SnowSure publishes live snow conditions, 14-day multi-model forecasts, powder rankings, and verified SnowSure™ ratings for 500+ ski resorts — and every part of it is built to be used programmatically. This guide collects all of SnowSure's developer resources in one place: the REST API, the MCP server for AI agents, the Python SDK and CLI, webhooks, and the agent-discovery files.
REST API and OpenAPI specification
The SnowSure REST API is free for reads and requires no authentication. Every endpoint is documented in the OpenAPI 3.1 specification — resorts, snow reports, forecasts, leaderboards, trends, and the grounded answer engine at POST /api/v1/ask. Responses include rate-limit headers, JSON error envelopes, cursor pagination, and Idempotency-Key support. Start at the developer documentation for quickstarts and the full endpoint catalog, and see the API deprecation policy for the /api/v1 stability guarantee.
MCP server for AI agents — Claude, ChatGPT, Cursor
SnowSure operates a hosted Model Context Protocol server at https://www.snowsure.ai/mcp (Streamable HTTP) with 40 tools: snow reports, resort guides, powder finders, road access, webcams, ML trends, and grounded Q&A. Reads need no auth; account tools use OAuth 2.1. Setup guides for Claude, ChatGPT, and Cursor are in the MCP integration guide. SnowSure is also a native ChatGPT App.
Python SDK and CLI
Install the official Python SDK with pip install snowsure. It ships a typed client (get_resorts, get_resort, get_snow_report, ask) and a snowsure command-line tool. Source, agent configs, and a Claude Code skill live in the snowsure-mcp GitHub repository. A JavaScript bridge is on npm as snowsure-mcp-server.
Webhooks
Subscribe to outbound snow events — snow-report.updated and powder-alert — with HMAC-signed deliveries. See the webhooks documentation for the subscribe API and signature verification.
Agent discovery: llms.txt, ARD, and agents.md
Agents can discover the whole surface from machine-readable files: llms.txt, the Agentic Resource Discovery (ARD) catalog at /.well-known/ai-catalog.json with a signed trust manifest, the A2A agent card, MCP server cards, and the agents.md quickstart. The homepage also supports Markdown content negotiation (Accept: text/markdown) and a dedicated agent view at /?mode=agent.
Start here: Developers · MCP guide · OpenAPI · ai-catalog.json
