Two tools with minimal definitions. Both have descriptions (good baseline), but schemas are incomplete and lack parameter descriptions. 'echo' has a single string parameter with description via Zod, but 'debug' accepts empty input with no parameters to describe. Neither tool documents output structure or error handling. Naming is generic ('echo', 'debug'), 'echo' lacks a verb prefix and 'debug' is vague about what it returns. No tool annotations (readOnlyHint, etc.). Output is unstructured text blobs, not typed objects. This is a starter template, not production-grade.
Lists all available tools and methods
Echoes back a message with 'Hello' prefix
Tool 'debug' has empty input schema and vague description. LLM cannot determine when to call it or what it returns. Description 'Lists all available tools and methods' is ambiguous, does it return tool names, full schemas, or capability metadata?
No output schemas documented. Both tools return unstructured text in a 'content' array. LLMs cannot parse or chain results. Should return typed objects (e.g., {message: string} for echo, {tools: string[], methods: string[]} for debug).
Tool names lack action verbs. 'echo' should be 'echo_message' or 'repeat_message'. 'debug' should be 'list_tools' or 'describe_capabilities'. Generic names reduce LLM clarity.
No error handling or recovery guidance. If 'echo' receives null or 'debug' fails, LLM gets no actionable error message.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Both are read-only but this is not declared in the schema.