FastMCP Data Viewer

A data viewer tool built with FastMCP that allows you to inspect the contents and summary of various data formats. This initial version supports Parquet files.

Features

• View a summary of a Parquet file, including column names, types, row count, file size, and a preview of the first few rows.
• Retrieve a specific row (item) from a Parquet file by its index.

Requirements

The project requires the following Python packages:

• fastmcp
• pandas
• pyarrow

You can install them using pip:

pip install -r requirements.txt

Installation

This project is designed to be used as a Gemini CLI tool. An installation script is provided for convenience.

1. Make sure you have the Gemini CLI and its dependencies installed.
2. Run the installation script:

bash install.sh

This will install the tool with the name data-viewer.

Usage

Once installed, you can interact with the data viewer through the Gemini CLI. The following tools are available:

get_summary

Provides a summary of a Parquet file.

Arguments:

• path (str): The path to the Parquet file.
• preview_rows (int, optional): The number of rows to include in the preview. Defaults to 2.
• max_col_width (int, optional): The maximum character width for text columns in the preview. Truncates long strings. Defaults to 200. Set to 0 to disable truncation.

Example:

@data-viewer get_summary path='test.parquet'

get_item

Retrieves a specific item (row) from a Parquet file by its index.

Arguments:

• path (str): The path to the Parquet file.
• index (int): The 0-based index of the item to retrieve.

Example:

@data-viewer get_item path='test.parquet' index=0