agent-email

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and reads email content from external ClawEmail accounts (via mail-cli commands such as claw_list, claw_body, claw_search and the claw_init flow that curls https://u.163.com/...), and SKILL.md/references state agents "拉取邮件 → 解析邮件内容 → 执行业务逻辑" and "Agent 理解内容、执行操作", so untrusted third‑party email bodies can directly influence agent decisions and actions.