stress-test

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly loads public market data via src.data.yahoo_client.get_stock_info and get_price_history (used in load_portfolio and the MACRO_FACTORS factor-fetch loop) and then prints/feeds that fetched data (raw JSON and analysis inputs) for the LLM to interpret and generate recommendations, so untrusted third-party web content can influence the agent's decisions.