Xlsx Specialist Skill

---
name: xlsx-specialist-skill
description: >-
  Spreadsheet-first tasks — read, edit, fix, create .xlsx/.xlsm/.csv/.tsv
  (formulas, formatting, charts, data cleaning). Not Word docs, HTML reports,
  scripts, or Google Sheets.
license: Apache-2.0
compatibility: opencode
category: Framework
---

## What I do

- Create, edit, and manipulate Excel spreadsheets (.xlsx, .xlsm) and other tabular formats (.csv, .tsv)
- Build financial models with industry-standard color coding, number formatting, and formula construction
- Recalculate all formulas using LibreOffice and verify zero Excel errors (#REF!, #DIV/0!, #VALUE!, #N/A, #NAME?)
- Analyze data with pandas and build spreadsheets with openpyxl for complex formatting and formulas
- Convert between tabular file formats while preserving data integrity
- Clean and restructure messy tabular data into properly formatted spreadsheets

## When to use me

Use me when you need to work with spreadsheet files as the primary deliverable:

- **Creating spreadsheets**: Build new Excel files from scratch or convert other data sources (JSON, CSV, databases) to .xlsx
- **Editing spreadsheets**: Modify existing .xlsx files (add columns, compute formulas, apply formatting, create charts, clean data)
- **Financial modeling**: Build professional financial models with proper color coding (blue=hardcoded, black=formulas, green=cross-sheet, red=external, yellow=assumptions)
- **Formula verification**: Recalculate and verify all formulas work correctly with zero errors
- **Data analysis**: Read and analyze Excel data with pandas for statistics, filtering, transformations
- **Data cleaning**: Fix malformed rows, misplaced headers, junk data in tabular files
- **Format conversion**: Convert between .xlsx, .xlsm, .csv, .tsv formats

Do NOT use me when the primary deliverable is:
- Word documents (.docx)
- PDFs
- HTML reports
- Standalone Python scripts
- Database pipelines
- Google Sheets API integration (even if tabular data is involved)

## Prerequisites

### Required Tools

- **LibreOffice** (for formula recalculation)
  - Automatically configured on first run of `scripts/recalc.py`
  - Works in sandboxed environments with Unix socket restrictions

### Python Libraries

- **openpyxl**: For creating, reading, and editing Excel files with formulas and formatting
- **pandas**: For data analysis, bulk operations, and simple data export

Install dependencies:
```bash
pip install openpyxl pandas
```

### LibreOffice Configuration

The `scripts/recalc.py` script automatically:
1. Sets up LibreOffice macro for formula recalculation on first run
2. Handles sandboxed environments with Unix socket restrictions
3. Works on both Linux and macOS

## Steps

### Step 1: Choose the Right Tool

- **Use pandas** for data analysis, bulk operations, and simple data export
- **Use openpyxl** for complex formatting, formulas, and Excel-specific features

### Step 2: Create or Load the Spreadsheet

#### Creating New Files
```python
from openpyxl import Workbook
from openpyxl.styles import Font, PatternFill, Alignment

wb = Workbook()
sheet = wb.active
```

#### Loading Existing Files
```python
from openpyxl import load_workbook

wb = load_workbook('existing.xlsx')
sheet = wb.active  # or wb['SheetName'] for specific sheet
```

### Step 3: Add Data and Formulas

**CRITICAL: Use Excel Formulas, Not Hardcoded Values**

Always use Excel formulas so the spreadsheet remains dynamic and updateable.

❌ WRONG - Hardcoding Calculated Values
```python
total = df['Sales'].sum()
sheet['B10'] = total  # Hardcodes 5000
```

✅ CORRECT - Using Excel Formulas
```python
sheet['B10'] = '=SUM(B2:B9)'
```

Add formulas:
```python
sheet['B2'] = '=(C4-C2)/C2'      # Growth rate
sheet['C5'] = '=AVERAGE(D2:D19)'   # Average
sheet['D10'] = '=SUM(A1:A10)'      # Sum
```

