gemini-3-multimodal
Warn
Audited by Snyk on Feb 15, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.80). This skill directly ingests arbitrary user/third‑party media via genai.upload_file and GoogleAIFileManager.uploadFile (see scripts like process-video.py and examples using genai.upload_file / fileUri and model.generate_content([..., video_file])), so the agent will read and interpret untrusted images/audio/video/PDF content that could contain indirect prompt injections.
Audit Metadata