always-on-guidance

Enforcing policy coordination across multiple Claude-plugin agents requires loading explicit, rule-based guidance into context. This aligns behavior with defined policies, preventing drift and ensuring consistent decision-making across multi-agent tasks.

Always-on guidance is a persistent, rule-oriented layer for Claude-plugin agents. It actively enforces global rules and model defaults during execution to guarantee adherence to organizational policies and ensure safe tool usage.

Apply model defaults by loading explicit guidance into the agent context. This mitigates behavior drift by enforcing constant, rule-based alignment with safe defaults throughout the agent's operational execution.

Yes, this guidance layer supports multi-agent task coordination. It loads global rules and governance patterns into context, ensuring multiple agents adhere to defined policies and maintain consistent tool usage across workflows.

The primary limitation of rule-based guidance is its dependence on explicitly loaded context. It requires predefined policies for every scenario and cannot dynamically adapt to situations outside its defined global rules and model defaults.