The Agent Skills Directory

[COMMAND_EXECUTION]: The script set_up_reminders_automation.py and its shell counterpart use subprocess.run to execute osascript with a hardcoded, read-only AppleScript string. This is a standard and benign practice in macOS automation to trigger the system's permission prompts for controlling applications.
[INDIRECT_PROMPT_INJECTION]: The skill contains logic to read and search through reminder titles and notes, creating an attack surface where malicious content stored in the Reminders app could influence the agent.
Ingestion points: list_reminders.py and complete_reminders.py fetch reminder titles and completion status from the local macOS Reminders database.
Boundary markers: None; the scripts retrieve and output raw text from the Reminders database.
Capability inventory: The skill can read, create, delete, and modify reminders using JXA and PyXA.
Sanitization: No sanitization is performed on the retrieved text, but the data is not executed as code nor used in shell command construction.
[DYNAMIC_EXECUTION]: The skill utilizes osascript to execute JXA and AppleScript at runtime. This is the core functionality of the skill and is used safely with static or template-based scripts to interact with the local Reminders application.

automating-reminders