The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the native macOS shortcuts command-line tool to list, view, and execute system automations. This is a legitimate use of system utilities for their intended purpose.
[INDIRECT_PROMPT_INJECTION]: The skill processes output from external sources that could contain untrusted instructions, such as shortcut names or the text output generated by a shortcut.
Ingestion points: Output from the shortcuts list command (shortcut names) and the shortcuts run command (shortcut execution results) in SKILL.md.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content within shortcut names or output.
Capability inventory: The skill can execute macOS Shortcuts, read files via the --input-path argument, and perform shell operations as shown in examples.md.
Sanitization: There is no explicit sanitization or filtering of the output received from the shortcuts binary before it is passed to the agent's context.

shortcuts