CLAUDE.md Writer
A skill for creating and optimizing CLAUDE.md files following official Anthropic best practices.
Golden Rule
Para cada linea, pregunta: *"Eliminar esto causaria que Claude cometa errores?"* Si no, eliminalo. CLAUDE.md inflados causan que Claude ignore instrucciones.
"Think of CLAUDE.md as the 'unwritten knowledge' in your codebase"
Quick Tools
/init- Genera CLAUDE.md inicial basado en la estructura del proyecto#key - Agregar instrucciones dinamicamente durante la sesion
@import Syntax
CLAUDE.md puede importar otros archivos:
See @README.md for overview and @package.json for commands.
Git workflow: @docs/git-instructions.mdReference Documentation
For examples and validation checklist, see:
references/good-example.md- Well-structured CLAUDE.md examplereferences/bad-example.md- Common anti-patterns to avoidreferences/checklist.md- Quality validation checklistreferences/dead-weight-audit.md- Dead weight audit (5 criterios para recortar instrucciones)
What to Include vs Exclude
| Include | Exclude |
|---|---|
| Bash commands Claude can't guess | Anything Claude can figure out by reading code |
| Code style rules that differ from defaults | Standard language conventions |
| Testing instructions and preferred test runners | Detailed API docs (link instead) |
| Repository etiquette (branch naming, PR) | Information that changes frequently |
| Architectural decisions specific to project | Long explanations or tutorials |
| Developer environment quirks (env vars) | File-by-file descriptions |
| Common gotchas or non-obvious behaviors | Self-evident practices |
CLAUDE.md vs Skills
| CLAUDE.md | Skills |
|---|---|
| Reglas amplias que aplican siempre | Conocimiento de dominio solo relevante a veces |
| Code style y workflow | Workflows especializados reutilizables |
| Convenciones del repo | Info que cambia por contexto |
| Se carga en cada sesion | Se cargan on-demand |
Content Triage
Antes de agregar contenido a CLAUDE.md, evaluar:
| Si el contenido es... | Entonces... |
|---|---|
| Regla que aplica a TODAS las tareas | Agregar a CLAUDE.md |
| Domain knowledge especifico | Sugerir crear skill |
| Workflow que solo aplica a veces | Sugerir crear skill |
| Info que cambia frecuentemente | No agregar, linkear a docs |
Si el usuario quiere agregar info condicional ("Cuando trabajes con X...", "Para tareas de tipo Y..."), sugerir crear un skill separado.
Usage Modes
| Invocacion | Comportamiento |
|---|---|
/claude-md-writer | Proceso completo: analizar proyecto, escribir/mejorar CLAUDE.md |
/claude-md-writer actualiza con la sesion | Extraer info relevante de la sesion actual y agregar a CLAUDE.md |
/claude-md-writer agrega regla: [descripcion] | Agregar una regla especifica con el formato correcto |
/claude-md-writer revisa y optimiza | Auditar CLAUDE.md existente, eliminar redundancias, mejorar enfasis |
/claude-md-writer audita | Dead weight audit: aplicar los 5 criterios contra el setup actual (ver references/dead-weight-audit.md) |
/claude-md-writer convierte a skill | Identificar contenido que deberia ser skill y extraerlo |
Process
Step 1: Analyze the Project
Identify tech stack, project structure, and existing documentation.
Step 2: Apply the Three Dimensions
| Dimension | Question to Answer |
|---|---|
| WHAT | What is this project? Tech stack? Structure? |
| WHY | Why does it exist? What problem does it solve? |
| HOW | How do I work on it? Commands? Workflows? |
Step 3: Write Following Best Practices
Length: Keep under 200 lines (ideal ~100 lines)
Structure (in order):
- Repository Overview (1 paragraph)
- Why This Exists (purpose, users)
- Quick Start & Commands (bash with descriptions)
- Architecture Overview (key patterns)
- Key Development Rules (with emphasis)
- References (links to detailed docs)
Emphasis for Critical Rules:
- **CRITICAL**: [Never violate this]
- **IMPORTANT**: [Always do this]Progressive Disclosure:
- Use file references instead of code snippets
- Link to external docs for detailed topics
- Keep CLAUDE.md focused on universal instructions
Step 4: Verify with Checklist
Before finalizing, run through the checklist in references/checklist.md.
Template
# CLAUDE.md
This file provides guidance to Claude Code when working with this repository.
## Repository Overview
[One paragraph: what this project is, tech stack, main purpose]
## Why This Exists
[What problem it solves, who uses it, business context]
## Quick Start & Commands
\`\`\`bash
npm ci # Install dependencies
npm start # Start dev server
npm test # Run tests
npm run build # Production build
\`\`\`
## Architecture Overview
[Key patterns, folder structure, main concepts - keep brief]
## Key Development Rules
- **CRITICAL**: [Most important rule that must never be violated]
- **IMPORTANT**: [Second most important rule]
## References
- `/docs/` - Detailed documentation
- `README.md` - Project readmeDead Weight Audit
Cuando el usuario pida auditar su setup, aplicar los 5 criterios de references/dead-weight-audit.md contra cada regla:
- ¿Default? — Claude ya lo hace sin que se lo digan
- ¿Conflicto? — Contradice otra regla en el setup
- ¿Redundancia? — Cubierto por otra regla o proceso
- ¿Parche puntual? — Agregado para arreglar un output malo específico
- ¿Vago? — Se interpreta distinto cada vez
Output esperado:
- Lista de cortes con razón en una línea cada uno
- Lista de conflictos encontrados entre archivos
- Versión limpia del CLAUDE.md sin dead weight
Proceso post-audit: eliminar → probar 3 tareas típicas → re-agregar solo lo que se rompió.
Anti-Patterns to Avoid
| Anti-Pattern | Why It's Bad | Do This Instead |
|---|---|---|
| Code snippets | Become stale, waste tokens | Use file:line references |
| Styling rules | LLM is slow for deterministic tasks | Configure linters |
| Kitchen sink | Dilutes instruction effectiveness | Keep < 200 lines |
| No emphasis | Critical rules same weight as minor | Use CRITICAL/IMPORTANT |
| Auto-generated | Misses project nuances | Manually craft |
| Domain knowledge | Inflates every session | Move to skills |