The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's test suite (scripts/tests/) uses the subprocess module to execute internal Python scripts for verification purposes.\n
The calls in test_core.py, test_design_system_stack.py, and test_catalog_refresh.py pass arguments as lists directly to the executable, avoiding shell invocation.\n
These processes are limited to running the skill's own maintainer scripts (search.py, refresh-google-fonts.py, etc.) using the verified Python interpreter.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied query strings and project names, which are interpolated into search results and generated markdown files.\n
Ingestion points: query and project_name arguments in search.py and design_system.py.\n
Boundary markers: SKILL.md contains instructions for the agent to treat results as advisory recommendations.\n
Capability inventory: The skill can write files to the project directory via the persist flag.\n
Sanitization: A safe_slug function sanitizes project and page names (allowing only [a-z0-9_-]), which prevents path traversal or unauthorized file overwrites outside the design-system/ folder.\n- [EXTERNAL_DOWNLOADS]: The catalog maintenance scripts (referenced in tests) include functionality to fetch data from well-known official sources.\n
Scripts fetch font and icon metadata from Google's and Phosphor's official domains, which are recognized as trusted services.

ui-ux-pro-max