Spreadsheet

When to Use

User needs spreadsheet operations: reading data, writing cells, analyzing tables, generating reports, or tracking structured information across Google Sheets, Excel, or CSV files.

Architecture

Memory lives in ~/spreadsheet/. See memory-template.md for setup.

~/spreadsheet/
  memory.md           # Preferences, recent sheets, format rules
  projects/           # Per-project schemas and configs
    {name}.md         # Sheet IDs, columns, formulas
  templates/          # Reusable structures
  exports/            # Generated files

Quick Reference

Scope

This skill ONLY:

• Reads/writes spreadsheets user explicitly requests
• Stores schemas and preferences in ~/spreadsheet/
• Processes files user provides

This skill NEVER:

• Accesses sheets without user request
• Stores passwords, API keys, or sensitive financial data
• Modifies files outside ~/spreadsheet/ or user paths

Data Storage

All data stored in ~/spreadsheet/. Create on first use:

mkdir -p ~/spreadsheet/{projects,templates,exports}

Self-Modification

This skill NEVER modifies its own SKILL.md. All user data stored in ~/spreadsheet/ only.

Core Rules

1. Schema First

On first access to any sheet:

1. Document columns (name, type, sample)
2. Save to projects/{name}.md
3. Reference schema in future ops

2. Format Preservation

3. Large Data Strategy

4. Integration Priority

1. Google Sheets - if API configured
2. Excel (.xlsx) - local files, use openpyxl
3. CSV - universal fallback

5. Memory Updates

Common Traps

• Truncating without warning - Always confirm before loading >1000 rows
• Losing formulas - Use data_only=False in openpyxl, read formulas separately
• Schema drift - Re-verify if last access >7 days
• Rate limits - Batch Google Sheets requests, max 100/100s
• Encoding - Default UTF-8, check for BOM on European files
• Empty cells - Google API omits them; pandas fills with NaN