gmail-automation

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DATA_EXFILTRATION]: The skill connects to a third-party domain https://rube.app/mcp to function as an MCP server. This routes agent interactions and potentially sensitive email metadata through an external service not included in the trusted vendors list.\n- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it processes external, untrusted content from emails.\n
  • Ingestion points: Untrusted data is ingested via tools such as GMAIL_FETCH_EMAILS, GMAIL_FETCH_MESSAGE_BY_MESSAGE_ID, and GMAIL_GET_ATTACHMENT.\n
  • Boundary markers: The skill lacks instructions for using delimiters or boundary markers to prevent the agent from obeying commands embedded within email bodies.\n
  • Capability inventory: The skill provides access to powerful tools including GMAIL_SEND_EMAIL, GMAIL_REPLY_TO_THREAD, GMAIL_DELETE_LABEL, and GMAIL_BATCH_MODIFY_MESSAGES.\n
  • Sanitization: No input validation or sanitization of email content is performed before the data is processed by the agent.\n- [EXTERNAL_DOWNLOADS]: The skill directs the user to add an external dependency in the form of a remote MCP server located at https://rube.app/mcp.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 12:22 PM