MCP server for e-commerce backend operations, providing user management tools
This MCP server has severe definition quality issues across both tools. The first tool 'getListUsers' lacks any description and parameter documentation entirely. The second tool 'list-users' has a basic description but critical schema and naming problems. Both tools return unstructured JSON text rather than properly typed responses. No error handling guidance, no output schemas documented, and parameter descriptions are minimal. The server does not follow the verb_noun naming convention consistently (mixing camelCase 'getListUsers' with kebab-case 'list-users'). No pagination, no structured output, no recovery guidance on errors. This is well below production quality.
List users optionally filtered by firstname
Tool 'getListUsers' has NO description whatsoever. Per hard scoring rules, description score MUST be 0. LLMs cannot determine when or why to select this tool.
Tool 'getListUsers' has NO input schema visible in the source code. The tool registration shows empty object {} as parameters with no property definitions. Per hard scoring rules, schema score MUST be 0.
Inconsistent tool naming convention: 'getListUsers' uses camelCase while 'list-users' uses kebab-case. Both should follow verb_noun pattern consistently (e.g., 'list_users' for both). This forces LLMs to reason about naming variations.
Both tools return raw JSON text wrapped in a content array with type='text', rather than structured, typed responses. LLMs must parse unstructured output, wasting tokens and increasing error likelihood. Response schema is not documented.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 26 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 20 | - | v1 |
Tool 'list-users' accepts an optional 'firstname' parameter with description 'Filter by firstname', but does not validate the input. No constraints (min/max length), no regex pattern, and error responses are generic. If the API fails or returns empty, the error message is basic.
No pagination support or result limits. Tool 'getListUsers' fetches ALL users without limit, risking context window exhaustion. No page/offset/limit parameters, no total count, no cursor.
Error handling is minimal. Tool 'list-users' checks if users.length === 0 and returns a message, but does not categorize errors (retryable vs fatal), provide recovery guidance, or include constraint violations. A network error or API 500 returns a raw stack trace with no actionable next step.
The 'firstname' parameter in 'list-users' accepts any string without validation. No enum constraint, no length bounds, no format hint. An LLM could pass special characters, SQL injection payloads, or extremely long strings without friction.
No output schema documented. Tools return JSON text, but the structure (fields, types, required keys) is not declared. LLMs must guess what fields exist in the response and how to extract relevant data.
Tool names do not convey clear intent. 'getListUsers' is ambiguous, does 'get' mean retrieve, display, or fetch? 'list-users' is clearer but inconsistently named.