Cross-client automated testing tool for Model Context Protocol (MCP) servers
Two tools with complete input schemas and descriptions, but descriptions are minimal (under 100 chars) and lack actionable context. Both tools are READ_ONLY and well-named with verb_noun pattern. No output schemas documented. Parameters have types and descriptions, meeting baseline requirements. Error handling present but generic. No tool annotations (readOnlyHint/destructiveHint). Schemas are properly structured JSON Schema with required fields declared.
Get carrier tracking details for an order
Lookup an order by numeric order ID in the e-commerce database
Output schemas not documented. LLMs cannot plan downstream calls or extract fields without knowing response structure.
Tool descriptions lack context on WHEN to use each tool and what distinguishes them. Both are order-related but no guidance on which to call first or in what sequence.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Both tools are READ_ONLY but this is not declared in the tool definition, forcing LLMs to infer safety.
Error handling returns generic error objects without recovery guidance. 'Unknown tool' error gives LLM no actionable next step.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
lookup_order accepts only integer order_id but no guidance on valid range, format, or what happens for non-existent IDs. get_order_tracking accepts any string with no validation hints.