Spreadsheets Are Still the Biggest Supply Gap
Spreadsheets are one of the most common things people want an agent to do. After this batch, we have one excellent spreadsheet skill and two mediocre ones.
| Skill | Score | What it covers |
|---|---|---|
| xlsx | 9.8 | Excel workbook files: read, edit, create, format |
| google-sheets | 6.6 | Google Sheets API operations |
| sheets-automation | 5.4 | Generic Sheets automation |
The contrast with the document skills is stark. docx scores 9.7, pptx 9.7, pdf 8.6. The Office file formats have mature, well-engineered skills. The Google Sheets side does not.
Why it matters
The gap is not "no spreadsheet skills". It is "no top-tier skill for the spreadsheet most people actually use in a browser". A Google Sheets task usually involves reading a schema, finding the right range, applying a formula or transformation, writing back, and verifying the result. That is a workflow problem, and the current files stop before the workflow.
xlsx succeeds because it encodes the file-level procedure: inspect the workbook, identify sheets and ranges, preserve formulas and formatting, write, and verify. google-sheets scores 4.5 on workflow because it describes API calls without that sequence.
What a good Google Sheets skill needs
- Schema discovery. Read the sheet names, headers, and data types before writing.
- Range discipline. Never write to a guessed range; resolve it from the schema.
- Formula versus value. Decide whether to write a formula or a computed value, and say why.
- Destructive-write gate. Confirm before overwriting a populated range.
- Verification. Re-read the affected range after writing and report what changed.
None of this is exotic. It is the same pattern as jira applied to a spreadsheet.
The opportunity
If you are writing a skill in this space, the bar is visible. A Google Sheets skill that scores 9+ would fill one of the largest demand gaps in the directory. The existing files show what not to stop at.