Skip to main content
MCP support is experimental. Enable with experimental.mcp configuration.
Model Context Protocol (MCP) servers provide a standardized way to expose tools to AI applications like Claude Desktop, Cursor, and other MCP-compatible clients.

Quick Start

Creating MCP Servers

create()

string
required
Unique name for the MCP server
MCPConfigCreationParams
required

Managing MCP Servers

list()

List all MCP servers:

get()

Retrieve a specific server:

update()

Update server configuration:

delete()

Delete an MCP server:

Generating User Instances

generate()

Create a user-specific MCP server instance:

Client Configuration

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

Cursor

Add to Cursor MCP settings:

Windsurf

Use the setup command from server.commands.windsurf.

Server Response

Complete Example

Best Practices

  1. Tool Selection: Only include necessary tools
  2. Naming: Use descriptive server names
  3. Auth Management: Choose appropriate connection management
  4. Documentation: Document setup for your users
  5. Updates: Update servers when adding new tools

Limitations

  • MCP support is experimental
  • Requires MCP-compatible clients
  • Some features may not work in all environments
  • API may change in future versions

Next Steps

Tool Router

Alternative approach

Tools API

Learn about tools

Toolkits

Browse toolkits

Connected Accounts

Manage connections