mirror-multi-agent-framework-iterative
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to generate and execute Python solver code (e.g., using the PuLP, Gurobi, or OR-Tools libraries) as part of its 'Iterative Adaptive Revision' workflow. This execution is used to validate the model's feasibility and correct implementation errors, which is the intended purpose of the optimization assistant.
- [PROMPT_INJECTION]: The skill processes untrusted natural language input from users to generate executable code, creating an indirect prompt injection surface.
- Ingestion points: User-provided problem descriptions (e.g., 'A factory makes two products...', 'Minimize the makespan...').
- Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands in the user input.
- Capability inventory: Access to Python execution tools for running generated solver scripts.
- Sanitization: The framework relies on the 'Modeling Revision' and 'Code Revision' agents to diagnose and fix errors during execution rather than performing input sanitization.
Audit Metadata