qwwiwi/public-architecture-claude-code · Archived

superpowers

Agentic skills framework (pcvelz fork): TDD, debugging, planning, code-review, git-worktrees, parallel agents. 15 skills with .tasks.json persistence between sessions.

First seen May 29, 2026

Installation

$ npx skills add qwwiwi/public-architecture-claude-code --skill superpowers

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from qwwiwi/public-architecture-claude-code.

npx skills add qwwiwi/public-architecture-claude-code

Browse all from qwwiwi/public-architecture-claude-code

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Declared
Cursor Declared
Codex Declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Declared

Repository health

Stars 6
License LICENSE
Default branch main
Open issues 0
Status Archived

Skill metadata

Parsed from SKILL.md frontmatter.

Declared agents claude-code cursor codex opencode

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,461 B
  • docs SUMMARY.md 186 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 2 installs

SKILL.md

Superpowers (pcvelz)

Superpowers -- extended fork of obra/superpowers for Claude Code. 15 built-in skills that auto-activate by development context. Key advantage: .tasks.json persistence -- plans survive between sessions.

Install

claude plugins marketplace add pcvelz/superpowers
claude plugins install superpowers@superpowers-marketplace

Skills Included (15)

Testing & Quality

  • test-driven-development -- RED-GREEN-REFACTOR cycle
  • verification-before-completion -- verify fix actually works before claiming done
  • systematic-debugging -- 4-phase root cause analysis

Planning

  • brainstorming -- 9-step design-first flow, hard gate before code
  • writing-plans -- plans with .tasks.json (goal, files, acceptanceCriteria, verifyCommand)
  • executing-plans -- loads .tasks.json, continues from last completed task

Collaboration

  • requesting-code-review -- pre-review checklist
  • receiving-code-review -- implement review feedback with rigor
  • dispatching-parallel-agents -- concurrent subagent workflows
  • subagent-driven-development -- two-stage review (spec compliance + code quality)

Git

  • using-git-worktrees -- isolated development branches
  • finishing-a-development-branch -- merge/PR decisions

Meta

  • writing-skills -- TDD framework for creating new skills
  • using-superpowers -- meta-skill, auto-invocation rules
  • shared -- shared utilities for other skills

Key Differences from obra/superpowers

  • .tasks.json persistence -- plans survive session restarts
  • Pre-commit hook -- blocks commit if tasks incomplete
  • Task metadata: verifyCommand, acceptanceCriteria per task
  • PlanMode explicitly banned (EnterPlanMode/ExitPlanMode)
  • Multi-IDE: Claude Code + Codex + Cursor + OpenCode
  • Built-in code-reviewer subagent in agents/

Integration

Works as a plugin layer on top of CLAUDE.md memory architecture. Does not replace existing memory/rules -- complements them with proven development workflows that activate automatically.

Reference