The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze rendered UI and screenshots provided as input. These inputs represent an untrusted data source that could contain embedded instructions (e.g., in text elements within the UI) designed to override the agent's behavior.
Ingestion points: User-provided screenshots and rendered UI components specified in the audit workflow (SKILL.md).
Capability inventory: The agent is tasked with generating design critiques, usability scores, and implementation-ready improvement proposals based on the analyzed content.
Boundary markers: The skill does not define specific delimiters or instructions to ignore potential commands within the visual UI content.
Sanitization: There is no mention of sanitizing or filtering the visual or text content extracted from the UI before processing.
[EXTERNAL_DOWNLOADS]: The skill documentation points to external resources, specifically layout-audit.js and sub-skills (design-spatial, design-ux) hosted on the author's GitHub repository. While the skill does not include commands to automatically download or execute this code, the intended functionality relies on these external, unbundled dependencies which must be manually vetted by the user.

deterministic-design