longbridge-catalyst-radar

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

• Third-party content exposure detected (high risk: 1.00). The skill's execution flow and API map (Step 3 "获取数据" and references/longbridge-api-map.md) explicitly use Web Search as a fallback and MCP endpoints (e.g., SocialMCPServiceGetTickerNewsArticles, news detail, filings and community topics) to fetch public news, community/third‑party articles and forum/report content—which are untrusted user-generated or open‑web sources that the agent reads and uses to generate signals and influence its recommendations.