R MCP Server (github.com)

RMCP v0.3.2 is a new Model Context Protocol (MCP) server that exposes 33 statistical and ML tools across eight categories — from OLS and logistic regression to ARIMA, VAR, panel/IV econometrics, clustering, random forests, and a broad set of visualization and data-prep utilities. It’s designed so AI assistants (e.g., Claude) can call statistical workflows via natural language using full JSON‑RPC 2.0 MCP semantics over stdio, HTTP, or WebSocket. The release emphasizes production-ready integrations (pip install rmcp), safe R execution, comprehensive error reporting, and example-driven conversational flows for business analytics, economics, time-series forecasting, and A/B testing. Technically, RMCP bridges Python (3.8+) and R (4.0+) ecosystems: tools are implemented as callable MCP methods (e.g., linear_model, arima_model, panel_regression, instrumental_variables, kmeans_clustering, random_forest) and require common R packages (jsonlite, plm, lmtest, sandwich, forecast, vars, rpart, randomForest, ggplot2, etc.). Outputs include diagnostics (R², p-values, AIC), forecasts with confidence intervals, model diagnostics and plots, and CSV import/export. For developers it’s transport-agnostic, JSON-RPC compliant, and scriptable (example server create_server and direct tool calls provided). This lowers the barrier to integrating rigorous statistical workflows into conversational AI assistants and automates reproducible analyses for analysts, researchers, and product teams.