opencli-adapter-author

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md runbook and api-discovery sections explicitly instruct the agent to fetch and inspect open/public third-party websites and responses (e.g., using "opencli browser network", "opencli browser eval", "page.fetchJson()", "page.evaluate()", and the §5 installInterceptor flow) and cites public sites like eastmoney/xueqiu/bilibili/tonghuashun, meaning untrusted web content is ingested and can directly change authentication strategy, endpoint selection, and subsequent actions.