Skip to content
Get packs

Guide · Agent skill

How to use the XLSX Spreadsheet Toolkit agent skill (xlsx-official)

Updated · written with AI from the project's own documentation and checked against our catalogue data

The xlsx-official skill is a bundle of guidance documents and small Python helper scripts that an AI agent loads when a task involves a workbook file such as .xlsx, .xlsm, .xltx, .csv or .tsv. It is for finance and ops staff who want an agent to build or update a model or report, analysts cleaning messy spreadsheets, and developers who need to inspect or repair .xlsx files. After this guide you will know how to install the skill, how to phrase a request so the agent uses it, what it needs on your machine, and what the scripts can do to your files.

XLSX Spreadsheet ToolkitXiaomiMiMo/MiMo-Code · MIT

Quick start

  1. 1Download the ZIP. Download the xlsx-official ZIP from this skill's page on the directory.
  2. 2Unzip the skill. Unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
  3. 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
  4. 4Install Python libraries. Install openpyxl, pandas and lxml with pip. Add xlsxwriter if you want it.
  5. 5Install LibreOffice if needed. Install LibreOffice (soffice) if you need formula recalculation or PDF export.
  6. 6Ask in plain language. Describe the workbook, its path and the result you want, such as a formula-driven model or a CSV export.
  7. 7Check the result. Ask the agent to validate the file and review the output before you rely on it.

What an agent skill is and what this one does

An agent skill is a folder of instructions and scripts that an AI agent can load when a task matches the skill's description. This one is named xlsx-official and is licensed under MIT on the source page. Its SKILL.md file describes spreadsheet work as a live calculation surface: the agent should write formulas rather than pasted numbers, put assumptions in dedicated input cells, and keep one sheet per idea.

The skill also tells the agent to check its own output. The bundled scripts can describe a workbook, recalculate formulas and flag errors such as #REF! and #DIV/0!, validate the ZIP and XML structure, convert sheets to CSV or PDF, and unpack and repack the raw XML inside an .xlsx for detailed edits.

  • Builds workbooks with live formulas, such as =SUM(...), instead of values computed in Python
  • Applies named styles, number formats and frozen header rows
  • Recalculates and checks for broken formulas (needs LibreOffice)
  • Exports sheets to CSV or a workbook to PDF (PDF needs LibreOffice)
  • Unpacks and repacks .xlsx internals for surgical edits

Install the skill

You can install the skill in two ways. For a personal setup, download the ZIP from the skill's page on this directory and unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder instead.

If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

  • Personal: unzip into ~/.claude/skills/
  • Project only: unzip into .claude/skills/ inside the project folder
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Install the Python requirements and LibreOffice

The skill needs Python 3, openpyxl, pandas and lxml. xlsxwriter is optional. LibreOffice (soffice) is optional but required for formula recalculation and PDF export, through bake.py and pdf_out.py.

If your environment provides a bundled Python with these libraries, the source says to set the MIMO_PYTHON variable and run commands through it instead of running pip. The source also mentions a MIMO_SOFFICE variable for a bundled LibreOffice.

  • Required: Python 3, openpyxl, pandas, lxml
  • Optional: xlsxwriter
  • Needed for recalc and PDF: LibreOffice (soffice)
python3 -m pip install --upgrade openpyxl pandas lxml python3 -m pip install --upgrade xlsxwriter

Trigger the skill with natural requests

You do not need to name the skill. The skill's description lists typical triggers, so the agent will usually load it when the workbook is the thing being produced, changed, cleaned or read. Describe the file and the result you want.

The skill is meant to be skipped when a spreadsheet is only source material and the real output is a Word doc, an HTML page, a standalone Python script, a Google Sheets integration or a database ingestion pipeline. In those cases ask for the deliverable you actually want.

  • 'Build me a model with revenue, costs and a margin row, all formulas.'
  • 'Update this sheet so the Q3 column pulls from the Inputs tab.'
  • 'Add a column for variance % and compute the totals as formulas.'
  • 'Sanity-check this xlsx for broken formulas.'
  • 'Export sheet 2 of report.xlsx to CSV.'
  • 'Render the workbook as PDF so I can look at it.'
  • 'The spreadsheet in ~/Downloads is a mess, fix it.'

