Spend governance

Spend is measured in dollars, not tokens — the number you set is the number on your invoice, not a conversion you have to do yourself.

Two ceilings

Per-team daily ceiling — enforced by the runtime itself, inside the worker. This is the same mechanism a scheduled process’s own maxUsdPerDay uses (see Processes & scheduling) — it pauses that team’s automatic runs, without failing them, once the rolling 24-hour spend hits the cap.
Org-wide rollup ceiling — enforced by the platform across every team in the org, checked continuously.

What happens on breach

Breaching the org-wide ceiling stops every worker in that org — visibly. The dashboard shows a “paused: daily budget reached” banner with the actual figures, not a silent failure someone has to notice on their own. Clearing the breach is a deliberate action an operator or owner takes, not something that resolves itself when spend drops.

There’s also a manual kill switch, independent of any budget — for when you need to stop a team right now, for a reason that isn’t spend.

What this isn’t, yet

This is ceiling-and-breach governance, not a spend analytics dashboard. There’s no per-agent cost breakdown or spend-over-time view today — if you need to understand which agent or process is driving spend, that’s a gap, not a solved problem yet.

Next