use-design-md

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

• Third-party content exposure detected (medium risk: 0.30). use-design-md takes external network content at runtime from getdesign.kr (including llms.txt index/category links and then the selected services//llms.txt design.md), which is freeform text authored by third parties, so it is subject to indirect prompt injection whenever the outsider can influence which catalog entry is fetched via their query/slug selection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly performs runtime fetches of remote design briefs (e.g., curl https://getdesign.kr/services//llms.txt and the optional token sidecar https://raw.githubusercontent.com/CaesiumY/ko-design-md/main/services/.tokens.json) which are required inputs and directly drive the agent's instructions for how to restyle the user's UI.