Language: 日本語English

superpowers Codex Plugin

Author
Jesse Vincent
Category
Developer Tools
Topics
Claude Code Customization & Workflow · Code Quality, Review & Testing
Version
5.1.3
First seen
2026-07-12
Last confirmed
2026-07-13
Explanation last updated
2026-07-12

The explanation below is AI-generated. Please verify it against the sources.

Superpowers is a plugin providing an agentic skills framework and software development methodology for coding agents, covering planning, TDD, debugging, and collaboration workflows, according to its plugin.json and README. It is built by Jesse Vincent (obra) and Prime Radiant, and is distributed under the MIT license. The plugin guides an agent through a fixed sequence: brainstorming a spec, creating a git worktree, writing an implementation plan, executing that plan via subagent-driven development or batch execution, enforcing red/green TDD, requesting code review, and finishing the development branch. It ships a library of composable 'skills' organized into Testing, Debugging, Collaboration, and Meta categories, and states these skills trigger automatically once installed. It can be installed into multiple coding-agent harnesses, including Claude Code, Codex CLI/App, Cursor, Factory Droid, GitHub Copilot CLI, Gemini CLI, OpenCode, Kimi Code, and Pi, per the README installation instructions.

Overview

Superpowers is not a SaaS product; it is an open-source plugin/skills framework (MIT licensed) that installs into coding-agent tools (e.g., Claude Code, Codex, Cursor) to enforce a structured software development methodology, according to the README and plugin.json.

What you can do with superpowers

  • Have the agent brainstorm and refine a spec through questions before writing code (brainstorming skill)
  • Create an isolated git worktree and verify a clean test baseline (using-git-worktrees skill)
  • Break approved designs into small, verifiable implementation tasks (writing-plans skill)
  • Execute the plan via subagent-driven-development (fresh subagent per task with two-stage review) or executing-plans (batch execution with human checkpoints)
  • Enforce RED-GREEN-REFACTOR test-driven development, deleting code written before tests (test-driven-development skill)
  • Run systematic 4-phase root-cause debugging (systematic-debugging skill) and verify fixes before completion (verification-before-completion skill)
  • Request and respond to code review against the plan, with issues reported by severity (requesting-code-review, receiving-code-review skills)
  • Finish a branch by verifying tests and choosing merge/PR/keep/discard, then cleaning up the worktree (finishing-a-development-branch skill)
  • Install the plugin across supported coding-agent harnesses such as Claude Code, Codex CLI/App, Cursor, Factory Droid, GitHub Copilot CLI, Gemini CLI, OpenCode, Kimi Code, and Pi

Sources

Original description (English)

An agentic skills framework & software development methodology that works: planning, TDD, debugging, and collaboration workflows.

History of superpowers

Back to list