When the agent reaches for it
The agent keeps every installed skill's description in view and loads the full instructions when a request matches. This is the description it reads:
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
One gate, three sizes of process
The skill scales the ceremony to the task and keeps the approval gate fixed. A spike is a feasibility question: two or three sentences on what the agent will try, a nod from you, then an answer rather than kept code. A bounded change to existing code gets a short design in the chat. An architectural change gets the full process: questions, alternative approaches, a sectioned design, a written spec, and a handoff to writing-plans.
The agent says which path it picked out loud, so you can override it. When it is unsure it takes the heavier path, and a task that turns out bigger than it looked moves up a path. Nothing moves down mid-task.
What the instructions cover
The sections of its SKILL.md, in order.
- Three Paths
- Anti-Pattern: "Too Simple To Need Approval"
- Red Flags
- Checklist
- Process Flow
- The Process
- After the Design (architectural path)
- Visual Companion
What ships with it
Besides SKILL.md, the folder carries these. The agent opens them only when a step in the instructions calls for one.
-
scripts/ -
spec-document-reviewer-prompt.md -
visual-companion.md
Install the brainstorming skill
In Claude Code, run this in a session. It comes from the publisher's own instructions.
/plugin install superpowers@claude-plugins-official License: MIT.
In Longwave, add Superpowers from the marketplace. It installs all 14 of its skills, and your Claude Code, Codex and OpenCode threads can use them.