Last released Oct 22, 2026
The safety layer for the AGI era. Runtime guardrails for autonomous AI agents. Loop detection, cost limits, and telemetry.