The Agent Skills Directory

[COMMAND_EXECUTION]: The persistence mechanism in scripts/design_system.py uses the project_name argument to construct file system paths. Due to the behavior of the pathlib library, if an absolute path is provided as the project name, the script will write files to that absolute location rather than staying within the intended design-system/ subdirectory. This allows for potential path traversal and unauthorized directory creation on the host system depending on the permissions of the execution environment.
[PROMPT_INJECTION]: The skill processes external input that could be used to exploit the path traversal vulnerability through indirect manipulation of command parameters.
Ingestion points: User-provided query and project name parameters in scripts/search.py (query, --project-name, --page) which are often derived from untrusted external content or session context.
Boundary markers: Absent. The script does not use delimiters or instructions to ignore potential commands embedded in the input data.
Capability inventory: File system write access, directory creation, and the ability to persist structured markdown content across sessions.
Sanitization: Input is normalized for search indexing but not sanitized for file path safety; characters like .. or leading slashes in names are not filtered.
[EXTERNAL_DOWNLOADS]: The skill references resources and documentation from established and trusted platforms.
Documentation links in the stack guidelines (data/stacks/*.csv) point to official repositories and domains for community-standard frameworks like Astro, Flutter, Tailwind CSS, and Next.js.
The typography database (data/typography.csv) contains links to Google Fonts for CSS imports and font family selection, which is a well-known and standard service.

ui-ux-pro-max