voice-segment-selector

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill provides tooling and step-by-step automation to extract, prepare, export, and even run/publish fine-tuned TTS/voice-clone models (including inference smoke tests and HF publish helpers), which is dual‑use and potentially misuseable for impersonation though not explicitly malicious.

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime /review flow serves an HTTP review UI and accepts outsider-authored transcript_final text via POST /api/transcript/{clip_id}, which the server writes into candidates.jsonl and later exports into metadata.jsonl/Orpheus datasets.