home / mcp / databeak mcp server

DataBeak MCP Server

Model Context Protocol server providing 40+ specialized CSV data manipulation tools for AI assistants. Enables seamless data loading, transformation, analysis, validation, and export workflows with session management and auto-save capabilities.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "jonpspri-databeak": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/jonpspri/databeak.git",
        "databeak"
      ],
      "env": {
        "DATABEAK_MAX_ROWS": "YOUR_VALUE",
        "DATABEAK_SESSION_TIMEOUT": "YOUR_VALUE",
        "DATABEAK_MAX_MEMORY_USAGE_MB": "YOUR_VALUE",
        "DATABEAK_URL_TIMEOUT_SECONDS": "YOUR_VALUE",
        "DATABEAK_MAX_DOWNLOAD_SIZE_MB": "YOUR_VALUE",
        "DATABEAK_HEALTH_MEMORY_THRESHOLD_MB": "YOUR_VALUE"
      }
    }
  }
}

DataBeak is an MCP server that lets you process and analyze CSV data through a flexible, AI-assisted interface. It supports loading data from URLs or string content, performing transformations, running analytics, validating data, and managing sessions securely in a stateless design. You interact with it via an MCP client, enabling powerful data workflows in natural language or structured prompts without exposing the underlying data handling details.

How to use

Set up your MCP client to connect to DataBeak using the provided runtime configuration. You will run the server through a standard MCP pattern that keeps the external context separate and supports multiple concurrent sessions. Use the following approach to start and interact with DataBeak in your client.

How to install

Prerequisites: you need a runtime capable of executing MCP server configurations. The fastest way to begin is to use the runtime that DataBeak provides for MCP integration.

1. Create a settings file for your MCP client and add the server configuration block shown below. This config uses a stdio runtime to load the MCP server from a remote source.

{
  "mcpServers": {
    "databeak": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/jonpspri/databeak.git",
        "databeak"
      ]
    }
  }
}

HTTP Mode (Advanced)

If you prefer HTTP-based access or a custom deployment, you can run DataBeak in HTTP mode and expose the MCP interface over a network port. The following command starts the server in HTTP mode and binds to all interfaces on port 8000.

uv run databeak --transport http --host 0.0.0.0 --port 8000

Practical usage patterns

Once DataBeak is running, you can ask your MCP client to load and analyze CSV data with simple prompts. For example you can load CSV content directly, or provide a URL for the data source, and then request operations such as removing duplicates or computing statistics. You can query for outliers in a numeric column and retrieve a data profile to understand data quality and distribution.

Notes on security and limits

DataBeak is designed to be web-safe with no local file system access for hosted deployments. It supports asynchronous I/O and streaming downloads, with session isolation to support multi-user scenarios. Be mindful of predefined limits on downloads, memory, and rows, and adjust environment variables accordingly for production workloads.