### Step 4: Apply Formatting

#### Professional Font
Use a consistent, professional font:
```python
from openpyxl.styles import Font

sheet['A1'].font = Font(name='Arial', size=11)
```

#### Financial Model Color Coding

Apply industry-standard color conventions:
```python
from openpyxl.styles import Font, PatternFill

# Blue text (RGB: 0,0,255): Hardcoded inputs
sheet['B5'].font = Font(color='0000FF')

# Black text (RGB: 0,0,0): ALL formulas and calculations
sheet['B6'].font = Font(color='000000')

# Green text (RGB: 0,128,0): Links from other worksheets
sheet['C10'].font = Font(color='008000')

# Red text (RGB: 255,0,0): External links to other files
sheet['D15'].font = Font(color='FF0000')

# Yellow background (RGB: 255,255,0): Key assumptions
sheet['E20'].fill = PatternFill('solid', start_color='FFFF00')
```

#### Number Formatting

**Years**: Format as text strings (e.g., "2024" not "2,024")

**Currency**: Use `$#,##0` format, specify units in headers ("Revenue ($mm)"):
```python
# Apply custom number format
sheet['B10'].number_format = '$#,##0;($#,##0);-'

# For zeros, use number format to display as "-"
sheet['B15'].number_format = '$#,##0;($#,##0);-'
```

**Percentages**: Default to 0.0% format (one decimal):
```python
sheet['C5'].number_format = '0.0%'
```

**Multiples**: Format as 0.0x for valuation multiples (EV/EBITDA, P/E):
```python
sheet['D20'].number_format = '0.0x'
```

**Negative numbers**: Use parentheses `(123)` not minus `-123`:
```python
sheet['E10'].number_format = '#,##0;(#,##0)'
```

### Step 5: Formula Construction Rules

#### Assumptions Placement
Place ALL assumptions (growth rates, margins, multiples, etc.) in separate assumption cells. Use cell references instead of hardcoded values in formulas.

Example:
```python
# WRONG: =B5*1.05
sheet['C5'] = '=B5*1.05'  # Hardcoded growth rate

# CORRECT: =B5*(1+$B$6)  where B6 contains 0.05
sheet['C5'] = '=B5*(1+$B$6)'  # References assumption cell
```

#### Documentation for Hardcodes
Comment or note beside cells with hardcoded values:

Format: "Source: [System/Document], [Date], [Specific Reference], [URL if applicable]"

Examples:
- "Source: Company 10-K, FY2024, Page 45, Revenue Note, [SEC EDGAR URL]"
- "Source: Company 10-Q, Q2 2025, Exhibit 99.1, [SEC EDGAR URL]"
- "Source: Bloomberg Terminal, 8/15/2025, AAPL US Equity"
- "Source: FactSet, 8/20/2025, Consensus Estimates Screen"

### Step 6: Save the Spreadsheet

```python
wb.save('output.xlsx')
```

### Step 7: Recalculate Formulas (MANDATORY IF USING FORMULAS)

Excel files created or modified by openpyxl contain formulas as strings but not calculated values. Use the `scripts/recalc.py` script to recalculate formulas:

```bash
python scripts/recalc.py output.xlsx 30
```

The script:
- Automatically sets up LibreOffice macro on first run
- Recalculates all formulas in all sheets
- Scans ALL cells for Excel errors (#REF!, #DIV/0!, #VALUE!, #NAME?, #NULL!, #NUM!, #N/A)
- Returns JSON with detailed error locations and counts
- Works on both Linux and macOS

### Step 8: Verify and Fix Any Errors

Interpret the script output:

```json
{
  "status": "success",           // or "errors_found"
  "total_errors": 0,              // Total error count
  "total_formulas": 42,           // Number of formulas in file
  "error_summary": {              // Only present if errors found
    "#REF!": {
      "count": 2,
      "locations": ["Sheet1!B5", "Sheet1!C10"]
    }
  }
}
```

If `status` is `"errors_found"`:
1. Check `error_summary` for specific error types and locations
2. Fix identified errors:
   - `#REF!`: Invalid cell references
   - `#DIV/0!`: Division by zero
   - `#VALUE!`: Wrong data type in formula
   - `#NAME?`: Unrecognized formula name
3. Recalculate again until `status` is `"success"` and `total_errors` is 0

