design-system-reference

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The required workflow instructs the agent to read “the relevant section of [the cross-industry catalog]” (references/catalog.md), which is outsider-authored free text (public design-system index content) that will be ingested into the LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). This skill explicitly instructs the agent to "open three to five primary design-system sources and verify that their guidance is current," meaning the agent will fetch and use live documentation (e.g., ui.shadcn.com, radix-ui.com, headlessui.com, spectrum.adobe.com, carbondesignsystem.com) at runtime to shape its recommendations, so external content directly influences prompts/results.