Superpowers Bundle of 15 skills
- Use it when:
- You want the agent to follow a full development process instead of jumping straight to code.
- It does:
- Chains its skills into one workflow: brainstorm, plan, build test-first on its own branch, review, then merge.
- You get:
- An approved design, a written plan, tested commits and a review before anything merges.
- Runs:
- By itself: before every reply, the agent checks which of its skills applies.
- Works with:
- Claude Code, Codex, Cursor, Gemini CLI, OpenCode and more, each installed separately.
TL;DR
Superpowers makes its skills fire on their own: using-superpowers, loaded at the start of each session, has the agent check for a matching skill before every reply. Building something then follows one path. brainstorming asks what you actually want and ends with a design you approve, in chat for small changes or as a document for projects. using-git-worktrees sets up an isolated branch with a clean test baseline, and writing-plans breaks the design into small tasks, each with exact files, code and checks. The plan runs either through subagent-driven-development, a fresh subagent per task with a review after each, or through executing-plans, inline with one review at the end. Throughout, test-driven-development makes every change start from a failing test, and finishing-a-development-branch verifies the tests, then merges, opens a PR, keeps or discards the branch. Other skills step in when needed: systematic-debugging for bugs, verification-before-completion before claiming anything works, requesting-code-review and receiving-code-review around reviews, dispatching-parallel-agents for independent tasks, and writing-skills for writing new skills test-first.
Summarizes the repo at b739a8a. Its skills have changed by 5,767 words since.
Any agent, with the skills CLI
npx skills add https://github.com/obra/superpowers --skill '*' Claude Code plugin
/plugin install superpowers@claude-plugins-official Source on GitHubAll skills from obra/superpowers
- Total Downloads
- 422,260 of brainstorming, its most installed skill, on skills.sh, Nov 9, 2026
- Repo stars
- 314,664 obra/superpowers
- First seen
- Jan 19, 2026 on skills.sh, its first skill
Its skills, in the order they're used
The ones marked “Only in the bundle” need the rest of it. The others also work alone.
-
Makes the agent look for and invoke a matching skill before every response.
Makes the agent load the other Superpowers skills before it replies; alone, it has nothing to load.
-
brainstorming
422K total downloadsExplores what you actually want before any creative work starts.
-
using-git-worktrees
230K total downloadsIsolates feature work in a git worktree.
-
writing-plans
298K total downloadsWrites an implementation plan from a spec before touching code.
-
Executes a plan with one fresh subagent per task and a review after each.
Executes a plan such as writing-plans writes, and has its final reviewer use the prompt that ships with requesting-code-review.
-
Executes an implementation plan inline in the current session.
Runs scripts that ship with subagent-driven-development, and loads test-driven-development and verification-before-completion.
-
test-driven-development
274K total downloadsWrites the failing test first, then the code.
-
requesting-code-review
276K total downloadsSends finished work to a reviewer subagent before merging.
-
receiving-code-review
234K total downloadsEvaluates review feedback technically before implementing it.
-
finishing-a-development-branch
227K total downloadsVerifies tests, then merges, opens a PR or cleans up a finished branch.
-
systematic-debugging
325K total downloadsFinds the root cause before proposing any fix.
-
verification-before-completion
269K total downloadsRequires running verification commands before claiming work is done.
-
dispatching-parallel-agents
231K total downloadsSplits independent tasks across parallel agents.
-
writing-skills
231K total downloadsWrites skills test-first: pressure-test with subagents, then document.
-
Investigates why a Superpowers session went wrong from its transcripts, and drafts an evidence-backed bug report for the maintainers.
Only analyzes sessions run with the other Superpowers skills, so it has nothing to look at without them.