agentmail-toolkit

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs users to install the agentmail-toolkit package via standard package managers (npm and pip). These are vendor-provided resources intended for integrating the AgentMail service.
  • [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted external data by allowing the agent to read emails through tools like list_threads and get_thread. This creates a surface for indirect prompt injection if the agent processes malicious instructions embedded in email bodies or attachments.
  • Ingestion points: Email content is retrieved via get_thread, list_threads, and get_attachment (SKILL.md).
  • Boundary markers: No specific delimiters or safety instructions are provided in the documentation snippets to help the agent distinguish between system instructions and email content.
  • Capability inventory: The agent is granted capabilities to take actions based on processed data, including send_message, reply_to_message, and delete_inbox (SKILL.md).
  • Sanitization: The provided integration examples do not include explicit sanitization or filtering of email content before it is passed to the LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 03:47 AM
Security Audit — agent-trust-hub — agentmail-toolkit