chartforge-data-mcp

chartforge-data-mcp

MCP server for data preparation, cleaning, transformation, and statistical analysis — the essential first step before creating charts and visualizations with ChartForge.

Tools

Tool	Description
parse_data	Parse CSV or JSON data and profile columns — types, null counts, unique values, samples
clean_data	Remove duplicates, handle missing values (drop/mean/median/mode/fill), normalize formats
analyze_data	Statistics for numeric columns — mean, median, std dev, percentiles, outlier detection, correlations
transform_data	Aggregate (group by + sum/avg/count), sort, filter, pivot, compute new columns, select/rename
suggest_chart	Recommend the best chart types based on your data structure and purpose

Install

npx chartforge-data-mcp

Configuration

Claude Desktop

{
  "mcpServers": {
    "chartforge-data": {
      "command": "npx",
      "args": ["-y", "chartforge-data-mcp"]
    }
  }
}

Examples

Parse and profile a dataset

parse_data({ data: "Month,Revenue,Expenses\nJan,45000,32000\nFeb,48000,33500\nMar,52000,35000" })

Clean data with missing values

clean_data({ csv: "Name,Score\nAlice,95\nBob,\nCarol,88\nDave,null", missingStrategy: "median" })

Aggregate by category

transform_data({ csv: "Region,Product,Sales\nNorth,A,100\nNorth,B,200\nSouth,A,150", operation: "aggregate", groupBy: "Region", valueColumn: "Sales", aggregation: "sum" })

Get chart recommendations

suggest_chart({ csv: "Quarter,Revenue,Profit\nQ1,2.3M,0.5M\nQ2,2.8M,0.7M\nQ3,3.1M,0.8M\nQ4,3.9M,1.2M", purpose: "show revenue growth" })

Related

• chartforge-generate-mcp — Generate charts from natural language
• chartforge-template-mcp — Chart templates and presets
• chartforge-analytics-mcp — Chart performance analytics

License

MIT