name: osaurus-xlsx description: Read, create, and modify Excel (.xlsx) spreadsheets using the Osaurus XLSX plugin. Use when working with .xlsx files, converting between CSV and Excel, or performing batch spreadsheet operations. metadata: author: tpae version: "0.1.0"

Osaurus XLSX

This plugin reads, creates, and modifies Excel (.xlsx) files. It uses a stateful in-memory model — you load or create a workbook, manipulate it, then save it to disk.

Workflow

Always follow this sequence:

1. read_xlsx(path) or create_xlsx(sheets) — get a workbook_id
2. write_cells / modify_xlsx — mutate the workbook in memory (optional, repeat as needed)
3. save_xlsx(workbook_id, path) — write the final .xlsx file to disk

For one-shot read operations, list_sheets and xlsx_to_csv read directly from disk without needing the full workflow.

Quick Reference

Reading

Writing

Transformation

Cell References

All cell references use standard Excel notation:

• Single cell: A1, B5, Z100
• Range: A1:D10, B2:B50
• Column letters: A=1, Z=26, AA=27, AZ=52, etc.

Value Types

When writing cell values, types are auto-detected from the string value:

You can override auto-detection with the type parameter in write_cells:

• "string" — force string
• "number" — force number
• "boolean" — force boolean
• "formula" — force formula (adds = prefix if missing)

Tool Tips

read_xlsx

• Returns all sheets by default. Use sheet_name to read a single sheet.
• Use range (e.g. "A1:D10") to limit the data returned — useful for large files.
• The returned workbook_id is needed for write_cells, modify_xlsx, get_cell_value, and save_xlsx.
• Sheet name matching is case-insensitive.

get_cell_value

• Requires a workbook_id from a prior read_xlsx or create_xlsx call.
• You must provide either cell (e.g. "B5") or range (e.g. "A1:C3") — omitting both returns an error.
• Sheet name matching is case-insensitive.

list_sheets

• Reads sheet names directly from disk — no workbook_id needed.
• Useful for discovering sheet names before calling read_xlsx with a specific sheet_name.

create_xlsx

• Pass an array of sheet definitions, each with a name, optional headers, and optional rows.
• headers are always stored as strings.
• rows values are auto-detected (numbers, booleans, strings).
• Remember to call save_xlsx afterwards — nothing is written to disk until you save.

write_cells

• If the specified sheet_name doesn't exist, a new sheet is created automatically.
• Each cell in the cells array needs a ref (e.g. "B5") and a value.
• Use type to force a specific value type if auto-detection isn't what you want.

save_xlsx

• The .xlsx extension is added automatically if missing.
• Always call this when you're done. The workbook only exists in memory until saved.

xlsx_to_csv

• Reads directly from a file on disk — no need to read_xlsx first.
• Defaults to the first sheet. Use sheet_name to specify a different one.
• Set format to "tsv" for tab-separated output.

csv_to_xlsx

• Accepts raw CSV/TSV text in the csv_data parameter.
• Set has_header: true to keep the first row as strings (prevents number detection on headers).
• Set delimiter to "tab" or "\t" for TSV input.
• Returns a workbook_id — call save_xlsx to write to disk.

modify_xlsx

• Batch multiple operations in a single call for efficiency.
• If sheet_name is omitted, operations target the first sheet in the workbook.
• Available operations:
  • set_cell — set a cell value (ref + value)
  • set_formula — set a formula (ref + formula)
  • delete_row — delete a row by number and shift rows up (row)
  • delete_column — delete a column by letter and shift columns left (column)
  • rename_sheet — rename the target sheet (new_name)
  • add_sheet — add a new empty sheet (name)

Limitations

1. No cell formatting. The plugin reads and writes cell values only. Fonts, colors, borders, and other formatting from existing files are not preserved.

2. No images or charts. Embedded objects in existing files are not read or preserved.

3. Formulas are stored, not evaluated. When you write =SUM(A1:A10), the formula text is saved. Excel will evaluate it when the file is opened.

4. Large files. Reading very large spreadsheets (100k+ rows) loads all data into memory. Use the range parameter with read_xlsx to limit data.

5. Round-trip fidelity. Reading an existing .xlsx and saving it back will preserve cell values but may lose formatting, charts, images, and other advanced features.

Example: Read and Summarize

1. read_xlsx(path="sales.xlsx")
   → workbook_id, sheets with cell data

2. Agent analyzes the data and responds with a summary

Example: Create a New Spreadsheet

1. create_xlsx(sheets=[
     {name: "Employees", headers: ["Name", "Department", "Salary"],
      rows: [["Alice", "Engineering", "120000"],
             ["Bob", "Marketing", "95000"]]}
   ])
   → workbook_id

2. save_xlsx(workbook_id, path="employees.xlsx")
   → file written to disk

Example: Read, Modify, and Save

1. read_xlsx(path="budget.xlsx")
   → workbook_id

2. write_cells(workbook_id, sheet_name="Q4",
     cells=[{ref: "B10", value: "=SUM(B1:B9)", type: "formula"}])
   → cell written

3. save_xlsx(workbook_id, path="budget_updated.xlsx")
   → file written to disk

Example: CSV to Excel

1. csv_to_xlsx(csv_data="Name,Age\nAlice,30\nBob,25", has_header=true)
   → workbook_id

2. save_xlsx(workbook_id, path="people.xlsx")
   → file written to disk

Example: Batch Modifications

1. read_xlsx(path="data.xlsx")
   → workbook_id

2. modify_xlsx(workbook_id, sheet_name="Sheet1", operations=[
     {type: "set_cell", ref: "A1", value: "Updated Title"},
     {type: "set_formula", ref: "D10", formula: "=AVERAGE(D1:D9)"},
     {type: "delete_row", row: 5},
     {type: "add_sheet", name: "Summary"},
     {type: "rename_sheet", new_name: "Data"}
   ])
   → 5 operations applied

3. save_xlsx(workbook_id, path="data_modified.xlsx")
   → file written to disk