research

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs agents to "Use WebSearch" and "WebFetch on key URLs" and to read/cite external sources (official docs, GitHub READMEs, blog posts and production experience reports) in the Wave 1/Wave 2 templates and Quick Research Mode, so untrusted public web content is fetched and used to drive synthesis and decisions.