gmail

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill directly reads and returns full email bodies and headers from a user's Gmail via scripts/gmail_operations.py (e.g., read_email, list_emails, forward_email, reply_to_email) which ingests arbitrary user-generated/untrusted email content that the agent will read and incorporate into drafts/forwards.