convez.ai
SkillProdutividadeIntermediário
stefan-stepzero

Shipkit

stefan-stepzero/shipkit

Framework completo para shippar produtos rápido com Claude Code

Ver no GitHub
InicialPluginScore · de 0 a 100

Framework de desenvolvimento de produtos assistido por IA para Claude Code, com 38 skills e 11 personas especializadas. Otimizado para entregar MVPs rapidamente, cobrindo desde descoberta de produto até code review, testes e auditoria de qualidade em um workflow estruturado.

Instalação

$npx convez-ai stefan-stepzero/shipkit

1 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

Estrelas
1
Forks
0
Acompanhando
0
Saúde
57%
Último commit
há 4 meses
Versão
v1.12.0
Issues abertas
0
Contribuidores
2

Casos de uso

  • Orquestração de desenvolvimento de MVPs do zero ao deploy
  • Descoberta de produto, especificação e planejamento técnico automatizados
  • Quality assurance, code review e auditoria de produção antes do ship

Veredito

Shipkit é um sistema robusto e opinado que estrutura todo o ciclo de desenvolvimento de produtos em skills reutilizáveis. Recomendado para times que querem padronizar workflows de MVP com suporte a cada etapa (discovery → planning → execution → qa → shipping), reduzindo decisões ad-hoc.

README

A tradução para o português deste README ainda está na fila.

Shipkit - Streamlined Product Development Framework

A focused collection of Claude Code skills for efficient product development, from discovery to shipped code.

38 skills organized for streamlined workflow:

  • Spec & Planning (5) - spec-roadmap, spec, feedback-bug, ...
  • Knowledge & Memory (2) - claude-md, work-memory
  • Orchestration (3) - orch-direction, orch-planning, orch-shipping
  • Execution (1) - test-cases
  • Quality & Communication (12) - review-direction, review-planning, review-shipping, ...
  • System (3) - update, get-skills, get-mcps

Plus 11 agent personas that specialize behaviors for different workflow stages.


What's Inside

Skills (38 total)

All skills use the shipkit- prefix for clarity.

Core Workflow (4 skills):

  • shipkit-master - Meta skill for workflow orchestration
  • shipkit-project-context - Codebase scanning, stack detection
  • shipkit-codebase-index - Semantic codebase indexing
  • shipkit-claude-md - CLAUDE.md management

Orchestration (3 skills):

  • shipkit-orch-direction - Direction loop — strategic artifacts + coherence review
  • shipkit-orch-planning - Planning loop — definitions/specs + alignment review
  • shipkit-orch-shipping - Shipping loop — implementation + verification + release gate

Discovery & Planning (14 skills):

  • shipkit-why-project - Strategic vision definition
  • shipkit-product-discovery - Personas, journeys, user needs
  • shipkit-product-definition - Product blueprint (features, patterns, differentiators)
  • shipkit-engineering-definition - Engineering blueprint (mechanisms, components, stack)
  • shipkit-design-system - Design system scaffold (principles, tokens, aesthetic direction)
  • shipkit-stage - Project stage, constraints, and graduation criteria
  • shipkit-product-goals - User-outcome success criteria (P-*)
  • shipkit-engineering-goals - Technical performance criteria
  • shipkit-spec-roadmap - Prioritize spec backlog from definitions + goals
  • shipkit-spec - Feature specification
  • shipkit-feedback-bug - Process feedback into investigated bug specs (5 Whys root cause)
  • shipkit-plan - Implementation planning
  • shipkit-metrics - Capture metric values for goal evaluation
  • shipkit-thinking-partner - Think through decisions with cognitive frameworks

Execution (1 skill):

  • shipkit-test-cases - Generate code-anchored test case specs

Quality & Documentation (13 skills):

