Not for the local agent. Agents have no knowledge of the backing infrastructure's resource usage, relying instead on the backing platform to have some job queuing mechanism. For example, the k8s agent creates kubernetes jobs - creating 1000 kubernetes jobs won't immediately start 1000 kubernetess jobs, kubernetes will queue the jobs running them within the cluster resource limits. Local processes have no such queuing, users are currently responsible for scheduling jobs in such a way as to not overcommit their resources.