MCP server providing weather alerts and forecasts via the National Weather Service API
Two tools with clear verb-noun naming (get-alerts, get-forecast) and basic descriptions. Input schemas are present with Zod validation and constraints (latitude/longitude bounds, state length). However, descriptions are minimal (10-40 chars), lack context on when to use each tool, and do not explain prerequisites or error recovery. Output schemas are not documented, LLMs cannot predict response structure. No error handling guidance. Parameters lack depth: 'state' description is adequate but 'latitude'/'longitude' descriptions are generic. No mention of API limitations (US-only for NWS), rate limits, or failure modes. Tool composition is sound (single responsibility), but descriptions fall short of production baseline (194 chars average for A+ tools).
Get weather alerts for a state
Get weather forecast for a location
Descriptions are too brief (10-40 chars) and lack context. 'Get weather alerts for a state' does not explain when to use this tool, what format the response takes, or that it only works for US states. Baseline for A+ tools is 194 chars average.
Output schemas are not documented. The code returns structured text responses with 'content' arrays, but LLMs cannot predict the exact structure or extract fields for downstream tool calls. Baseline: 100% of A+ tools document return types.
No error handling guidance. When NWS API fails or coordinates are out of bounds, responses are plain text ('Failed to retrieve...') without actionable recovery steps. LLMs cannot determine if the error is retryable, user-fixable, or fatal.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
API limitations not documented. get-forecast only works for US locations (NWS API constraint), but the description does not mention this. Agents will waste calls on non-US coordinates and receive cryptic 'grid point data' errors.
No rate limiting or timeout handling visible. External API calls (fetch to NWS) lack explicit timeouts. An agent in a retry loop could hammer the NWS API without backoff.