---

name: data-connector description: Connect to and inspect data sources. Use this skill when you need to verify data access, inspect table schemas, check row counts, or understand the structure of a dataset before performing analysis.

Data Connector

Connect to data sources and retrieve basic information about datasets.

When to Use

• Before starting any data analysis task
• To verify data is accessible and readable
• To inspect column names and types
• To check dataset size (rows, columns, file size)

Available Scripts

data-connect - Inspect Data Source

Connects to a data source and returns schema and summary information.

# Basic usage (outputs to stdout)
data-connect --source <path>

# Save to file
data-connect --source <path> --output report.md

Arguments:

• --source (required): Path to data file or connection string
• --output: Output file path (default: stdout)
• --type: Override source type detection (parquet, csv, json)

Output Format

The script produces a markdown report with:

• Source path and type
• Row count and column count
• File size (if applicable)
• Column listing with data types

Example Output

# Data Connection Report

- **source**: data/sales.parquet
- **type**: parquet
- **row_count**: 1,234,567
- **column_count**: 15
- **file_size**: 45.2 MB

## Columns

| Column | Type |
|--------|------|
| id | INTEGER |
| date | DATE |
| amount | DOUBLE |
| category | VARCHAR |

Supported Data Sources

The connector auto-detects source type from file extension:

• .parquet - Apache Parquet files
• .csv - CSV files (auto-detects delimiter)
• .json, .jsonl - JSON files
• .db, .duckdb - DuckDB database files