synthesis-meeting-transcripts

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly searches and fetches user-generated content from Gmail and Google Drive (see SKILL.md Steps 2–3 and optional-workspace-mcp/fetch-meeting.py which calls search_drive_files and get_drive_file_content) and then reads/interprets the Notes/Transcript to update indices and commit files (Steps 4–5), so untrusted third-party document content can materially influence agent decisions and actions.