extract-design

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

• Third-party content exposure detected (high risk: 0.85). This skill’s runtime workflow invokes a third-party tool (npx designlang <url>) that crawls/reads the provided website URL and then “reads the generated markdown file” (cat design-extract-output/*-design-language.md), which is derived from outsider-authored public web page content and therefore can include free text that gets into the agent/LLM context.