agentmail

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly receives and processes arbitrary incoming emails via webhooks (see SKILL.md and references/WEBHOOKS.md message.received examples and the webhook handler code) and uses email subject/body/attachments to drive actions like auto-replies, forwarding, GitHub issue creation, and file processing, so untrusted third-party content can materially influence agent behavior.