deep-research-skill

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 performs web discovery and then “fetch → extract content” from arbitrary outsider-authored webpages/sources during Phase 3 (Iterative Querying), so injected instructions contained in fetched page text are ingested as untrusted free text.