screenshot
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on executing bundled local scripts (Python, Bash, PowerShell, and Swift) and system utilities to perform its core functions. It invokes native tools such as 'screencapture' and 'osascript' on macOS, and 'scrot', 'gnome-screenshot', or ImageMagick's 'import' on Linux. On Windows, it uses PowerShell to interface with Win32 APIs for window management and screen capture. These actions are appropriate for a tool designed for OS-level screenshot capabilities.
- [PROMPT_INJECTION]: A surface for indirect prompt injection exists because the skill processes application names and window titles retrieved from the operating system. These strings are used in command arguments and displayed back to the agent, potentially allowing an untrusted application to influence agent behavior through its window title. This is a common architectural risk for system-monitoring tools.
Audit Metadata