Connectivity (MCP) • aisdk

The Model Context Protocol (MCP) is a standard for connecting AI systems to data sources. aisdk includes a built-in MCP Client and Server.

Using the MCP Client

You can connect to any MCP-compliant server (Node.js, Python, or R) and use its tools.

Example: GitHub MCP Server

This example connects to the official GitHub MCP server to let the AI interact with your repositories.

Prerequisites:
- Node.js installed (to run the server).
- GITHUB_TOKEN environment variable set.
Connect Client:

library(aisdk)

# Start the Node.js GitHub server
github_client <- create_mcp_client(
  command = "npx",
  args = c("-y", "@modelcontextprotocol/server-github"),
  env = c(GITHUB_PERSONAL_ACCESS_TOKEN = Sys.getenv("GITHUB_TOKEN"))
)

# List available tools
tools <- github_client$list_tools()
print(tools[[1]]$name) 
# e.g., "create_issue", "list_pull_requests"...

Use with AI:

Convert the MCP tools to aisdk Tools and pass them to the model.

sdk_tools <- github_client$as_sdk_tools()

# Create a session (manages history and tool loops)
# max_steps=10 enables automatic ReAct loop for tool calling
chat <- create_chat_session(
  model = "openai:gpt-4o",
  tools = sdk_tools,
  max_steps = 10
)

# Option 1: Run programmatically
result <- chat$send("List my recent pull requests")
print(result$text)

# Option 2: Run interactively with streaming
# console_chat(chat)

Cleanup:

github_client$close()