What is Superpowers
Superpowers is an open-source Agentic skills framework and software development methodology developed by Jesse Vincent and the Prime Radiant team, designed for AI Coding Agents such as Claude, Cursor, and GitHub Copilot. Through 14 composable skills and initial instructions, it stops the AI coding assistant from jumping straight into writing code; instead, it first refines requirements through Socratic questioning, generates verifiable design documents, and creates a detailed implementation plan before entering a subagent-driven development workflow. It core-emphasizes RED-GREEN-REFACTOR test-driven development, the YAGNI principle, and the DRY principle, and supports advanced workflows such as Git Worktrees isolated development, parallel subagent scheduling, and two-phase code review, enabling agents like Claude to run autonomously for hours without deviating from the established plan.
Superpowers' main features
-
Socratic brainstorming: refine requirements through follow-up questions, present design options in chunks for user confirmation before entering coding
-
Design document generation: turn rough ideas into structured, verifiable specifications and save them automatically
-
Detailed implementation plan (writing-plans): break work down into 2-5 minute micro-tasks, including precise file paths, complete code, and verification steps
-
Subagent-driven development: dispatch an independent subagent for each task, and perform two-phase review (specification compliance + code quality)
-
Test-driven development: enforce the RED-GREEN-REFACTOR cycle, write tests before code, and delete any code written before the tests
-
Systematic debugging: a four-phase root-cause analysis process, including root-cause tracing, defense in depth, conditional waiting, and other techniques
-
Git Worktrees workflow (using-git-worktrees): develop in an independent branch and workspace, and verify a clean test baseline
-
Parallel agent scheduling (dispatching-parallel-agents): execute multiple subagent workflows concurrently to improve efficiency on large projects
-
Code review mechanism (requesting-code-review / receiving-code-review): automatic review between tasks, report issues by severity, and block progress on critical issues
-
Branch wrap-up (finishing-a-development-branch): verify tests, provide merge/PR/keep/discard options, and clean up the workspace
-
Skill self-extension (writing-skills): create and test new skills by following best practices, supporting teams in accumulating internal methodologies
-
Multi-platform plugin support: full coverage of the official Claude Code marketplace, Cursor, OpenAI Codex CLI/App, GitHub Copilot CLI, Gemini CLI, and OpenCode
How to use Superpowers
-
Official Claude Code marketplace: run
/plugin install superpowers@claude-plugins-official。 -
Claude Code Superpowers marketplace: first execute
/plugin marketplace add obra/superpowers-marketplace, then execute/plugin install superpowers@superpowers-marketplace。 -
OpenAI Codex CLI: open the plugin search interface, search for "Superpowers", and select install.
-
OpenAI Codex App: click Plugins in the sidebar, find Superpowers in the Coding category, and click
+to install. -
Cursor: in the Agent chat, search for "superpowers" through the plugin marketplace and install it.
-
OpenCode: Input
Fetch and follow instructions from https://raw.githubusercontent.com/obra/superpowers/refs/heads/main/.opencode/INSTALL.md。 -
GitHub Copilot CLI: Run
copilot plugin marketplace add obra/superpowers-marketplaceandcopilot plugin install superpowers@superpowers-marketplace。 -
Gemini CLI: Run
gemini extensions install https://github.com/obra/superpowers, when updating usegemini extensions update superpowers。 -
Basic Workflow: Brainstorming → Design Confirmation → Git Worktree Creation → Writing Plan → Sub-agent Execution/Plan Execution → TDD Implementation → Code Review → Branch Wrap-up.
Key Information and Usage Requirements of Superpowers
-
Open Source License: MIT License, free to use, modify, and apply commercially.
-
Maintainer: Jesse Vincent and the Prime Radiant team.
-
Core Philosophy: TDD first, systematic process, complexity reduction, evidence verification.
-
Number of Skills: 14 composable skills, covering testing, debugging, collaboration, plan execution, parallel agents, and Git workflows.
-
Installation Prerequisites: Requires a Coding Agent that supports a plugin system (Claude Code, Cursor, Codex, etc.).
-
Workflow Mandatoriness: The Agent must check relevant skills before executing tasks; this is a mandatory process, not a suggestion.
-
Community Support: Discord community, GitHub Issues, email subscription for release notifications.
-
Contribution Guidelines: Arbitrary new skills are not accepted; modifications must be verified across all supported platforms, following the
writing-skillsguidelines. -
Update Mechanism: Some platforms update automatically, depending on the Coding Agent's plugin system.
Core Advantages of Superpowers
-
Don't Write Code Directly, Understand Requirements First: When the Agent sees a development task, it first steps back and clarifies the real goal through questions, avoiding hallucinations and rework.
-
Systematic Rather Than Arbitrary: A process-based methodology replaces guesswork programming, with complexity reduction as the primary goal.
-
Evidence Over Assertion: Success is declared only after verification passes, avoiding false completion that "looks like it runs".
-
Automatic Skill Triggering: Skills activate automatically when relevant tasks are detected, with no manual invocation or memorization of complex instructions required.
-
Long Autonomous Operation: Agents such as Claude can work autonomously for hours under an established plan without deviating from the goal.
-
Mandatory Test First: TDD as a mandatory workflow rather than an optional suggestion, fundamentally ensuring code quality.
-
Multi-Agent Ecosystem Compatibility: The same skill library works seamlessly across platforms such as Claude, Cursor, Copilot, Codex, and Gemini.
-
Micro-task Breakdown: Task descriptions at a 2-5 minute granularity let even an "enthusiastic junior engineer" execute accurately.
-
Isolated Development: Git Worktrees avoid polluting the main branch, supporting parallel experiments and safe rollback.
Application Scenarios of Superpowers
-
Building New Features from Scratch: Ensure correct understanding of requirements through brainstorming and design documents, avoiding large-scale rework later.
-
Large-scale Refactoring Projects: Use Git Worktrees to isolate experiments, with sub-agents processing the transformation of different modules in parallel.
-
Bug Fixing and Root Cause Analysis: Systematic debugging skills guide the Agent through four-phase root cause tracing instead of blind trial and error.
-
Code Review Automation: Before committing, have the Agent perform a two-phase review to proactively detect spec deviations and quality issues.
-
Complex Multi-File Changes: Break large requirements into micro-tasks, dispatch sub-agents to implement step by step, and continuously verify.
-
Cross-Platform Development Collaboration: The same set of Superpowers skills provides a consistent experience across Claude, Cursor, and Copilot.
Superpowers FAQ
A: Both. It is a complete software development methodology based on composable skills, and it is also distributed as an open-source plugin to major AI coding platforms.
A: No. Skills are triggered automatically based on the current task. Users only need to interact with the Coding Agent as usual, and Superpowers takes over the workflow in the background.
A: Claude Code (official marketplace), Cursor, OpenAI Codex CLI/App, GitHub Copilot CLI, Gemini CLI, OpenCode.
A: Yes, but you must follow the writing-skills skill guidelines, and you must ensure compatibility across all supported platforms. Official contributions of new skills are generally not accepted directly.
A: Generating code directly easily leads to hallucinations and rework. Superpowers ensures the Agent truly understands the requirements through design documents and plans before starting, reducing the cost of later corrections.
A: No. RED-GREEN-REFACTOR is a mandatory workflow. The Agent deletes any code written before the tests, ensuring test coverage.