google-drive-automation

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md workflows explicitly call Google Drive APIs (e.g., GOOGLEDRIVE_FIND_FILE and GOOGLEDRIVE_DOWNLOAD_FILE) to locate and download arbitrary user-generated files from Google Drive via the Composio/Rube MCP toolkit, which the agent is expected to read and then take follow-up actions (share/move/modify), exposing it to untrusted third‑party content that could carry indirect instructions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill requires configuring and calling the Rube MCP endpoint (https://rube.app/mcp) at runtime—specifically instructing to call RUBE_SEARCH_TOOLS to fetch tool schemas—so remote content from that URL can directly control available tools/prompts and thus influence agent behavior.