The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on local Python scripts (scripts/search.py, scripts/core.py, and scripts/design_system.py) to perform its core functions. These scripts only use Python standard libraries and operate on local files.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured data from 23 CSV files. These files contain design rules and AI prompt templates (e.g., in styles.csv). While this data enters the agent context, it is static content bundled with the skill and poses no risk of secondary instruction injection from untrusted sources.
[SAFE]:
File Access: The skill reads from its own data/ directory and writes generated documentation to a design-system/ folder in the current working directory. It does not access sensitive system files or credentials.
Network: No network operations were found in the source code. External URLs present in the data files are limited to documentation (e.g., nextjs.org, tailwindcss.com) and font assets (e.g., fonts.googleapis.com), which are all well-known and legitimate services.
Obfuscation: No encoded strings, homoglyphs, or hidden character techniques were identified.
Dependencies: The Python scripts have no external package dependencies, using only built-in modules like csv, json, and pathlib.

ui-ux-pro-max