mcp-runner-node

name	mcp-runner-node
description	Guide for using the MCP Runner Node to execute external tools via the Model Context Protocol.

MCP Runner Node Skill

The MCP Runner Node allows Litepik AI Suite to interact with external tools and services using the Model Context Protocol (MCP).

Capabilities

• Server Discovery: Automatically detects available MCP servers running via the backend.
• Tool Execution: Lists tools from connected servers and allows execution with JSON arguments.
• AI Auto-Configuration: Uses Gemini Flash to automatically configure tool arguments based on natural language requests.
• Self-Healing: Automatically fixes invalid arguments or execution errors using AI.

Usage

1. Add the Node: Drag "MCP Runner" from the Sidebar (Developer Section) onto the canvas.
2. Select Server: Choose a server from the dropdown (e.g., filesystem, github).
3. Select Tool: Choose a tool provided by that server (e.g., read_file, list_directory).
4. Configure Arguments:
   • Manually enter JSON arguments.
   • OR click AI Auto-Configure, type your intent (e.g., "Read package.json"), and let AI generate the config.
5. Run: Click "Run Tool" to execute. The output will be displayed in the terminal view.

Integration

• Backend: Requires the Litepik backend (server/index.mjs) to be running on port 3001.
• Frontend: The node communicates via /api/mcp/* endpoints.

Extending

To add new servers, edit server/index.mjs and add entries to the SERVERS object:

const SERVERS = {
    'filesystem': { command: 'npx', args: ['-y', '@modelcontextprotocol/server-filesystem', process.cwd()] },
    'my-custom-server': { command: 'node', args: ['./path/to/server.js'] }
};