engineering / skill

Superpowers

Add skills to your coding agent that guide it from brainstorming and planning through test-driven development to code review.

Repository owner obra

Not rated yet
0 comments

Like 0 Save

Published 11 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: Add skills to your coding agent that guide it from brainstorming and planning through test-driven development to code review.
Runs in: Local agent with shell
Operating systems: Not stated
Needs: Runtime: Not stated; Needs internet: Not stated; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Yes
Tool type: Skill
Delivery: From GitHub
Version: Latest on the main branch
Access: Code on GitHub
Compatibility: Claude Code, Antigravity, Codex App, Codex CLI, Cursor, Devin CLI, Factory Droid, Gemini CLI, GitHub Copilot CLI, Grok Build CLI, Kimi Code, OpenCode, Pi, Qwen Code, Hermes Agent and Muse, according to the README. The README says to install it separately for each one.
External costs: The repository is MIT-licensed. The README states no price for the skills. It mentions commercial support, additional tooling and managed spending for enterprise users, without prices. Your coding agent or model provider may charge for usage.
Permissions: The skills act through your coding agent with the access you give it. According to the README, they create an isolated workspace on a new branch and run project setup, dispatch subagents, commit work as part of the test-driven development cycle, delete code written before tests, offer merge, pull request, keep or discard choices when a branch is finished, and clean up the worktree. For OpenCode, the README has you tell the agent to fetch and follow an install instructions file from the repository. The README says a startup bootstrap is injected by a session-start hook on several agents, and that the diagnosing-superpowers skill reads the session transcript and, if you want, packages a scrubbed bundle for a bug report.
Usage terms: MIT License, as stated in the repository. See its LICENSE file for the exact terms.

Code on GitHub

obra/superpowers

MIT

Shell298,226 stars26,652 forksLast pushed 10 Oct 2026 UTC

An agentic skills framework & software development methodology that works.

Topics: ai brainstorming coding obra sdlc skills subagent-driven-development superpowers

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

Skill folders in this repository (15)

Folders that contain a SKILL.md file, from GitHub's file list on 13 Oct 2026 UTC.

View on GitHub ↗

What it does

Superpowers is a set of skills that adds a software development workflow to your coding agent. According to the README, it is a software development methodology built on composable skills and some initial instructions that make sure the agent uses them. The README says the skills trigger automatically.

The README describes this flow. The agent asks what you are trying to build, draws a spec out of the conversation and shows it in chunks short enough to read, then writes an implementation plan after you sign off on the design. When you say go, it starts subagent-driven development, with agents working through each task and reviewing their work.

The README groups the repository's skills as testing (test-driven-development), debugging (systematic-debugging, verification-before-completion, diagnosing-superpowers), collaboration (brainstorming, writing-plans, executing-plans, dispatching-parallel-agents, code review, git worktrees, finishing a branch, subagent-driven-development) and meta skills (writing-skills, using-superpowers).

Who it is for

People who build software with a coding agent and want it to follow a design, plan, test and review process.

Example tasks

Brainstorm a design with your agent before any code is written.
Turn an approved design into a plan of small tasks with exact file paths.
Work through the plan with a fresh subagent per task and a review after each.
Use red, green, refactor test-driven development while implementing.
Ask your agent to figure out what went wrong with superpowers in a session.

Getting started

The README gives separate steps for each agent. For Claude Code it gives /plugin install superpowers@claude-plugins-official, or /plugin marketplace add obra/superpowers-marketplace followed by /plugin install superpowers@superpowers-marketplace. For Cursor it gives /add-plugin superpowers. Other agents use their own plugin, extension or marketplace commands. The README says updates depend on the coding agent and are often automatic. The README has the exact commands.

Limitations

It needs a supported coding agent. The README says installation differs by harness, so you install Superpowers separately for each one you use. It says some harnesses need extra steps: restart Hermes sessions after installing, and approve the plugin's hooks on first install in Muse. It says Hermes has no post-compaction hook, so a very long session that compacts over its first turn loses the bootstrap. It says subagent and task-list tools remain optional Pi companion packages. The README names no model requirement. The repository can change at any time; check its README before you install.

Data practices

The README has a section called Visual companion telemetry. It says the optional visual companion in the brainstorming skill loads a logo from the authors' website by default, that the request includes the Superpowers version, and that the authors use this to get a rough idea of how many people use Superpowers and which version. According to the README, it includes no details about your project, prompt or coding agent. To turn it off, set the SUPERPOWERS_DISABLE_TELEMETRY environment variable to any true value. The README says Superpowers also honors Claude Code's DISABLE_TELEMETRY and CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC opt-outs. It says the diagnosing-superpowers skill can package a scrubbed bundle for a bug report if you want. Beyond that, the README does not describe data handling. What is sent where depends on your coding agent and model provider.