Model Context Protocol

MCP Server.

A fully working MCP weather server built with Nimbus. Connect from Cursor, Claude Desktop, or any MCP-compatible client and call tools with a single line of Go.

Server

Online

Weather Demo v1.0.0

SSE /mcp/weather

Tools

2 registered

get_weather Get current weather conditions for any location. Returns temperature, humidity, wind speed, and condition.

hello Say hello to someone by name. A simple smoke-test tool to verify your MCP connection is working.

Resources

1 registered

weather://forecast/{location} Weather forecast JSON for a given location. Readable by AI agents as structured context.

Connect from Cursor

Open or create ~/.cursor/mcp.json

Add the server config and save

{
  "mcpServers": {
    "nimbus-weather": {
      "url": "http://localhost:3333/mcp/weather"
    }
  }
}

Adjust the port if your Nimbus app runs on something other than 3333.

Reload Cursor and ask the AI about the weather

mcp/weather.go

// Register MCP server
mcp.Register("weather", func(s *mcp.Server) {
  s.SetInfo("Weather Demo", "1.0.0")

  // Tool: get_weather
  s.Tool("get_weather",
    "Get current weather",
    func(args WeatherArgs) string {
      return fetchWeather(args.Location)
    },
  )

  // Resource: forecast URI
  s.Resource(
    "weather://forecast/{location}",
    forecastHandler,
  )
})

Capabilities

Tools — callable functions for AI agents

Resources — URI-addressable structured data

SSE transport — streaming over HTTP

Type-safe args — Go structs, zero boilerplate

Auto-discovery — schema generated from types

Compatible clients

Cursor IDE-native AI with tool calling

Claude Claude Desktop & claude.ai

Any MCP Any MCP 1.0-compliant client

Built with

mcp.Register() mcp.Server s.Tool() s.Resource() SSE