Claude Skills - Agentes IA Especializados
DhanushNehru/claude-skills
Skills prontos para copiar-colar que dão superpoderes ao Claude
Ver no GitHubBiblioteca de skills prontos para uso que transformam Claude em agentes IA especializados. Cada skill é um arquivo de instruções estruturado, testado e otimizado para tarefas específicas (revisão de código, debugging, design de APIs, etc), permitindo usar Claude como um copiloto profissional sem engenharia de prompts complexa.
Instalação
npx convez-ai DhanushNehru/claude-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
- Revisão de código em produção
- Debugging sistemático de bugs
- Design de APIs RESTful
- Automação de commits Git
- Análise de segurança
- Escrita e edição técnica
Veredito

Abordagem inteligente para reutilizar prompts de qualidade via skills modulares. Melhor que coleções genéricas, mas funciona apenas com Claude — dependência do ecossistema Anthropic reduz portabilidade.
README
A tradução para o português deste README ainda está na fila.
🧠 Claude Skills
Ready-to-use skill files that transform Claude into specialized AI agents.
No prompt engineering required. Just copy → paste → go.
Think of these as plugins for your brain's AI copilot. Each skill is a self-contained instruction set — battle-tested, community-reviewed, and ready for production use.
Get Started · Browse Skills · Create a Skill · Contribute
🤔 What Are Claude Skills?
A Claude Skill is a portable, copy-paste-ready instruction file that gives Claude a specialized capability. Unlike generic prompt collections, each skill is:
| Feature | Description |
|---|---|
| 🎯 Purpose-built | Designed for a specific real-world task, not a toy demo |
| 📋 Structured | Uses the proven Role → Goal → Constraints → Output contract format |
| 🧪 Tested | Every skill includes example inputs/outputs so you know what to expect |
| 🔌 Plug-and-play | Works with Claude.ai, Claude API, Claude Code (CLAUDE.md), and any Claude-powered tool |
| 🏷️ Tagged | Difficulty level, estimated token usage, and compatibility info included |
⚡ Quick Start
Method 1: Claude.ai (Recommended for beginners)
- Browse the Skill Catalog below
- Open the skill file (e.g.,
skills/developer/code-reviewer.md) - Copy the content inside the
<system_prompt>tags - Paste it into Claude.ai → Project Knowledge or Custom Instructions
- Start chatting — Claude now has that skill!
Method 2: Claude Code (CLAUDE.md)
- Copy the skill file to your project root as
CLAUDE.md(or append to an existing one) - Run
claudein your terminal — it auto-reads the file - Claude Code now operates with that specialized skill
Method 3: API / SDK
import anthropic
# Load the skill
with open("skills/developer/code-reviewer.md") as f:
skill = f.read()
client = anthropic.Anthropic()
message = client.messages.create(
model="claude-sonnet-4-20250514",
max_tokens=4096,
system=skill, # ← The skill becomes the system prompt
messages=[{"role": "user", "content": "Review this pull request: ..."}]
)
📚 Skill Catalog
🛠️ Developer Skills
| Skill | Description | Difficulty |
|---|---|---|
| Code Reviewer | Production-grade code review with security, performance & maintainability analysis | ⭐⭐ |
| Git Commit Crafter | Generate perfect conventional commits from diffs | ⭐ |
| Bug Hunter | Systematic debugging agent that traces root causes | ⭐⭐⭐ |
| API Designer | Design RESTful APIs following OpenAPI best practices | ⭐⭐ |
| Regex Wizard | Build, explain, and test regular expressions in any flavor | ⭐ |
🔒 Security Skills
| Skill | Description | Difficulty |
|---|---|---|
| Threat Modeler | STRIDE-based threat modeling for any system architecture | ⭐⭐⭐ |
| Dependency Auditor | Deep-dive analysis of package vulnerabilities and supply chain risks | ⭐⭐ |
