MCP Server Template

Prerequisites

LLM CLI Installation

This project requires the LLM CLI to be installed. You can install it using Homebrew:

brew install llm

After installation, ensure the llm command is available in your PATH by running:

llm --version

Getting Started

Development

1. Install dependencies:

   npm install
   

2. Start the development server with hot reload:

   npm run dev
   

3. Build the project:

   npm run build
   

4. Run tests:

   npm test
   

5. Start the production server:

   npm start
   

Architect Tool

The Architect tool (src/tools/architect.ts) provides an interface to interact with the LLM CLI for architectural design feedback. It maintains conversation context and handles the communication between your application and the LLM CLI.

Features

- Maintains conversation context across multiple interactions
- Handles command execution through the LLM CLI
- Provides error handling and logging
- Supports both new conversations and continued discussions

Requirements

- The LLM CLI must be installed and available in your PATH (see Prerequisites section)
- Environment variables should be properly configured for the LLM CLI

Testing

The template includes a built-in TestClient for local testing and the MCP Inspector for visual debugging.

Using TestClient

The TestClient provides a simple way to test your tools:

import { TestClient } from "./utils/TestClient";
describe("YourTool", () => {

  const client = new TestClient();
it("should process data correctly", async () => {

    await client.assertToolCall(

      "your-tool-name",

      { input: "test" },

      (result) => {

        expect(result.toolResult.content).toBeDefined();

      }

    );

  });

});

Using MCP Inspector

The template includes the MCP Inspector for visual debugging of your tools:

1. Start the inspector:

   npx @modelcontextprotocol/inspector node dist/index.js
   

2. Open the inspector UI at <http://localhost:5173>

The inspector provides:

- Visual interface for testing tools
- Real-time request/response monitoring
- Tool metadata inspection
- Interactive testing environment

Local Testing with Cursor

To test your MCP server locally with Cursor:

1. Build and link the package:

   npm run build
   npm run link
   

2. Verify the binary works:

   npx architect-test-mcp-tool
   

3. Add the server to Cursor:

- Open Cursor settings
- Navigate to the Features tab
- Scroll down to MCP Servers section
- Click "Add Server"
- Select "Command" type
- Give it a name (e.g., "Local Example Tool")
- Enter the command: npx architect-test-mcp-tool
- Click Confirm

4. Verify the server starts correctly in Cursor by checking the MCP Servers section shows your server as running.

Note: If you make changes to your code, remember to rebuild and relink:

npm run build
npm run link

When you're done testing, you can unlink the package:

npm run unlink

This will remove the global symlink created during development.

Troubleshooting

Using MCP Inspector

The MCP Inspector is a helpful tool for debugging and inspecting your MCP server. To use it:

1. First, build your project:

   npm run build
   

2. Run the inspector:

   npx @modelcontextprotocol/inspector node dist/index.js
   

The inspector provides a web interface that allows you to:

- View all available tools and their schemas
- Test tool calls interactively
- Inspect request/response payloads
- Debug communication issues between your server and clients