Source: obra/superpowers · Maintainer: community: Jesse Vincent (Prime Radiant) · License: MIT · Checked: 24 October 2026
What is the using-superpowers skill?
using-superpowers is the skill that makes the rest of Superpowers work: it tells your coding agent to check for and invoke relevant skills before it does or says anything, including before it asks a clarifying question. The Superpowers plugin loads it automatically at the start of every session.
It lives in obra/superpowers, the skills library by Jesse Vincent and Prime Radiant. The README lists it under "Meta" as the "Introduction to the skills system." Its description reads: "Use when starting any conversation - establishes how to find and use skills, requiring skill invocation before ANY response including clarifying questions."
As of 24 October 2026, its skills.sh page listed 270.5K installs (snapshot, 24 October 2026). The latest Superpowers release on GitHub was v7.0.1, published 21 October 2026.
For the full library and the workflow it drives, see our Superpowers skills guide.
What does it do?
The SKILL.md is short. It sets one rule, a priority order, a list of excuses to reject, and a precedence rule for your own instructions.
The rule. "Invoke relevant or requested skills BEFORE any response or action." The skill frames it as a threshold: "If you think there is even a 1% chance a skill might apply to what you are doing, you ABSOLUTELY MUST invoke the skill." If a skill turns out to be wrong for the situation, the agent doesn't have to keep using it.
Brainstorm before plan mode. If the agent hasn't brainstormed yet, it invokes the brainstorming skill before entering plan mode.
Announce and follow. The agent says "Using [skill] to [purpose]" and follows the skill exactly. If the skill has a checklist, it creates a todo per item.
Process skills first. When several skills apply, process skills set the approach and implementation skills carry it out. The two examples it gives:
- "Let's build X" goes to brainstorming first, then implementation skills.
- "Fix this bug" goes to systematic-debugging first, then domain skills.
Red flags. A table of twelve thoughts that mean the agent is rationalizing, each with a reply. A few of them:
| Thought | Reality (from the skill) |
|---|---|
| "This is just a simple question" | Questions are tasks. Check for skills. |
| "Let me explore the codebase first" | Skills tell you HOW to explore. Check first. |
| "I remember this skill" | Skills evolve. Read current version. |
| "The skill is overkill" | Simple things become complex. Use it. |
Subagents opt out. The file opens with a block telling a subagent "dispatched as a subagent to execute a specific task" to ignore the skill.
You come first. The last section says user instructions (CLAUDE.md, AGENTS.md, GEMINI.md, direct requests) take precedence over skills, which in turn override default behavior.
What else is in the folder?
A references/ folder holds one note per harness. The SKILL.md points to Claude Code, Codex, Pi, Antigravity, Hermes Agent and Muse; the folder also has a Gemini note. The Claude Code note describes an opt-in way to run a whole subagent-driven plan more cheaply through one orchestrator subagent. The Codex note covers turning on multi-agent support and spawning subagents with a clean context.
How is it loaded at session start?
In Claude Code, the plugin's hooks/hooks.json registers a SessionStart hook with the matcher startup|clear|compact. The hook script reads skills/using-superpowers/SKILL.md and returns its full text as session context, wrapped in a note that begins "You have superpowers."
Claude Code's hooks reference explains the mechanism: the startup, clear and compact matchers fire on a new session, after /clear and after compaction, and text a hook returns as additionalContext is wrapped in a system reminder; for a new session, that reminder goes at the start of the conversation, before the first prompt. That is why the rule is in force before you type anything, and why it comes back after a compaction. Our Claude Code hooks guide covers hooks in general.
Other harnesses differ. The README says Pi injects the bootstrap at startup and again after compaction. It also warns that Hermes has no post-compaction hook, so a very long session can lose the bootstrap; start a fresh session if skills stop triggering. In Antigravity, the README says the skill's description is what prompts the model to load it.
How do you install it?
Install the Superpowers plugin. using-superpowers is not meant to run on its own, because its job is to route the agent to the other skills.
Claude Code, official marketplace. From the README:
/plugin install superpowers@claude-plugins-official
Claude Code, Superpowers marketplace.
/plugin marketplace add obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplace
Claude Code's plugin docs explain that /plugin install opens the plugin's details so you can review it and pick a scope, and that plugin skills appear as /<plugin>:<skill>, so this one is /superpowers:using-superpowers.
Codex, Cursor and others. The README covers 16 harnesses. In Codex, open /plugins, search for Superpowers and select Install Plugin. In Cursor Agent chat, run /add-plugin superpowers. See our guides to Codex skills and Cursor skills for how each one finds skills.
The folder alone. The skills CLI can copy just this skill:
npx skills add https://github.com/obra/superpowers --skill using-superpowers
Two caveats follow from the files. No hook ships with a bare skill folder, so it loads only when the agent matches its description. And it names other Superpowers skills, such as brainstorming and systematic-debugging, which won't exist unless you install them too.
To check the plugin works, the README suggests starting a fresh session and sending "Let's make a react todo list": a working install triggers brainstorming before any code is written.
How do you use it?
You don't call it directly. It shapes how the agent treats every request. These prompts are suggestions for seeing it work, not quotes from the repo:
Let's build a small CLI that renames photos by the date they were taken.Expect "Using brainstorming to..." and questions before code.The login test started failing after the last merge. Fix it.Expect systematic-debugging before any edit.Which Superpowers skills apply to adding rate limiting to this API?Skip the Superpowers workflow for this one: rename userId to accountId in utils.ts.Your request outranks the skill.Before you start, tell me which skill you're using and why.
Tips and limits
- It is strict by design. The 1% threshold means more questions and more tokens on small jobs. Put a line in CLAUDE.md or AGENTS.md if you want small edits handled without the full process; the skill says your instructions win.
- Reloads after compaction in Claude Code. The hook matcher includes
compact, so the rule returns after a long session is summarized. - Turning it off. The skill has no switch of its own. In Claude Code, disable the Superpowers plugin from the Installed tab in
/plugin, or tell the agent to skip the workflow. - Security audit. On 24 October 2026, the skills.sh page showed Gen Agent Trust Hub: Warn, Socket: Pass and Snyk: Pass. The page gave no details for the Warn. The skill runs no scripts; the plugin's hook only reads the SKILL.md and prints it as context.
- When a skill misfires. The diagnosing-superpowers skill reads the session transcript and reports which skills fired and when.
Save it in your Kipt library
The MIT License lets you copy and change the skill as long as you keep the copyright notice. In Kipt, open Skills and paste the folder's GitHub link; Kipt reads the SKILL.md and its references/ files, and for a repo with many skills it shows a list to pick from. Kipt stores your skills as Markdown in a private GitHub repo you own and installs them into Claude Code or Codex with one command:
npx skills add https://github.com/<you>/<repo>.git --all -g
A saved copy won't include the plugin's session-start hook, so keep the plugin installed if you want the bootstrap behavior. Kipt is not affiliated with Jesse Vincent, Prime Radiant or obra/superpowers.
Similar skills
- Brainstorming skill: the process skill using-superpowers sends new work to first.
- Systematic debugging skill: the first stop for bug reports.
- Diagnosing Superpowers skill: finds out why a skill did or didn't fire.
- Find skills skill: Vercel's skill for searching and installing skills from skills.sh.
- Writing skills skill: how Superpowers writes and tests new skills.