Skip to main content
Version: 2026.8.0

Metrics and health

The console's read-only dashboards, from the at-a-glance Overview down to per-workspace product metrics. None of these change anything. They are how you tell whether the instance is healthy and being used.

Overview​

The landing page of /admin is a set of counts, cheap enough to leave open and refresh:

  • Workspaces and users: how many exist, and how many users have signed in over the last seven days.
  • Activity: how much there was in the last day.
  • Capability grants and installed bundles in play.
  • The running build's commit.
  • "Needs attention" numbers sit alongside, like open feedback and pending waitlist entries, which stay at zero on an instance with no public form.

Product metrics​

The Product Metrics tab reports two things per workspace, both windowed over the last 7 and 30 days:

  • Time to a working app: the gap between when the workspace was created and when its first real item was made.
  • Walls hit: a count of the moments a member ran into a hard edge, such as a permission denial, a rejected value, or a failed automation.

Read together, they answer whether people get to something useful quickly and how often the product gets in their way afterward.

AI usage​

If the built-in AI features are on, two views cover what they cost:

  • The Overview carries a 24-hour roll-up of AI calls and their cost.
  • The AI tab is the full cross-workspace call log: each call's workspace, the member who triggered it, the capability, the model, token counts, cost, and whether it was served from cache. Open a single call to see its full prompt and response.
  • A companion Barcodes tab shows the product-lookup cache built up as people scan.

Health​

The Health tab is a live snapshot:

  • A database ping with its latency.
  • A count of activity in the last hour.
  • A placeholder for backup status that reminds you to confirm backups on the host itself (the console cannot see your restic state).
  • Two further tabs report on sandboxed module code: per-workspace CPU use against the quota, and per-module invocation counts, error rates, and latency percentiles.

For the raw event stream behind these numbers, the Activity tab is a cross-workspace feed of every logged action, filterable by workspace, user, or action type. Backups themselves are covered in Operating and Backup and export.