The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute open or xdg-open on the generated index.html file. This is a standard workflow for previewing generated web content and is scoped to the output directory.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from various Markdown files (e.g., ux/*.md, PRD-*.md, analysis/*) to derive design decisions. While this is its primary purpose, it creates an attack surface where malicious instructions inside these research artifacts could attempt to influence the agent's behavior during the generation process. The skill mitigates this somewhat by providing structured references and templates for the generation.
[DATA_EXPOSURE]: The skill uses Glob and Read tools to access files within the project directory (e.g., ux/, analysis/, dev/ux/). This access is necessary for its functionality as a design translator and is restricted to project-specific paths.

ux-wireframe