agentmail-toolkit

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill requires the installation of the agentmail-toolkit package from npm and PyPI. This package is not maintained by a pre-approved trusted organization (e.g., OpenAI, Anthropic, Google) and its runtime behavior has not been verified.
  • PROMPT_INJECTION (LOW): The skill exhibits a surface for Indirect Prompt Injection (Category 8).
  • Ingestion points: External data enters the agent context via list_threads, get_thread, and get_attachment tools, which retrieve content from incoming emails.
  • Boundary markers: There are no visible boundary markers or instructions to the agent to disregard embedded commands within email bodies.
  • Capability inventory: The agent possesses sensitive capabilities including send_message, delete_inbox, and create_inbox, which could be abused if an attacker sends a malicious email that overrides the agent's instructions.
  • Sanitization: No sanitization logic is described for the content returned by the email tools before it is processed by the LLM.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:02 PM