The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses a bash process substitution pattern to execute a remote script: bash <(curl -Ls https://raw.githubusercontent.com/MHSanaei/3x-ui/v3.4.1/install.sh). This allows the execution of arbitrary code from a third-party source without allowing the AI or the user to verify the script's integrity or contents before execution.
[COMMAND_EXECUTION]: User-provided inputs, such as the host address (<host>), usernames (<u>), and passwords (<p>), are directly interpolated into shell command strings throughout the script. Without explicit sanitization or validation logic, this poses a risk of command injection if a malicious string is provided as an argument.
[EXTERNAL_DOWNLOADS]: The skill downloads and installs the 3x-ui panel and xray binaries from the MHSanaei GitHub repository. Since this is an external third-party source not listed as a trusted vendor, it introduces a supply chain risk where the downloaded content could be modified at the source.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via the processing of untrusted data in shell commands.
Ingestion points: User-defined variables for host, credentials, and configuration targets in SKILL.md.
Boundary markers: Absent; there are no delimiters or instructions provided to the agent to treat these variables as literal data rather than executable instructions.
Capability inventory: The skill has extensive permissions to execute administrative shell commands (apt-get, sysctl, systemctl, ufw, iptables) and modify system configuration files (/etc/ssh/sshd_config, /etc/fstab).
Sanitization: Absent; the instructions do not contain logic to escape or sanitize external content before it is used in subprocess calls.

ray-vpsinit