SSE-based Server and Client for
About
It demonstrates a Server-Sent Events (SSE) based MCP server and a standalone MCP client that uses tools from the server. This pattern enables decoupled, cloud‑native communication between MCP servers and clients, as opposed to the STDIO pattern where the client spawns the server…
Details
- Author
- goyatparmod
- Downloads
- 235
- Categories
- Other
Jump to
- Demo SSE‑based MCP server with weather tools
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
SSE-based Server and Client forCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Set the ANTHROPIC_API_KEY environment variable. Run the server with uv run weather.py (optionally with --host and --port flags) then run the client with uv run client.py http://<server>:<port>/sse. The client connects to the SSE endpoint and lists available tools. You can also install via Smithery for Claude Desktop using npx -y @smithery/cli install @sidharthrajaram/mcp-sse --client claude.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"sse-based server and client for": {
"sse-mcp": {
"command": "npx",
"args": [
"-y",
"@smithery/cli",
"install",
"@sidharthrajaram/mcp-sse",
"--client",
"claude"
]
}
}
}
}
McpServers
{
"sse-mcp": {
"command": "npx",
"args": [
"-y",
"@smithery/cli",
"install",
"@sidharthrajaram/mcp-sse",
"--client",
"claude"
]
}
}
SSE-based Server and Client for MCP
This demonstrates a working pattern for SSE-based MCP servers and standalone MCP clients that use tools from them. Based on an original discussion here.
Usage
Note: Make sure to supply ANTHROPIC_API_KEY in .env or as an environment variable.
uv run weather.py
uv run client.py http://0.0.0.0:8080/sse
Initialized SSE client...
Listing tools...
Connected to server with tools: ['get_alerts', 'get_forecast']
MCP Client Started!
Type your queries or 'quit' to exit.
Query: whats the weather like in Spokane?
I can help you check the weather forecast for Spokane, Washington. I'll use the get_forecast function, but I'll need to use Spokane's latitude and longitude coordinates.
Spokane, WA is located at approximately 47.6587° N, 117.4260° W.
[Calling tool get_forecast with args {'latitude': 47.6587, 'longitude': -117.426}]
Based on the current forecast for Spokane:
Right now it's sunny and cold with a temperature of 37°F and ...
Why?
This means the MCP server can now be some running process that agents (clients) connect to, use, and disconnect from whenever and wherever they want. In other words, an SSE-based server and clients can be decoupled processes (potentially even, on decoupled nodes). This is different and better fits "cloud-native" use-cases compared to the STDIO-based pattern where the client itself spawns the server as a subprocess.
Installing via Smithery
To install SSE-based Server and Client for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install @sidharthrajaram/mcp-sse --client claude
Server
weather.py is a SSE-based MCP server that presents some tools based on the National Weather Service APIs. Adapted from the MCP docs' example STDIO server implementation.
By default, server runs on 0.0.0.0:8080, but is configurable with command line arguments like:
uv run weather.py --host <your host> --port <your port>
Client
client.py is a MCP Client that connects to and uses tools from the SSE-based MCP server. Adapted from the MCP docs' example STDIO client implementation.
By default, client connects to SSE endpoint provided in the command line argument like:
uv run client.py http://0.0.0.0:8080/sse
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.



