Habilidades writing-skills
✍️

writing-skills

Seguro

Write High-Quality AI Agent Skills

Também disponível em: Cygnusfear,Cycleaddict,ZhanlinCui,davila7,obra,Dimon94,DYAI2025,CodingCossack

This skill provides templates, decision trees, and best practices for creating and improving AI agent skills. It helps users avoid common mistakes and follow standardized patterns.

Suporta: Claude Codex Code(CC)
🥉 75 Bronze
1

Baixar o ZIP da skill

2

Upload no Claude

Vá em Configurações → Capacidades → Skills → Upload skill

3

Ative e comece a usar

Testar

A utilizar "writing-skills". I want to create a skill for handling API errors

Resultado esperado:

A complete SKILL.md with: name: api-error-handler, description: 'Use when API request fails or returns error status', metadata with triggers: api-error, network-failure, 503, with references/ folder for detailed handling procedures

A utilizar "writing-skills". My skill description is 'This skill helps with file operations' - it's not triggering

Resultado esperado:

Optimized description: 'Use when user asks to read, write, or modify files. Triggers on: file-create, file-edit, file-delete, touch, mkdir, rm command'

Auditoria de Segurança

Seguro
v1 • 2/25/2026

This is a documentation and reference skill providing guidance on writing AI agent skills. All static findings are false positives - the scanner detected patterns in markdown documentation (code formatting with backticks, mentions of cryptographic terms, URLs in documentation) rather than actual executable code. The render-graphs.js is a build script for documentation generation, not runtime functionality.

21
Arquivos analisados
4,088
Linhas analisadas
0
achados
1
Total de auditorias
Nenhum problema de segurança encontrado
Auditado por: claude

Pontuação de qualidade

36
Arquitetura
100
Manutenibilidade
87
Conteúdo
50
Comunidade
100
Segurança
100
Conformidade com especificações

O Que Você Pode Construir

Creating a New Skill

Use the decision tree to select the right architecture tier and template for a new skill, then follow the template structure.

Improving Existing Skills

Diagnose why a skill is not being triggered or ignored by agents, then apply anti-rationalization or CSO techniques.

Standardizing Team Skills

Ensure all team skills follow consistent naming conventions, metadata standards, and directory structures.

Tente Estes Prompts

Create a Simple Skill
I need to create a new skill called [skill-name]. It should [what it does]. Use the Tier 1 template from the writing-skills reference.
Debug Skill Triggering
My skill [skill-name] is not being triggered by Claude. The description is: [description]. Apply CSO optimization from writing-skills.
Add Anti-Rationalization
I have a discipline skill with rules that Claude keeps ignoring. The rules are: [rules]. Apply anti-rationalization techniques.
Structure Complex Skill
I need to create a complex skill with multiple concepts (200-1000 lines). Use Tier 2 expanded architecture with proper references.

Melhores Práticas

  • Always start descriptions with 'Use when...' followed by specific trigger conditions
  • Include at least 3 keywords in metadata.triggers for better discoverability
  • Keep single-file skills under 200 lines; use references/ for longer content
  • Split monolithic skills into multiple files using the tier system

Evitar

  • Describing workflow instead of triggers (e.g., 'This skill helps you create files' instead of 'Use when user wants to create a file')
  • Using generic skill names like 'helper' or 'utility' instead of descriptive gerunds
  • Missing metadata.triggers field or having fewer than 3 trigger keywords
  • Creating long monolithic SKILL.md files instead of modular references/

Perguntas Frequentes

What is CSO in skill writing?
CSO stands for 'Concept Search Optimization'. It is the practice of writing skill descriptions that maximize the likelihood of being triggered by matching user intent patterns.
What is anti-rationalization?
Anti-rationalization refers to techniques that make skill rules difficult for agents to ignore or rationalize away. It involves making rules specific, unavoidable, and tied to concrete outcomes.
What is the difference between Tier 1 and Tier 2 skills?
Tier 1 skills are simple, single-file skills under 200 lines. Tier 2 skills are complex, multi-concept skills between 200-1000 lines that use a references/ directory structure.
How do I test my skill?
Use the RED-GREEN-REFACTOR cycle: Write a test case (RED), implement the skill to pass (GREEN), then refactor for clarity and efficiency.
What triggers should I include?
Include specific keywords, error messages, command names, and symptoms that users would type. Aim for at least 3-5 diverse triggers per skill.
Can I use this skill for any AI tool?
Yes, this skill provides guidance that applies to Claude, Codex, Claude Code, and other AI agent systems that support skill-based functionality.