  • shipkit-review-direction - Assess strategic artifact coherence
  • shipkit-review-planning - Assess planning artifact alignment
  • shipkit-review-shipping - QA and acceptance criteria
  • shipkit-preflight - MVP production readiness audit
  • shipkit-scale-ready - Scale & enterprise readiness audit
  • shipkit-prompt-audit - LLM prompt architecture audit
  • shipkit-codebase-audit - Dead code, orphans, unused deps & unwired seams (knip-class + intent↔code)
  • shipkit-semantic-qa - Semantic QA for API outputs and UI screenshots
  • shipkit-qa-visual - Visual QA with Playwright: UI goals + autonomous test generation
  • shipkit-ux-audit - UX analysis and patterns
  • shipkit-user-instructions - User-facing documentation
  • shipkit-communications - Communication and formatting
  • shipkit-work-memory - Session memory and context

System (3 skills):

  • shipkit-update - Install or update Shipkit from GitHub
  • shipkit-get-skills - Discover and install Claude Code skills
  • shipkit-get-mcps - Discover and install MCP servers

Agent Personas (11)

AgentUsed For
shipkit-orch-master-agentMaster loop orchestrator — sequential dispatch of direction/planning/shipping
shipkit-orch-direction-agentDirection loop orchestrator — strategic artifact dispatch + review cycle
shipkit-orch-planning-agentPlanning loop orchestrator — definition/spec dispatch + review cycle
shipkit-orch-shipping-agentShipping loop orchestrator — implementation/verification dispatch + review cycle

Producers:

AgentUsed For
shipkit-visionary-agentStrategic visionary — sets stage, vision, constraints, business goals
shipkit-product-owner-agentProduct manager — definitions, specs, feedback, product goals
shipkit-architect-agentEngineering manager — architecture, plans, engineering goals
shipkit-thinking-partner-agentCognitive discussion & thinking partner

Reviewers:

AgentUsed For
shipkit-reviewer-direction-agentDirection judgment — strategic coherence assessment
shipkit-reviewer-planning-agentPlanning judgment — definition/spec alignment assessment
shipkit-reviewer-shipping-agentShipping judgment — implementation quality + QA dispatch

Quick Start

Installation

cd your-project
npx shipkit-dev init

Pre-publish note: Until the npm package is published, use: npx github:stefan-stepzero/shipkit init

Add -y for non-interactive mode (uses sensible defaults): npx shipkit-dev init -y

Update an existing installation:

npx shipkit-dev update
curl -O https://raw.githubusercontent.com/stefan-stepzero/shipkit/main/installers/install.py
python install.py --from-github

The installer will:

  • Install all 38 skills
  • Set up 11 agent personas
  • Configure session hooks
  • Create .shipkit/ workspace

After Installation

Your project will have:

your-project/
├── CLAUDE.md                    # User preferences & learnings (editable)
├── .claude/
│   ├── settings.json            # Permissions + hooks
│   ├── rules/
│   │   └── shipkit.md           # Framework rules (managed by /shipkit-update)
│   ├── skills/                  # 38 skill definitions
│   ├── agents/                  # 11 agent personas
│   └── hooks/                   # Session hooks
└── .shipkit/                    # Your workspace
    ├── specs/                   # Feature specifications
    ├── plans/                   # Implementation plans
    └── [context files...]       # Stack, architecture, etc.

Basic Workflow

flowchart LR
    SKL_WHY["/why-project"] --> SKL_DISCOVERY["/product-discovery"]
    SKL_DISCOVERY --> SKL_PRODDEF["/product-definition"]
    SKL_PRODDEF --> SKL_ENGDEF["/engineering-definition"]
    SKL_ENGDEF --> SKL_GOALS["/goals"]
    SKL_GOALS --> SKL_ROADMAP["/spec-roadmap"]
    SKL_ROADMAP --> SKL_SPEC["/spec"]
    SKL_SPEC --> SKL_PLAN["/plan"]
    SKL_PLAN --> IMPL["implement"]
    IMPL --> SKL_VERIFY["/review-shipping"]
    SKL_VERIFY --> SKL_WORKMEM["/work-memory"]

Shortcut: If you already know what to build, skip straight to /shipkit-spec. The discovery chain (why → discovery → product-definition → engineering-definition → goals) is most valuable for new products or when exploring problem space.


