Reasonix Superpowers
SunNull/reasonix-superpowers
Metodologia completa de dev para agent Reasonix com TDD, code review e subagents
Ver no GitHubPack de 14 skills de metodologia de desenvolvimento adaptado do Superpowers original para o agent Reasonix. Implementa fluxo estruturado: brainstorm → planejamento → TDD com subagents → code review → ship, com enforcement total de práticas de qualidade e debugging sistemático.
Instalação
npx convez-ai SunNull/reasonix-superpowers1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Estruturar desenvolvimento com TDD obrigatório em projetos Reasonix
- Implementar code review automático e verificação de qualidade antes de merge
- Orquestrar múltiplos subagents paralelos para execução de tarefas independentes
- Debugging sistemático e investigação de root cause em falhas de teste
Veredito

Solução madura e bem-estruturada para impor disciplina metodológica em projetos com Reasonix. Exigente em termos de conformidade, mas oferece proteção real contra código de baixa qualidade e processos ad-hoc. Recomendado para equipes que valorizam reprodutibilidade e rigor.
README
A tradução para o português deste README ainda está na fila.
Superpowers for Reasonix
A complete software development methodology for your Reasonix coding agent — adapted from obra/superpowers (228k⭐).
English · 简体中文
What This Is
Superpowers is an agentic skills framework and software development methodology originally built for Claude Code. It enforces structured workflows: brainstorm before code → write a plan → execute via subagents with TDD → review → ship.
This repo adapts its 14 core skills to Reasonix's skill system — preserving the full methodology while mapping every Claude Code tool reference to its Reasonix equivalent.
Skills
🧭 Core Workflow
| Skill | When it triggers | What it does |
|---|---|---|
using-superpowers | Start of every conversation | Bootstrap — makes all other skills auto-trigger |
brainstorming | Before any creative work | Socratic design refinement; no code until design is approved |
writing-plans | After design approval | Break work into 2-5 min TDD tasks with exact file paths and code |
subagent-driven-development | When executing a plan | Fresh subagent per task + two-stage review (spec → quality) |
executing-plans | Alternative to above | Batch execution with human checkpoints |
finishing-a-development-branch | When tasks complete | Verify tests → present merge/PR/keep/discard options |
🔬 Quality & Debugging
| Skill | When it triggers | What it does |
|---|---|---|
test-driven-development | Before writing implementation code | RED-GREEN-REFACTOR enforcement, no exceptions |
systematic-debugging | On any bug, test failure, unexpected behavior | 4-phase root cause investigation; no guessing |
verification-before-completion | Before claiming work is done | Evidence before claims — run it, read it, then say it |
requesting-code-review | Between tasks / before merge | Dispatch code reviewer subagent with crafted context |
receiving-code-review | When receiving feedback | Technical evaluation, not performative agreement |
🛠 Meta & Collaboration
| Skill | When it triggers | What it does |
|---|---|---|
using-git-worktrees | Before feature work needing isolation | Ensure isolated workspace via git worktree |
dispatching-parallel-agents | 2+ independent problems | Concurrent subagent dispatch via run_in_background |
writing-skills | Creating/editing skills | TDD applied to process documentation |
Installation
These are general-purpose skills — a development methodology, not tied to any specific project. Install GLOBALLY (recommended) so every project benefits. Project-scope install is supported but optional (use it only when you want a project-specific skill set or version pin).
Option 1: Install via Reasonix (recommended — global)
In a Reasonix session, run:
install skills from https://github.com/SunNull/reasonix-superpowers
Or use the install-capability skill:
/install-capability https://github.com/SunNull/reasonix-superpowers
To ensure a global install, run the command from your home directory (no project workspace open), or pass scope=global if your Reasonix version supports it.
Option 2: Manual clone (global — recommended)
# GLOBAL (recommended) — all projects can use these skills
git clone https://github.com/SunNull/reasonix-superpowers.git \
~/.reasonix/skills/superpowers
# Project-scope (optional) — this project only
git clone https://github.com/SunNull/reasonix-superpowers.git \
.reasonix/skills/superpowers
Reasonix auto-discovers skills under ~/.reasonix/skills/ (global) and .reasonix/skills/ (project). Each <name>/SKILL.md is picked up automatically; references/*.md files are auto-appended to the skill body. Global skills apply to every project; project skills override globals on name collision.
Verify installation
After install, restart your Reasonix session and check:
/skill list
You should see all 14 skills in the index.
The Workflow
User: "Let's build X"
│
▼
brainstorming ──── design doc saved to docs/specs/
│
▼
writing-plans ───── plan saved to docs/plans/
│
▼
subagent-driven-development
├── Task 1: implementer → spec-reviewer → code-quality-reviewer
├── Task 2: implementer → spec-reviewer → code-quality-reviewer
└── ...
│
▼
finishing-a-development-branch ──── merge / PR / keep / discard
Along the way:
- test-driven-development fires on every implementation step
- systematic-debugging fires when a test fails or a bug appears
- verification-before-completion fires before any "done" claim
Key Adaptations from Claude Code
| Claude Code (original) | Reasonix (this pack) |
|---|---|
Task tool (dispatch subagent) | task tool with prompt parameter |
Skill tool (invoke skill) | run_skill tool |
TodoWrite | todo_write |
Read / Write / Edit / Bash | read_file / write_file / edit_file / bash |
EnterPlanMode tool | Present plan as text; user approves before changes |
EnterWorktree tool | git worktree add (no native equivalent in Reasonix) |
@file.md inline references | references/*.md auto-appended to skill body |
superpowers:skill-name cross-refs | run_skill({name: "skill-name"}) |
| Visual Companion (HTTP server) | ASCII diagrams + ask tool for structured choices |
Parallel Task() calls | task with run_in_background: true + wait |
Full mapping: using-superpowers/references/reasonix-tools.md
Reasonix-Specific Enhancements
continue_fromfor iterative review loops: Reasonix'stasktool supports session continuation (continue_from: "sa_...") — when a reviewer finds issues, resume the implementer's session instead of starting fresh. More efficient than the original Claude Code workflow.run_in_backgroundfor parallel agents: Dispatch multiple independent subagents concurrently, collect results viawait.asktool for structured choices: Replaces open-ended questions with multiple-choice (better UX in the terminal).
File Structure
reasonix-superpowers/
├── brainstorming/
│ ├── SKILL.md
│ └── references/
│ └── spec-document-reviewer-prompt.md
├── subagent-driven-development/
│ ├── SKILL.md
│ └── references/
│ ├── implementer-prompt.md
│ ├── spec-reviewer-prompt.md
│ └── code-quality-reviewer-prompt.md
├── systematic-debugging/
│ ├── SKILL.md
│ └── references/
│ ├── root-cause-tracing.md
│ ├── defense-in-depth.md
│ ├── condition-based-waiting.md
│ ├── condition-based-waiting-example.ts
│ └── find-polluter.sh
├── test-driven-development/
│ ├── SKILL.md
│ └── references/
│ └── testing-anti-patterns.md
├── ... (10 more skills)
├── using-superpowers/
│ ├── SKILL.md
│ └── references/
│ └── reasonix-tools.md ← tool mapping reference
├── README.md
├── README.zh-CN.md
└── LICENSE
Credits
- Original Superpowers by Jesse Vincent and the Prime Radiant team. MIT License.
- Adapted for Reasonix.
License
MIT