agentmail-toolkit

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt includes insecure examples that hard-code or pass an API key literal (e.g., apiKey: "your-api-key" and client=AgentMail(api_key="your-api-key")), which can lead an agent to embed real secret values verbatim in generated code or commands even though it also mentions the safer env-var pattern.

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

• Third-party content exposure detected (high risk: 1.00). This skill's SKILL.md shows integrating AgentMailToolkit tools (e.g., get_inbox, get_thread, get_attachment, send_message) so the agent will fetch and read arbitrary incoming emails and attachments from external users, i.e., untrusted third‑party content that can materially influence actions.