ai-rag-pipeline

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow for the RAG examples uses belt to fetch web search results and extract web page content (e.g., tavily/search-assistant and tavily/extract) before sending that retrieved text into an LLM, so outsider-authored free text can be ingested via search/extracted pages.