The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides an interface to define and execute arbitrary system commands via the command and args fields in the script configuration. The add script and start <name> functionalities allow for the execution of any binary or script available on the host system.
[REMOTE_CODE_EXECUTION]: By allowing the registration of new scripts with custom commands, the skill can be used to execute payloads downloaded from external sources (e.g., via curl or wget commands provided in the args).
[DATA_EXFILTRATION]: The logs <name> command enables viewing the last 500 lines of stdout and stderr from any managed process. This capability could be used to harvest sensitive information, credentials, or environment data emitted by processes into their logs.

orchestrator-mgmt