Loading ToolBench
Single tool 'fetch-weather' has critical definition gaps. Tool name lacks action verb prefix (should be 'get_weather' or 'fetch_current_weather'). Input parameter 'city' has null description, violates mandatory parameter documentation rule. No output schema documented. Tool description missing entirely from visible code. No error handling guidance. Naming is vague ('fetch-weather' could mean historical data, forecast, or current conditions).
Tool description is completely missing. LLMs cannot determine when or why to select this tool without a description explaining what it does, when to use it, and what it returns.
Input parameter 'city' has null description. Every parameter must have a non-empty description explaining what it controls. LLMs cannot infer that 'city' expects a city name vs. city code vs. coordinates.
Tool name 'fetch-weather' lacks action verb prefix and is ambiguous. Does it return current conditions, forecast, or historical data? Should be 'get_current_weather' or 'get_weather_current'. Baseline: 90% of A+ tools start with action verb.
Output schema is not documented. LLMs need to know what fields to expect (e.g., temperature, condition, humidity, wind_speed) so they can plan downstream operations and extract the right data.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 29 | <=2025-11-25 | v2 |
No error handling guidance. If the API key is missing, the city is invalid, or the API is down, the tool returns a raw error. LLMs need actionable recovery hints like 'City not found. Try a larger city name or country code.'