china-ai-readiness

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

• Third-party content exposure detected (high risk: 0.78). The skill’s runtime workflow calls external data providers (iFind MCP / AkShare MCP) and also references secondary sources like 巨潮 filings and broker/industry reports; these are outsider-authored web/market data that can include free-form text (e.g., company descriptions, filings, report narratives) which the agent may ingest into the LLM context.