Rmcp

Name: Rmcp
Author: gojiplus

@gojiplus

代码仓库主页

关于 Rmcp

R MCP Server

基本信息

分类

其他

许可证

MIT

运行时

python

传输方式

stdio

发布者

gojiplus

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "rmcp": {
      "command": "docker",
      "args": [
        "run",
        "-e",
        "RMCP_HTTP_HOST=0.0.0.0",
        "-e",
        "RMCP_HTTP_PORT=8000",
        "rmcp:latest"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Rmcp?

Rmcp is a Model Context Protocol (MCP) server that enables AI assistants to perform sophisticated statistical analysis through natural conversation. It provides 52 statistical analysis tools across 11 categories and integrates 429 R packages from systematic CRAN task views.

How to use Rmcp?

Install Rmcp via pip install rmcp, then run rmcp start to start the MCP server. Rmcp can be used with Claude Desktop, Claude web, or any MCP client. For web applications, use rmcp serve-http --host 0.0.0.0 --port 8080 for a local HTTP server, or connect to the hosted server at https://rmcp-server-394229601724.us-central1.run.app/mcp. Configuration is supported via environment variables, a config file (~/.rmcp/config.json), or command-line options.

Key features of Rmcp

Natural conversation interface for statistical analysis
429 R packages with a 4‑tier security system
Professional output with markdown tables and inline visualizations
Full MCP protocol compliance with stdio and HTTP transports
Flexible configuration via env vars, config files, and CLI options
100% test success rate across real-world scenarios

Use cases of Rmcp

Business analysis: measure marketing ROI with regression
Economic research: test macroeconomic relationships like Okun’s Law
Customer analytics: predict churn with logistic regression
Time series forecasting: forecast sales with ARIMA models
Statistical testing: evaluate A/B tests with t-tests and ANOVA

FAQ from Rmcp

What dependencies does Rmcp require?

Rmcp requires Python 3.10+ and R 4.4.0+ with a comprehensive package ecosystem. Core R packages include jsonlite, dplyr, ggplot2, broom, plm, forecast, randomForest, rpart, caret, AER, vars, and mgcv.

How do I troubleshoot “R not found” or missing R packages?

Run R --version to verify R is installed. For missing R packages, use rmcp check-r-packages to check what’s missing. On macOS install R with brew install r; on Ubuntu use sudo apt install r-base.

How can I test the MCP connection?

Run echo '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}' | rmcp start to verify the server responds. For the HTTP server, use curl against the health endpoint or the MCP initialization endpoint.

What transports does Rmcp support?

Rmcp supports stdio (for Claude Desktop) and HTTP (for web applications). The HTTP server can be run locally or connected to the hosted production server.

Is authentication required?

The README does not specify authentication requirements. The hosted server endpoints are publicly accessible for testing; local servers may be secured with your own infrastructure.