using-superpowers

A skill-first workflow mandates that the AI invokes relevant skills via the Skill tool before generating any response, ensuring deterministic behavior and reducing misinterpretations in multi-skill environments.

To ensure the AI invokes skills before every reply, instruct it at the start of the conversation to first detect and execute relevant skills, using clarifying questions when needed to guide selection and execution order.

The AI handles multi-skill selection by evaluating the conversation context to identify relevant skills, determining their correct execution order, and applying a safe fallback mechanism when no specific skill applies.

Yes, you can use this approach for AI safety and instruction following, as it enforces mandatory skill execution to coordinate behavior, deferring non-skill-based content until a relevant skill is engaged.

When no relevant skills apply during a conversation, the workflow specifies a safe fallback mechanism, deferring non-skill-based content generation until a relevant skill is properly engaged.