using-superpowers

Enforcing skill invocation before an AI response requires applying a consistent rule to locate and use skills upfront. This ensures the agent calls the Skill tool to determine applicable skills before answering any prompt, including clarifications and follow-up questions.

Skill lookup before answering prompts guides AI behavior and decision-making by establishing how to locate and use skills upfront. This mechanism ensures the agent consults relevant skills for complex tasks, preventing unsafe, ad-hoc actions during conversations.

Formatting a SKILL.md file requires placing a YAML frontmatter with name and description at the root. You can also include optional scripts, references, and assets to provide extended resources for governing how the agent locates and uses skills.

Workflow governance applies to clarification and follow-up questions to maintain consistent skill-driven behavior. This ensures the agent calls the Skill tool before every reply, providing guardrails across all conversations to prevent ad-hoc, unsafe actions.

The limitation of enforcing early skill invocation is that it strictly requires the agent to call the Skill tool before generating any text. This mandates a YAML frontmatter at the SKILL.md root and may delay immediate responses to ensure safety and governance.

The best way to provide guardrails for safety and governance is enforcing early skill invocation across all prompts. This requires the agent to call the Skill tool to determine applicable skills before answering, preventing ad-hoc, unsafe actions during complex tasks.