How It Works

1. Session Hooks Enforce Skills

When Claude Code starts:

  1. Runs session-start hook
  2. Loads shipkit-master into context
  3. Skills become the primary workflow method

2. Context Lives in .shipkit/

Your project context is stored in:

.shipkit/
├── specs/               # Feature specifications
├── plans/               # Implementation plans
├── architecture.json    # Architecture decisions
├── stack.json           # Technology stack
├── implementations.json   # What's been built
└── engineering-definition.json  # Technical mechanisms & data contracts

3. Skills Chain Together

Skills naturally flow from one to another. Each produces an artifact the next step reads:

flowchart LR
    SKL_WHY["why-project"] --> SKL_DISCOVERY["product-discovery"]
    SKL_DISCOVERY --> SKL_PRODDEF["product-definition"]
    SKL_PRODDEF --> SKL_ENGDEF["engineering-definition"]
    SKL_ENGDEF --> SKL_GOALS["goals"]
    SKL_GOALS --> SKL_ROADMAP["spec-roadmap"]
    SKL_ROADMAP --> SKL_SPECPLAN["spec / plan"]
    SKL_SPECPLAN --> IMPL["implement"]
    IMPL --> SKL_VERIFY["verify"]

Repository Structure

shipkit/
├── cli/                             # npx CLI (zero dependencies)
│   ├── bin/shipkit.js               # Entry point
│   └── src/                         # Commands, prompts, utilities
│
├── install/                         # Everything that gets installed
│   ├── skills/                      # <!-- sync:skill_count -->38<!-- /sync:skill_count --> shipkit-* skill definitions
│   ├── agents/                      # <!-- sync:agent_count -->11<!-- /sync:agent_count --> shipkit-*-agent personas
│   ├── rules/
│   │   └── shipkit.md               # Framework rules (auto-loaded)
│   ├── profiles/
│   │   └── shipkit.manifest.json    # Skill manifest
│   ├── settings/
│   │   └── shipkit.settings.json    # Permissions + hooks
│   ├── claude-md/
│   │   └── shipkit.md               # CLAUDE.md template (user-editable)
│   ├── shared/
│   │   ├── hooks/                   # Session hooks
│   │   └── scripts/                 # Shared utilities
│   └── templates/                   # Queue templates
│
├── installers/                      # Alternative installers
│   ├── install.py                   # Python installer (legacy)
│   └── uninstall.py                 # Uninstaller
│
├── docs/                            # Documentation
│   ├── creating-skills.md           # Contributor guide
│   └── generated/                   # Interactive HTML docs
│       ├── shipkit-overview.html
│       └── orchestration-pipeline.html
│
├── package.json                     # npm package config
└── README.md                        # This file

Key Features

Streamlined Workflow

  • All skills use shipkit- prefix for clarity
  • Context stored in single .shipkit/ folder
  • No complex workspace structure

Specification-Driven

  • Features start with /shipkit-spec
  • Plans reference specs
  • Implementation follows plans
  • Quality checks verify alignment

Session Persistence

  • /shipkit-work-memory maintains context
  • /shipkit-project-context detects your stack
  • Context files persist between sessions

Natural Capabilities

Implementation, debugging, testing, refactoring, and code documentation are natural Claude capabilities that don't need skills. Skills focus on:

  • Human decisions that must be explicit
  • Persistence that survives sessions

Common Workflows

New Feature

/shipkit-spec "Add user authentication"
/shipkit-plan
# implement the feature
/shipkit-review-shipping

Explore Codebase

/shipkit-project-context

Get External Resources

/shipkit-get-skills     # Find Claude Code skills
/shipkit-get-mcps       # Find MCP servers

Updating

npx shipkit-dev update

Or use the shipkit-update skill from within Claude Code:

/shipkit-update

This preserves your .shipkit/ context files and custom settings while updating skill definitions.


License

MIT — See LICENSE for details.


Ready to ship faster?

npx shipkit-dev init

Streamlined product development, guided by AI.

Ver o original no GitHub

Bom para

Relacionados