Start dashboardr MCP Server — dashboardr_mcp

Launches an MCP server that exposes dashboardr documentation, function reference, example code, and visualization guides to LLM-powered coding assistants like Claude Desktop, Claude Code, Cursor, and VS Code Copilot.

Usage

dashboardr_mcp_server(
  backend = c("mcptools", "mcpr"),
  transport = c("stdio", "http"),
  port = 8080L
)

Arguments

backend: Character, which MCP backend to use. "mcptools" (default, Posit-maintained) or "mcpr" (alternative). If the chosen backend is not installed, falls back to the other.
transport: Character, either "stdio" (default) or "http".
port: Integer, port for HTTP transport (default 8080). Ignored when transport = "stdio".

Value

Called for its side effect (starts a blocking MCP server). Does not return.

Details

Tools are defined using ellmer::tool() and served via mcptools::mcp_server() (default) or mcpr::serve_io() as a fallback.

The server exposes five tools:

dashboardr_guide: Returns the full dashboardr API guide covering the three-layer architecture, function index, and quick start.
dashboardr_function_help: Look up detailed help for any exported dashboardr function.
dashboardr_list_functions: List exported functions, optionally filtered by category.
dashboardr_example: Get runnable example code for common dashboard patterns.
dashboardr_viz_types: Quick reference of all visualization types with key parameters and use cases.

Configuration

Claude Code:

claude mcp add dashboardr -- Rscript -e "dashboardr::dashboardr_mcp_server()"

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "dashboardr": {
      "command": "Rscript",
      "args": ["-e", "dashboardr::dashboardr_mcp_server()"]
    }
  }
}

VS Code / Cursor:

{
  "mcp": {
    "servers": {
      "dashboardr": {
        "type": "stdio",
        "command": "Rscript",
        "args": ["-e", "dashboardr::dashboardr_mcp_server()"]
      }
    }
  }
}