Weather for Grown Ups

io.github.janhelclPublished by GitHub account janhelclv0.5.5Updated Oct 24, 2026

Operational NWP, ensembles, AI weather models, regional forecasts, history, and verification.

VerifiedSTDIODesktop onlyLocation & LifestyleData & Analytics

Overview

AI-generated overview

Local MCP server giving assistants a unified query language for operational weather models, ensembles, AI forecasts, history and verification.

What you can do
  • Browse the model catalog to see datasets, grids, members, fields and provenance
  • Query forecasts by dataset, geometry, time and variable/pressure-level selection
  • Compare deterministic, ensemble, physics and AI models on the same atmospheric question
  • Retrieve archived runs and verify past forecasts against analyses or radiosondes
  • Expose the same core through a CLI or an MCP server
When to use it
Useful when simple current-weather lookups are not enough and you want an assistant to compare global and regional guidance, inspect vertical structure, quantify ensemble spread, or check how an old forecast verified.
Requirements
Runs as a local process over stdio; requires Node.js 20 or later, installed via the npm package or run with npx. No authentication, environment variables or headers are declared. Network access is needed to reach the weather data sources.
Before you install
The server can run forecast queries and catalog lookups against external weather data services, so requests and locations are sent to third parties. No credentials, payments or write actions are declared.

Installation

In SourceWeft

  1. Open Weather for Grown Ups in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Weather for Grown Ups

Weather is the hello-world of agent tools. WFG is for when “temperature tomorrow” stops being enough.

One query language for operational NWP: global and regional models, deterministic and ensemble forecasts, physics and AI, current runs and history.

Ask the same atmospheric question across models without flattening what makes the models different.

30-second start

Requires Node.js 20+.

bash
npx weather-for-grown-ups catalog --dataset all --search wind --json
npx weather-for-grown-ups query \  --dataset gfs \  --lat 50.08 --lon 14.43 \  --at 2026-08-24T12:00:00Z \  --vars temperature,wind \  --levels 850,700,500 \  --json

For repeated use:

bash
npm install -g weather-for-grown-upswfg --help

The contract

Normal atmospheric access is four orthogonal choices:

text
dataset × geometry × time × selection
json
{  "dataset": "gefs",  "geometry": {    "type": "point",    "latitude": 50.08,    "longitude": 14.43  },  "time": {    "at": "2026-08-30T12:00:00Z"  },  "selection": {    "variables": ["temperature", "wind"],    "pressureLevelsHpa": [850, 700, 500]  },  "ensemble": {    "quantiles": [0.1, 0.5, 0.9]  }}

Change the dataset; keep the question. Unsupported combinations fail rather than being coerced into fake symmetry.

Read the unified atmospheric API →

Ask harder questions

Start with the question. Let the agent build the forecast investigation from scratch.

01 — Give me the serious forecast

What does the atmosphere over Prague look like tomorrow afternoon? Cover the surface and vertical structure, compare the main global guidance, and show which parts are robust across the ensembles.

GFS + IFS → pressure profiles → GEFS + IFS ENS → regional model if useful

02 — Is a front coming?

Is a frontal passage expected near Prague tomorrow? If so, when does it arrive, what changes through the column, and how consistent are the models?

GFS + IFS → time evolution → pressure profiles → GEFS + IFS ENS → align latest vs previous runs

03 — AI vs physics

How do AIFS and IFS describe the atmosphere over Prague tomorrow? Where do they disagree through time and height, and is that disagreement large relative to their ensemble uncertainty?

align IFS, AIFS → pressure profiles → align IFS ENS, AIFS ENS

04 — Plan a paragliding day

I’m planning to paraglide around Bassano tomorrow. How does the convective window develop through the day, what do the wind profile and stability look like, and how robust is the picture across ensembles and global/regional guidance?

profiles → parcel diagnostics → wind through the column → time evolution → ensembles → global + regional models

05 — Verify an old forecast

What did GFS predict three days ago for Prague today, and how well did that forecast verify?

archived forecast → analysis / radiosonde → error → lead provenance

Explore the investigation gallery →

Models

GFS · GEFS · AIGFS · AIGEFS · HGEFS · IFS · IFS ENS · AIFS · AIFS ENS · ICON-D2 · ICON-D2-EPS · AROME · PE-AROME · GFS analysis · GEFSv12 reforecast

Capabilities come from the catalog. Native grids, cadence, domains, members, fields and provenance remain explicit.

bash
wfg catalog --dataset all --json

Agents

WFG exposes the same core through CLI and MCP.

Shell available → CLI.
No shell / remote → MCP.

bash
npx weather-for-grown-ups mcp

Add the portable WFG skill, then pick a setup guide: Codex · Claude Code · Hermes · OpenClaw · other agents

Go deeper

Examples · Installation · API contract · Catalog · History · Architecture · All docs

License

MIT. See LICENSE.

Source: README.md at commit d5d13d0

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.5.5LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts