drafting

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime LLM context is populated with outsider-authored free text from public sources during Step A evidence resolution: it calls superpower-writing:research-lookup / citation-management (network fallback) and may ingest scraped/web page text or retrieved document excerpts as prose to resolve claims.