| Security Headers Checker | Analyze and fix HTTP security headers for web applications | ⭐ |
📊 Data & Analysis Skills
| Skill | Description | Difficulty |
|---|---|---|
| SQL Query Optimizer | Analyze and optimize slow SQL queries with execution plan analysis | ⭐⭐ |
| Data Storyteller | Transform raw datasets into compelling narratives with visualizations | ⭐⭐ |
| CSV Detective | Profile, clean, and analyze messy CSV/JSON datasets | ⭐ |
✍️ Writing & Content Skills
| Skill | Description | Difficulty |
|---|---|---|
| Technical Writer | Write clear, structured technical documentation and ADRs | ⭐⭐ |
| README Generator | Generate stunning, complete README files from codebases | ⭐ |
| Changelog Author | Generate human-readable changelogs from git history | ⭐ |
🏗️ Architecture & DevOps Skills
| Skill | Description | Difficulty |
|---|---|---|
| System Design Coach | Interactive system design interview prep and architecture review | ⭐⭐⭐ |
| Dockerfile Optimizer | Analyze and optimize Dockerfiles for size, security, and build speed | ⭐⭐ |
| CI/CD Pipeline Builder | Generate production-ready GitHub Actions / GitLab CI pipelines | ⭐⭐ |
🧩 Productivity Skills
| Skill | Description | Difficulty |
|---|---|---|
| Meeting Summarizer | Extract action items, decisions, and key points from meeting transcripts | ⭐ |
| Email Diplomat | Craft professional emails for sensitive situations | ⭐ |
| Learning Path Generator | Create personalized, structured learning roadmaps for any topic | ⭐⭐ |
🎨 Creative Skills
| Skill | Description | Difficulty |
|---|---|---|
| Color Palette Generator | Generate accessible, harmonious color palettes from any inspiration | ⭐ |
| Naming Consultant | Generate memorable names for projects, products, and companies | ⭐ |
🧬 Skill Anatomy
Every skill follows a consistent structure:
skills/
├── developer/
│ └── code-reviewer.md ← The skill file
├── security/
│ └── threat-modeler.md
└── ...
Each skill file contains:
# 🔍 Skill Name
> One-line description
## Metadata
- **Category**: Developer | Security | Data | Writing | DevOps | Productivity | Creative
- **Difficulty**: ⭐ | ⭐⭐ | ⭐⭐⭐
- **Works With**: Claude.ai, Claude Code, API
- **Estimated Tokens**: ~500 system prompt tokens
## System Prompt
<system_prompt>
... the actual instruction set ...
</system_prompt>
## Example Usage
**User**: [example input]
**Claude**: [example output]
## Tips & Variations
- Modification ideas for different use cases
🔨 Create Your Own
Use our Skill Template to create your own skill:
- Fork this repo
- Copy
SKILL_TEMPLATE.mdinto the appropriateskills/<category>/folder - Fill in your skill following the template structure
- Test it thoroughly (include real example I/O)
- Submit a PR!
Pro tips:
- Use XML tags (
<context>,<constraints>,<output_format>) for complex instructions - Keep system prompts under 1000 tokens for optimal performance
- Include at least 2 example interactions
- Add edge cases in your "Tips & Variations" section
🤝 Contributing
We love contributions! Whether it's a new skill, an improvement to an existing one, or fixing a typo — every PR is welcome.
Please read our Contributing Guide and Code of Conduct before submitting.
What makes a great skill?
- ✅ Solves a real problem people encounter regularly
- ✅ Includes tested examples with realistic inputs/outputs
- ✅ Is self-contained — no external dependencies or setup
- ✅ Has a clear, specific scope — does one thing extremely well
- ✅ Uses structured formatting (XML tags, markdown) for reliability
🌟 Star History
If this repo helped you, please consider giving it a ⭐ — it helps others discover these skills!
📜 License
This project is licensed under the MIT License — see the LICENSE file for details.
Made with 🧠 by the community, for the community.