mcpr is an R implementation of the Model Context Protocol (MCP), enabling R applications to expose capabilities (tools, resources, and prompts) to AI models through a standard JSON-RPC 2.0 interface. It also provides client functionality to connect to and interact with MCP servers.
See the official MCP documentation for more information, particularly on schemas and capabilities. Visit package docs for more information on the R implementation.
Installation
You can install mcpr from GitHub using the pak package:
pak::pkg_install("devOpifex/mcpr")Basic Usage
Server
Here’s a simple example that creates an MCP server with a calculator tool:
library(mcpr)
calculator <- new_tool(
name = "calculator",
description = "Performs basic arithmetic operations",
input_schema = schema(
properties = properties(
operation = property_enum(
"Operation",
"Math operation to perform",
values = c("add", "subtract", "multiply", "divide"),
required = TRUE
),
a = property_number("First number", "First operand", required = TRUE),
b = property_number("Second number", "Second operand", required = TRUE)
)
),
handler = function(params) {
result <- switch(params$operation,
"add" = params$a + params$b,
"subtract" = params$a - params$b,
"multiply" = params$a * params$b,
"divide" = params$a / params$b
)
response_text(result)
}
)
mcp <- new_mcp(
name = "R Calculator Server",
description = "A simple calculator server implemented in R",
version = "1.0.0"
)
mcp <- add_capability(mcp, calculator)
serve_io(mcp)You can return multiple responses by returning a list of response objects:
response(
response_text("Hello, world!"),
response_image(system.file("extdata/logo.png", package = "mcpr")),
response_audio(system.file("extdata/sound.mp3", package = "mcpr")),
response_video(system.file("extdata/video.mp4", package = "mcpr")),
response_file(system.file("extdata/file.txt", package = "mcpr")),
response_resource(system.file("extdata/resource.json", package = "mcpr"))
)You can also serve via HTTP transport with serve_http:
# Serve via HTTP on port 3000
serve_http(mcp, port = 3000)See the Get Started guide for more information.
MCP Roclet for Automatic Server Generation
mcpr includes a roxygen2 roclet that can automatically generate MCP servers from your R functions using special documentation tags. This provides a convenient way to expose existing R functions as MCP tools.
Usage
- Add
@mcpand@typetags to your function documentation:
#' Add two numbers
#' @param x First number
#' @param y Second number
#' @type x number
#' @type y number
#' @mcp add_numbers Add two numbers together
add_numbers <- function(x, y) {
x + y
}- Generate the MCP server using roxygen2:
# Generate documentation and MCP server
roxygen2::roxygenise(roclets = c("rd", "mcpr::mcp_roclet"))This will create an MCP server file at inst/mcp_server.R that includes: - Tool definitions for all functions with @mcp tags - Proper input schemas based on @type tags - Handler functions that call your original R functions - A complete, runnable MCP server
Supported Types
The @type tag supports these parameter types: - string - Text values - number - Numeric values (integers and decimals) - integer - Integer values - boolean - True/false values - array - Lists/vectors - object - Complex R objects - enum:value1,value2,value3 - Enumerated values
Example Generated Server
# Generated MCP server code
add_numbers_tool <- new_tool(
name = "add_numbers",
description = "Add two numbers together",
input_schema = schema(
properties = properties(
x = property_number("x", "First number", required = TRUE),
y = property_number("y", "Second number", required = TRUE)
)
),
handler = function(params) {
result <- add_numbers(params$x, params$y)
response_text(result)
}
)
mcp_server <- new_server(
name = "Auto-generated MCP Server",
description = "MCP server generated from R functions with @mcp tags",
version = "1.0.0"
)
mcp_server <- add_capability(mcp_server, add_numbers_tool)
serve_io(mcp_server)Client
Here’s a simple example of using the client to interact with an MCP server:
library(mcpr)
# Create a client that connects to an MCP server
# For HTTP transport
client <- new_client_http(
"http://localhost:8080",
name = "calculator",
version = "1.0.0"
)
# Or for standard IO transport
# client <- new_client_io(
# "Rscript",
# "/path/to/server.R",
# name = "calculator",
# version = "1.0.0"
# )
# List available tools
tools <- tools_list(client)
print(tools)
# Call a tool
result <- tools_call(
client,
params = list(
name = "calculator",
arguments = list(
operation = "add",
a = 5,
b = 3
)
),
id = 1L
)
print(result)
# List available prompts
prompts <- prompts_list(client)
print(prompts)
# List available resources
resources <- resources_list(client)
print(resources)
# Read a resource
resource_content <- resources_read(
client,
params = list(
name = "example-resource"
)
)
print(resource_content)ellmer integration
Now supports ellmer tools, you can use ellmer’s or mcpr’s tools, interchangeably.
See the example below, taken from the ellmer documentation
# create an ellmer tool
current_time <- ellmer::tool(
\(tz = "UTC") {
format(Sys.time(), tz = tz, usetz = TRUE)
},
"Gets the current time in the given time zone.",
tz = ellmer::type_string(
"The time zone to get the current time in. Defaults to `\"UTC\"`.",
required = FALSE
)
)
mcp <- new_server(
name = "R Calculator Server",
description = "A simple calculator server implemented in R",
version = "1.0.0"
)
# register ellmer tool with mcpr server
mcp <- add_capability(mcp, current_time)
serve_io(mcp)Using mcpr
Claude Code Integration
To use your MCP server with Claude Code, see the documentation
Cursor Integration
To integrate with Cursor see the documentation
VS Code Agent Mode Integration
To integrate with VS Code Agent mode see the documentation
"mcp": {
"servers": {
"my-mcp-server-calculator": {
"type": "stdio",
"command": "Rscript",
"args": [
"path/to/calculator_server.R"
]
}
}
}More integrations in the docs