Vercel Connect adds observability support

Vercel Connect has been extended with an observability layer that surfaces token lifecycle events at the line level. Previously, token activity was only visible through indirect logs or external monitoring; now each connector includes a dedicated tab that lists creation, usage, refresh, revocation, and delivery actions. The UI also offers a quick-access filter to view configuration changes for a given connector. This change gives developers a single place to audit token behavior without building custom instrumentation.

The observability data is enriched with two persistent identifiers, one for the token itself and another for the authorization request. These identifiers remain constant across events, enabling engineers to match Connect activity with logs from their own services. By using these IDs, teams can construct end-to-end traces that include both Vercel-managed events and internal processing steps, reducing the time spent correlating disparate log sources.

Retention of the captured events varies by subscription tier: a short window on the free tier, a few days on the mid-level plan, and a month on the highest tier. For use cases that require longer history, the platform allows forwarding the event stream to a custom webhook, a capability that is only available on the paid tiers. Consequently, organizations needing audit-level retention must either upgrade or implement the webhook drain to satisfy compliance requirements.

The feature is rolled out in beta and is included on all current plans at no extra charge, but its behavior and supported connectors may evolve before full release. Enterprise customers also receive role-based access controls and audit logs, extending the security posture for larger teams. Engineers should treat the observability UI as a supplemental tool and plan for external archiving if their retention needs exceed the built-in limits.

Adopting the new observability does not require code changes; it is enabled through the connector’s UI and optional webhook configuration. However, integrating the stable IDs into existing monitoring pipelines will involve updating log parsers or correlation logic. The observability layer stops providing data after the plan-specific retention window unless events are exported, so any long-term analysis must rely on the webhook or external storage.