Lance's AI Skills & Plugins
lancewillett/ai-plugins
Marketplace de skills reutilizáveis para Claude e Codex
Ver no GitHubMarketplace de skills reutilizáveis para Claude Code e Codex, contendo ferramentas práticas para promover documentação em specs, auditar performance de JS, capturar screenshots de UI e manter configurações limpas. Funciona como um repositório centralizado de comandos e automações que potenciam fluxos de desenvolvimento.
Instalação
npx convez-ai lancewillett/ai-plugins1 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
- Converter notas markdown privadas em specs compartilháveis no GitHub
- Auditar e otimizar performance de aplicações JavaScript
- Capturar screenshots before/after de mudanças visuais em PRs
- Manter e limpar configurações de Claude Code e Codex
Veredito

Coleção prática de skills focada em dev workflow real: documenting, auditing e visual validation. Útil para quem trabalha com Claude Code/Codex e precisa automatizar tarefas repetitivas de desenvolvimento, sem overhead de instalação complexa.
README
A tradução para o português deste README ainda está na fila.
Lance's AI plugins
Public skills, commands, and scripts wrapped into a simple plugin marketplace that works with both Claude Code and Codex.
Install
Claude Code
Add this marketplace once:
/plugin marketplace add https://github.com/lancewillett/ai-plugins.git
Then install any plugin by name:
/plugin install kb-to-spec@ai-plugins
Codex
Add this marketplace once:
codex plugin marketplace add https://github.com/lancewillett/ai-plugins.git
Then install any plugin by name:
codex plugin install kb-to-spec@ai-plugins
Skills
kb-to-spec
Promote a private Markdown note into a shareable spec committed to a GitHub repo. It de-privatizes the content, writes a standalone spec with an Authors header, links back from the private note, and keeps the two files independent, with no duplicated content. Use it when publishing a plan as a shared design doc, committing notes into a project repo, or when you say "promote this to a spec", "turn this note into a repo spec", or "make this shareable".
js-perf-audit
Audit any JS-heavy codebase for performance issues. It checks bundle weight, dependency hygiene, code splitting, rendering architecture, runtime re-render patterns, and performance infrastructure, then writes a findings report with severity ratings and prioritized recommendations. Use it when you want to assess or improve a JavaScript app's performance, or when you say "audit this codebase for performance", "check our bundle size", or "why is this app slow".
visual-screenshots
Capture before/after screenshots for PRs with visual UI changes using Playwright. It reads the repo's dev-server setup, captures matched before/after shots at the right viewports, and can upload them straight into the PR comment as inline images (works on github.com and GitHub Enterprise). Use it when a PR changes CSS, components, layouts, or themes, or when you say "screenshot this change" or "add before/after screenshots to the PR". Requires the Playwright MCP server.
claude-cleanup
Audit and clean up your Claude Code configuration. It checks skills, hooks, settings, plugins, configs, commands, and stale data for broken references, outdated paths, version drift, and orphaned files, then fixes them interactively (or previews with --dry-run). Use it for periodic maintenance, after a folder reorganization, or when you say "clean up my Claude config" or "audit my skills and plugins".
Contributing
This repo is for personal use, but pull requests that improve or generalize existing skills are welcome. New skills should be general-purpose and not private or environment-specific.
License
MIT: see LICENSE.