The Agent Skills Directory

[SAFE]: The skill provides documentation and templates for creating specialized AI assistants. It follows standard practices for configuring subagents within the platform's supported directory structure.
[COMMAND_EXECUTION]: The skill includes instructional shell commands such as mkdir and touch. These are used appropriately to demonstrate how to create the necessary directories and files for subagent configuration in expected locations (.cursor/agents and ~/.cursor/agents).
[PROMPT_INJECTION]: Example subagent definitions contain instructional system prompts that focus on task-specific behaviors (e.g., code review, debugging). These instructions do not contain patterns for bypassing safety filters, overriding core agent behavior, or extracting system prompts.

create-subagent