builder-product — AI Product Skills Pack
RBraga01/builder-product
Skills e agents que enforçam qualidade de produto antes do engineering
Ver no GitHubPacote de 6 skills e 3 agents que garantem qualidade em trabalho com produtos assistidos por IA, desde definição de problemas e métricas até validação de features antes da aprovação. Enfatiza gates de qualidade em PRDs, scoping, pesquisa sintetizada em decisões e validação de hallucinations em features de IA.
Instalação
npx convez-ai RBraga01/builder-product1 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
- Validar PRDs e specs de features contra gates de qualidade
- Desenhar frameworks de métricas com baselines verificados
- Sintetizar pesquisa de usuários em cadeias de insight→decisão
- Validar features de IA para hallucinations e confiança do usuário
- Escrever A/B tests com hipóteses, amostras e regras de parada
Veredito

Conjunto maduro e bem estruturado focado em disciplina de produto — excelente para times que já trabalham com Claude e querem enforçar gates de qualidade antes de engineering. A integração com o ecossistema builder-* amplifica seu valor em projetos maiores.
README
A tradução para o português deste README ainda está na fila.
builder-product — AI Product Skills Pack v1.0.0
Skills and agents that enforce product quality before engineering time is committed: problems defined before scopes, metrics instrumented before builds, research synthesised into decisions before planning, and AI features validated for hallucination UX, trust calibration, and reversibility before approval.
Works standalone or alongside A Team, builder-ai, builder-design, and builder-growth.
Skills
| Skill | What It Enforces |
|---|---|
prd-quality-gate | User problem + success metric + scope boundary + anti-goals before any estimate |
feature-scoping | Written scope before estimate; estimate before commit; no open-ended "we'll see" specs |
metric-definition | North star, guardrail, and diagnostic metrics with baselines before build begins |
user-research-synthesis | Insight → decision chain documented; no "users said X" without "therefore we will Y" |
ab-test-design | Hypothesis, metric, sample size, duration, stopping rule, decision rule before any test runs |
ai-feature-validation | Hallucination UX, trust calibration, capability scope, and reversibility in every AI PRD |
Agents
| Agent | Role |
|---|---|
product-critic (Sonnet) | Reviews PRDs and feature specs against all skill gates; PASS / CONDITIONAL / BLOCK |
metric-designer (Sonnet) | Designs measurement frameworks with verified baselines before features are built |
research-synthesiser (Opus) | Synthesises user research into an insight → decision chain |
Install
# bash (macOS / Linux / WSL)
bash <(curl -fsSL https://raw.githubusercontent.com/RBraga01/builder-product/master/install.sh)
# PowerShell (Windows)
irm https://raw.githubusercontent.com/RBraga01/builder-product/master/install.ps1 | iex
Or clone directly:
git clone https://github.com/RBraga01/builder-product
cp -rn builder-product/skills your-project/
cp -rn builder-product/.claude your-project/
The builder-* Ecosystem
| Pack | Domain | Skills | Agents |
|---|---|---|---|
| A Team | Engineering baseline | 18 | 25 |
| builder-ai | LLM engineering | 8 | 5 |
| builder-design | AI UI design | 8 | 5 |
| builder-product | Product quality | 6 | 3 |
| builder-growth | Growth & messaging | 6 | 3 |
All packs share the same enforcement model: Completion Statement Formats that require real values, not summaries.
License
MIT — Ricardo Romão Marques Braga