The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/deploy-line-server.sh downloads and executes code from remote sources using patterns that bypass standard security reviews. Specifically, it pipes content from https://get.acme.sh directly to sh and uses bash -c with command substitution to execute an installer from https://github.com/XTLS/Xray-install/raw/main/install-release.sh.
[COMMAND_EXECUTION]: The script scripts/deploy-line-server.sh performs destructive operations by running rm -rf on several system configuration directories, including /etc/sing-box, /etc/v2ray, /etc/hysteria, and /etc/x-ui, to purge existing proxy components during the installation process.
[PRIVILEGE_ESCALATION]: The deployment workflow requires the scripts to be run with root privileges and performs significant system-level modifications: it modifies the system firewall using ufw, adjusts kernel networking parameters in /etc/sysctl.conf to enable IP forwarding, and manages system packages using apt-get purge.
[DYNAMIC_EXECUTION]: The script scripts/deploy-line-server.sh uses a Python heredoc to dynamically construct and execute a Python script that generates the final Xray configuration JSON at runtime based on environment variables.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits vulnerability to indirect prompt injection due to its processing of untrusted data: 1. Ingestion points: User-provided environment variables such as DOMAIN, CF_TOKEN, and REALITY_TARGET. 2. Boundary markers: Absent. 3. Capability inventory: Root command execution, system service management, and network operations. 4. Sanitization: Absent; environment variables are interpolated directly into shell commands and scripts without validation.
[EXTERNAL_DOWNLOADS]: The skill fetches content from external sources, including downloading configuration data from Cloudflare's public IP list.

xray-line-server