**The data behind agent permission rules.** Verified facts about what each CLI command does — destructive flags, network reach, BSD vs GNU differences — with a re-runnable check behind every row, so a policy is written from evidence instead of intuition.
-
Updated
Sep 5, 2026 - Python