simple

This MCP server exposes three utility tools over a streamable HTTP (MCP) transport. The server is authored in Python and is started with a Python invocation, serving MCP requests at the /mcp endpoint. The available tools are: generate_uuid, which returns a UUID (version 4 or 1 as implemented); convert_temperature, which converts temperatures between Celsius and Fahrenheit; and text_statistics, which computes basic text metrics like character count, word count, and line count. Clients using the included sample code (FastMCP client, httpx client, or the curl-based client) can list tools, invoke each tool, and view results via the server-sent events (SSE) stream. To start the server, use the uv tooling as described in the README, which operates the python script in this project and exposes the MCP endpoint at /mcp.

Prerequisites:

• Python 3.13+
• uv (as described in the project README)

Installation steps:

1. Install uv if you haven't already (follow the project’s prerequisites in the README).
2. Install project dependencies via uv sync, which will install the required Python packages (e.g., fastmcp, httpx) listed in the repository’s configuration.
3. Start the server using uv by running the server script as described: uv run server.py
4. (Optional) Verify the server is reachable at http://localhost:8000/mcp and that SSE sessions are established via the Mcp-Session-Id header.

Notes and tips:

• Transport is streamable HTTP with MCP over SSE. The server exposes the /mcp endpoint and uses the Mcp-Session-Id header to manage sessions.
• When testing with curl-based clients or httpx, ensure you parse SSE events correctly to extract tool responses.
• If you modify server.py or the tool implementations, re-run uv sync to ensure dependencies stay aligned with your changes.
• The three tools are implemented in server.py and are invoked by name (generate_uuid, convert_temperature, text_statistics). You can extend with more tools following the project’s pattern.
• The curl client relies on curl and jq and is intended to demonstrate raw HTTP requests and parsing of the MCP protocol payload.