What it does
Given coordinates, and optionally a peak name and summit elevation, the server fetches hourly forecasts for temperature, precipitation, wind, gusts, cloud cover and snow line. It picks the highest-resolution model for the location: GeoSphere Austria AROME at 2.5 km for Central Europe, MeteoSwiss ICON-CH2 at 2 km for the Alpine region, Météo-France AROME at 2.5 km for France, and Open-Meteo at about 11 km elsewhere. If a provider fails, it moves to the next one. Temperatures are lapse-corrected to summit elevation using model terrain derived from surface pressure, at -6.5 °C per km. Each hour is scored against alpine climbing thresholds, and contiguous good-weather windows are identified.
The output suits anyone comparing timing for an ascent or a trailhead day across the coming forecast period. Along with scored text, it serves an interactive UI built with React and uPlot, with a horizon tape, a mountain profile and chart panels, which renders inline in a compatible host.
The server runs as a standard MCP server over stdio or SSE. Claude Desktop is configured through its config file, and the repository also documents an HTTP server setup and a development setup.
Limitations and requirements
Rendering the interactive UI requires an MCP host that supports the MCP Apps protocol. The publisher names Claude Desktop (macOS and Windows), ChatGPT via the OpenAI Apps SDK, and any chat client implementing the MCP Apps spec. Without such a host, only the scored text output is available.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
io.github.ByteOverDev/peak-window
Ranks the best weather windows to climb any peak worldwide, on the sharpest regional model.
v1.0.2As shown on registry.modelcontextprotocol.io when we last read it