convez.ai
MCP ServerDesenvolvimento WebIniciante
VersoXBT

Claude Code Starter Kit

VersoXBT/claude-initial-setup

75 skills + 14 agents + 15 comandos para Claude Code pronto em produção

Ver no GitHub
InicialPluginScore · de 0 a 100

Kit plug-and-play com 75 skills, 14 agents e 15 comandos para Claude Code, cobrindo TypeScript, Python, Go, Rust, React, Next.js, FastAPI, Django, Docker, CI/CD e segurança. Inclui configurações compatíveis com Cursor, GitHub Copilot e OpenAI Codex para setup produtivo imediato.

Instalação

$npx convez-ai VersoXBT/claude-initial-setup

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
4
Forks
5
Acompanhando
0
Saúde
42%
Último commit
há 6 meses
Versão
v1.0.0
Issues abertas
0
Contribuidores
2

Casos de uso

  • Configurar Claude Code com melhores práticas e padrões de codificação em até 60 segundos
  • Automatizar code reviews, testes, auditorias de segurança e debug com agents especializados
  • Padronizar desenvolvimento em múltiplas linguagens (TS, Python, Go, Rust) com skills reutilizáveis

Veredito

Excelente acelerador para equipes que usam Claude Code, eliminando setup manual e fornecendo foundation sólido. A cobertura ampla de stacks (web, backend, DevOps) e suporte cross-AI tornam viável para projetos com múltiplas ferramentas.

README

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

Stars License: MIT Version Built for Claude Code Skills Agents Commands

Claude Code Starter Kit

A plug-and-play Claude Code plugin with 75 skills, 14 agents, 15 slash commands, 8 rules, and 6 hooks covering TypeScript, Python, Go, Rust, React, Next.js, FastAPI, Django, Docker, CI/CD, security, testing, and more. Includes cross-AI configuration files for Cursor, GitHub Copilot, and OpenAI Codex.


Table of Contents


Why Use This

Most developers start Claude Code with a blank slate -- no skills, no agents, no coding standards. This plugin gives you a production-ready foundation in under 60 seconds.

  • 75 curated skills across 25 categories so Claude Code knows your stack's best practices out of the box
  • 14 specialized agents for code review, security audits, TDD, debugging, architecture, and more -- ready to spawn as subagents
  • 15 slash commands that map to the workflows developers actually use: /review, /tdd, /debug, /security-check, /plan
  • 8 rules that enforce coding standards (immutability, conventional commits, 80%+ test coverage) without manual configuration
  • 6 hooks for automatic formatting, security checks, and lint reminders on every tool event
  • Cross-AI support -- the same best practices are exported to Cursor, GitHub Copilot, and OpenAI Codex configuration formats

Whether you write TypeScript, Python, Go, Rust, or Java, this starter kit gets Claude Code productive immediately.


Quick Start

Install as a Claude Code Plugin

/plugin marketplace add VersoXBT/claude-initial-setup
/plugin install claude-initial-setup@claude-initial-setup

Install Manually

git clone https://github.com/VersoXBT/claude-initial-setup.git
cd claude-initial-setup
chmod +x install.sh
./install.sh

The install script symlinks all 75 skills into ~/.claude/skills/.

Install a Single Skill

mkdir -p ~/.claude/skills
ln -sfn /path/to/claude-initial-setup/skills/typescript/strict-typescript ~/.claude/skills/strict-typescript

What Is Included

ComponentCountDescription
Skills75SKILL.md files across 25 categories
Agents14Specialized subagents for delegated tasks
Commands15Slash commands for common workflows
Rules8Coding standards and guidelines
Hooks6Automation triggers for tool events

Claude Code Agents

Agents are specialized subagents you can spawn for delegated tasks. Each agent has a defined role, toolset, and process.

AgentPurpose
plannerImplementation planning for complex features
architectSystem design and architectural decisions
code-reviewerCode quality, security, maintainability review
security-reviewerOWASP vulnerabilities, secrets, injection detection
tdd-guideTest-driven development, 80%+ coverage enforcement
debuggerSystematic debugging with hypothesis testing
build-error-resolverFix build/type errors with minimal diffs
refactor-cleanerDead code cleanup and consolidation
doc-updaterDocumentation and README updates
performance-optimizerProfiling, caching, optimization
database-reviewerSchema design, query optimization, migration review
api-designerREST/GraphQL API design and review
devops-engineerDocker, CI/CD, infrastructure
e2e-runnerPlaywright end-to-end testing

Slash Commands

Slash commands for common development workflows.

