The Agent Skills Directory

[SAFE]: The skill instructions and reference materials do not contain any malicious patterns such as prompt injection, obfuscated code, or persistence mechanisms.
[DATA_EXPOSURE]: The skill's operations are confined to reading its own reference files and writing documentation to the project's local directory (.claude/docs/). No hardcoded credentials or access to sensitive system paths (e.g., SSH keys, environment variables) were found.
[INDIRECT_PROMPT_INJECTION]: The skill processes user input and web research results. Although these are external data sources, the skill's specific objective of extracting technical design properties (hex codes, CSS values) according to a strict output schema provides a structured environment that reduces the risk and impact of indirect prompt injection.

aesthetic-guide