MCP server providing weather information, forecasts, and alerts for locations worldwide using OpenWeatherMap API
This server exhibits significant quality gaps across naming, descriptions, and schema consistency. The TypeScript implementation (tools 1-3) has better-structured schemas with proper constraints (latitude/longitude bounds, enum units, day limits), but descriptions are generic and lack context for LLM selection. The Python implementation (tools 4-7) has minimal schemas, only location strings with no type validation, bounds, or format guidance. Tool naming is inconsistent: 'get-forecast' vs 'get_forecast', mixing kebab and snake case. Descriptions are uniformly terse (10-50 chars), below the 50-200 char baseline for LLM-optimized tools. No output schemas are documented. Error handling is absent from tool definitions. The server exposes 7 tools but lacks coherent composition, tools 1-3 and 4-7 appear to be duplicate implementations with different quality levels.
Get weather alerts for a location anywhere in the world
Get weather forecast for a location anywhere in the world
Get current weather for a location anywhere in the world
Get weather alerts for a location
Get current time for a location
Get weather forecast for a location
Get current weather for a location anywhere in the world
Duplicate tool implementations with inconsistent quality. Tools 1-3 (TypeScript) have structured schemas; tools 4-7 (Python) have minimal schemas with only location strings and no validation. This forces LLMs to guess which variant to use and creates maintenance burden.
Inconsistent tool naming: kebab-case (get-forecast) vs snake_case (get_forecast). LLMs rely on consistent naming conventions to parse intent. This ambiguity increases selection errors.
Descriptions are uniformly terse (10-50 chars, e.g., 'Get weather forecast for a location'). Baseline is 50-200 chars. Missing context on WHEN to use each tool, what it returns, and how it differs from similar tools. LLMs cannot distinguish get-forecast from get_forecast based on these descriptions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | 2026-07-28+ | v2 |
Python implementation (tools 4-7) has no input schema validation. 'location' parameter is a bare string with no format, length, or example constraints. This invites hallucinated values like 'xyz123' or empty strings. TypeScript tools have better constraints (lat/lon bounds, enum units) but Python tools lack any.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., does get-forecast return a list of daily forecasts or a single summary?). This forces agents to guess and wastes tokens on clarification.