Review Crew
zwrose/zwr-agent-skills
Revisão multi-agente de código com especialistas e auto-fix estruturado
Ver no GitHubPainel de revisão especializado com cinco reviewers (arquitetura, código, segurança, testes e failure-modes) que analisam PRs, planos e dívida técnica em paralelo. Calibra-se ao seu projeto via profile, executa auto-fix loops e mede rigorosamente falsos positivos versus achados reais.
Instalação
npx convez-ai zwrose/zwr-agent-skills1 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
- Revisar PRs com perspectivas especializadas (segurança, testes, arquitetura)
- Red-team de planos e specs antes de implementação
- Auditoria periódica de dívida técnica em repositórios
- Auto-fix de issues encontradas mantendo contexto do projeto
Veredito

Skill robusta que vai além de prompts genéricos: calibra-se ao projeto, executa evals determinísticos e gera saídas estruturadas. Ideal para times que querem revisão sistemática com rigor mensurável, mas exige setup inicial e manutenção do profile.
README
A tradução para o português deste README ainda está na fila.
superheroes
Your team of superheroes, powered by superpowers — agent tools for Claude Code.
A marketplace of Claude Code plugins, each a character with a job to do. They team up to take real development work off your plate — reviewing, testing, and (more and more) running the loop themselves. Add the marketplace once, then install whichever heroes you want:
/plugin marketplace add zwrose/superheroes
review-crew
A standing review panel for your code, plans, and tech debt.
Most AI review is one model skimming a diff for "anything wrong?" review-crew is built differently: a panel of five specialist reviewers — architecture, code, security, test, and failure-mode (premortem) — each with its own methodology, running in parallel under a shared severity rubric. An orchestrator compiles their findings, triages each one, and (for code) drives an auto-fix loop that applies the safe fixes and re-reviews until nothing Critical or Important remains.
Two things make it more than a clever prompt:
- Calibrated to your project.
review-initgenerates a.claude/review-profile.md— your threat model, verify command, scope, and canonical patterns — so reviews match your codebase instead of generic best practices. Severity rules, diff-scope discipline, and "citefile:lineor drop the finding" are enforced when findings are compiled, not left to hope. - Measured, not vibes. The reviewer agents ship with a frozen eval harness
(planted findings + decoy traps, a deterministic scorer) and a non-regression
gate: a change has to prove it catches real issues without inflating false
positives before it lands. See
plugins/review-crew/eval/.
It's also context-frugal — the orchestrator never loads the full diff or raw agent output into its own conversation; subagents do the heavy reading and write structured results to disk.
Commands
| Command | Use it to… |
|---|---|
/review-crew:review-init | Generate or refresh a project's review profile (run this first). |
/review-crew:review-code | Review an open PR or local branch and auto-fix what it finds — commits locally, never pushes. |
/review-crew:review-plan | Red-team a draft plan or design spec before any code is written. |
/review-crew:audit-debt | Periodically sweep a whole repo for accumulated debt → a prioritized set of GitHub issues. |
Install & first run
/plugin marketplace add zwrose/superheroes
/plugin install review-crew@superheroes
Then, in any project:
/review-crew:review-init # calibrate to this repo
/review-crew:review-code # review the current branch / PR
test-pilot
Behavioral proof that a change actually works — not just that it compiles.
review-crew reads your code; test-pilot drives your app. It seeds realistic test data, writes a manual test plan onto the PR as a checklist, then — when you ask — pilots that plan in a real browser, fixes the bugs it trips over, and hands you a results comment plus a short spot-check. The goal is a trustworthy "here's it working" before a human ever clicks anything.
Like review-crew, it's calibrated per project (test-pilot-init sets up a
profile, seeding blocks, and browser tooling) so the plans and data fit your app.
Commands
| Command | Use it to… |
|---|---|
/test-pilot:test-pilot-init | Set up (or refresh) a project's testing profile, seed blocks, and browser tooling (run this first). |
/test-pilot:test-pilot-plan | Seed test data for a PR/branch and post a checkbox test plan to the PR. |
/test-pilot:test-pilot-execute | Drive the plan in a real browser, fix what breaks, and post a results comment before your spot-check. |
Install & first run
/plugin marketplace add zwrose/superheroes
/plugin install test-pilot@superheroes
Then, in any project:
/test-pilot:test-pilot-init # calibrate to this app
/test-pilot:test-pilot-plan # seed data + post a plan to the PR
Where this is going
superheroes is growing into a band that runs much of a project's development loop for you. See the roadmap for the phases, and CONVENTIONS.md for the cross-plugin contracts.
Contributing
Issues and pull requests are welcome. Fork the repo, open a PR, and we'll help get it merged. See CONTRIBUTING.md.
License
MIT © Zach Rose