Current weather and 5-day forecast for any location, powered by the free OpenWeatherMap API.
1 field · 1 secret
Credentials are injected into an ephemeral sandbox at boot and destroyed with it.
Light — starts in seconds
Sessions run up to 60 minutes before the sandbox is reclaimed.
60+ AI models
Chat with 60+ AI models on the same workflow — switch to a different model mid-conversation and re-run the same prompt, or use Compare mode to put several side-by-side.
How models use it and what it is built for.
The OpenWeather MCP server gives a model current conditions and a 5-day forecast for any location on earth, backed by the OpenWeatherMap API.
It is the cleanest example of why MCP exists. A language model cannot know the weather — it is not in the training data and never will be. One small tool call closes that gap completely, and the pattern generalises to every other piece of live data an agent needs.
It is also the server most worth building against first. The free tier is generous, setup is a single API key, and failures are obvious rather than subtle.
Concrete operations exposed as tools.
In order. Each step assumes the previous one worked.
Register at openweathermap.org/api and copy the key from the API keys tab. The free tier covers 60 calls a minute, which is far more than an agent conversation needs.
This trips up nearly everyone: a brand-new OpenWeatherMap key can take up to a couple of hours to become active. Until then every call returns 401. If your first attempt fails, wait rather than regenerating the key.
Paste it into the API key field. It is stored as a secret and injected into the sandbox at boot.
Ask for the weather in London. A successful answer with a real temperature confirms the key is active.
What you provide before the sandbox starts.
OpenWeatherMap API KeyRequiredSecretGet a free key at https://openweathermap.org/api. New keys can take up to a few hours to activate.
Paste any of these into Agent Studio once the server is connected.
What is the weather in Mumbai right now, and will it rain in the next 48 hours?
Compare the 5-day forecast for Berlin, Lisbon and Athens and tell me where to go hiking.
I am flying to Tokyo on Friday. What should I pack based on the forecast?
Which of my three office cities is coldest this week?
Known constraints, stated plainly.
Deploy the server hosted, then watch which tools a model actually reaches for — with full JSON input and output on every call. Switch models mid-conversation to compare how each one uses the same server.
About the OpenWeather MCP server.
Yes for this use. The free tier covers current conditions and 5-day forecasts at 60 calls per minute, which comfortably covers agent usage. Historical data, minute-level forecasts and bulk endpoints are paid.
Almost always activation delay. A newly created OpenWeatherMap key is inactive for up to a few hours after signup. Regenerating it restarts that clock, so wait rather than creating another.
Because weather is live data and a language model only knows its training set. Without a tool it will either refuse or invent a plausible-sounding temperature. This is the canonical demonstration of what MCP fixes.
Yes — the underlying API accepts city names, coordinate pairs and postcodes with a country hint. Coordinates are the most reliable for anywhere with a common place name.