The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes a crawler script, scripts/crawl_apple_hig.py, which retrieves Human Interface Guidelines documentation from developer.apple.com. As this operation targets a well-known, official technical service for the purpose of updating reference content, it is considered safe.
[COMMAND_EXECUTION]: The repository contains local Python scripts (generate_routing_index.py, package_runtime_zip.py) used by the developer to manage the skill's data. These scripts perform safe local operations, and the packaging script specifically implements security checks to ensure hidden files and sensitive metadata are excluded from the distribution.
[DATA_EXPOSURE]: The skill does not access sensitive system paths or credentials. Network activity is limited to documentation scraping from trusted sources with no patterns suggesting data exfiltration.
[PROMPT_INJECTION]: Instructions in SKILL.md are focused on accurate documentation retrieval and platform-specific design guidance. No patterns of behavioral override, safety filter bypass, or adversarial role-play were detected.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests a user request to trigger file loading, the files being loaded are local, static documentation provided with the skill package. This represents a standard reference surface without high-risk capability exposure.

apple-hig