using-superpowers

To enforce AI tool invocation before generating responses, apply a frontmatter-driven rule that checks for applicable skills and mandates loading their instructions prior to any reply.

Conversational AI guardrails for preventing prompt injection are operational workflows that force the system to evaluate and invoke relevant skills first, ensuring safe, tool-guided behavior across all domains.

You set up this AI workflow by loading operational rules that specify mandatory Skill tool invocation at the start of an interaction, automatically checking for and applying relevant skills before answering.

Yes, this skill invocation workflow applies cross-domain to routine queries, clarifications, and planning, ensuring every interaction begins by evaluating and following relevant skill guidance.

The fallback behavior when no relevant skill applies is defined in the operational workflow loaded from the skill configuration, dictating how the system proceeds safely without tool-based reasoning.

You can use this pre-invocation rule to improve AI safety across different domains by enforcing tool-based reasoning at the outset of any conversation, reducing misaligned responses and prompt injection risks.