using-superpowers

A pre-response skill check prevents premature answers by requiring the Skill tool to be invoked whenever there is even a 1% chance a skill applies. This mandatory workflow stops rationalizations that skip the process before any conversation output.

Instruction priority establishes that user instructions take precedence, while specialized skills override default agent behavior when conflicts occur. This deterministic model ensures conversation control and proper tool routing across execution steps.

Yes, the skill invocation workflow defines strict tool-usage rules by platform, clarifying how to access and invoke skills across Claude Code, Copilot CLI, Gemini CLI, and other environments to ensure consistent conversation control.

You need to perform a skill check at all conversation starts, before asking clarifying questions, and during task execution steps. The probabilistic rule applies whenever there is any possibility a skill is relevant to the workflow.

Agents skip specialized guidance by rationalizing the bypass of the skill check process. Enforcing a mandatory pre-response skill invocation workflow with a deterministic instruction priority model stops this behavior and ensures proper tool routing.