AI Agent Hub
Back to skills
🤖

Using Superpowers

AI Agent Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please install @org-02qudk26/using-superpowers-zh by following the guide at https://skillhub.cn/install/skillhub.md.

About this skill

Problem

using-superpowers addresses the failure mode where an agent skips existing workflows and acts on intuition. It makes skill invocation a mandatory preflight check: if a task could relate to any skill, even with only a 1% chance of fit, the agent must invoke the skill before deciding what to do next. It places skill checks before clarification, searching, and editing, so behavior relies on the workflow rather than ad hoc judgment. This reduces excuses such as “this is a simple question” or “I already remember the skill.”

How It Works

  • Check and invoke relevant skills before replying or taking action, rather than first asking for context, exploring code, or collecting information.
  • In Claude Code, load the skill content through the Skill tool and follow it directly; do not open skill files with Read.
  • Priority order: use process skills first, such as brainstorming and debugging; then use implementation skills, such as frontend-design and mcp-builder.
  • Distinguish rigid skills from flexible skills: rigid skills must be followed precisely, while flexible skills can be adapted to context.

Boundaries

The skill does not replace domain-specific skills; it constrains the agent's workflow. User instructions specify what to do, not how to do it; “add X” or “fix Y” still requires checking the relevant skill first. If another platform lacks a skill-loading mechanism, confirm the equivalent invocation method in that platform's documentation.

Use Cases

  • Fix a bug in Claude Code: call debugging before changing code.
  • Build a feature: call brainstorming before frontend-design.
  • When multiple skills fit, follow process skills first.
  • On 'add X' or 'fix Y', check skills before reading files.

Best For

  • Claude Code workflow engineers who want agents to stop skipping debugging and design.
  • Agent architects who need skill-invocation priority encoded in prompts.
  • Tech leads debugging unstable Claude Code behavior and excuse-based skips.
  • Skill-repo maintainers who want consistent rigid vs flexible skill rules.