Loading ToolBench
CLI tool to quickly scaffold new MCP servers for Cursor AI
This is a template/scaffold tool, not a production MCP server. The 10 math tools have inconsistent quality: add/subtract have good descriptions and parameter docs (10 tools total, 2 well-documented); multiply/divide/power/sqrt/sin/cos/tan/log lack any tool-level descriptions and have null parameter descriptions. 8 of 10 tools violate the critical rule that every parameter needs a description. Tool names are clear (add, subtract, multiply) but lack action verbs (should be 'compute_sum', 'compute_difference'). No output schemas documented. Error handling exists in code (divide-by-zero, sqrt of negative) but not surfaced in tool descriptions. Overall: 2 tools score ~65, 8 tools score ~25.
Adds two numbers together
8 of 10 tools lack descriptions entirely (multiply, divide, power, sqrt, sin, cos, tan, log).
8 of 10 tools have null parameter descriptions (multiply, divide, power, sqrt, sin, cos, tan, log). LLMs cannot infer parameter meaning from names alone. Violates critical rule: 'Every parameter needs a description.'
Tool names lack action verbs. 'add', 'multiply', 'sin' are nouns/functions, not verb_noun patterns. Should be 'compute_sum', 'compute_product', 'compute_sine' to match LLM expectations.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Subtracts the second number from the first
Error handling (divide-by-zero, sqrt of negative, log of non-positive) is implemented in code but not documented in tool descriptions. LLMs cannot plan error recovery without explicit guidance.
No output schemas documented. Tools return text content but LLMs need to know the structure (e.g., numeric result as string, range of values, units). Violates 'Document the output schema' rule.