Agent Security
Destructive Command
Also: dangerous command, risky shell command
A shell or database command that deletes, overwrites or exposes data in a way that is hard to undo, such as rm -rf, a forced git push or DROP TABLE.
In more detail
Coding agents run commands for you, and a misread task or injected instruction can turn a cleanup into a wipe. Guardrails check the command before it runs, inspect its target (is this the project directory or the home folder?) and either block it or ask a human. WipeBench measures how often agents attempt this class of action across 112 scenarios.