FelixYifeiWang Felix MCP Smithery

What FelixYifeiWang-felix-mcp-smithery MCP server does

FelixYifeiWang-felix-mcp-smithery MCP server provides a small set of general-purpose tools rather than connecting to a project-management platform or code repository. Its tools are:

• hello(name), which returns a greeting for the supplied name.
• randomNumber(max?), which produces a random integer; the default maximum is 100.
• weather(city), which retrieves current weather information through wttr.in.
• summarize(text, maxSentences?, model?), which sends text to OpenAI for a concise summary.

The server is a fit for agents that need simple utility operations, current city weather, or basic text reduction without adding a larger external service integration.

How it works

FelixYifeiWang-felix-mcp-smithery uses the MCP SDK with an Express application and exposes a Streamable HTTP transport on /mcp. Clients submit JSON-RPC requests with POST /mcp, including the initial initialize request and subsequent tool calls. GET /mcp is used for server-to-client notifications over SSE, while DELETE /mcp closes a session.

Sessions use the Mcp-Session-Id header. The application allows requests from all origins and exposes that session header for hosted clients. The deployed Smithery endpoint is available at https://server.smithery.ai/@FelixYifeiWang/felix-mcp-smithery/mcp.

The summarization implementation calls OpenAI’s /v1/chat/completions endpoint through a fetch wrapper. Its default model is gpt-4o-mini, and callers can provide a different model through the tool argument.

Setup and configuration

For local installation, use Node.js 18 or newer, clone the repository, and run npm install. Start the service with node index.js; the local server listens on 0.0.0.0:8081 and serves MCP at /mcp. The README demonstrates sending an initialization request with protocol version 2025-06-18 using curl.

Set OPENAI_API_KEY in the server environment before calling summarize locally. The key is not needed for the other listed tools. Do not place secrets directly in client configuration or source files.

Smithery deployment packages the application as a container and uses a Streamable HTTP start command on port 8081. Smithery’s platform configuration can attach an api_key query parameter and an optional profile value to the hosted endpoint; those values are configured in the platform UI.

Tools and capabilities

FelixYifeiWang-felix-mcp-smithery supports ordinary MCP tool discovery and invocation. Tool calls use the MCP tools/call method with a tool name and JSON arguments. The examples cover greeting a named user, limiting a random number to 10, checking Boston weather, and requesting a two-sentence summary.

Limitations and notes

The README only documents four tools. Weather data depends on wttr.in, so the weather result depends on that external service and the requested city. Summarization depends on OpenAI availability, a valid API key, and the selected model. No license is specified in the provided material.

The hosted Smithery endpoint uses platform-managed authentication. For local use, the documented setup is an HTTP server rather than a local stdio command, so the MCP client must be able to connect to a Streamable HTTP endpoint.