The Agent Skills Directory

[SAFE]: The skill functions as a developer guide for creating skills. It does not contain executable scripts or perform any automated operations outside of providing instructional context to the agent.
[DYNAMIC_CONTEXT_INJECTION]: The skill documents the use of dynamic context injection syntax (!command). The examples provided (e.g., git log, gh issue view) are standard development tasks and are correctly formatted within markdown code blocks to prevent execution during the skill's own lifecycle.
[COMMAND_EXECUTION]: The skill provides documentation on how a user can manually set up a skill directory and files using standard shell commands like mkdir and cat. These are well-documented, transparent operations intended for the developer.
[METADATA_POISONING]: The skill's metadata (name, description) accurately reflects its content and purpose as a creation guide.

creating-skills