An MCP server example with tools, resources, and prompts for end-to-end testing
This demo server has critical definition quality gaps. Only 1 tool (echo) is present. The tool has a Zod schema with minimal inline type definition (message: z.string()), but parameter descriptions are null/missing entirely. Tool description is absent. The prompt and resource implementations show reasonable Zod usage but no descriptions are provided for those either. This server prioritizes e2e testing patterns over production-grade agent-facing definitions.
Tool 'echo' has no description. LLMs cannot determine when or why to select this tool without documentation.
Parameter 'message' in 'echo' tool has null/missing description. LLMs cannot infer what the message parameter controls.
Tool name 'echo' lacks a clear action verb. While 'echo' is understandable as 'return the input', a verb_noun pattern like 'return_message' or 'echo_text' better matches production conventions and improves LLM selectivity.
No output schema documentation. The tool returns structured output (text content), but there is no description of the response schema for the LLM to understand what fields to expect.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 37 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 34 | - | v1 |
Prompts and Resources defined but lack descriptions. The 'greet' prompt and 'config://app' resource have no docstrings to guide LLM discovery or usage.