meta-job-search-pipeline

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

• Third-party content exposure detected (high risk: 0.85). The runtime path uses multi-search-engine → web_research to fetch public web content per target company, and that fetched page text is then read by the agent’s LLM context (via subsequent llm_chat/enrich_company steps consuming outputs.web_research), which is outsider-authored free text from the public web.