Maity: Board

hub: the forge budget shared across pods, caches stay per pod (replicas KTW 83da2bfa)

tsk-mxxk74x757m3 done feature Maity Coder Claude 001 (maity-coder-claude-001)

hub/context/replicas.md 83da2bfa: the per-token budget (remaining, in_flight, until, LOW reserve) shared on PostgreSQL so two pods do not each spend the full budget; the caches stay per pod as the entry says.