Loading ToolBench
Two simple tools with basic schemas and minimal descriptions. Both tools have identical input schemas (MathOperation) and descriptions under 30 chars, which violates the 10-1024 char guideline and provides insufficient context for LLM selection. No output schema documented. No error handling guidance. No parameter constraints or validation hints. Tool names are verb-based (add, multiply) but descriptions lack WHEN/WHY context. Prompts feature is implemented but not relevant to tool quality.
Adds two numbers.
Multiplies two numbers.
Tool descriptions are too short (16 chars each). Descriptions must be 10-1024 chars and explain WHAT, WHEN, and WHY. Current descriptions ('Adds two numbers.', 'Multiplies two numbers.') lack context for LLM selection and do not distinguish when to use each tool.
No output schema documented. LLMs cannot plan downstream calls or extract results without knowing the response structure. Tool descriptions should state what fields are returned.
No error handling guidance. Error responses (e.g., invalid input) do not tell the LLM what to do next or how to self-correct. Missing recovery hints.
Parameter descriptions are generic ('First number', 'Second number'). No constraints on range, format, or valid values. Descriptions should specify expected input format and any limits.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |
Tool composition is minimal. Both tools accept identical parameters and perform similar operations. No guidance on when to use add vs multiply, or how they chain with other tools.