Skip to main content
Glama
alejandro-ao

mcp-streamable-http

by alejandro-ao
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Remote

MCP Servers over Streamable HTTP β€” Step-by-Step Guide

πŸ“ Read the full article here: MCP Servers over Streamable HTTP (Step-by-Step)

This repository contains a complete, working example of how to build and run an MCP (Model Context Protocol) server using Python, mcp, and FastAPI. You’ll learn how to:

  • Expose tools and functions over HTTP using the MCP protocol

  • Connect those tools to AI assistants like Cursor

  • Use streamable HTTP as the transport

  • Mount multiple MCP servers in a FastAPI app

πŸ“ Folder Structure

.
β”œβ”€β”€ docs/                        # Diagrams and assets (e.g., mcp-client-server.png)
β”œβ”€β”€ fastapi_example/            # Example mounting multiple MCP servers in FastAPI
β”‚   β”œβ”€β”€ echo_server.py          # A server exposing a simple echo tool
β”‚   β”œβ”€β”€ math_server.py          # A server exposing a math tool
β”‚   └── server.py               # FastAPI app that mounts both echo and math servers
β”œβ”€β”€ .gitignore
β”œβ”€β”€ .python-version             # Python version (for tools like pyenv or uv)
β”œβ”€β”€ pyproject.toml              # Project config and dependencies
β”œβ”€β”€ readme.md                   # You're here!
β”œβ”€β”€ runtime.txt                 # Python runtime for platforms like Render
β”œβ”€β”€ server.py                   # Basic standalone MCP server using Tavily search
β”œβ”€β”€ uv.lock                     # Lockfile for uv dependency manager

βΈ»

πŸ›  Quickstart

  1. Install uv (recommended Python package manager)

curl -LsSf https://astral.sh/uv/install.sh | sh

  1. Install dependencies and set up environment

uv venv && source .venv/bin/activate
uv pip install -r pyproject.toml

  1. Run the basic MCP server This uses the Tavily API to expose a simple web_search tool.

uv run server.py

  1. Run the FastAPI app with multiple MCP servers

uv run fastapi_example/server.py

This will mount:

βΈ»

πŸ§ͺ Debug with MCP Inspector

  1. Install CLI support

uv add 'mcp[cli]'

  1. Launch the inspector

uv run mcp dev server.py

Then go to: http://localhost:6274/?MCP_PROXY_AUTH_TOKEN=...

βΈ»

πŸ”Œ Connect to Cursor

In Cursor, add your MCP server under Chat Settings > MCP Servers:

{
  "mcpServers": {
    "tavily": {
      "url": "http://localhost:8000/mcp/"
    }
  }
}

βœ… Note: You must include the trailing / in the URL.

βΈ»

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • Simple Python MCP Server

    rb58853
    A
    license
    Not graded
    quality
    D
    maintenance
    A simple MCP server implementation in Python using FastMCP and FastAPI, providing streamable HTTP transport with optional OAuth authorization for tools like echo and math.
    8
    MIT

  • Simple MCP Server with

    jonigl
    F
    license
    Not graded
    quality
    D
    maintenance
    Demonstrates how to create a simple MCP server with streamable HTTP transport, featuring tools, prompts, and resources.
    4
    -

  • remote-mcp

    mo3tssem
    F
    license
    Not graded
    quality
    D
    maintenance
    Enables building and running MCP servers over streamable HTTP, exposing tools to AI assistants like Cursor, with examples of mounting multiple servers in FastAPI.
    -