/

Superpowers Docs

A complete software development methodology for your coding agents, built on composable skills that trigger automatically when you need them.

Superpowers is a complete software development methodology for your coding agents, built on a set of composable skills and initial instructions that make sure your agent uses them.

The core idea is simple: your coding agent doesn’t just jump into writing code. Instead, it follows a structured workflow from idea to shipping that emphasizes test-driven development, systematic debugging, and evidence over claims.

What it is

Superpowers is a collection of Markdown skills, shell hooks, and workflow conventions. Each skill gives your agent a specific behavior that triggers automatically at the right moment.

SkillCategoryPurpose
brainstormingCollaborationRefine rough ideas through questions before writing code
using-git-worktreesCollaborationCreate isolated workspace on new branch
writing-plansCollaborationBreak work into bite-sized tasks with exact file paths
subagent-driven-developmentCollaborationFast iteration with two-stage review
executing-plansCollaborationBatch execution with human checkpoints
requesting-code-reviewCollaborationReview against plan before proceeding
receiving-code-reviewCollaborationRespond to feedback systematically
finishing-a-development-branchCollaborationMerge/PR decision workflow
dispatching-parallel-agentsCollaborationConcurrent subagent workflows
test-driven-developmentTestingRED-GREEN-REFACTOR cycle with testing anti-patterns reference
verification-before-completionTestingEnsure it’s actually fixed
systematic-debuggingDebugging4-phase root cause process
using-superpowersMetaIntroduction to the skills system
writing-skillsMetaCreate new skills following best practices

Who it is for

  • Developers who want structured agent workflows instead of blank prompts.
  • Teams running multiple AI coding sessions in parallel and needing a common process.
  • Engineers who want test-driven development, systematic debugging, and review discipline around agent-written code.
  • Anyone who has watched an agent confidently write broken code and wants a safety net.

What you can do with it

  • Think through a product idea before implementation with Socratic questioning.
  • Turn a vague feature request into a reviewed, tested implementation plan.
  • Run autonomous subagent workflows that can work for hours without deviating.
  • Enforce true red/green TDD: write failing test, watch it fail, write minimal code, watch it pass.
  • Debug systematically with 4-phase root cause analysis instead of guessing.
  • Review code against the original plan with severity-based issue reporting.
  • Ship with clean worktrees and proper merge/PR discipline.

Start here

If you are new to Superpowers, follow the shortest useful path:

  1. Read the Quick start to see the core workflow in action.
  2. Install Superpowers with the Install guide for your platform.
  3. Learn the mental model in How it works.
  4. Use the Skills reference when you need to understand a specific skill.
  5. Browse the Skills guide for a tour of available skills.