weather-mcp
Provides current weather conditions and 5-day/3-hour forecasts for any location using OpenWeatherMap, accessible through natural language in MCP clients.
README

weather-mcp
TypeScript MCP server that exposes weather tools backed by OpenWeatherMap.
Tools
| Tool | Description |
|---|---|
get_current_weather |
Current conditions by city name or lat/lon |
get_forecast |
5-day / 3-hour forecast by lat/lon |
Transports
| Mode | Env | Use when |
|---|---|---|
stdio (default) |
TRANSPORT=stdio |
Local Cursor / Claude Desktop |
| Streamable HTTP | TRANSPORT=http |
Deploy (Render, etc.) / remote URL |
Official MCP docs: https://modelcontextprotocol.io/docs/
Live remote (Render)
Deployed on Render’s free tier:
| Base URL | https://weather-mcp-ts.onrender.com/ |
| MCP endpoint | https://weather-mcp-ts.onrender.com/mcp |
| Health check | https://weather-mcp-ts.onrender.com/health |
Cursor config (remote)
Add this to Settings → MCP or ~/.cursor/mcp.json:
{
"mcpServers": {
"weather": {
"url": "https://weather-mcp-ts.onrender.com/mcp"
}
}
}
No local build or API key in Cursor — the key lives in Render’s env vars.
Render free tier: the service shuts down after ~15 minutes of inactivity. The next request is a cold start and can take up to ~50 seconds before MCP responds. Hit
/healthonce if Cursor tools look stuck waking up.
Prerequisites (local only)
- Node.js 20+
- pnpm (or npm/yarn)
- An OpenWeatherMap API key (get one here)
Setup (local)
pnpm install
cp .env.example .env
Edit .env and set your key:
OPEN_WEATHER_MAP_API_KEY=your_key_here
TRANSPORT=stdio
Build the server:
pnpm run build
This writes the runnable entrypoint to dist/index.js.
Add to Cursor (local stdio)
Cursor does not start this process from the project root by default for env loading, so pass the env file with an absolute path via Node’s --env-file flag.
- Open Cursor Settings → MCP, or edit
~/.cursor/mcp.json. - Add a
weatherentry undermcpServers:
{
"mcpServers": {
"weather": {
"command": "node",
"args": [
"--env-file",
"/ABSOLUTE/PATH/TO/weather-mcp/.env",
"/ABSOLUTE/PATH/TO/weather-mcp/dist/index.js"
]
}
}
}
Replace /ABSOLUTE/PATH/TO/weather-mcp with your real project path (e.g. run pwd from the repo root). Keep TRANSPORT=stdio (or unset) in .env.
Alternative: env inline
{
"mcpServers": {
"weather": {
"command": "node",
"args": [
"/ABSOLUTE/PATH/TO/weather-mcp/dist/index.js"
],
"env": {
"OPEN_WEATHER_MAP_API_KEY": "your_key_here",
"TRANSPORT": "stdio"
}
}
}
}
Prefer .env + --env-file so the key stays out of mcp.json.
- Save, then toggle the server off/on in Settings → MCP (or restart Cursor).
- Confirm the server shows as connected and lists
get_current_weatherandget_forecast.
After code changes
pnpm run build
Then reload the MCP server in Cursor.
Streamable HTTP (local)
HTTP mode serves MCP at POST/GET /mcp (configurable) plus GET /health. Streamable HTTP is configured via the official SDK’s createMcpHandler + toNodeHandler.
pnpm run build
TRANSPORT=http PORT=3000 pnpm run start
# or
pnpm run start:http
You should see something like:
Weather MCP Server running on http://0.0.0.0:3000/mcp (Streamable HTTP)
Deploy on Render
- Create a Web Service from this repo (root of the project).
- Build:
pnpm install && pnpm run build(ornpmequivalent). - Start:
node dist/index.js(orpnpm run start). - Env vars:
TRANSPORT=httpOPEN_WEATHER_MAP_API_KEY=...PORTis set by Render automatically
- MCP URL:
https://<your-service>.onrender.com/mcp
This deployment:https://weather-mcp-ts.onrender.com/mcp
The HTTP endpoint is unauthenticated. Anyone with the URL can call the tools. Add auth before sharing widely.
Manual smoke check (OpenWeatherMap)
With .env set:
node --env-file=.env -e "
import { getApiKey, buildCurrentWeatherUrl, makeOWMRequest, formatCurrentWeather } from './dist/utils.js';
const apiKey = getApiKey();
const data = await makeOWMRequest(buildCurrentWeatherUrl({ apiKey, city: 'Yaounde,CM', units: 'metric' }));
console.log(data ? formatCurrentWeather(data, 'metric') : 'failed');
"
Remote health:
curl -s https://weather-mcp-ts.onrender.com/health
Repo layout
weather-mcp/
├── README.md
├── package.json
├── tsconfig.json
├── .env.example
└── src/
├── index.ts # stdio + Streamable HTTP entry
├── server.ts # tool registration
├── types.ts
└── utils.ts
Notes
- Never commit
.env(it is gitignored). - Stdio servers must not write to stdout except MCP JSON-RPC; logs go to stderr (
console.error). - Free OpenWeatherMap plans cover current weather and the 5-day / 3-hour forecast used here.
License
ISC
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。