Loading ToolBench
Scaffolder, validator, and test harness for local Model Context Protocol (MCP) servers, targeting MCP spec 2025-06-18.
Three tools with minimal definitions. All have schemas and descriptions present, but descriptions are extremely brief (10-20 chars), falling below the 34-char p10 baseline. Parameters lack descriptions entirely (word_count has no description for 'text'). No output schemas documented. Tools are trivial samples, not production-grade. Naming is acceptable but generic. No error handling guidance, no composition hints.
Add two integers and return the sum as structured content.
Echo back the provided text. A minimal sample tool.
Count the words in a string (defensive, read-only).
Parameter descriptions missing or incomplete. 'word_count' tool has a 'text' parameter with no description. LLMs cannot infer what the parameter controls without explicit text.
Tool descriptions are extremely brief (10-20 chars), well below the 34-char p10 baseline. 'Echo back the provided text' and 'Count the words in a string' lack context for when/why to use them. LLMs need 50-200 char descriptions to disambiguate from similar tools.
No output schemas documented. Callers cannot know what fields to expect from tool responses. LLMs need structured output documentation to plan downstream calls.
No error handling guidance. Tools provide no recovery hints if inputs are invalid or calls fail. LLMs receive no direction on what to do next.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | 2025-06-18+ | v2 |
Tools are trivial samples (echo, add, word_count) with no real-world utility. This is a scaffolding/testing library, not a production MCP server. Definition quality reflects sample-grade implementations.