Loading ToolBench
MCP server demonstrating streamable HTTP transport with tools, prompts, and resources
Server implements 1 tool (add_integers) with a clear, concise description and properly typed input schema. However, there are significant gaps in production-readiness. The tool lacks error handling guidance, has no output schema documentation, and the parameter descriptions, while present, are minimal. The server also implements prompts and resources, which add breadth but are not the focus of tool evaluation. Based on the rubric calibration, this falls into the 'Fair' (C) range: descriptions are present and adequate, schemas are properly typed, but error handling is absent and best practices for composition and chaining are not evident.
This function is used whenever we want to add two 'Integer' numbers together.
No error handling or recovery guidance documented. Tool lacks description of what happens on invalid input (e.g., integer overflow, null values) and provides no recovery hints for the LLM.
Output schema not documented. While the function returns an int, there is no explicit schema definition stating what the output structure is, preventing LLMs from planning downstream tool calls.
Parameter descriptions are minimal (e.g., 'The first Integer number to add'). Best practice would include constraints like min/max bounds, edge case behavior, or type clarifications beyond the schema.
No documentation of idempotency or side effects. The description does not clarify whether the tool is read-only or if repeated calls are safe.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 66 | 2026-07-28+ | v2 |
| 2026-03-09 | C | 69 | - | v1 |