Loading ToolBench
Executable MCP App production boundary laboratory for testing model-visible, component-visible, component-only, resource-delivery, and optional host-capability projections
Single tool 'inspect_boundary' has a well-structured schema with proper enum constraints and parameter descriptions. However, the tool description is vague about its actual purpose and use cases. The description mentions 'deterministic MCP App boundary specimen' and 'projections' without clearly explaining what the tool does, when to call it, or what the user should expect. Parameter descriptions are adequate but the overall tool lacks clarity on practical application. No output schema is documented in the source code.
Return one deterministic MCP App boundary specimen. Use it to inspect model-visible, component-visible, component-only, resource-delivery, and optional host-capability projections without claiming named-host acceptance.
Tool description lacks clarity on purpose and use cases. 'Return one deterministic MCP App boundary specimen' and references to 'projections' are jargon-heavy and do not explain what the tool actually does or when an LLM should call it.
No output schema documented. The tool returns data but the source code does not specify what fields, types, or structure the response contains. LLMs cannot plan downstream actions without knowing the output shape.
Tool name 'inspect_boundary' is vague. It does not start with a clear action verb (get_, list_, describe_, etc.) and does not convey what 'boundary' means in this context. Rename to something like 'describe_mcp_boundary' or 'get_boundary_projection'.
Parameter 'scenario' enum values ('resource-delivery', 'optional-capability') lack descriptions. The tool description does not explain what each scenario does or when to use one vs the other.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Tool description does not state whether this is a read-only inspection or if it has side effects. The 'Risk: READ_ONLY' annotation is present but not reflected in the description text, which LLMs rely on to understand tool safety.