Quick Data MCP Server

How to use

You interact with the Quick Data MCP server through Claude Desktop. Start the server inside Claude Desktop and then issue commands to load datasets, run analyses, and visualize results. Typical workflows include loading a CSV or JSON file, performing correlations or distributions analyses, and creating charts or dashboards.

Suggested usage patterns you can run in Claude Desktop after setup: - Load a dataset: Load my sales data: C:\Users\YourName\Documents\sales_data.csv as "sales" - Basic analysis: Show me correlations in the sales dataset - Visualization: Create a bar chart of sales by region - Advanced analytics: Generate dashboard with revenue trends and regional breakdown

How to install

# Prerequisites
- Windows 10/11
- Python 3.9+
- Claude Desktop

# 1) Download or clone the project
# (Use the exact commands shown here to reproduce the setup)
git clone https://github.com/Beaulewis1977/quick-data-for-windows-mcp.git
cd quick-data-for-windows-mcp

# 2) Install dependencies
install_dependencies.bat

# 3) Test the server standalone (optional but recommended)
test_server.bat

# 4) Configure Claude Desktop
copy claude_desktop_config_fixed.json "%APPDATA%\Claude\claude_desktop_config.json"

# 5) Edit the config to point to your installation (cwd) if needed
"IMPORTANT" Edit the config file and update the cwd path to your actual installation directory.

# 6) Restart Claude Desktop

Configuration and troubleshooting notes

If you need manual configuration, you can place a runtime configuration in Claude Desktop to point to the local Python script. The server is started via Python and expects the working directory to be the project path.

{
  "mcpServers": {
    "quick_data_win": {
      "command": "python",
      "args": [
        "C:\\path\\to\\your\\quick-data-for-windows-mcp\\main.py"
      ],
      "cwd": "C:\\path\\to\\your\\quick-data-for-windows-mcp",
      "env": {
        "LOG_LEVEL": "INFO",
        "PYTHONPATH": "C:\\path\\to\\your\\quick-data-for-windows-mcp\\src"
      }
    }
  }
}

Troubleshooting and testing

If you encounter ModuleNotFoundError: No module named 'mcp', verify Python is accessible and dependencies are installed. Ensure Python is in your PATH and that the required packages are present.

To test the server independently of Claude Desktop, run: - python main.py Expected output: `` Quick Data for Windows MCP v1.0.0 Server running on stdio... ``

load_dataset

Load CSV/JSON files with automatic schema discovery into memory for analysis.

list_loaded_datasets

List all datasets currently loaded in memory.

get_dataset_info

Retrieve detailed information about a loaded dataset.

clear_dataset

Remove a specific dataset from memory.

clear_all_datasets

Clear all loaded datasets from memory.

segment_by_column

Analyze data by categorical segments.

find_correlations

Compute relationships between variables across datasets.

analyze_distributions

Analyze statistical distributions of numeric columns.

detect_outliers

Identify anomalous data points.

suggest_analysis

AI-powered recommendations for additional analyses.

create_chart

Generate an interactive chart (bar, scatter, line, histogram).

generate_dashboard

Create a multi-chart dashboard from multiple datasets.

validate_data_quality

Score data quality across defined criteria.

compare_datasets

Compare two datasets to highlight differences.

merge_datasets

Join datasets with flexible strategies.

calculate_feature_importance

Analyze which features contribute most to a target variable.

export_insights

Export results in multiple formats (CSV/JSON/PNG/HTML).