Skip to main content

MCP Plugin

Model Context Protocol (MCP) client integration for connecting TokenRing agents with MCP servers.

Overview​

The @tokenring-ai/mcp package provides MCP client functionality to connect TokenRing agents with MCP servers, enabling access to external tools and resources through the Model Context Protocol.

Key Features​

  • Multiple transport types: Support for stdio, SSE, and HTTP transports
  • Automatic tool registration: MCP server tools are automatically registered with TokenRing agents
  • Seamless integration: Works with existing TokenRing agent architecture
  • Dynamic tool discovery from MCP servers

Core Components​

MCPService​

Main service for managing MCP server connections.

Key Methods:

  • register(name, config, team): Registers an MCP server with the TokenRing agent team
    • name: Unique identifier for the MCP server
    • config: Transport configuration object
    • team: AgentTeam instance to register tools with

Transport Types​

Stdio Transport: Process-based communication

{
  type: 'stdio',
  command: 'node',
  args: ['path/to/mcp-server.js']
}

SSE Transport: Server-Sent Events

{
  type: 'sse',
  url: 'http://localhost:3000/sse'
}

HTTP Transport: Standard HTTP

{
  type: 'http',
  url: 'http://localhost:3000/mcp'
}

Usage Example​

import { MCPService } from '@tokenring-ai/mcp';
import { AgentTeam } from '@tokenring-ai/agent';

const mcpService = new MCPService();
const team = new AgentTeam();

// Register an MCP server with stdio transport
await mcpService.register('myserver', {
  type: 'stdio',
  command: 'node',
  args: ['path/to/mcp-server.js']
}, team);

Configuration Options​

  • Transport Type: stdio, SSE, or HTTP
  • Connection Details: Command/args for stdio, URL for SSE/HTTP
  • TLS Support: Available for HTTP/SSE transports (future enhancement)

Dependencies​

  • @tokenring-ai/agent: Core agent framework
  • MCP protocol implementation libraries

Future Enhancements​

  • Enhanced transport options with TLS and authentication
  • Dynamic tool discovery and hot-reloading
  • Tool versioning and namespacing
  • Metrics and observability
  • Health checks and auto-reconnect
  • Caching layer for tool definitions
  • CLI utility for management