Loading ToolBench
Two tools with minimal descriptions (10-11 chars each, well below 20-char floor). Input schemas present but lack parameter descriptions entirely, parameters 'date' and 'sensor' have no guidance on format, constraints, or usage. No output schema documented. Error handling returns plain strings without recovery guidance. Tools lack idempotency hints and do not follow verb_noun naming (get_weather_data is acceptable, but get_humidity_sensor_data is verbose and unclear on distinction). No pagination, no structured output, no field-level documentation.
Get humidity sensor data for sensor name.
Get weather data for sensor name.
Tool descriptions are critically short (10-11 chars). 'Get weather data for sensor name.' and 'Get humidity sensor data for sensor name.' violate the 20-char minimum and provide no context on when to use each tool, what data is returned, or how they differ.
Input parameters lack descriptions. 'date' and 'sensor' parameters have no guidance on expected format (ISO 8601? Unix timestamp?), constraints (date range?), or examples. LLMs cannot infer parameter meaning from names alone.
No output schema documented. Tools return formatted strings with no structure. LLMs cannot parse results reliably or extract fields for downstream tool calls. Baseline: 100% of A+ tools document return types.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
Error handling returns plain strings ('No data found...', 'Unable to fetch...') without recovery guidance. LLMs cannot determine if errors are retryable, user-fixable, or fatal. No actionable next steps provided.
No tool annotations (readOnlyHint, idempotentHint). Both tools are read-only and idempotent, but this is not declared in the schema. Agents cannot optimize retry logic or understand safety properties.