# claude-r

> Execute R code through RStudio via MCP. Use for R programming tasks, statistical analysis, data visualization, or setting up/troubleshooting the RStudio integration with Claude Code.

- Author: Naoto Iwase
- Repository: naoto-iwase/Claude-R
- Version: 20260209193839
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-09
- Source: https://github.com/naoto-iwase/Claude-R
- Web: https://mule.run/skillshub/@@naoto-iwase/Claude-R~claude-r:20260209193839

---

---
name: claude-r
description: Execute R code through RStudio via MCP. Use for R programming tasks, statistical analysis, data visualization, or setting up/troubleshooting the RStudio integration with Claude Code.
---

# Claude-R - Execute R through RStudio

Connect Claude Code to RStudio, enabling natural language R programming. RStudio acts as the execution engine; Claude generates and runs the code.

## Quick Start

### Prerequisites Check

Verify RStudio server is running:

```bash
uv run scripts/check_r_connection.py
# or with system Python (if httpx installed)
python scripts/check_r_connection.py
```

If not running:
1. Open RStudio
2. File -> Open Project -> select your .Rproj file
3. Run: `library(ClaudeR)`
4. Run: `claudeAddin()`
5. Click "Start Server" in Viewer pane

### Basic Usage

Execute R code:
```r
x <- 1:10
mean(x)
```

Generate plots:
```r
library(ggplot2)
ggplot(iris, aes(x = Sepal.Length, y = Sepal.Width, color = Species)) +
  geom_point(size = 3) +
  theme_minimal()
```

## Available MCP Tools

### execute_r
Execute R code and return text output.

### execute_r_with_plot
Execute R code that generates graphics. Returns PNG image.

### get_r_info
Query R environment:
- `what: "packages"` - Installed packages
- `what: "variables"` - Workspace objects
- `what: "version"` - R version
- `what: "all"` - Complete summary

### get_active_document
Retrieve currently open file in RStudio.

### modify_code_section
Modify code in active RStudio document by pattern or line range.

## Setup and Installation

### Step 1: Install R and Dependencies

```bash
bash scripts/setup_clauder.sh
```

This installs:
- R via Homebrew
- System dependencies (harfbuzz, libtiff, libgit2, etc.)
- R packages (renv, devtools, ClaudeR)

### Step 2: Choose Python Environment

You'll need Python with `httpx` and `mcp` packages for the MCP server.

**Option A: uv (Recommended)**

No installation needed - dependencies loaded automatically:

```bash
claude mcp add r-studio --scope user \
  uv run --with httpx --with mcp python \
  /path/to/ClaudeR/renv/library/.../ClaudeR/scripts/persistent_r_mcp.py
```

**Option B: System Python**

Install dependencies first:

```bash
python3 -m pip install --user httpx mcp
claude mcp add r-studio --scope user \
  python3 \
  /path/to/ClaudeR/renv/library/.../ClaudeR/scripts/persistent_r_mcp.py
```

**Option C: Custom Python Environment**

Use your preferred environment manager (venv, conda, etc.):

```bash
# Example with venv
python3 -m venv ~/.claude-r-env
source ~/.claude-r-env/bin/activate
pip install httpx mcp

claude mcp add r-studio --scope user \
  ~/.claude-r-env/bin/python \
  /path/to/ClaudeR/renv/library/.../ClaudeR/scripts/persistent_r_mcp.py
```

### Step 3: Start R Server

1. Open RStudio
2. File -> Open Project -> select your .Rproj file
3. `library(ClaudeR)`
4. `claudeAddin()`
5. Click "Start Server"

### Step 4: Verify

Run `/mcp` in Claude Code - "r-studio" should show "connected"

## Architecture

Three-layer system:

1. **Claude Code**: Natural language interface
2. **MCP Server**: Python bridge (stdio ↔ HTTP)
3. **ClaudeR HTTP Server**: R execution in RStudio (port 8787)

RStudio acts passively - users interact with Claude Code, which generates and executes R code. RStudio's Environment, Plots, and Console panes show results.

See [references/architecture.md](references/architecture.md) for details.

## Troubleshooting

If R code execution fails, check connection:

```bash
uv run scripts/check_r_connection.py
```

Common issues:
- RStudio server not running
- MCP configuration incorrect
- Python dependencies missing

See [references/troubleshooting.md](references/troubleshooting.md) for complete guide.

## Key Features

- **State persistence**: R session persists across executions
- **RStudio integration**: View variables, plots, and console in RStudio
- **Natural language**: Generate R code through conversation

## Limitations

- **Single-threaded**: Sequential execution only
- **Local only**: Server binds to localhost
- **RStudio required**: Must be running with server active
- **No streaming**: Results return after complete execution

## Resources

- [Troubleshooting Guide](references/troubleshooting.md) - Connection issues, setup errors
- [Architecture](references/architecture.md) - System design and data flow
- [Setup Script](scripts/setup_clauder.sh) - Automated installation
- [Connection Checker](scripts/check_r_connection.py) - Verify server status
- [ClaudeR GitHub](https://github.com/IMNMV/ClaudeR) - Original project