Two tools with minimal definitions. Both have basic schemas with single required string parameters, but descriptions are extremely terse (10-14 chars each, well below the 20-char minimum). No parameter descriptions provided. No output schemas documented. No error handling guidance. Tool names lack action verbs (getWeather, calculate are imperative but not verb_noun pattern). This is a basic example implementation, not production-grade tooling.

• Expand tool descriptions to 50-200 chars. Example: 'getWeather' → 'Retrieve current weather conditions for a city, including temperature, humidity, and forecast. Returns structured data suitable for display or further processing.'
• Add parameter descriptions. Example: 'city' → 'The city name (e.g., San Francisco, London). Accepts common city names; if ambiguous, the tool returns the most populous match.'
• Document output schemas. Example for getWeather: {temperature: number, humidity: number, condition: string, forecast: array}. For calculate: {result: number, expression: string}.
• Add error handling guidance. Example: 'If city is not found, the tool returns a list of similar city names. If the weather API is unavailable, it returns a retryable error with a suggested delay.'
• Rename tools to snake_case verb_noun: getWeather → get_weather, calculate → evaluate_expression. This matches production baselines and improves LLM parsing.