MCP server for customer support automation with tracking and knowledge base search capabilities
Two tools with minimal descriptions (both ~50 chars, baseline 194 chars). Input schemas present but sparse: single string parameters with basic descriptions. No output schemas documented. No error handling guidance, no parameter constraints (enums, ranges), no idempotency hints. Tools lack context on when to use them vs alternatives. fetch_tracking_data returns mock data; search_policy calls external API but no error recovery documented. No tool annotations (readOnlyHint, destructiveHint). Naming is clear (verb_noun) but descriptions are generic.
Fetch tracking information from shipping API.
Search company policy and knowledge base for relevant information.
Missing output schemas. Neither tool documents what fields are returned or their types. LLMs cannot plan downstream calls or extract required data.
Descriptions are too short (35 chars vs baseline 194 chars). 'Fetch tracking information from shipping API' and 'Search company policy and knowledge base' lack context on WHEN to use each tool, what data they return, or prerequisites.
No error handling guidance. search_policy calls external API (http://localhost:8000/api/search-kb) but no documentation of failure modes, retry logic, or recovery steps for LLM.
No parameter constraints. 'tracking_number' and 'query' are bare strings with no format, length, or pattern guidance. LLMs may pass invalid values without validation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | 2026-07-28+ | v2 |
No tool annotations. Tools lack readOnlyHint, destructiveHint, or idempotentHint. LLMs cannot determine if tools are safe to retry or have side effects.