Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically..
Terminal score: 0–1000 raw, weighted across 4 dimensions. Public score: 0–10 normalized (shown in the 30-day stars chart above).