CommandDescription
/reviewRun code review on recent changes
/tddStart TDD workflow (red-green-refactor)
/debugSystematic debugging process
/build-fixIncrementally fix build errors
/security-checkOWASP security audit
/refactorDead code cleanup and refactoring
/test-coverageAnalyze and improve test coverage
/optimizeProfile and optimize performance
/planCreate implementation plan
/deploy-checkPre-deployment verification
/api-reviewReview API design
/db-reviewReview database schema/queries
/update-docsUpdate documentation
/lint-fixAuto-fix lint and format issues
/checkpointGit commit checkpoint

Skills Catalog

75 skills organized into 25 categories. Click any category to expand.

SkillDescription
commit-conventionsConventional commits format, types, scopes, breaking changes
branching-strategyGit Flow vs trunk-based, branch naming, merge vs rebase
pr-workflowPR templates, review process, CI gates, merge strategies
SkillDescription
immutability-patternsSpread operators, Object.freeze, structuredClone, immer
file-organizationFeature-based vs type-based, barrel exports, colocation
naming-conventionsVariables, functions, classes, files, database columns
SkillDescription
systematic-debuggingBinary search debugging, hypothesis testing, bisect
log-analysisStructured logging, log levels, correlation IDs
error-tracingStack trace reading, source maps, error boundaries
SkillDescription
strict-typescriptstrictNullChecks, noUncheckedIndexedAccess, type narrowing
type-patternsDiscriminated unions, branded types, template literal types
generics-masteryGeneric constraints, conditional types, mapped types, infer
SkillDescription
idiomatic-goError handling, interfaces, struct embedding, package organization
error-handling-goCustom error types, error wrapping, sentinel errors, errors.Is/As
concurrency-patterns-goGoroutines, channels, WaitGroup, context, worker pools
SkillDescription
ownership-borrowingOwnership rules, borrowing, lifetimes, Box/Rc/Arc
error-handling-rustResult/Option, ? operator, thiserror, anyhow
cargo-workflowsWorkspaces, feature flags, build.rs, clippy/fmt
SkillDescription
pythonic-patternsComprehensions, generators, context managers, decorators
type-hints-guideProtocol, TypeVar, ParamSpec, Generic, Literal
project-structure-pythonpyproject.toml, src layout, uv/rye/poetry, virtual environments
SkillDescription
async-fastapiAsync endpoints, background tasks, middleware, WebSockets
pydantic-validationBaseModel, field/model validators, computed fields
dependency-injection-fastapiDepends(), security deps, database sessions
SkillDescription
django-orm-masteryQuerySets, select/prefetch_related, F/Q objects, annotations
drf-api-patternsSerializers, ViewSets, permissions, pagination, filtering
django-admin-customizationModelAdmin, inlines, custom actions, list filters
SkillDescription
component-patternsCompound components, render props, HOCs, composition
hooks-masteryCustom hooks, useCallback, useMemo, useRef, useReducer
nextjs-app-routerServer components, route handlers, middleware, streaming
SkillDescription
middleware-patternsMiddleware chain, auth, rate limiting, CORS, validation
express-error-handlingAsync errors, custom error classes, centralized middleware
rest-api-nodeRESTful conventions, pagination, filtering, versioning
SkillDescription
spring-boot-patternsAnnotations, dependency injection, profiles, actuator
maven-gradle-guidepom.xml, build.gradle.kts, multi-module, dependency management
junit-testingJUnit 5, Mockito, MockMvc, Testcontainers
SkillDescription
dockerfile-best-practicesMulti-stage builds, layer caching, non-root users
docker-compose-guideServices, networks, volumes, health checks, profiles
multi-stage-buildsBuilder pattern, distroless images, cache mounts
SkillDescription
github-actions-guideWorkflow syntax, reusable workflows, matrix strategies
pipeline-patternsPipeline design, artifact management, environment promotion
deploy-strategiesBlue-green, canary, rolling updates, feature flags
SkillDescription
schema-design-guideNormalization, indexing strategy, constraints, naming
query-optimizationEXPLAIN ANALYZE, index tuning, N+1 detection
migration-patternsZero-downtime migrations, data migrations, rollbacks
SkillDescription
owasp-top-10-preventionSQL injection, XSS, CSRF, SSRF, broken auth
secret-managementEnvironment variables, vault patterns, secret rotation
input-validation-guideZod, Pydantic, allowlists, sanitization
SkillDescription
review-checklistWhat to check, severity levels, constructive feedback
automated-review-setupESLint, Prettier, husky/lint-staged, CODEOWNERS
pr-standardsPR size limits, description templates, review SLAs
SkillDescription
profiling-guideChrome DevTools, cProfile, pprof, flame graphs
caching-strategiesRedis, CDN, browser cache, stale-while-revalidate
lazy-loading-patternsCode splitting, dynamic imports, React.lazy, Suspense
SkillDescription
anthropic-sdk-guidePython/TS SDK, messages API, streaming, prompt caching
prompt-engineeringSystem prompts, few-shot, chain-of-thought, structured output
tool-use-patternsTool schemas, multi-tool orchestration, parallel tool use
SkillDescription
mcp-server-creationServer scaffolding, transport, lifecycle hooks, testing
mcp-tool-designNaming, input schemas, error responses, pagination
mcp-resource-patternsResource URIs, templates, subscriptions, dynamic resources
SkillDescription
multi-agent-orchestrationOrchestrator, pipeline, consensus, delegation patterns
context-managementContext window optimization, summarization, RAG, pruning
agent-communicationMessage passing, shared state, event-driven, pub/sub
SkillDescription
pre-post-hooksPreToolUse, PostToolUse, Stop hooks, matcher patterns
automation-triggersAuto-format, auto-lint, auto-test, file-change reactions
hook-recipesReady-to-use recipes: prettier, tsc, secret leak prevention
SkillDescription
subagent-patternsWhen to use subagents, types, result handling
parallel-task-executionFan-out/fan-in, parallel tool calls, background agents
task-coordinationTask dependencies, owner assignment, status tracking
SkillDescription
memory-patternsMEMORY.md structure, topic files, save vs skip decisions
context-optimizationToken budgeting, progressive disclosure, subagent firewalls
knowledge-persistenceLearning from corrections, pattern extraction
SkillDescription
tdd-workflowRed-green-refactor, test-first, 80%+ coverage targets
test-architectureTest organization, fixtures, factories, isolation
mocking-strategiesMocks vs stubs vs spies, DI for testing, when NOT to mock

