The Agent Skills Directory

[SAFE]: The skill is composed exclusively of markdown files that serve as documentation and procedural guidelines. There are no executable scripts, network requests, or sensitive data access patterns present in the provided files.
[NO_CODE]: The entire skill is implemented as a set of prompt-based instructions. It does not contain or depend on any programmatic code, external libraries, or automated scripts.
[PROMPT_INJECTION]: The skill uses authoritative language, such as "MANDATORY" and "NEVER set any range limits," to ensure the AI agent reads the full extent of the evaluation criteria. While these phrases use imperative syntax common in prompt injection, they are contextually appropriate for the auditing workflow and do not attempt to override the underlying model's safety protocols.

skill-judge