The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to incorporate external resources into the generated prototypes, specifically referencing the Tailwind CSS CDN (cdn.tailwindcss.com), FontAwesome for icons, and image services like Unsplash and Pexels. These are well-known technology services and their inclusion aligns with the skill's primary purpose of generating frontend code.
[INDIRECT_PROMPT_INJECTION]: The skill functions as a component in a multi-step chain where user-provided design requirements are processed to create a final prompt intended for another AI agent. This pattern introduces a vulnerability surface where a user could potentially supply requirements containing instructions that override the behavior of the downstream agent.
Ingestion points: User-provided requirements in SKILL.md Step 1, including application purpose, platform context, design preferences, and technical constraints.
Boundary markers: The references/prompt-structure.md file includes a template that provides a basic structure to delimit sections, which can help mitigate accidental instruction following but may not stop sophisticated adversarial input.
Capability inventory: The skill itself is restricted to generating text output (markdown and code snippets) and does not have the capability to execute commands, write files, or perform network operations directly.
Sanitization: The skill lacks explicit validation or sanitization mechanisms to filter or escape instructions embedded within the user-provided requirement strings before they are interpolated into the generated output prompt.

prototype-prompt-generator