### Step 9: Final Verification

Use the Formula Verification Checklist:

#### Essential Verification
- [ ] **Test 2-3 sample references**: Verify they pull correct values before building full model
- [ ] **Column mapping**: Confirm Excel columns match (e.g., column 64 = BL, not BK)
- [ ] **Row offset**: Remember Excel rows are 1-indexed (DataFrame row 5 = Excel row 6)

#### Common Pitfalls
- [ ] **NaN handling**: Check for null values with `pd.notna()`
- [ ] **Far-right columns**: FY data often in columns 50+
- [ ] **Multiple matches**: Search all occurrences, not just first
- [ ] **Division by zero**: Check denominators before using `/` in formulas (#DIV/0!)
- [ ] **Wrong references**: Verify all cell references point to intended cells (#REF!)
- [ ] **Cross-sheet references**: Use correct format (Sheet1!A1) for linking sheets

#### Formula Testing Strategy
- [ ] **Start small**: Test formulas on 2-3 cells before applying broadly
- [ ] **Verify dependencies**: Check all cells referenced in formulas exist
- [ ] **Test edge cases**: Include zero, negative, and very large values

## Best Practices

### Library Selection

| Task | Recommended Library | Why |
|------|---------------------|-----|
| Data analysis, bulk operations, simple export | **pandas** | Powerful data manipulation, easy to use |
| Complex formatting, formulas, Excel-specific features | **openpyxl** | Preserves formulas, supports styling |

### Working with openpyxl

- Cell indices are 1-based (row=1, column=1 refers to cell A1)
- Use `data_only=True` to read calculated values: `load_workbook('file.xlsx', data_only=True)`
- **Warning**: If opened with `data_only=True` and saved, formulas are replaced with values and permanently lost
- For large files: Use `read_only=True` for reading or `write_only=True` for writing
- Formulas are preserved but not evaluated - use `scripts/recalc.py` to update values

### Working with pandas

- Specify data types to avoid inference issues: `pd.read_excel('file.xlsx', dtype={'id': str})`
- For large files, read specific columns: `pd.read_excel('file.xlsx', usecols=['A', 'C', 'E'])`
- Handle dates properly: `pd.read_excel('file.xlsx', parse_dates=['date_column'])`

### Preserve Existing Templates

When updating templates:
- Study and EXACTLY match existing format, style, and conventions
- Never impose standardized formatting on files with established patterns
- Existing template conventions ALWAYS override these guidelines

## Design Aesthetics — Avoiding Generic AI Slop

**CRITICAL**: Spreadsheets must look intentionally designed and professionally formatted, not like raw data dumps or default Excel output. Apply these principles alongside the technical formatting rules above.

### Anti-Patterns to AVOID

- **Default Excel aesthetics**: Avoid spreadsheets that look like unformatted data — no default column widths, no Calibri 11pt for everything, no gridlines showing through data, no generic "Sheet1" tab names
- **Raw data dumps**: Do not produce spreadsheets where data is simply dumped into cells without structure. Every sheet needs a clear visual hierarchy: title row, header row, data rows, summary rows
- **Generic formatting**: Avoid using the same font size and color throughout. Financial models need blue/black/green color coding; dashboards need accent colors for KPIs
- **Template-looking layouts**: Default Excel tables with no styling, random column widths, and no clear sections scream "auto-generated." Design with intentional grouping, borders, and section headers
- **Overused color choices**: No default Excel blue (#4472C4) for all charts. No random rainbow palettes for data series. Choose colors that enhance readability and convey meaning
- **Missing visual hierarchy**: A flat wall of numbers is unreadable. Create clear distinction between headers, data, totals, and assumptions through font weight, color, borders, and shading
- **Chart defaults**: Never use default Excel chart styling (blue bars, gray gridlines, legend on right). Style every chart with intention — custom colors, data labels, minimal gridlines

### Signature Design Elements

Every spreadsheet should incorporate at least ONE distinctive design choice:

