Data Wrangler MCP Server

Interactive CSV/JSON data explorer with column type detection, sorting, filtering, and summary statistics.

Features

• Auto Format Detection - Automatically detects CSV or JSON format with fallback parsing
• Intelligent Type Detection - Identifies column types (number, string, date, boolean) with 80% threshold matching
• Advanced Filtering - Filter data by equality, comparison (gt/lt), text matching (contains/startsWith)
• Column Statistics - Compute unique counts, null counts, min/max values, mean, median, and top value frequencies
• Interactive UI - Browse parsed data in a responsive table with sortable columns
• Batch Analytics - Get comprehensive statistics for any column including distribution analysis

Installation

npm install @gongrzhe/server-data-wrangler

Usage

As a CLI

npx @gongrzhe/server-data-wrangler

Claude Desktop Configuration

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "data-wrangler": {
      "command": "npx",
      "args": ["-y", "@gongrzhe/server-data-wrangler"]
    }
  }
}

Claude Code Configuration

claude mcp add data-wrangler -- npx -y @gongrzhe/server-data-wrangler

Tools

parse-data

Parses CSV or JSON data with auto-detection of column types (number, string, date, boolean). Returns structured table data with statistics.

Input:

• data (string): The raw CSV or JSON data to parse
• format (optional): Data format - "csv", "json", or "auto" (default: "auto")

Output:

• columns: Array of column definitions with name, detected type, and null count
• rows: Array of objects representing data rows
• rowCount: Total number of rows parsed
• stats: Statistical summary for each column

filter-data

Applies a filter to the last parsed dataset. Supports equality, comparison, contains, and startsWith operators.

Input:

• column (string): Column name to filter on
• operator (string): Filter operator - "eq", "neq", "gt", "lt", "contains", "startsWith"
• value (string): Value to compare against

Output:

• Filtered dataset with updated row count and statistics

get-column-stats

Returns detailed statistics for a specific column including unique count, nulls, min/max, mean, median, and top values.

Input:

• column (string): Column name to get statistics for

Output:

• column: Column name
• type: Detected column type
• uniqueCount: Number of unique non-null values
• nullCount: Number of null/empty values
• min/max: Minimum and maximum values
• mean/median: Statistical averages (numeric columns only)
• topValues: Array of most frequent values with their counts

Example Prompt

Parse this CSV data:

name,age,salary
Alice,32,125000
Bob,28,85000
Carol,45,155000

Then filter for employees with salary > 100000 and show me the statistics for the age column.

License

MIT