Loading ToolBench
An MCP server that provides echo functionality through tools, resources, and prompts
The Echo server defines 3 artifacts (tool, resource, prompt) with minimal documentation. The single tool 'echo' has a basic schema but lacks depth in descriptions and parameter guidance. Descriptions are present but generic (10-50 chars), falling well below the 50-200 char production baseline. No error handling guidance, no output schema documentation, and parameter descriptions are minimal. The server is functionally correct but demonstrates poor LLM optimization, no actionable guidance, no recovery paths, and no clear distinction of when to use each artifact.
Tool echo: returns the input message prefixed with 'Tool echo: '
Tool description is only 58 characters and lacks actionable context. 'Tool echo: returns the input message prefixed with \'Tool echo: \'' tells LLMs WHAT it does but not WHEN to use it, WHAT it returns structurally, or dependencies. Production baseline for descriptions is 50-200 chars with explicit purpose, use cases, and return hints.
Parameter 'message' lacks a description (input schema shows only type: string). LLMs cannot infer whether this accepts free text, code, commands, or data. Description should specify format constraints (e.g., 'Free-form text up to 1000 chars', 'Markdown-formatted message'). Currently scores 0 on parameter descriptions.
No documented output schema. The tool returns {content: [{type: 'text', text: string}]} but this is not explicitly documented in tool metadata. LLMs need to know the response structure to plan downstream operations and extract fields reliably.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 46 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 54 | - | v1 |
No error handling guidance. Tool has no documented error cases, recovery suggestions, or error classification. If 'message' is omitted or malformed, LLM receives no guidance on what to do. Production tools must include: 'Returns error if message is empty, provide non-empty text and retry.'
Redundant artifacts. Tool 'echo', Resource 'echo://{message}', and Prompt 'echo' all do essentially the same thing, echo text with different prefixes ('Tool echo:', 'Resource echo:', 'Please process this message:'). This violates composition principles: multiple tools doing the same thing waste LLM reasoning cycles. Consolidate into one canonical 'echo' tool or clearly document distinct use cases for each.
Tool name 'echo' is generic and action-verb weak. Does not follow verb_noun convention (get_, create_, send_, etc.). 'echo' alone is ambiguous, is it a greeting, debug tool, or message relay? Better name: 'echo_message' or 'process_message' with explicit purpose in description.