gemini-omni
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing user-supplied natural language prompts and media files (images/videos) to generate video content. While this is an ingestion surface, the risk is negligible as the output is media content and does not influence agent control flow.
- Ingestion points: User-provided prompts,
reference_image_paths, andinput_video_pathparameters. - Boundary markers: None explicitly documented in the Markdown text.
- Capability inventory: API calls for video generation and editing; the skill does not instruct the agent to execute shell commands based on the generated content.
- Sanitization: The documentation assumes the use of the Gemini API, which employs built-in safety filters and guardrails.
- [SAFE]: The skill serves as an educational guide for legitimate Google DeepMind services. All external references point to trusted official domains (ai.google.dev, blog.google). The handling of API keys via environment variables follows standard security best practices.
Audit Metadata