Prototypes, notes, and experiments.
Every agent-governance tool slaps a single 0-to-1 trust score on actions. It looks scientific and it is nearly useless: uninterpretable, un-actionable, and it collapses completely different risks into the same number. We deleted it, and replaced it with a signed, structured verdict.
Put one workflow behind Xybern and watch every agent action authorised, and sealed to the vault.
Xybern