@pipeworx/noaa

Connect: https://gateway.pipeworx.io/noaa/mcp · Install: one-click buttons

Tools: 3

NOAA’s data services. Active weather alerts, forecasts, station observations, climate normals, hurricane / tropical-cyclone tracks, marine and aviation forecasts. The official US authority on weather, climate, and ocean data. Free, no auth (light rate limit).

Why this matters for AI agents

Where the Weather pack gives you generic global current-conditions / forecast, NOAA gives you US-grade authoritative data: severe-weather alerts, station-level historical observations, climate normals (30-year averages by location), and tropical-storm tracking. For US-specific weather questions or anything climate-historical, NOAA is the right answer.

Common flows:

• Active alerts. “What weather alerts are active in Texas?” → NWS active-alerts feed.
• Station observations. Historical hourly observations at a specific airport or weather station.
• Climate normals. 30-year averages of temperature/precipitation for a location — the baseline for “is this hot for here?”
• Tropical-cyclone tracking. Hurricane positions, intensities, forecast cones.

Auth

NOAA NWS APIs (alerts, forecasts, observations) are free and unauthenticated, with a polite User-Agent expected. NCEI Climate Data APIs may require a free token from https://www.ncdc.noaa.gov/cdo-web/token. Pipeworx forwards pipeworx-mcp/1.0 and supports _apiKey for token-required endpoints.

Datasets worth knowing

Common pitfalls

• Coverage is US + territories + ocean basins. NOAA covers Atlantic, eastern Pacific, and central Pacific tropical cyclones; western Pacific is JMA’s domain. For global weather, fall back to the Weather pack (Open-Meteo).
• Station vs. point forecasts. NWS point forecasts are model-derived for any lat/lon. Station observations are real measurements at specific locations. Don’t conflate “forecast for ZIP” with “observation at airport in ZIP.”
• Climate normals are 30-year averages. Currently 1991–2020 (updated each decade). They’re not “long-term averages” — climate change means current-decade averages already differ.
• Alert geometries. Active alerts have polygonal boundaries; checking “is point X in alert Y” requires geometric intersection. Filter to alerts you actually care about.
• NWS observations have gaps. Stations go offline; data is “preliminary” for ~24h before quality control finalizes it. For research-grade data, prefer NCEI’s archived versions over real-time NWS.

Tools

• get_forecast — Get a multi-day NOAA NWS weather forecast for a US location. US ONLY (50 states + territories) — for non-US coordinates use the “weather” pack instead. Returns temperature, precipitation, wind, and co
• get_alerts — Check active weather alerts for a US state (e.g., “CA”, “NY”, “TX”). Returns alert type, severity, affected areas, and descriptions.
• get_stations — Find weather observation stations in a US state (e.g., “CA”, “NY”, “TX”). Returns station IDs, names, locations, and observation types.