agent-architect

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly allows and instructs use of built-in WebSearch/WebFetch and RAG in Phase 2 (see "Built-in Tools" / "WebSearch / WebFetch" and the RAG section), so the agent may fetch and read untrusted public web pages/documents as part of its workflow.