multi-agent-patterns

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs multi-agent workflows that include a "Researcher (web search, document retrieval)" and discusses "searching multiple independent sources" (see Examples and the "Parallelization" / "Research Team Architecture" sections), which indicates the agents will fetch and interpret open/public web content that could be untrusted and materially influence downstream agent decisions.