superpowers

Product Introduction

1. Definition: Superpowers is an open-source, agentic skills framework and a comprehensive software development methodology designed for AI-powered coding agents. It is a technical system of composable "skills" and initial instructions that govern an AI agent's workflow.
2. Core Value Proposition: It exists to transform AI coding assistants from simple code generators into systematic, autonomous engineering partners. Its primary value is enforcing a disciplined, test-driven, and plan-driven development lifecycle (SDLC) directly within the agent's decision-making process, leading to higher-quality, more reliable software outputs.

Main Features

1. Composable Skills Library: A modular collection of predefined workflows that activate automatically based on context. These are not suggestions but mandatory procedures. Key technical skills include brainstorming for requirement refinement, writing-plans for granular task breakdown, test-driven-development for enforcing the RED-GREEN-REFACTOR cycle, and subagent-driven-development for orchestrating multi-agent review workflows.
2. Subagent-Driven Development (SDD): A core methodology where the primary agent dispatches fresh, context-limited subagents to execute individual tasks from the implementation plan. How it works: Each subagent's output undergoes a two-stage review—first for spec compliance, then for code quality—ensuring work adheres strictly to the approved design before integration.
3. Git Worktree Integration (using-git-worktrees): A technical feature that creates isolated git worktrees on new branches for each development session. This ensures a clean, separate workspace, runs project setup scripts, and establishes a verified test baseline, preventing environment contamination and providing a clear rollback point.
4. Systematic Debugging & Verification: The framework includes skills like systematic-debugging, which enforces a 4-phase root cause analysis process, and verification-before-completion, which mandates that fixes are empirically proven before a task is considered complete. This moves the agent beyond guesswork to evidence-based problem resolution.

Problems Solved

1. Pain Point: Unstructured and ad-hoc code generation by AI agents, leading to spaghetti code, lack of tests, deviation from requirements, and poor maintainability.
2. Target Audience: Software developers and engineering teams using AI coding assistants (Claude Code, Cursor, GitHub Copilot, etc.) who seek to scale agentic output for serious project work. It is particularly valuable for solo developers, tech leads, and startups aiming to maintain code quality with AI augmentation.
3. Use Cases: Essential for building new features from a vague idea, implementing complex refactoring, systematically debugging elusive issues, and onboarding an AI agent to an existing codebase with enforced testing and review protocols.

Unique Advantages

1. Differentiation: Unlike simple prompt templates or one-off agent instructions, Superpowers is a full, integrated methodology. It differs from other frameworks by its focus on enforcement (skills trigger automatically), its built-in SDLC (from design to merge), and its deep integration with development primitives like git and TDD.
2. Key Innovation: The "agentic skills framework" concept itself—treating development best practices as executable, composable modules that an AI agent can mandatorily invoke. Its integration of git worktrees for isolation and the subagent-driven-development model for self-review are specific technical innovations that enable reliable, autonomous multi-step development.

Frequently Asked Questions (FAQ)

1. What is the Superpowers AI agent framework? Superpowers is an open-source framework that equips AI coding assistants with a structured set of skills and a full software development methodology, enforcing practices like test-driven development, systematic debugging, and planned implementation to improve code quality and agent autonomy.
2. How do you install Superpowers for Claude Code? You can install the Superpowers plugin for Claude Code directly from the official Anthropic marketplace using the command /plugin install superpowers@claude-plugins-official, or from the community Superpowers marketplace.
3. Does Superpowers work with GitHub Copilot? Yes, Superpowers is compatible with GitHub Copilot CLI. You install it by adding the Superpowers marketplace (copilot plugin marketplace add obra/superpowers-marketplace) and then installing the plugin (copilot plugin install superpowers@superpowers-marketplace).
4. What is subagent-driven development in Superpowers? Subagent-driven development (SDD) is a core workflow where the main AI agent breaks work into tasks and dispatches each to a fresh, focused subagent. Each subagent's output is rigorously reviewed for specification compliance and code quality before being accepted, ensuring high-fidelity execution of the plan.
5. Is Superpowers free to use? Yes, Superpowers is open-source software released under the MIT license, making it free to use and modify. The project offers commercial support and services for enterprise users through its maintainer, Prime Radiant.