Python R-Server MCP

A secure, Docker-based Model Context Protocol (MCP) server for R data visualization and analysis, built with Python and FastMCP.

What is this? This tool allows you to perform statistical analysis in R through natural language conversations with LLM models like Claude. All R code is executed in secure Docker containers, ensuring complete isolation and security.

Key Benefits:

• 🗣️ Natural Language Interface: Chat with AI to perform complex statistical analyses
• 🐳 Secure Docker Execution: All R code runs in isolated containers with network disabled
• 📊 Automatic R Execution: AI runs R code and returns results without you writing code
• 📁 Multiple Directory Mounting: Mount multiple local directories for R to access
• 📈 Inline Plot Capture: Plots automatically returned as base64 PNG images
• 🔒 PII Protection: Optional auto-redaction of personal data before it reaches the LLM
• 💾 Session Persistence: R variables persist between calls
• ⚡ Smart Caching: Configurable result caching for instant repeated operations
• 📦 Auto Library Detection: Missing R packages automatically installed

🌐 Türkçe | English

Overview

This project is inspired by gdbelvin’s rlang-mcp-server but is a complete Python reimplementation using the FastMCP framework. While the original Go version provided basic R visualization tools, this Python version extends the functionality with comprehensive security, file management, and data privacy features.

Features

🎨 Visualization & Analysis

• ggplot2 Rendering: Execute R code with ggplot2 commands and return publication-ready visualizations
• Inline Plot Capture: Plots automatically detected and returned as base64 PNG
• R Script Execution: Run any R script with smart file handling and return formatted output
• Session Persistence: R variables and workspace persist between calls via .RData

📁 File Management

• Multiple Directory Mounting: Mount multiple local directories with custom mount points
• Smart File Discovery: Use R scripts to explore and analyze files in mounted directories
• Unmount Support: Cleanly unmount directories when done

📦 Package Management

• Auto Library Detection: library() and require() calls parsed, missing packages auto-installed
• Package Installation: Install R packages on-demand with version control
• Automatic Dependencies: Smart package dependency resolution
• Temporary Network: Network temporarily enabled only during package installation

🛡️ Security & Isolation

• Mandatory Docker: All R code execution in isolated containers
• Network Disabled by Default: Container cannot access internet during R execution
• Resource Limits: Configurable memory (default 2GB) and CPU limits