What does this skill do, and when should you use it?
This is the xlsx skill bundled with Open Cowork, an open-source Claude Cowork desktop app, located at .claude/skills/xlsx/ in the repository. It guides an AI agent to work with .xlsx, .xlsm, .csv, and .tsv files using pandas for data analysis and openpyxl for formulas and formatting. Its core discipline is writing real Excel formulas instead of hardcoding computed values, then running the bundled recalc.py script (which requires LibreOffice) to recalculate and scan for #REF!, #DIV/0!, and similar errors, ensuring zero-error deliverables. For financial models it enforces industry-standard color coding and number formatting conventions. It requires a Python environment and LibreOffice.
- Reads and analyzes .xlsx/.csv/.tsv files with pandas (preview, statistics, bulk read/write)
- Creates new workbooks with openpyxl, including formulas, fonts, fills, alignment, and column widths
- Edits existing files via load_workbook while preserving formulas and formatting, across multiple sheets
- Enforces real Excel formulas (=SUM, =AVERAGE, etc.) instead of Python-hardcoded values so spreadsheets stay dynamic
- Runs recalc.py (LibreOffice-based) to recalculate all formulas and return JSON error locations for fixing
- Applies blue/black/green/red text and yellow background color standards, separated assumption cells, and source documentation for hardcodes in financial models
- A data analyst with a pile of .csv/.xlsx exports who needs quick summary statistics and a formula-driven Excel report
- A finance or banking professional building or modifying financial models that must follow industry-standard color coding and number formats
- An operations person updating an existing template report without breaking its formulas or layout
- Anyone who needs formulas (totals, growth rates, averages) batch-added to spreadsheets and verified free of #REF!/#DIV/0! errors
- An Open Cowork desktop-app user who wants to process Excel files in their workspace from a single prompt
- Users without Python + LibreOffice who aren't running Claude Code or Open Cowork — recalc.py can't run, so formulas never recalculate
- Read-only data inspection with no write-back or formulas needed — the workflow is heavier than necessary
- Note: the SKILL.md itself is licensed as Proprietary (terms in LICENSE.txt), unlike the host repo's MIT license; confirm terms before commercial use
How do you install this skill?
- Shell / CLI
- Local filesystem
Python 3pandasopenpyxlLibreOffice (required for formula recalculation via recalc.py)
The skill ships bundled in the Open Cowork repository; no standalone install commands are documented. The direct route is to get the repo:
Claude Code / Open Cowork
git clone https://github.com/OpenCoworkAI/open-cowork.gitThe source does not document commands for installing this skill alone into other hosts; the Open Cowork desktop app automatically loads built-in skills under .claude/skills/, so no manual steps are needed there.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Read sales_2024.csv in this folder, analyze it with pandas, and generate an xlsx summary report with SUM formulas and formatting
- Open quarterly_model.xlsx and change the 2025 revenue growth rate to a formula referencing the assumptions sheet, then recalculate and check for formula errors
- Build a simple revenue projection model for this company: assumptions in separate cells, blue for inputs, black for formulas, then run recalc.py to confirm zero errors
- Update existing_report.xlsx without breaking its formulas or layout — append a totals row styled exactly like the existing template
In an Agent Skills-compatible host, the agent triggers the skill automatically when a request involves spreadsheets, based on the SKILL.md description. The typical flow: 1) choose a tool — pandas for data, openpyxl for formulas/formatting; 2) create or load the workbook; 3) modify data, formulas, and formatting; 4) save; 5) if formulas were used, recalculation is mandatory:
python3 recalc.py output.xlsxThe script auto-configures the LibreOffice macro on first run, recalculates formulas in all sheets, scans every cell for errors, and returns JSON with status (success / errors_found) plus error types and locations (e.g., #REF! at Sheet1!B5). Fix and rerun until clean. An optional timeout argument is supported, e.g. python3 recalc.py output.xlsx 30.
What are this skill's strengths and limitations?
- Enforces real Excel formulas instead of hardcoded values, producing dynamically updateable spreadsheets
- Built-in recalc.py recalculation and error scanning gives a process guarantee of zero formula errors
- Provides industry-standard color coding, number formatting, and assumption-cell conventions for financial models
- openpyxl editing preserves existing formulas, formatting, and template conventions
- Depends on LibreOffice for formula recalculation; without it the core verification loop is unavailable
- The SKILL.md license is Proprietary (see LICENSE.txt), differing from the host repo's MIT license
- Opening with openpyxl data_only=True and saving permanently destroys formulas — a real operational trap
- No automated test suite or evidence of handling large/corrupt files is provided in the source
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| XLSX Spreadsheet Skill (Open Cowork) this page | 69 · Recommended | ★ 2.2k | 12d ago | MIT |
| Three-Statement Model Builder ✓ Anthropic · Official | 55 · Use with care | ★ 40k | 1mo ago | Apache-2.0 |
| 3-Statements-Ultra | 55 · Use with care | ★ 85 | 5mo ago | — |
| Financial Model Auditor ✓ Anthropic · Official | 53 · Use with care | ★ 40k | 1mo ago | Apache-2.0 |
| Investment Banking Pitch Deck Filler ✓ Anthropic · Official | 53 · Use with care | ★ 40k | 1mo ago | Apache-2.0 |
The host README lists it alongside sibling skills pptx, docx, and pdf as built-in document skills; xlsx is specifically responsible for spreadsheet formulas, formatting, and data analysis. No third-party Excel tools are named for comparison.
How did FollowSkills review this skill?
Also a bundled copy of Anthropic's official xlsx skill — content is trustworthy but the version may lag the official source.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
This review was completed under FSRS 1.0. Dimension scores are not converted; it is queued for re-review under FSRS 2.0.
See the full review method →