together-audio

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

• Third-party content exposure detected (high risk: 0.90). The skill's STT workflow explicitly allows ingesting audio via arbitrary URLs (see references/stt-models.md "Input Methods" showing file="https://example.com/audio.mp3" and the transcribe examples in scripts/stt_transcribe.py/ts that call client.audio.transcriptions.create), so the agent can fetch and interpret untrusted third-party audio whose content could influence subsequent actions.