The Agent Skills Directory

[SAFE]: The skill performs legitimate service management tasks using standard system utilities. All operations are documented and consistent with the stated goal of providing a watchdog for the OpenClaw Gateway.\n- [COMMAND_EXECUTION]: The script guardian.sh uses commands like pgrep, pkill, and git to manage the lifecycle of the gateway process. These actions are restricted to the application's own workspace and process name.\n- [DATA_EXFILTRATION]: Health alerts and recovery status messages are sent to a user-provided Discord webhook. This uses the well-known Discord API for legitimate notification purposes and does not involve sensitive user data.

openclaw-guardian