voice-to-report

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

• Third-party content exposure detected (high risk: 0.90). The skill ingests arbitrary user-generated audio (e.g., Telegram voice messages shown in the n8n workflow's "Telegram Trigger" / "Download Voice" nodes and the mobile upload flow in SKILL.md), transcribes that untrusted content and feeds the transcript into the LLM to drive structured report creation, so third-party content can materially influence agent behavior.