inbound-opportunity-protocol

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to fetch and read untrusted public sources—e.g., "Search GitHub", "Search the web: company website, LinkedIn, Twitter/X, Crunchbase" (Phase 1.1) and "Read their docs, API references, and any public code" (Phase 2.1 and Phase 5.1)—and to use those findings to shape replies and conversation strategy, which clearly allows third-party content to materially influence actions.