mcpdirectory.dev
9,129 servers · updated 23 Sept

Gribstream MCP server

official from Gribstream Active last commit 36 d
{
  "mcpServers": {
    "gribstream": {
      "type": "http",
      "url": "https://gribstream.com/mcp"
    }
  }
}
file .mcp.json (in your project root)version 1.1.4

What it does

GribStream Weather MCP provides AI agents access to NOAA and ECMWF weather forecast data through discovery, validation, and query tools. It lets developers list datasets, look up variables, build API requests, and execute live weather queries after OAuth authorization.

  • Discover available weather forecast models like GFS, HRRR, NBM, and IFS, and find exact variable selectors and levels for a specific model.
  • Build and validate GribStream API requests for timeseries or model-run comparisons before executing them.
  • Query live weather forecast data by valid time or across multiple model runs after authenticating with an OAuth token.

Tools this server exposes

ToolDoes
gribstream_list_datasetsList public GribStream datasets and their metadata.
gribstream_get_datasetGet full public metadata for a specific dataset code.
gribstream_list_parametersList parameter summaries available in a dataset.
gribstream_get_parameterGet detailed metadata and selector variations for a dataset parameter.
gribstream_list_shared_parametersList normalized shared parameter presets across datasets.
gribstream_resolve_shared_parameterResolve a shared parameter preset to native variables and expressions.
gribstream_get_expression_referenceGet expression-language rules, examples, and math functions.
gribstream_build_timeseries_requestBuild and validate a timeseries API request without sending it.
gribstream_build_runs_requestBuild and validate a runs API request for comparing forecasts across model runs.
gribstream_validate_requestValidate request shape, selectors, and expression syntax.
gribstream_query_timeseriesExecute an authenticated timeseries query and return weather values.
gribstream_query_runsExecute an authenticated runs query to compare forecasts across model runs.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.