obra/superpowers
An agentic skills framework & software development methodology that works.
Deep Analysis
Complete AI programming agent workflow system that achieves hours of autonomous development through structured processes (brainstorming → planning → TDD → subagent execution → code review).
Core Features
Don't rush to write code; first understand real requirements through questioning, display design in chunks for confirmation
Each task is assigned to an independent subagent with two-stage review (spec compliance + code quality)
Enforces RED-GREEN-REFACTOR: write failing test first, then minimal code, then refactor
Each feature develops in an isolated worktree, merge or create PR after safe verification
Technical Implementation
- Autonomous work for hours: Structured process keeps Claude on track
- Two-stage code review: First check spec compliance, then check code quality
- Strict TDD: Write failing test first, then implement; delete code written before tests
- Cross-platform support: Works with Claude Code, Codex, and OpenCode
- From vague idea to clear design: Brainstorming skill guides requirement clarification
- Long-term autonomous development: Subagent-driven process enables hours of unattended work
- High-quality code delivery: TDD + two-stage review ensures code quality
- Safe feature iteration: Git worktree isolation + choose merge strategy upon completion
- Learning curve: Need to understand and follow the workflow
- Process enforcement: Skills are mandatory processes, not suggestions
- Subagent dependency: Requires Claude Code subagent support
Superpowers
Superpowers is a complete software development workflow for your coding agents, built on top of a set of composable "skills" and some initial instructions that make sure your agent uses them.
How it works
It starts from the moment you fire up your coding agent. As soon as it sees that you're building something, it doesn't just jump into trying to write code. Instead, it steps back and asks you what you're really trying to do.
Once it's teased a spec out of the conversation, it shows it to you in chunks short enough to actually read and digest.
After you've signed off on the design, your agent puts together an implementation plan that's clear enough for an enthusiastic junior engineer to follow. It emphasizes true red/green TDD, YAGNI, and DRY.
Next up, once you say "go", it launches a subagent-driven-development process, having agents work through each engineering task, inspecting and reviewing their work.
Installation
Claude Code (via Plugin Marketplace)
/plugin marketplace add obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplace
The Basic Workflow
- brainstorming - Refines rough ideas through questions
- using-git-worktrees - Creates isolated workspace on new branch
- writing-plans - Breaks work into bite-sized tasks (2-5 minutes each)
- subagent-driven-development - Dispatches fresh subagent per task
- test-driven-development - Enforces RED-GREEN-REFACTOR
- requesting-code-review - Reviews against plan
- finishing-a-development-branch - Verifies tests, presents options
Related Skills
obra/superpowers-lab
obraExperimental skills for Claude Code Superpowers - new techniques and tools
Innovative experimental plugin library addressing real Claude Code pain points.
obra/superpowers-skills
obraCommunity-editable skills for Claude Code's superpowers plugin
obra/superpowers-marketplace
obraCurated Claude Code plugin marketplace
obra/the-elements-of-style
obraWilliam Strunk Jr.'s Elements of Style (1918) in markdown format for AI agents
obra/superpowers-chrome
obraClaude Code plugin for direct Chrome browser control via DevTools Protocol - zero dependencies
obra/double-shot-latte
obraClaude Code plugin that automatically evaluates whether Claude should continue working instead of stopping prematurely