MCP SSE Client Python

Name: MCP SSE Client Python
Author: zanetworker

@zanetworker

Repository

About MCP SSE Client Python

Simple MCP Client for remote MCP Servers 🌐

#mcp-sse-client #python-client #server-sent-events

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "mcp-sse-client-python": {
      "command": "python",
      "args": [
        "-m",
        "mcp_sse_client.examples.llm_example",
        "--provider",
        "openai",
        "--endpoint",
        "http://localhost:8000/sse"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is MCP SSE Client Python?

MCP SSE Client Python is a Python toolkit for connecting to and interacting with remote Model Context Protocol (MCP) endpoints over Server-Sent Events (SSE). It is designed for developers building AI-driven applications that need to list and invoke MCP tools from a remote server.

How to use MCP SSE Client Python?

Install the package via pip install -e . from source or pip install mcp-playground once published. Set environment variables for LLM provider API keys (e.g., OPENAI_API_KEY). Use the MCPClient class with an endpoint URL ending in /sse. Optionally use OpenAIBridge or other LLM bridges for natural-language tool selection. Run the included Streamlit app with streamlit run app.py for interactive testing.

Key features of MCP SSE Client Python

Full SSE implementation for real-time remote MCP communication
LLM-powered tool selection using OpenAI, Anthropic, Ollama, or OpenRouter
Pythonic MCPClient with timeout and retry logic
Interactive Streamlit testing interface with multiple chat modes
Command-line interface for quick tool listing and invocation
Automatic error handling with bespoke exception types

Use cases of MCP SSE Client Python

Connect an AI agent to a remote MCP tool server and let the LLM decide which tool to invoke
Test and debug a remote MCP endpoint interactively without writing full client code
Build a chat interface that transparently uses MCP tools when needed
Scripted automation of MCP tool calls with configurable retry and timeout

FAQ from MCP SSE Client Python

What protocols does MCP SSE Client Python support now?

Currently only Server-Sent Events (SSE). Streamable HTTP, WebSocket, and gRPC are planned for future releases.

Does the endpoint URL require a specific format?

Yes. All MCP endpoints must end with /sse (e.g., http://localhost:8000/sse). The client validates this requirement.

Which LLM providers are supported?

OpenAI, Anthropic, Ollama (local), and OpenRouter. Each has a dedicated bridge class: OpenAIBridge, AnthropicBridge, OllamaBridge, and OpenRouterBridge.

What are the runtime requirements?

Python 3.8 or later, plus mcp>=0.1.0, pydantic>=2.0.0, and optionally openai, anthropic, ollama, and streamlit for their respective features.

How does MCP SSE Client Python handle connection errors?

The client includes automatic retry with exponential backoff, configurable timeout (default 30 s), and maximum retries (default 3). Specific exception types (MCPConnectionError, MCPTimeoutError) are provided.