <control-plane-url>/v1/handoff/start?next=%23%2Fauth — if you're still signed in to the control plane, it lands you straight on your Dashboard's Credentials page (#/auth); if you're signed out, you'll hit the control-plane sign-in gate first, then land on the same place.
2. In the Credentials list, find the row whose health badge reads invalid (not merely "not authenticated" — that's a Credential your Stack has on file but can no longer use, distinct from one never provisioned at all). Its own row also names which Provider and kind it is. Health is tracked per named Credential, not per Provider, so one dead Credential says nothing about another on the same Provider.
3. Reconnect it by re-entering it in the Add a Credential form with the same name, Provider and kind — this re-provisions the existing Credential in place rather than creating a second one:
   • An expired or revoked vendor API key — paste the fresh key.
   • An expired personal-subscription token or auth file — see Connect a personal subscription credential for the consent-checkbox step that path requires.

Why this only ever emails once per problem

You'll get exactly one email per Credential per transition into "invalid" — not one every time the control plane happens to notice it's still broken. Fixing it re-arms the alert, so a later relapse emails you again; nothing about this depends on you dismissing or acknowledging the email itself.

Related

• Auth & consent — what a Credential is and how its health is established.
• Connect a personal subscription credential — the consent step for an OAuth token or an auth file.
• Quickstart — connecting a Provider for the first time with a vendor API key.