google-drive-automation
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as file names, folder structures, and file metadata retrieved from Google Drive, which can serve as a vector for indirect prompt injection attacks.
- Ingestion points: Data enters the context through
GOOGLEDRIVE_FIND_FILE,GOOGLEDRIVE_LIST_FILES, andGOOGLEDRIVE_GET_FILE_METADATAas described inSKILL.md. - Boundary markers: The instructions do not define strict delimiters or 'ignore embedded instructions' warnings when handling file metadata or search results.
- Capability inventory: The skill provides high-privilege capabilities including file sharing (
GOOGLEDRIVE_ADD_FILE_SHARING_PREFERENCE), file movement (GOOGLEDRIVE_MOVE_FILE), and permission deletion (GOOGLEDRIVE_DELETE_PERMISSION). An attacker could name a file with instructions to share itself with an external email. - Sanitization: No specific sanitization or validation steps are outlined for the data returned from Google Drive queries.
- [EXTERNAL_DOWNLOADS]: The skill configuration requires the addition of a remote Model Context Protocol (MCP) server endpoint at
https://rube.app/mcp. This allows the agent to interact with a third-party service to execute its operations.
Audit Metadata