cali-product-multi-method-market-analysis

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

• Third-party content exposure detected (high risk: 0.85). Yes—Variant 2’s required workflow instructs the agent to “Search across” public web/social sources (e.g., news sites, LinkedIn/X, Reddit) and then incorporate the resulting scraped/collected text into the LLM’s generated “Weekly Intelligence Canvas,” which is outsider-authored free text at runtime.