The Agent Skills Directory

[COMMAND_EXECUTION]: Uses system-level commands to detect environment details and manage software installation. This includes executing uname, pgrep, defaults read, and find to check OS versions and existing software installations.
[COMMAND_EXECUTION]: Executes administrative commands using sudo on Linux systems (dpkg -i, rpm -ivh, apt-get install -f) to install software packages.
[REMOTE_CODE_EXECUTION]: Generates and executes dynamic scripts at runtime. It creates a PowerShell script (install_opend.ps1) and runs it with -ExecutionPolicy Bypass to automate the download, extraction, and execution of Windows installers.
[REMOTE_CODE_EXECUTION]: Downloads and executes third-party installers (DMG, EXE, DEB, RPM) and scripts (fixrun.sh) from remote servers to set up the development environment.
[EXTERNAL_DOWNLOADS]: Fetches software components from vendor-owned domains (futunn.com, moomoo.com, softwaredownload.futunn.com) and utility sites (7-zip.org, github.com/ip7z/7zip).
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by ingesting untrusted data through the $ARGUMENTS parameter (specifically the -path value). This data is interpolated into shell and PowerShell script templates without instructions for sanitization or escaping, which could allow a malicious user to perform command injection by providing a crafted path string.
Ingestion points: $ARGUMENTS in SKILL.md.
Boundary markers: Absent.
Capability inventory: Bash, Write, and powershell execution.
Sanitization: Absent; the instructions direct the agent to replace placeholders with raw input values.

install-opend