manage-email
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill includes a dedicated safety block that explicitly instructs the agent to treat scraped content (emails, transcripts, files) as untrusted data rather than instructions. This provides a strong defense against indirect prompt injection by telling the AI to ignore any commands found within email bodies.
- [COMMAND_EXECUTION]: The skill implements a robust human-in-the-loop pattern. It is designed to pause for user confirmation (e.g., "go", "draft", "mark them read") before moving between stages like reading, drafting, and cleanup. This prevents the agent from taking autonomous actions on sensitive email data without oversight.
- [DATA_EXFILTRATION]: Analysis shows no unauthorized data transfer mechanisms. The workflow is contained within the user's existing Gmail environment, where drafts are placed for manual review rather than being sent automatically or transmitted to external third-party servers.
Audit Metadata