/quick-start

Quick start

Get Superpowers running and complete your first workflow in under 10 minutes.

The fastest way to understand Superpowers is to complete one full workflow from idea to tested code. This guide takes you through a 10-minute example.

Before you start

You need a coding agent that supports Superpowers:

  • Claude Code
  • OpenAI Codex CLI or App
  • Cursor
  • OpenCode
  • GitHub Copilot CLI
  • Gemini CLI

See Install for platform-specific setup.

Try the core workflow

Superpowers triggers automatically when it detects you’re starting work. The typical first experience looks like this:

1. Start with an idea

Open your project and tell your agent what you want to build:

I want to add a feature that lets users export their data as CSV

2. Let brainstorming run

Instead of immediately coding, the agent activates the brainstorming skill. It will:

  • Ask clarifying questions about the export format
  • Explore edge cases (what about empty datasets? large files?)
  • Present a design in short chunks you can validate
  • Save the approved design to a design document

This is the first Superpowers skill firing. You don’t trigger it manually—it detects that you’re starting work and have no approved design yet.

3. Approve the design

Review the design chunks. When you’re satisfied, tell the agent:

This design looks good. Let's proceed.

4. Watch the worktree skill activate

The using-git-worktrees skill automatically creates an isolated workspace on a new branch and runs your project setup. The agent verifies a clean test baseline before proceeding.

5. See the implementation plan

The writing-plans skill breaks the work into bite-sized tasks (2-5 minutes each). Every task includes:

  • Exact file paths
  • Complete code to implement
  • Verification steps

Review the plan. Approve it to continue.

6. Observe subagent-driven development

The subagent-driven-development skill dispatches fresh subagents for each task with two-stage review:

  1. Spec compliance check
  2. Code quality review

Critical issues block progress until fixed.

7. Watch TDD in action

The test-driven-development skill enforces RED-GREEN-REFACTOR:

  • Write failing test first
  • Watch it fail
  • Write minimal code to pass
  • Watch it pass
  • Commit

If the agent tries to write code before tests, the skill will make it delete that code.

8. Get code review

Between tasks, the requesting-code-review skill reviews work against the plan and reports issues by severity.

9. Finish the branch

When complete, the finishing-a-development-branch skill:

  • Verifies tests pass
  • Presents options: merge, PR, keep, or discard
  • Cleans up the worktree

What success looks like

After this workflow, you will have:

  • A written design document saved in your project
  • Clean git history with isolated branch
  • Tested implementation following TDD
  • Code reviewed against the original plan
  • A merge or PR ready for submission

The agent worked autonomously for most of this, checking in at decision points you controlled.

Next steps

  • Read How it works to understand the philosophy behind the workflow.
  • Browse the Skills reference to see all available skills.
  • Check Install for platform-specific installation details.