email-compose

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill implements strict negative constraints and a mandatory 'draft-review-approve' sequence that prevents the agent from sending emails autonomously or bypassing human oversight.
  • [DATA_EXFILTRATION]: Risk is mitigated through a multi-stage workflow where the user must explicitly confirm recipients and content. The skill also includes a specific checkpoint to detect and block the transmission of sensitive patterns like credentials or secrets in the email body.
  • [CREDENTIALS_UNSAFE]: The skill follows secure secret management practices by utilizing environment variables for SMTP authentication instead of hardcoding credentials in the configuration file.
  • [COMMAND_EXECUTION]: While the skill utilizes the Bash tool, its scope is limited to standard library operations and predefined validation logic, with no evidence of arbitrary or untrusted command execution.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 05:16 PM
Security Audit — agent-trust-hub — email-compose