How to use the Excel workbook helper agent skill (xlsx)

What the xlsx skill makes the agent do

Once installed, the agent can open, read, create and modify Excel workbooks (.xlsx and .xlsm) and bulk CSV or TSV tables. It writes Python code to do this, working with sheets, formulas, styles, charts and multi-sheet workbooks.

The skill uses openpyxl for cells, formulas, styles, charts, merged cells and number formats. It also includes guidance on common pitfalls so the output is more trustworthy.

• Read sheet data and analyze or summarize it
• Add columns, formulas, formatting and charts to an existing workbook
• Clean messy tables such as junk rows or a header that is not on row 1
• Build a new workbook from scratch, including summary sheets

Install the xlsx skill

Download the ZIP for this skill from the page where you found this guide. Unzip it so the skill folder ends up in your personal skills directory ~/.claude/skills/, or in a project's .claude/skills/ folder if you want it limited to one project.

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

The skill's package contains a single file, SKILL.md, which holds the instructions the agent reads.

• Personal install: ~/.claude/skills/ (unzip here)
• Project install: .claude/skills/ inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload

What you need before it works

The skill needs a Python runtime with openpyxl. The source says every supported DeepTutor installation declares openpyxl, but pandas is not guaranteed, so the skill avoids depending on it. It also needs a sandbox shell capability, because the agent runs Python code to do the work.

LibreOffice (soffice) is optional. If it is on your path, the agent can recalculate live formulas. Without it, the agent warns you that formulas will populate when you open the file in Excel.

• Python with openpyxl
• A sandbox shell capability for running code
• Optional: LibreOffice (soffice) on the path, for recalculating live formulas

Trigger the skill with natural requests

You do not need special syntax. Name the spreadsheet and say what you want done with it. The skill is meant to be used whenever a spreadsheet is the input or the deliverable, so requests that mention an Excel or CSV file usually trigger it.

Be specific about the input file, the change you want and where the result should go. The agent will create or revise a file under a relative filename, so state the name you want if it matters to you.

• "Open sales.xlsx and total the Amount column by Region on a new Summary sheet."
• "Clean up contacts.csv: the header is on row 3, drop empty rows, and save the result as contacts_clean.csv."
• "Build a workbook from this table with a bar chart of monthly revenue."
• "Add a formula column to budget.xlsx that calculates the variance against the plan column."

Choose static numbers or live formulas

The most important thing to understand is that openpyxl writes formulas but does not calculate them. A formula is stored as text, so anything that reads cached values before Excel opens the file will see blanks or stale data.

For most jobs the skill computes the result in Python and writes the number, which is correct immediately. Ask for a live model only when the sheet must keep recalculating as someone edits inputs. In that case the agent writes real cell-referenced formulas and, if LibreOffice is available, recalculates them before saving.

• Static numbers: the default choice when correct values are all you need
• Live formulas: use when the user will change inputs later; expect blank cached values unless LibreOffice recalculates or Excel opens the file
• The agent checks output for error values such as #REF!, #DIV/0!, #VALUE!, #NAME? and #N/A

Review the output and avoid common mistakes

The agent writes new files and may overwrite outputs under the same name, so keep a backup of any important or sensitive file before you edit it. The skill also warns that inserting or deleting rows does not update formulas that point at shifted cells, so check formula-heavy sheets after structural changes.

Reading a workbook with data_only=True returns the last values Excel cached, and saving that loaded copy discards its formulas. The skill is written to avoid that trap, but verify the file opens in Excel without a repair prompt.

• Keep a backup of financial or other sensitive files
• Check formula-heavy sheets after inserting or deleting rows or columns
• Open the result in Excel to confirm values and that there is no repair prompt