Type-safe tools, on-device inference, multi-agent workflows, memory, and guardrails, in one Swift-native runtime
Terminal score: 0–1000 raw, weighted across 4 dimensions. Public score: 0–10 normalized (shown in the 30-day stars chart above).
