using-superpowers

To enforce AI assistant skill invocation before generating a response, apply a skill-first governance workflow that mandates loading applicable skills into context before any action or clarifying question is taken. This ensures consistent task handling and prevents ad-hoc actions.

Skill governance in conversational AI workflows is a pre-checking process that requires agents to search for and load relevant skills before composing a response. This mechanism ensures traceability by announcing the active skill and prevents inconsistent ad-hoc task handling.

To stop your CLI-managed AI agent from taking ad-hoc actions, implement a strict skill enforcement checklist that requires checking for applicable skills even at a 1% likelihood before responding. This process guidance mandates loading specialized skills first.

You should load specialized skills in an AI workflow whenever a user request might involve brainstorming, debugging, or implementation. Process skills must be loaded and prioritized before implementation to determine which rigid or flexible skill discipline applies.

Enforcing skill-first behavior works with flexible AI task handling by defining priority ordering and distinguishing between rigid and flexible skill types. This allows conversational agents to follow appropriate discipline while still mandating initial skill checks.

The limitation of mandating skill checks before responding is that it requires the environment to have the ability to load skill content into context before response composition. Without this capability, agents cannot announce which skill is being used for traceability.