The Agent Skills Directory

[COMMAND_EXECUTION]: The script wshot.py executes system commands such as gdbus, convert, which, and xdg-open using subprocess.run. These executions are performed using list-based arguments rather than shell strings, which correctly prevents command injection vulnerabilities.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted window metadata.
Ingestion points: Window titles and application names are retrieved from the GNOME desktop environment via D-Bus calls in the list_windows function in wshot.py and returned to the agent.
Boundary markers: The skill does not employ delimiters or warnings to instruct the agent to ignore potentially malicious instructions embedded in window titles.
Capability inventory: The skill allows for listing windows, capturing screenshots, and basic filesystem operations (moving and deleting temporary screenshot files).
Sanitization: There is no sanitization or filtering of the window titles before they are included in the agent's context.

wshot