Skip to main content
Runs consume credits. Both credit commands are read-only.

​
Check the balance

Without --workspace, this reports your active workspace. Credits are held per workspace, so a team that runs a lot draws down its own balance rather than a shared pool.

​
See where they went

The ledger lists entries newest first, showing what consumed credits and when. Reach for it when a balance drops faster than expected. The usual causes are a schedule firing more often than intended, or an agent staged at live retrying a failing step.

​
Keeping consumption predictable

Check what’s scheduled.
The listing shows each installation’s schedule and next fire time. A */5 * * * * schedule left over from testing is the most common surprise. Prefer dry runs while iterating. Dry Run exercises the agent without production side effects, and it’s the right mode while you’re still shaping a spec. Look for retry loops.
A MUST_HAPPEN step that fails retries with backoff before giving up. A step failing on every run is doing that work repeatedly. Clear schedules you no longer need.

​
Scripting

​
See also