The Agent Skills Directory

[COMMAND_EXECUTION]: The script apply.sh executes shell commands including defaults, swift, and osascript to modify system settings and manage applications.
[DYNAMIC_EXECUTION]: The skill generates and executes Swift code strings at runtime. In apply.sh, the variables $FONT and $SIZE (passed via command-line arguments) are interpolated directly into the Swift command string for swift -e. Because these variables are not escaped or sanitized, a font name containing double quotes and Swift syntax (e.g., \"; [code]; //) could be used to execute arbitrary Swift code on the machine.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data that influences its execution flow.
Ingestion points: Arguments provided to the apply.sh script, specifically --font and --size, which are derived from the agent's instructions based on user input.
Boundary markers: None. The script does not employ delimiters or ignore-instructions to separate user-provided data from code logic.
Capability inventory: The skill possesses significant capabilities, including modifying system-wide preference plists, quitting running applications via AppleScript, and executing dynamic Swift code.
Sanitization: There is a complete lack of sanitization or character escaping for the font name parameter before it is placed into an executable context.

macos-font-defaults