web-search-exa

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

• Third-party content exposure detected (high risk: 0.95). The skill’s runtime workflow calls Exa endpoints like /search, /answer, /findSimilar, and /contents with user-supplied queries/URLs; Exa then fetches and returns scraped web page text (outsider-authored) which is used as LLM context (e.g., via context: true / text: true).