Loading ToolBench
Single tool 'example_tool' with minimal definition. Description is 18 chars ('Just example tool'), below the 20-char threshold for meaningful guidance. Input schema present but sparse: only 'name' parameter with basic string type and minimal description ('name parameter'). No output schema documented. No error handling guidance. Tool name lacks action verb (should be 'get_greeting' or similar). No indication of what the tool returns or when to use it. Fails baseline expectations for LLM-optimized descriptions (target 50-200 chars) and parameter clarity.
Just example tool
Tool description too short (18 chars). Must be 20+ chars and explain WHAT the tool does, WHEN to use it, and WHAT it returns. Current: 'Just example tool' provides no actionable context for LLM selection.
Tool name lacks action verb. 'example_tool' is generic and non-descriptive. Should start with verb (get_, create_, search_, etc.) to signal intent. E.g., 'get_greeting' or 'generate_greeting'.
Parameter description insufficient. 'name parameter' is circular and adds no value. Should explain: what kind of name (user, product, etc.), expected format, constraints, and examples of valid input.
No output schema documented. LLM cannot predict return type or structure. Must document: returns string greeting? JSON object? What fields? What format?
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 34 | <=2025-11-25 | v2 |
No error handling guidance. Tool provides no recovery hints if 'name' is empty, null, or invalid. Should specify: what happens on bad input? What should LLM do next?