The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill uses a 'Master + Overrides' pattern to persist design recommendations. User-supplied queries and project names are written into Markdown files (e.g., MASTER.md) that the agent is explicitly instructed to retrieve and obey as a source of truth in subsequent sessions. This creates a vector for instructions to be embedded in the design system and executed by the AI without its knowledge.
Ingestion points: User-provided query and project_name in scripts/search.py.
Boundary markers: Absent. The skill generates standard Markdown without delimiters or instructions to ignore potential commands embedded in the data.
Capability inventory: The skill has the ability to write to the local file system (scripts/design_system.py) and directs the agent to read these files to influence its behavior.
Sanitization: There is no validation or filtering of user input before it is written to the persisted Markdown files.
[COMMAND_EXECUTION]: The persistence logic in scripts/design_system.py uses a project name to create a directory structure without path normalization. By using directory traversal characters (e.g., ../), a user could influence the script to write generated files to unintended locations on the file system.

ui-ux-pro-max