Inicio / Skills / Ofimática y documentos / Spreadsheet Skill
Ofimática y documentos

Spreadsheet Skill

OpenAIApache-2.0Idioma del Skill: Inglés
Licencia verificadaIncluye scriptsRiesgo medio
DemostraciónEjecución real · claude-sonnet-5-5 · octubre de 2026
Please create an Excel workbook for tracking my monthly household budget. I want one sheet where I enter income and expenses by category for 12 months (January to December), with automatic totals per month and per category, a savings row (income minus expenses), a savings rate row (savings divided by income), sensible number formats, and a highlight on any month where expenses are higher than income. Use sample numbers to start. I'll open it in Excel. Write the Python that builds it and tell me how to check it.
Skill: spreadsheet

I can't run code or write files in this session, so the script below is untested. Run it once and use the checks at the end.

build_budget.py

from openpyxl import Workbook
from openpyxl.formatting.rule import FormulaRule
from openpyxl.styles import Alignment, Border, Font, PatternFill, Side
from openpyxl.utils import get_column_letter

OUT = "household_budget.xlsx"
MONTHS = ["Jan", "Feb", "Mar", "Apr", "May", "Jun",
          "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"]

INCOME = {
    "Salary":       [5200] * 12,
    "Side income":  [300, 250, 400, 350, 300, 200, 0, 300, 350, 250, 300, 200],
    "Other income": [0, 0, 0, 100, 0, 0, 0, 0, 0, 0, 0, 0],
}
EXPENSES = {
    "Rent / Mortgage": [1800] * 12,
    "Utilities":       [220, 210, 190, 160, 150, 170, 190, 185, 160, 170, 200, 230],
    "Groceries":       [620, 590, 640, 600, 610, 580, 650, 620, 600, 630, 640, 720],
    "Transport":       [180, 170, 190, 200, 210, 220, 260, 200, 190, 185, 180, 200],
    "Insurance":       [250] * 12,
    "Dining out":      [240, 200, 260, 230, 250, 280, 350, 260, 240, 270, 300, 420],
    "Entertainment":   [120, 100, 150, 130, 140, 160, 300, 120, 110, 140, 150, 350],
    "Health":          [60, 40, 80, 50, 45, 60, 70, 300, 50, 60, 40, 90],
    "Subscriptions":   [65] * 12,
    "Other":           [100, 80, 120, 90, 150, 110, 1500, 100, 90, 130, 110, 1400],
}

CURRENCY = '$#,##0;[Red]($#,##0);"-"'
PERCENT = '0.0%;[Red]-0.0%;"-"'

FIRST_COL, LAST_COL, TOTAL_COL = 2, 13, 14          # B..M months, N total
HEADER_FILL = PatternFill("solid", fgColor="1F3864")
SECTION_FILL = PatternFill("solid", fgColor="D9E1F2")
TOTAL_FILL = PatternFill("solid", fgColor="F2F2F2")
FLAG_FILL = PatternFill("solid", bgColor="F8CBAD", fgColor="F8CBAD")
INPUT_FONT = Font(color="0000FF")
BOLD = Font(bold=True)
TOP_BORDER = Border(top=Side(style="thin"))

wb = Workbook()
ws = wb.active
ws.title = "Budget"

ws["A1"] = "Household Budget"
ws["A1"].font = Font(bold=True, size=16)
ws["A2"] = "Blue = enter your numbers. Black = formulas. Months with expenses above income are highlighted."
ws["A2"].font = Font(italic=True, color="595959")

# Header row
HDR = 4
ws.cell(HDR, 1, "Category")
for i, m in enumerate(MONTHS):
    ws.cell(HDR, FIRST_COL + i, m)
ws.cell(HDR, TOTAL_COL, "Year Total")
for c in range(1, TOTAL_COL + 1):
    cell = ws.cell(HDR, c)
    cell.font = Font(bold=True, color="FFFFFF")
    cell.fill = HEADER_FILL
    cell.alignment = Alignment(horizontal="left" if c == 1 else "right")

first, last = get_column_letter(FIRST_COL), get_column_letter(LAST_COL)
tot = get_column_letter(TOTAL_COL)


def section(row, title):
    for c in range(1, TOTAL_COL + 1):
        ws.cell(row, c).fill = SECTION_FILL
    ws.cell(row, 1, title).font = BOLD


def data_rows(start, items):
    r = start
    for name, vals in items.items():
        ws.cell(r, 1, name).alignment = Alignment(indent=1)
        for i, v in enumerate(vals):
            cell = ws.cell(r, FIRST_COL + i, v)
            cell.font = INPUT_FONT
            cell.number_format = CURRENCY
        ws.cell(r, TOTAL_COL, f"=SUM({first}{r}:{last}{r})").number_format = CURRENCY
        ws.cell(r, TOTAL_COL).font = BOLD
        r += 1
    return start, r - 1


def total_row(row, label, a, b):
    ws.cell(row, 1, label)
    for c in range(FIRST_COL, TOTAL_COL + 1):
        L = get_column_letter(c)
        ws.cell(row, c, f"=SUM({L}{a}:{L}{b})").number_format = CURRENCY
    for c in range(1, TOTAL_COL + 1):
        cell = ws.cell(row, c)
        cell.font = BOLD
        cell.fill = TOTAL_FILL
        cell.border = TOP_BORDER


r = HDR + 1
section(r, "INCOME")
inc_a, inc_b = data_rows(r + 1, INCOME)
INC_TOT = inc_b + 1
total_row(INC_TOT, "Total Income", inc_a, inc_b)

r = INC_TOT + 2
section(r, "EXPENSES")
exp_a, exp_b = data_rows(r + 1, EXPENSES)
EXP_TOT = exp_b + 1
total_row(EXP_TOT, "Total Expenses", exp_a, exp_b)

