golem-quota-scala

Define a Rate limit under resourceDefaults in golem.yaml, then acquire a QuotaToken for that resource and wrap calls in withReservation. Golem enforces the limit across all agents in the deployment.

Reserve the maximum expected cost with withReservation, then commit the actual tokens used from the response. Unused reserved capacity returns to the pool automatically.

Reject returns Left(FailedReservation) that the agent must handle, optionally with an estimated wait time. Throttle suspends the agent until capacity is available, requiring no error-handling code.

Yes. Call token.split(amount) to create a child QuotaToken, pass it as a parameter to the child agent, and merge the returned token back with token.merge. Split traps if the amount exceeds the parent's expected use.

Yes. Use the golem resource update CLI command with a new limit definition and target environment. Changes take effect immediately for running agents without redeployment.