MCP Server for Data Exploration

MCP Server is a tool for interactive data exploration, acting as a personal data scientist assistant to transform datasets into actionable insights.

Try it Out

1. Download Claude Desktop.
2. Install and set up using python setup.py on macOS.
3. Load prompt templates and tools in Claude Desktop.
4. Select the explore-data prompt and provide the CSV path and exploration topic.

Examples

MCP Server can explore data without human intervention.

• Case 1: Analyzed a California real estate dataset (2,226,382 entries) to identify housing price trends.
• Case 2: Explored a UK weather dataset (2,836,186 entries) to analyze weather patterns in London, generating reports and graphs.

Components

• Prompts: explore-data for data exploration.
• Tools:
  • load-csv: Loads a CSV file into a DataFrame, requiring csv_path and optionally df_name.
  • run-script: Executes a Python script, requiring the script argument.

Modifying the Server

Configurations are stored in claude_desktop_config.json. Development and publishing instructions are provided.

Development

Instructions for building and publishing the project using uv sync, uv build, and uv publish are included.

Contributing

Contributions are welcome. Report issues, suggest improvements, or contribute code.

License

Licensed under the MIT License.

Get in Touch

Open an issue or contact the maintainers for questions or feedback.

About

An open-source project by ReadingPlus.AI LLC.