The Agent Skills Directory

[SAFE]: The skill is designed for maintaining project documentation in the ai/diagrams/ directory. It uses standard tools like Bash and Read/Write/Edit to manage local files, ensuring the implementation matches the documentation.
[COMMAND_EXECUTION]: The skill provides various shell command snippets and an automated audit script (in references/AUDIT.md) for checking file integrity, naming conventions, and synchronization with git history. These commands are localized to the project directory and are standard for developer productivity tools.
[INDIRECT_PROMPT_INJECTION]: The skill involves reading project source code (src/) and existing diagram files to generate updates. While this represents a potential surface for indirect injection if project files contain adversarial instructions, the skill defines strict quality standards, templates, and human-in-the-loop review steps (e.g., 'Confirm actions and show diagram preview') that effectively mitigate this risk.
[METADATA_POISONING]: The metadata provided in the skill frontmatter is consistent with the functionality described in the documentation and instructions.

diagram