SAV = EXP_TOT + 2
RATE = SAV + 1
ws.cell(SAV, 1, "Savings (Income - Expenses)")
ws.cell(RATE, 1, "Savings Rate (Savings / Income)")
for c in range(FIRST_COL, TOTAL_COL + 1):
    L = get_column_letter(c)
    ws.cell(SAV, c, f"={L}{INC_TOT}-{L}{EXP_TOT}").number_format = CURRENCY
    ws.cell(RATE, c, f"=IF({L}{INC_TOT}=0,0,{L}{SAV}/{L}{INC_TOT})").number_format = PERCENT
for row in (SAV, RATE):
    for c in range(1, TOTAL_COL + 1):
        ws.cell(row, c).font = BOLD
for c in range(1, TOTAL_COL + 1):
    ws.cell(SAV, c).border = TOP_BORDER

# Highlight months where expenses > income (header + summary rows only,
# so the input cells stay clean).
for rng, top in ((f"{first}{HDR}:{last}{HDR}", HDR),
                 (f"{first}{INC_TOT}:{last}{INC_TOT}", INC_TOT),
                 (f"{first}{EXP_TOT}:{last}{EXP_TOT}", EXP_TOT),
                 (f"{first}{SAV}:{last}{RATE}", SAV)):
    ws.conditional_formatting.add(
        rng,
        FormulaRule(formula=[f"{first}${EXP_TOT}>{first}${INC_TOT}"],
                    fill=FLAG_FILL, font=Font(bold=True, color="9C0006")),
    )

ws.column_dimensions["A"].width = 34
for c in range(FIRST_COL, TOTAL_COL + 1):
    ws.column_dimensions[get_column_letter(c)].width = 11
ws.column_dimensions[tot].width = 13
ws.freeze_panes = ws.cell(HDR + 1, FIRST_COL)
ws.sheet_view.showGridLines = False

wb.save(OUT)
print("Saved", OUT)

Run it

python3 -m pip install openpyxl
python3 build_budget.py
open household_budget.xlsx

How to check it

  1. Formulas calculate. Excel calculates on open. If you see blanks or zeros, press Ctrl+Alt+F9 (Cmd+= on Mac). No cell should show #DIV/0! or #NAME?.
  2. Spot-check totals. January income should be $5,500 and January expenses $3,655, so savings are $1,845. That's a 33.5% savings rate. Year Total (column N) should equal the sum of the twelve months, both down the Total rows and across the category rows.
  3. Highlight. The sample data overspends in Jul (income $5,200, expenses $6,435) and Dec (income $5,400, expenses $6,325). Those two columns should turn red in the header and summary rows, and no other months should. Change a Jul input, such as Salary to 9000, and the highlight should disappear.
  4. Formats. Currency shows as $1,234, negatives are red in parentheses, and zeros show as -. Savings rate shows as a percentage, and it goes red when negative.
  5. Optional automated check. If you have LibreOffice, run soffice --headless --convert-to pdf household_budget.xlsx and look at the PDF. You can also recalculate and read the values back with openpyxl.load_workbook(..., data_only=True).

Notes

  • Blue cells are inputs, so overwrite them freely.
  • To add a category, insert a row inside the existing block, between the first and last row of that section. The SUM ranges expand automatically. Inserting after the last row won't extend the total.
  • Savings rate uses IF(income=0, 0, ...) to avoid divide-by-zero errors in empty months.

Real run in an isolated folder with only this skill installed. Only the Skill and Read tools were enabled, so no command was run and no file was written. The scenario is fictional. The model could not run code in this run. The script was run by hand afterwards and builds a valid workbook, and the months that overspend (July and December) are right. However, the July and December expense totals quoted in step 3 are wrong: the script gives 5,435 and 5,525. Check the workbook itself, not the quoted figures.

Qué hace

Crea y modifica archivos .xlsx, .csv y .tsv con openpyxl y pandas. Los libros nuevos incorporan fórmulas para cada valor derivado (sin resultados codificados de forma fija), encabezados limpios, formatos de número y de fecha, anchos de columna legibles y una convención de color opcional (entradas en azul, fórmulas en negro). Los libros existentes se editan sin romper fórmulas, referencias ni formato, y se renderizan para comprobarlos visualmente cuando LibreOffice y Poppler están disponibles. También cubre convenciones financieras (números negativos entre paréntesis en rojo, unidades en los encabezados, fuentes en comentarios de celda) y diseños al estilo bancario.

Conviene saber

openpyxl no calcula fórmulas, así que los valores aparecen cuando el archivo se abre en Excel o Sheets. Se incluyen cuatro ejemplos ejecutables de openpyxl.

Ideal para

Presupuestos, seguimientos, informes y modelos; limpiar y resumir datos tabulares; ordenar un libro desordenado.

Notas y riesgos

Riesgo medio: escribe archivos .xlsx y .csv, puede instalar paquetes de Python (openpyxl, pandas) y, de forma opcional, LibreOffice y Poppler (las instrucciones para Ubuntu usan sudo apt-get; macOS usa brew), e indica a la IA que elimine sus propios archivos temporales al terminar. Los ejemplos incluidos sobrescriben un archivo con el mismo nombre sin preguntar, así que haz una copia de seguridad antes de editar un libro real. openpyxl no calcula fórmulas, así que comprueba los números en Excel. En la prueba, la IA no pudo ejecutar código y citó dos totales incorrectos; el script en sí se ejecutó a mano y funciona. Esta es una revisión anterior de la Skill de OpenAI, que OpenAI retiró más tarde de su repositorio. Apache-2.0, sin modificaciones.

SKILL.md original9 archivos en este paquete