LlamaSheets CLI

A command-line tool for parsing spreadsheet files using LlamaSheets.
Supports .xls and .xlsx files, with options for advanced extraction and metadata generation.

Installation

pip install llamasheets-cli

Or, with uv, you can do directly:

uv tool install llamasheets-cli

Usage

1. Authenticate

Before using the CLI, you must authenticate with your LlamaCloud API key.

sheets auth --token <YOUR_API_KEY>

Or, for more secure input:

sheets auth --stdin

You will be prompted to enter your API key securely.

If an API key is already configured, you will be warned and asked if you want to overwrite it.

2. Parse a Spreadsheet

Once authenticated, you can parse a spreadsheet:

sheets parse --file path/to/your.xlsx

Options

• --file, -f (required): Path to the .xls or .xlsx file to process.
• --sheet, -s: Name of the sheet(s) to process. Can be used multiple times. Defaults to all sheets.
• --include-hidden-cells/--no-include-hidden-cells: Include hidden cells in extraction. Default: --include-hidden-cells
• --extraction-range: A1 notation of the range to extract (e.g., A1:D20). Default: entire sheet.
• --generate-additional-metadata/--no-generate-additional-metadata: Generate extra metadata (title, description) for each region. Default: --generate-additional-metadata
• --use-experimental-processing/--no-use-experimental-processing: Enable experimental processing. Default: --no-use-experimental-processing
• --flatten-hierarchical-tables/--no-flatten-hierarchical-tables: Flatten hierarchical tables. Default: --no-flatten-hierarchical-tables
• --table-merge-sensitivity [strong|weak]: How aggressively to merge similar regions. Default: strong
• --save-csv/--no-save-csv: Save results as CSV files in the parsed/ folder. Default: --save-csv

Example

sheets parse -f data.xlsx -s Sheet1 -s Sheet2 --extraction-range A1:D20 --no-save-csv

Output

• Results are displayed in your terminal with markdown formatting.
• If --save-csv is enabled, CSV files are saved in the parsed/ directory, named after the extracted region titles.

License and Contributions

This project is distributed under the MIT License.

Contributions are welcome and encouraged, and should follow the guidelines in CONTRIBUTING.md.