Skills — Biblioteca de 722 Instruções Reutilizáveis para Agentes
mouadja02/skills
722 skills modulares para potencializar agentes de código e assistentes
Ver no GitHubColeção massiva de skills estruturadas em 31 categorias para Claude Code, Cursor e outros clientes. Cada skill é um pacote de instrução reutilizável com frontmatter YAML que cobre engenharia, design, DevOps, estratégia empresarial, dados e muito mais. Instalável via script ou diretamente nos diretórios de configuração dos clientes.
Instalação
npx convez-ai mouadja02/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
- Expandir capacidades de Claude Code com instruções especializadas em TDD, code review e qualidade
- Implementar workflows DevOps, CI/CD e infraestrutura como código através de skills pré-estruturadas
- Adicionar expertise em design UI, estratégia empresarial e engenharia de contexto aos agentes
Veredito

Biblioteca extensível e bem organizada que resolve um problema real: compartilhar instruções estruturadas entre ferramentas de agentes. A escala (722 skills) é impressionante, embora a qualidade e manutenção de cada skill exija verificação individual. Recomendado para quem busca acelerar setup de agentes com instruções testadas.
README
A tradução para o português deste README ainda está na fila.
skills
A curated collection of 722 Agent Skills across 31 categories for Claude Code, Cursor, and other clients that discover SKILL.md files recursively.
Each skill is a reusable instruction package with YAML frontmatter and an agent-readable body. Category folders keep the repository browsable without changing recursive skill discovery.
Browse the searchable site at mouadja02.github.io/skills, the
complete generated Markdown index at SKILLS.md, or the focused category indexes under
skills/.
Quick Start
Install a single skill with the provided scripts.
curl -fsSL https://raw.githubusercontent.com/mouadja02/skills/main/install.sh \
| bash -s -- engineering-craft/test-driven-development \
-d ~/.claude/skills
$content = irm https://raw.githubusercontent.com/mouadja02/skills/main/install.ps1
iex $content
Install-Skill engineering-craft/test-driven-development -Dest $HOME\.claude\skills
Selectors may be an exact install path, a category, a quoted glob, or --all / -All. Use
--dry-run / -DryRun to preview an installation.
Skills may also be installed from the Pages ZIP artifacts:
https://mouadja02.github.io/skills/zips/skill/<install-path>.zip
https://mouadja02.github.io/skills/zips/category/<category>.zip
https://mouadja02.github.io/skills/zips/all.zip
Client Paths
| Client | Project path | User path |
|---|---|---|
| Claude Code | .claude/skills/ | ~/.claude/skills/ |
| Cursor | .cursor/skills/ | ~/.cursor/skills/ |
Categories
This table is generated from docs/manifest.json.
| Category | Skills | Scope |
|---|---|---|
agent-design | 58 | Agent architecture, orchestration, harnesses, safety, scaffolding, and coding-agent interfaces. |
agent-eval | 24 | Agent evaluation, RAG evaluation, memory, autoresearch, benchmarking, and lifecycle improvement. |
api-backend | 11 | API design, backend implementation, OpenAPI, TypeSpec, FastAPI, and integrations. |
business-strategy | 63 | Executive advisory, board preparation, operating systems, and strategic decision support. |
cloud-azure | 18 | Azure, AWS, cloud architecture, IoT, pricing, deployment, and operations. |
code-quality | 17 | Code review, refactoring, static analysis, security review, and integrity checks. |
coding | 33 | Language-agnostic implementation workflows, planning, debugging, security, and shipping. |
context-engineering | 17 | Context design, compression, evolving memory, provenance, and codebase knowledge acquisition. |
databases | 22 | Database design, SQL optimization, migrations, analytics, Snowflake, PostgreSQL, and dbt. |
design-and-ui | 31 | Frontend design, UI systems, visual artifacts, animation, branding, and accessibility. |
dev-workflow | 30 | Git, GitHub, CLI tooling, release workflows, local automation, and developer productivity. |
devops | 38 | CI/CD, containers, infrastructure as code, Linux operations, observability, and security. |
diagrams-slides | 14 | Diagrams, presentations, meeting artifacts, and professional visual communication. |
documentation | 28 | READMEs, ADRs, project documentation, Markdown tooling, conversion, and publishing. |
dotnet | 19 | .NET, C#, WinUI, MVVM, NuGet, testing, and VS Code extension development. |
engineering-craft | 37 | Senior engineering practices, planning, mentoring, verification, and cross-cutting craft. |
go-to-market | 11 | Launch planning, positioning, pricing, partnerships, enterprise sales, and PLG. |
java-kotlin | 11 | Java, Kotlin, Spring Boot, testing, refactoring, and migration workflows. |
llm-tooling | 46 | LLM observability, evaluation, serving, vector search, OpenRouter, Phoenix, Arize, Qdrant, and vLLM. |
marketing-and-growth | 26 | Marketing strategy, content, acquisition, SEO, CRO, and lifecycle growth. |
mcp | 13 | Model Context Protocol server generation, tooling, deployment, and security. |
messaging | 3 | Messaging integrations and relay workflows. |
microsoft-agents | 11 | Microsoft Copilot agents, declarative agents, Foundry, Entra, and MCP tooling. |
microsoft-data | 17 | Power BI, Power Apps, Power Automate, Dataverse, and Power Platform architecture. |
personal-productivity | 9 | Personal productivity, reminders, communication, notes, and connected tools. |
product-management | 29 | Product discovery, specifications, delivery planning, analytics, and agile workflows. |
prompting | 15 | Prompt engineering, optimization, safety review, and creative-thinking frameworks. |
react-frontend | 17 | React, Vue, Next.js, mobile frontend frameworks, migrations, and testing. |
skills-management | 24 | Skill authoring, discovery, cleanup, Copilot configuration, and terse interaction modes. |
streamlit | 18 | Streamlit applications, dashboards, chat UIs, components, layouts, and performance. |
testing | 12 | Testing, QA, Playwright, pytest, debugging, and evaluation strategy. |
Local MCP Discovery
Use the local stdio MCP server when an agent should discover or install skills without loading the full library into context. See Local Skills MCP Quickstart.
The intended flow is:
list_categories -> list_skills or search_skills -> get_skill -> read_skill_doc -> install_skill
Machine-Readable Index
The canonical generated indexes are:
Regenerate them, the category READMEs, and the root catalog with:
npm run build:manifest
Local Development
npm install
npm run build:manifest
npm run check:docs
npm run build:zips
npm run preview
npm run build performs manifest generation, documentation validation, and ZIP packaging in order.
Adding Or Editing Skills
Follow CONTRIBUTING.md. Every skill must contain a SKILL.md file with name and
description frontmatter. Category READMEs are generated from the manifest; do not hand-edit them.
Repository Layout
skills/<category>/<skill>/SKILL.md skill instructions
skills/<category>/README.md generated category index
scripts/build-manifest.mjs manifests and documentation indexes
scripts/check-docs.mjs documentation coverage audit
scripts/build-zips.mjs ZIP artifacts for Pages
docs/ Pages site and machine-readable manifests
License
Skills come from a mix of original work and adapted public sources. Check individual skill folders for license files and attribution before redistributing a selected package.