FastAPI application hosting multiple MCP servers for various public APIs including weather, news, currency exchange, and quotes
The server defines 16 tools across 4 domains (weather, news, currency, quotes) with consistent naming patterns and good parameter type coverage. Most tools follow verb_noun conventions (get_*, search_*). However, descriptions are generic and often fail to explain WHEN to use each tool, what distinguishes it from similar tools, or what the output contains. Parameter descriptions exist but lack actionable constraints (ranges, enums, formats). Output schemas are not documented anywhere in the visible code. Error handling is minimal, no recovery guidance, no invalid-value examples. Security posture is reasonable (API keys use environment injection via settings) but audit logging is absent. The tools are well-composed and mostly idempotent (all READ_ONLY), but the lack of output schema documentation and shallow descriptions cost significant points.
Convert amount from one currency to another.
Get current weather for a specific city.
Get current exchange rates for a base currency.
Get historical exchange rates for a specific date.
Get news articles by specific category.
Get quotes by a specific author.
Get quotes by specific category/tag.
Output schemas undocumented. No visible documentation of what fields or structure each tool returns. LLMs cannot plan downstream tool calls or extract required data without knowing response shape.
Descriptions lack actionable constraints and decision guidance. 'Get current weather for a specific city' does not explain when to use get_current_weather vs get_weather_by_coordinates vs get_weather_forecast. No guidance on what units mean or when to use 'metric' vs 'imperial'.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | C | 65 | 2026-07-28+ | v2 |
| 2026-03-09 | C | 62 | - | v1 |
Get available quote categories/tags.
Get a random interesting fact.
Get a random inspirational quote.
Get list of all supported currencies.
Get top news headlines.
Get current weather by geographical coordinates.
Get weather forecast for a specific city.
Search for news articles by keyword.
Search for quotes containing specific keywords.
Parameter descriptions lack formal constraints. 'Temperature units (metric, imperial, kelvin)' is given as prose examples instead of JSON Schema enum. 'Number of days for forecast (1-5)' lacks a documented minimum/maximum. 'Number of articles to return (max 100)' implies no minimum.
No error handling guidance. Source code shows exceptions raised (e.g., 'Failed to get exchange rates: {str(e)}') but no recovery instructions. No distinction between retryable (network timeout) vs user-fixable (invalid currency code) vs fatal (API key missing) errors.
No pagination or result-limiting documented. Tools like 'search_news' and 'search_quotes' accept unbounded limits. No indication of default behavior, max results returned, or whether pagination (page/offset) is supported.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are READ_ONLY by design, but this is not declared in the tool definitions. LLMs cannot infer retry safety without explicit annotations.