Felix MCP Smithery Server

You can run Felix MCP (Smithery) as a compact, streamable HTTP MCP server with four built-in tools. It exposes a public HTTP endpoint for clients to interact with tools like hello, randomNumber, weather, and summarize, and it also supports a local, script-driven startup via npx for quick testing or embedding in your own tooling. Use this MCP to perform quick actions, fetch weather, generate random numbers, or generate concise OpenAI-style summaries in a hosted or local environment.

How to use

You enable an MCP client to connect to either the public Smithery-hosted endpoint or your own local instance. Once connected, you can list available tools, call each tool with the required arguments, and request summaries when you have an OpenAI API key configured. The server handles sessions and streams responses as events when supported by your client.

How to install

Prerequisites: ensure you have Node.js version 18 or newer installed on your system.

Clone the project, install dependencies, and run the server locally.

# Clone the MCP server repository
git clone https://github.com/FelixYifeiWang/felix-mcp-smithery
cd felix-mcp-smithery

# Install dependencies
npm install

If you plan to use the summarize tool locally, set your OpenAI API key in the environment before starting the server.

export OPENAI_API_KEY="sk-..."

Start the MCP server. The following command runs the server on the default port and exposes the MCP endpoint for POST/GET/DELETE /mcp.

node index.js
# ✅ MCP Streamable HTTP server on 0.0.0.0:8081 (POST/GET/DELETE /mcp)

Test the local MCP endpoint with a simple initialize request using curl.

curl -s -X POST "http://localhost:8081/mcp" \
  -H 'Content-Type: application/json' \
  --data '{"jsonrpc":"2.0","id":0,"method":"initialize","params":{"protocolVersion":"2025-06-18"}}'

Additional configuration and endpoints

Public server endpoint you can use from any MCP client (streaming HTTP) is available at the Smithery public page and the MCP URL provided by Smithery with your API key and profile.

Public MCP endpoint (HTTP) for Smithery hosting (streamable) is:

https://server.smithery.ai/@FelixYifeiWang/felix-mcp-smithery/mcp

When using Smithery, you attach your API key and profile via query parameters, and you should not embed secrets directly in your code or configuration.

Example tool usage

The server exposes these tools: hello, randomNumber, weather, summarize. You invoke them through the MCP client by calling the corresponding tool name and providing the required arguments.

Security and best practices

Do not hardcode API keys or secrets in your client configurations. Use the platform-provided authentication mechanism (api_key and profile) as shown in the examples.

NANDA integration notes

You can register Felix MCP Smithery in your NANDA context as a Streamable-HTTP MCP endpoint. Use the public endpoint with your key and profile for testing, then switch to your own deployment as needed.

Available tools

hello

Returns a quick greeting using the provided name.

randomNumber

Generates a random integer up to a specified maximum (default 100).

weather

Fetches current weather information for a city via wttr.in.

summarize

Produces an OpenAI-powered summary of the provided text. Requires OPENAI_API_KEY.