gemini-omni
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data and remote documentation which represents a potential injection surface.
- Ingestion points: User-supplied video generation prompts and the remote markdown file at
https://www.pixazo.ai/models/gemini-omni.md. - Boundary markers: No specific delimiters are used to separate user-provided content from agent instructions.
- Capability inventory: Performs network POST and GET requests to
gateway.pixazo.aiandwww.pixazo.ai; writes to and modifies permissions of the local file~/.pixazo/api-key. - Sanitization: No explicit sanitization or validation logic is defined for user prompts or the content of the remote markdown file.
- [EXTERNAL_DOWNLOADS]: The skill fetches operational configuration and parameter definitions from an external URL.
- Evidence: Instructs the agent to retrieve
https://www.pixazo.ai/models/gemini-omni.mdto determine the exact schema for API calls. - [COMMAND_EXECUTION]: The skill uses shell commands to manage local security settings for credentials.
- Evidence: The agent is instructed to use
chmod 600on the~/.pixazo/api-keyfile to ensure the API key is not accessible by other users on the system.
Audit Metadata