telegram-voice-to-voice-macos

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly ingests Telegram user-generated voice notes saved under ~/.openclaw/media/inbound/ (see SKILL.md and scripts/transcribe_telegram_ogg.sh) and transcribes/uses them to generate replies, so untrusted third-party content can influence the agent's outputs and behavior.