For Five Coffee

This MCP Server gives you fast access to For Five Coffee menu data and exposes a REST API for web apps plus an MCP endpoint for AI assistants. You can use it directly from MCP clients or call the HTTP API to fetch and search the menu, categories, and items.

How to use

You run the MCP Server locally and connect your MCP client or web app to it. The server offers two access methods: an MCP JSON-RPC endpoint via stdio for AI assistants and an HTTP API for web applications.

Connect an MCP client to the local server: use the HTTP endpoint to fetch the full menu, search items, or list categories. Example: open your client and point it at the HTTP MCP URL when prompted. Use the provided commands in your client’s MCP configuration to start and connect to the server.

Use the REST API to integrate the menu into your app. You can fetch the full menu, search by query, or retrieve categories. The HTTP API runs on port 3000 by default.

# Get full menu
curl http://localhost:3000/api/menu

# Search for items
curl "http://localhost:3000/api/menu/search?q=latte"

# Get categories
curl http://localhost:3000/api/menu/categories

How to install

Prerequisites: Node.js and npm installed on your system.

Step 1: Clone the project repository to your local machine.

Step 2: Change to the project directory.

Step 3: Install dependencies.

Step 4: Start the MCP server and HTTP API.

# Step 1
git clone https://github.com/Kong/for-five-mcp.git

# Step 2
cd for-five-mcp

# Step 3
npm install

# Step 4
npm start

Configuration and notes

Two ways to connect to the server are available: an internal MCP endpoint via stdio for AI assistants and a public HTTP API for web clients.

MCP client example (stdio) configuring a local node process to run the server script.

{
  "mcpServers": {
    "for_five_coffee": {
      "command": "node",
      "args": ["/path/to/for-five-mcp/server.js"]
    }
  }
}

Troubleshooting

If the MCP server doesn’t respond, verify the server path and that you restarted the MCP client after any config changes.

If the HTTP API isn’t responding, ensure the server is running and port 3000 is not in use.

No menu data can occur if there is no internet connection or if the server is unable to fetch data from its source.

Endpoints overview

Server info and health checks are available, along with menu endpoints for full data, search, and category-based queries. The MCP endpoint is exposed at /mcp when using the http transport.

Available tools

get_full_menu

Retrieve the complete menu data from the server.

search_menu_items

Search for menu items using a query string.

get_menu_categories

List all menu categories available.

get_items_by_category

Filter items by a specific category.