The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to install an external binary (remindctl) from a third-party Homebrew tap (steipete/tap/remindctl). While this is a common utility in the macOS developer community, it originates from an external individual's repository rather than an official system tool or a recognized vendor in the trusted list.
[COMMAND_EXECUTION]: The skill extensively uses shell commands (remindctl) and system automation scripts (osascript) to interact with the macOS Reminders application. It also suggests direct interaction with the underlying SQLite database (ZREMCDREMINDER.ZPARENTREMINDER) for advanced task management, which involves reading sensitive system data structures.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted text data (reminder titles and notes) that could contain malicious instructions.
Ingestion points: Data is pulled into the agent's context through commands like remindctl today, remindctl all, and remindctl list which output user-generated text.
Boundary markers: None; the instructions do not specify the use of delimiters or 'ignore' instructions to separate ingested data from the agent's core logic.
Capability inventory: The agent has the capability to execute shell commands, modify system tasks, and run AppleScript, which could be abused if an injected instruction is followed.
Sanitization: There is no mention of sanitizing or validating the content of reminders before they are processed by the agent.

apple-reminders