Rules

Rules enforce coding standards across your entire project. They are loaded automatically by Claude Code.

RuleContent
coding-styleImmutability, file organization, naming, small functions
git-workflowConventional commits, branch naming, PR process
testingTDD, 80% coverage, test types, isolation
securitySecrets, input validation, injection prevention
performanceProfile first, caching, lazy loading
api-designREST conventions, error format, pagination
databaseNormalization, indexing, migrations, N+1 prevention
agentsAgent orchestration, when to use which agent

Hooks

Automation triggers that run on Claude Code tool events. All hooks are non-blocking (warnings only) since this is a starter kit. Edit hooks/hooks.json to customize behavior.

  • Tmux reminder for long-running commands
  • Review reminder before git push
  • Package install security check
  • Prettier reminder after JS/TS edits
  • Console.log warning on edited files
  • TypeScript check reminder

Cross-AI Compatibility

This plugin includes configuration files for other AI coding assistants, so the same best practices apply regardless of which tool you use.

FileAI ToolDescription
AGENTS.mdOpenAI CodexDevelopment conventions and agent context
.github/copilot-instructions.mdGitHub CopilotShort instruction statements
.cursor/rules/claude-initial-setup.mdcCursor (modern)MDC rules with metadata
.cursorrulesCursor (legacy)Plain markdown rules

Repository Structure

claude-initial-setup/
  .claude-plugin/
    plugin.json              # Plugin manifest
    marketplace.json         # Marketplace metadata
  agents/                    # 14 specialized subagents
  commands/                  # 15 slash commands
  hooks/
    hooks.json               # Automation triggers
  rules/                     # 8 coding standards
  skills/                    # 75 skills across 25 categories
    <category>/
      <skill-name>/
        SKILL.md
  .github/
    copilot-instructions.md  # GitHub Copilot
  .cursor/
    rules/
      claude-initial-setup.mdc  # Cursor
  AGENTS.md                  # OpenAI Codex
  .cursorrules               # Cursor (legacy)
  install.sh                 # Symlink installer
  uninstall.sh               # Symlink remover

Customization

Modifying Skills

Edit any SKILL.md directly. Changes take effect in the next Claude Code session.

Adding Reference Files

For skills that need deeper content, add a references/ subdirectory:

skills/typescript/strict-typescript/
  SKILL.md
  references/
    compiler-options.md
    migration-guide.md

Creating New Skills

  1. Create a directory under the appropriate category in skills/
  2. Add a SKILL.md with YAML frontmatter (name and description are required fields)
  3. Run ./install.sh to symlink the new skill into ~/.claude/skills/

Disabling a Skill

Remove its symlink without deleting the source file:

rm ~/.claude/skills/<skill-name>

Uninstall

Remove all symlinked skills:

./uninstall.sh

Or remove individual skills:

rm ~/.claude/skills/<skill-name>

Contributing

Contributions are welcome. To add a new skill, agent, or command:

  1. Fork the repository
  2. Create a feature branch (git checkout -b feat/new-skill-name)
  3. Add your files following the existing directory structure
  4. Ensure any new SKILL.md has valid YAML frontmatter with name and description
  5. Test by running ./install.sh and verifying the skill loads in Claude Code
  6. Open a pull request with a clear description of what the skill covers and why it is useful

For bug reports or feature requests, open an issue on the GitHub Issues page.


License

MIT

Contributors

Ver o original no GitHub

Bom para

Relacionados