Useful commands the skill runs

The skill's SKILL.md lists these commands. The agent runs them on your files, so it helps to know what they do. Each script is a small self-contained Python file and its top lines list the full CLI options.

Recalculation writes results into the file, so keep a copy of anything important before asking for a bake or a PDF export.

  • overview.py: describes sheets, dimensions, formula count and sample rows
  • bake.py: recalculates formulas with LibreOffice and flags residual errors; default timeout 30s, adjustable with --timeout
  • audit.py: validates ZIP integrity, XML well-formedness and openpyxl load
  • csv_out.py: writes one CSV per sheet or a single sheet with --sheet
  • pdf_out.py: writes a PDF next to the workbook
  • explode.py and assemble.py: unpack an .xlsx to XML parts and repack it
python scripts/overview.py input.xlsx python scripts/bake.py output.xlsx --timeout 60 python scripts/audit.py output.xlsx python scripts/csv_out.py input.xlsx out.csv --sheet 0

Safety notes: scripts and network

The skill runs scripts on your files. The source lists this skill as running scripts: yes and using network: no. Before you let the agent run bake.py or pdf_out.py on an important file, make a copy, because those steps recalculate and rewrite the workbook.

The source also warns that on machines without LibreOffice, bake.py and pdf_out.py will not work, so the agent may run them on your files and fail partway. Check that soffice is installed, or skip recalculation and PDF requests.

  • Back up the original workbook before recalculation or repacking
  • Confirm LibreOffice is installed before asking for PDF or recalc
  • Review the scripts in the skill folder if you need to know exactly what runs

Tips for better results

Name the file path and the sheet you mean. Say which cells or columns are inputs. If you want the agent to keep formulas, say so, because the skill's own guidance is to avoid writing computed numbers into cells.

For template fills, give the template and the data source. For cleanups, describe the mess and what a clean result looks like. The skill notes that openpyxl does not evaluate formulas, so ask for a recalc and check step when the numbers matter.

  • Give the file path, such as ~/Downloads/file.xlsx
  • Say which cells are assumptions the user will change
  • Ask for a validation pass at the end
  • Ask for CSV output when another tool needs the data

What people use it for

  • Building a financial model with an inputs sheet, a calculations sheet and an output sheet linked by formulas
  • Updating an ops report by adding a column and filling totals as live formulas
  • Cleaning a messy spreadsheet in ~/Downloads and writing a tidy version back
  • Exporting one sheet of a workbook to CSV for another tool
  • Checking a workbook for broken formulas or corrupt structure before sharing it

Frequently asked questions

Is the xlsx-official skill free to use?

The source page lists the skill's licence as MIT. The skill's SKILL.md mentions Apache-2.0 for its own toolkit and attributions, so read the LICENSE file in the ZIP for the exact terms before reusing it.

What do I need to install before using it?

You need Python 3, openpyxl, pandas and lxml. xlsxwriter is optional. LibreOffice (soffice) is needed only for formula recalculation with bake.py and PDF export with pdf_out.py.

Where do I put the skill folder?

Unzip the downloaded ZIP into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps you can instead upload the ZIP under Settings → Capabilities → Skills.

Does the skill use the internet?

The source page lists the skill as not using the network. It does run Python scripts on your files, so the main risk is local changes to the workbook rather than data leaving your machine.

Why does the agent need LibreOffice for recalculation?

The skill's guidance says openpyxl does not evaluate formulas, so freshly written formulas have no cached value until LibreOffice or Excel recalculates them. bake.py uses LibreOffice for that step.

Can I use it when the real output is a Word doc or a database pipeline?

The skill says to skip it in those cases, because the spreadsheet is then only a means and not the deliverable. Ask for the Word doc, web page, script or pipeline directly.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you regularly have an agent build, update, clean or check Excel-style workbooks and you can install LibreOffice where recalculation and PDF export are needed. Skip it if your spreadsheet is only input to a Word doc, web page, standalone script, Google Sheets integration or database pipeline, or if you cannot run LibreOffice and do not want the agent running scripts on your files.

Download the skill