The Agent Skills Directory

[SAFE]: The skill is a standard technical documentation file that guides the AI in generating boilerplate code for a specific repository structure.
[COMMAND_EXECUTION]: The verification step includes running pnpm test on the newly created hook files. This is a standard development practice for ensuring code quality and does not involve arbitrary or unsafe command execution.
[PROMPT_INJECTION]: While the skill ingests user input for hook names and logic, it utilizes clear markdown templates and directory boundaries which mitigate the risk of indirect instruction obedience. Findings for this category are assessed as safe surface area for a code-generation tool.

add-hook