email-audit

Fail

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands (e.g., dig, curl, openssl) by directly interpolating the user-provided domain variable into the command string. This creates a high risk of command injection if the domain contains shell metacharacters like semicolons or pipes.
  • Evidence: dig txt +short
  • Evidence: curl https://mta-sts./.well-known/mta-sts.txt
  • [REMOTE_CODE_EXECUTION]: A Python execution pattern interpolates user input directly into a Python list literal, which is susceptible to code injection. An attacker could craft a domain input to execute arbitrary Python logic.
  • Evidence: python -c "import checkdmarc; print(checkdmarc.check_domains(['']))"
  • [EXTERNAL_DOWNLOADS]: The skill uses curl to retrieve policy files from external domains, which involves fetching and potentially processing data from unverified remote sources.
  • Evidence: curl https://mta-sts./.well-known/mta-sts.txt
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 29, 2026, 09:41 PM