email-for-ai-agents

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly subscribes to AgentMail inboxes (see SKILL.md code using client.websockets.connect and MessageReceivedEvent) and processes event.message.text from arbitrary inbound emails (public, user-generated content) as input to LLM-driven workflows, so untrusted third-party content can materially influence agent actions.