The Agent Skills Directory

[Privilege Escalation] (HIGH): The skill explicitly mentions features that automate or facilitate root access, such as x smart (root privilege handling automation) and x mac tidsudo (TouchID authentication for sudo). This allows an agent to bypass interactive security prompts.
[Command Execution] (HIGH): The skill enables the agent to modify critical security settings, including firewall configuration (x mac fw) and SSH server settings (x mac sshd), which could be used to weaken system security or create backdoors.
[Indirect Prompt Injection] (HIGH): The skill creates a dangerous vulnerability surface where the agent processes untrusted external data.
Ingestion points: The agent reads dynamic data from the environment, such as process lists (x ps) and network discovery results (x ip map).
Boundary markers: None. There are no delimiters or instructions to treat the output of these tools as untrusted data.
Capability inventory: The skill provides full 'write' and 'execute' capabilities (sudo, file modification, network configuration).
Sanitization: None. The agent directly interprets output from these tools.
[External Downloads] (MEDIUM): The skill requires the x-cmd CLI, a third-party dependency from an untrusted source (x-cmd.com), to be installed and present on the system.
[Data Exposure] (MEDIUM): Tools like x ip geolite and x ip map facilitate the discovery and external transmission of network topology and geolocation data.

x-cmd-system