The Agent Skills Directory

[SAFE]: The skill provides a structured procedure for executing tasks via the 'codex' CLI tool. It includes explicit instructions to utilize sandboxing (e.g., '-s read-only') and confirms tool status before execution. No evidence of data exfiltration, credential harvesting, or unauthorized privilege escalation was found.
[COMMAND_EXECUTION]: The skill is designed to facilitate command execution through the 'codex exec' utility. This is the primary stated purpose of the skill and is documented with security controls such as explicit sandbox modes and non-TTY execution guidelines to ensure auditability and stability.

codex-exec