- **Professional color scheme**: A consistent 3-5 color palette applied across all sheets — not random Excel defaults
- **Clear section architecture**: Sheets divided into labeled sections (Assumptions, Calculations, Output, Summary) with bold section headers and visual separation
- **Styled headers**: Header rows with background fills and white bold text, frozen panes, and filter-enabled tables
- **KPI dashboard section**: Key metrics highlighted in large font with conditional formatting or colored backgrounds
- **Intentional number formatting**: Consistent decimal places, currency symbols aligned, parentheses for negatives, dash for zeros
- **Clean chart styling**: Charts with branded colors, data labels instead of legends, no unnecessary gridlines, meaningful titles

### Differentiation Strategy

Before formatting, ask:
1. **Who reads this spreadsheet?** A CEO wants summary KPIs; an analyst wants detailed formulas
2. **What decisions does this support?** Design to make the decision point obvious and accessible
3. **Does this look professional or auto-generated?** If it looks like raw data output, it needs more formatting
4. **Can someone understand the structure in 10 seconds?** If not, add section headers, grouping, and visual hierarchy

### Color Palette Recommendations

| Use Case | Recommended Palette | Characteristics |
|---|---|---|
| Financial Model | Blue/Black/Green coding | Blue=inputs, Black=formulas, Green=cross-sheet |
| Dashboard | 3-4 accent colors + neutrals | One dominant, one accent for alerts, neutrals for structure |
| Data Analysis | Muted professional tones | Gray structure, blue highlights, green for positive, red for negative |
| Project Tracker | Clean 2-tone + status colors | Header color, alternating row tint, RAG status colors |
| Reporting Pack | Corporate branded | Match company colors, consistent across all sheets |

### Chart Styling Guidelines

- **Color**: Use 2-3 colors maximum per chart. Match the spreadsheet's overall palette
- **Data labels**: Prefer data labels over legends. Place labels directly on data points
- **Gridlines**: Minimal or no gridlines. If needed, use light gray (#E0E0E0) horizontal gridlines only
- **Titles**: Every chart needs a clear, descriptive title. No "Chart 1" defaults
- **Axes**: Clean axis labels with proper number formatting. Avoid cluttered tick marks
- **Background**: White chart background, no default Excel border

## Common Issues

### LibreOffice Not Installed

**Error**: Failed to setup LibreOffice macro

**Solution**: Install LibreOffice:
```bash
# Linux (Ubuntu/Debian)
sudo apt-get install libreoffice

# macOS
brew install --cask libreoffice
```

### Sandbox Environment Issues

**Symptom**: LibreOffice fails to start or macro doesn't work

**Solution**: The script automatically handles sandboxed environments with Unix socket restrictions. Ensure you're using the latest version of the script that includes sandbox detection from `scripts/soffice.py`.

### Formula Errors After Recalculation

**Common Errors**:
- `#REF!`: Invalid cell references - check for deleted cells/rows/columns
- `#DIV/0!`: Division by zero - add checks for zero denominators
- `#VALUE!`: Wrong data type - ensure cells referenced in formulas contain correct types
- `#NAME?`: Unrecognized formula name - check for typos in formula names

**Solution**: Use the error locations from `scripts/recalc.py` output to identify and fix specific cells.

### Formulas Not Calculating

**Symptom**: Formulas show as text strings (e.g., "=SUM(A1:A10)") instead of calculated values

**Solution**: Always run `scripts/recalc.py` after saving the file to calculate formula values.

### Performance Issues with Large Files

**Symptom**: Slow processing of large Excel files

**Solution**:
- Use `read_only=True` when reading: `load_workbook('large.xlsx', read_only=True)`
- Use `write_only=True` when writing: `Workbook(write_only=True)`
- For pandas: Read specific columns with `usecols` parameter

## Verification Commands

After completing your spreadsheet work:

```bash
# Recalculate formulas and check for errors
python scripts/recalc.py output.xlsx

# Verify file exists
ls -lh output.xlsx

# Open in LibreOffice for manual review (if needed)
soffice output.xlsx
```