Last released Jun 25, 2026
Runtime security for AI agents. Inspect every action, learn normal behaviour, flag what falls outside it, and block known-bad.
Supported by