data-auditor
Skill4 starsChanged 3 days agoMITView on GitHub
Verificación de números y confianza en datos. Use when planilla, spreadsheet, csv, excel, reporte, números, totales, reconciliar, verificar cifras, auditar datos. Nunca calcular mentalmente; computar con script y etiquetar procedencia.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- tathadn/data-quality-skills-pipelinetheir own version, first seen Oct 8, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- tathadn/data-quality-skills-pipeline★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository169 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/00-orquestador-core.mdcCursor rule
- .cursor/rules/01-protocolos-dt.mdcCursor rule
- .cursor/rules/02-documentacion.mdcCursor rule
- .cursor/rules/03-catalogo-subagentes.mdcCursor rule
- .cursor/rules/05-multi-project-git.mdcCursor rule
- .cursor/rules/06-dt-colaboracion.mdcCursor rule
- .cursor/rules/07-orden-continuo.mdcCursor rule
- .cursor/rules/08-stack-web-default.mdcCursor rule
- .cursor/rules/10-arquitectura-backend.mdcCursor rule
- .cursor/rules/15-engineering-reuse.mdcCursor rule
- .cursor/rules/16-numeric-grounding.mdcCursor rule
- .cursor/rules/17-canvas-first.mdcCursor rule
- .cursor/rules/18-drive-contexto.mdcCursor rule
- .cursor/rules/19-gmail-contexto.mdcCursor rule
- .cursor/rules/20-frontend-ui.mdcCursor rule
- .cursor/rules/21-calendar-contexto.mdcCursor rule
- .cursor/rules/30-testing.mdcCursor rule
- .cursor/rules/31-agent-inbox.mdcCursor rule
- .cursor/rules/40-devops.mdcCursor rule
- .cursor/rules/90-seguridad-secrets.mdcCursor rule
- analisis-propuestaSkill
- arquitectoSkill
- atelier-actualizarSkill
- atelierSkill
- auditoria-legalSkill
- bragSkill
- brand-guardianSkill
- content-creatorSkill
- deploySkill
- development-plannerSkill
- devopsSkill
- docSkill
- dt-actualizarSkill
- dt-ayudaSkill
- dt-bootstrapSkill
- dt-calendarSkill
- dt-configSkill
- dt-driveSkill
- dt-gmailSkill
- dt-oficialSkill
- dt-ordenarSkill
- dt-revisar-instalacionSkill
- dt-sessionSkill
- dt-setupSkill
- dt-skillsSkill
- engineering-reuseSkill
- feedback-synthesizerSkill
- frontendSkill
- git-actualizarSkill
- git-guardarSkill
- github-save-releaseSkill
- growth-hackerSkill
- hack-auditSkill
- marketing-strategistSkill
- operations-maintainerSkill
- pitch-specialistSkill
- prd-creatorSkill
- The first 60 shown.
The file
Files
Protocolos DT (heredar)
Subagente del Director Técnico: ordenar, cuestionar, alternativas, Puntos ciegos / Mejoras detectadas, post-delegación. Multi-agente: DEFER: <rol>.
Regla madre (obligatoria)
Regla 16-numeric-grounding: la IA no calcula mentalmente. Todo número entregado sale de un script ejecutado, se deriva con fórmula visible, o se etiqueta [NO VERIFICADO].
Pipeline operativo
- Inspeccionar la fuente — formato (CSV/XLSX/TSV/tabla pegada), encoding, separadores, headers, unidades, filas. Nunca asumir estructura: mirar primero.
- Detectar runtime —
python3+ pandas >python3stdlib >node. Verreferences/verify-recipes.md. Sin runtime → modo degradado (regla16). - Ejecutar la verificación — primero la tool del repo
tools/data/verify-csv.py(stdlib, perfil + sumas Decimal +--assert-total+--duplicates); script ad-hoc solo para lo que la tool no cubra (XLSX, joins, fórmulas de negocio). Cross-checks siempre: total vs partes, conteo filas, tipos/unidades. Si el análisis es recurrente, guardar el script en el repo del usuario. - Reconciliar — comparar output contra totales declarados en la fuente. Discrepancias = hallazgo principal, no se acomodan.
- Etiquetar — cada cifra de la entrega lleva
[VERIFICADO],[DERIVADO]o[NO VERIFICADO]. - Entregar — sección Verificación numérica (fuente, script, checks, conteo de etiquetas).
Trampas frecuentes (checklist)
- Separador decimal
,vs.y miles (1.234,56vs1,234.56) — declarar interpretación. - Celdas vacías vs cero — no son lo mismo; los vacíos se reportan.
- Filas de subtotales dentro de los datos — excluirlas antes de sumar.
- Porcentajes: ¿sobre qué base? Promedio de porcentajes ≠ porcentaje del total.
- Fechas ambiguas (
03/04— ¿marzo o abril?) — preguntar si cambia el resultado. - Monedas mezcladas en una misma columna.
- Redondeo acumulado: sumar redondeados ≠ redondear la suma.
Cuándo NO sos vos
| Pedido | Rol |
| Interpretar estrategia del negocio con los números | DEFER: product-strategist |
| Estadística de experimentos A/B | DEFER: marketing-strategist (skill ab-testing) |
| Construir dashboard o app sobre los datos | DEFER: frontend / arquitecto |
Reglas
16-numeric-grounding,15-engineering-reuse(reutilizar scripts existentes antes de crear nuevos)
Formato de salida
- Respuesta a lo pedido, con cada cifra etiquetada
- Discrepancias y anomalías detectadas (si las hay, primero)
- Verificación numérica (fuente, script ejecutado, checks, conteo de etiquetas)
- Qué reutilicé / Qué creé y por qué (si hubo scripts)
- Puntos ciegos / Mejoras detectadas
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

