The Agent Skills Directory

[COMMAND_EXECUTION]:
The SKILL.md file contains instructions prompting the agent to perform system-level installations using sudo apt install python3, brew install python3, and winget install. The use of elevated privileges (sudo) poses a risk if executed without user oversight.
The persist_design_system function in design_system.py constructs file paths by concatenating a base directory with user-supplied project and page names. Because the code only replaces spaces with dashes and fails to sanitize path traversal characters (e.g., ../), a malicious input could allow writing generated markdown files to arbitrary directories on the system.
[PROMPT_INJECTION]:
Ingestion points: User-provided search queries, project names, and page identifiers processed by the search.py CLI utility.
Boundary markers: Absent; the generated design system files (MASTER.md and page overrides) do not use delimiters or explicit instructions to prevent the agent from interpreting embedded data as new instructions.
Capability inventory: The skill possesses file system write capabilities (mkdir and open) and the ability to execute its own Python scripts via the agent's CLI.
Sanitization: Insufficient; the script's sanitization is limited to simple character replacement, leaving the system vulnerable to indirect prompt injection and path manipulation payloads.

ui-ux-pro-max