Last released Jun 3, 2026
A policy & audit firewall for agent actions — authorise before a credential is used, require human approval for sensitive actions, and prove every action traces back to the instruction that authorised it.
Supported by
