narova-google

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

• Third-party content exposure detected (medium risk: 0.30). In both tool/worker.py (TTS) and tool/video-worker.py (Veo), the required runtime main() reads JSONL requests from sys.stdin and then LLM-ingests outsider-provided free text fields (text or prompt) when an external user submits a synthesis/generation request via the Narova provider protocol.