xlsx · Spreadsheet Creation & Editing

Read, write, and fix Excel workbooks: new spreadsheets, bulk data processing, formula computation, formatting, data cleaning, and format conversion.

openpyxl handles formulas and formatting, pandas handles bulk data, markitdown gives quick previews. Hard output standards: professional fonts, formulas (never hardcoded results), zero formula errors (recalc-verified), literal spec compliance, and every assumption and hardcoded number documented with its source. Financial models follow color and number-format conventions.

Example invocation: "Make me a budget sheet with formulas."

Full brief

Positioning

xlsx is the creation, editing, and analysis tool for spreadsheet files: any task whose primary input or output is a spreadsheet goes through it. The deliverable is the spreadsheet file itself — not a report or a script.

Core capabilities

Create & edit: openpyxl writes formulas/formatting (=SUM(B2:B9), never hardcoded totals); pandas moves bulk data in and out.
Quick look: markitdown previews by sheet (no cell coordinates — not for planning edits).
Model reading: reading formulas and values needs two load_workbook passes (both data_only modes).
Mandatory recalc: any file with formulas must pass recalc.py (real LibreOffice evaluation); no delivery while errors_found.
Formula selection: prefer Excel-2007-era functions (SUMIFS/INDEX/MATCH/IFERROR); six newer functions need the _xlfn. prefix; XLOOKUP/SORT/FILTER/UNIQUE/SEQUENCE are banned (the runtime LibreOffice can't evaluate them).
Financial-model conventions: colors (blue=hardcoded input, black=formula, green=cross-sheet link, red=cross-file, yellow fill=key assumptions) and number formats (currency / percentages stored as fractions / multiples / years as text).
Structural discipline: every assumption in its own labeled cell, referenced by formulas; row-consistent formulas; guard zero denominators.
Literal spec compliance: exact tab names, column headers, and user-spelled formulas; a redesign that computes something else fails.
Assumption documentation: every assumption and hardcoded number sourced where the reader can see it (cell comment or adjacent cell).

Workflow

Confirm the need: create / edit / clean / convert; preview existing files with markitdown first
Write the script: openpyxl formulas + pandas bulk data
Run recalc.py; fix and re-run on errors
Write 2-3 formulas and verify values before building the full grid (prevents grid-wide misalignment)
Deliver the spreadsheet file

Inputs & outputs

Input	Required	Notes
Spreadsheet file / need	Yes	Existing .xlsx/.xlsm/.xltx/.csv/.tsv, or a new-build description

Output: the processed spreadsheet file (formulas recalc-verified).

Boundaries with adjacent skills

Skill	Lane
xlsx (this)	Deliverable is a spreadsheet file
Report-type tools	Deliverable is a report/HTML (spreadsheet is just intermediate data)

Fit

New budget/model sheets with formulas
Column adds, formula computation, formatting, charts on existing sheets
Cleaning and structuring messy tabular data
Format conversion (csv/tsv/xlsx)

Before you start

openpyxl / pandas / markitdown preinstalled, no pip needed; LibreOffice powers formula recalculation.
Delivery red lines: no delivery while recalc reports errors_found; formulas must recalculate, numbers must never be hardcoded.
Editing existing files: find its input-cell conventions first (color/shading markers), write only to input cells, leave existing formulas untouched; .xlsm needs keep_vba=True to keep macros.

xlsx is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.

← All 155 skills