The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is primarily designed to facilitate the execution of shell commands and automated workflows via the lev exec CLI tool. It implements significant security controls to mitigate risks, such as enforcing pre-execution validation (lev task validate), requiring machine-checkable 'receipt' evidence for task completion, and prohibiting the use of unverified command flags. The 'Adversarial Exec Contract' section specifically addresses the risk of untrusted task data by requiring explicit verification gates and preventing workers from accessing sensitive holdout data.

exec