The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/native_shell_executor.gd utilizes OS.execute to run shell commands. The run_shell_command function allows an agent to pass arbitrary strings as commands and arguments to the host system, which could lead to unauthorized system access or modification.
[DYNAMIC_EXECUTION]: The skill implements OS.create_process in SKILL.md and scripts/platform_desktop_patterns.gd. This allows the application to launch external executables or spawn detached background processes, potentially bypassing user oversight.
[INDIRECT_PROMPT_INJECTION]: The skill contains functions that ingest external data, creating a surface for indirect prompt injection.
Ingestion points: NativeShellExecutor.run_shell_command accepts external strings as commands; platform_desktop_patterns.gd contains a drag-and-drop handler (_on_files_dropped) that processes external file paths.
Boundary markers: No delimiters or protective instructions are used when handling these external inputs.
Capability inventory: The skill possesses powerful system capabilities including OS.execute, OS.create_process, and OS.shell_open.
Sanitization: There is no validation or sanitization performed on the commands or file paths before they are used in system-level calls.

godot-platform-desktop