survey

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow explicitly dispatches sub-agents to “gather sources from the web, papers, social media” and then consolidates their outputs into an indexed Markdown report, which implies outsider-authored free text from public web/social content is fetched and ingested into the agent’s LLM context for summarization/quotation.