That makes sense. We’re less concerned with community size and more interested in architecture for a multi-model, receipt-backed coding agent. Claurst’s coordinator mode, worker agents, hooks, and verifier concepts seem closer to what we want: a local-first agent runtime that can later become a node in a broader agent network. The main thing we’re evaluating against OpenCode is whether Claurst can expose a strong headless server/SDK layer for orchestration, receipts, and remote node execution. Is that direction aligned with where you want Claurst to go?