Hey, have you considered putting JevRouter directly between the agent and tool execution?

Hey, have you considered putting JevRouter directly between the agent and tool execution? The LLM would propose an action, but a controller would use JevRouter’s decision and policy checks to decide whether to execute it. The agent wouldn’t have a separate path to call the tool directly. That would make routing an enforced step rather than something the agent is instructed to consult. Is that a direction you’re considering?