omni-ai-optimizer
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the official
omnicommand-line interface to interact with model configurations, manage user attributes, and handle authentication profiles. This is the primary and intended mechanism for the skill's functionality. - [INDIRECT_PROMPT_INJECTION]: The skill manages an attack surface where instructions for the Omni Agent (e.g.,
ai_context,sample_queries) are stored within model files. - Ingestion points: Model configuration files are read via
omni modelscommands (SKILL.md), and usage analytics data is retrieved from the platform (references/sample-queries.md). - Boundary markers: The skill explicitly instructs the agent to use isolated branches for testing changes before promotion to ensure the shared model remains stable (SKILL.md).
- Capability inventory: The skill can modify model behavior through
omni models yaml-createandomni context-files update(SKILL.md). - Sanitization: The skill relies on the official
omni models validateutility to check for schema errors, deprecated syntax, and context budget constraints (references/context-window.md).
Audit Metadata