using-superpowers

To enforce AI safety guardrails before generating a reply in Claude Code, you orchestrate mandatory skill invocation at the start of a conversation, loading relevant instructions and guardrails into context before responding.

Mandatory skill invocation is a workflow mechanism that activates at the start of a new conversation to locate and load relevant skills. It loads the skill body into context and presents its instructions for execution to ensure correct guided responses.

You load relevant skills by invoking the Skill tool at the start of a conversation. This action locates the required skills, verifies their frontmatter includes a valid name and description, and loads the skill body into context for execution.

Yes, activating workflow guidance skills requires frontmatter that includes a valid name and description. This metadata ensures the skill can be properly located, loaded into context, and presented with the correct instructions for execution.

Yes, the skill invocation workflow supports optional scripts, references, and assets as needed. These components are loaded alongside the skill body to provide battle-tested workflows and clear guidance for clarifications and tasks.

Manual skill loading lacks mandatory enforcement, risking unguided or unsafe AI responses. An automated orchestration approach is required to consistently apply guardrails and ensure the right skill is invoked before any reply is generated.