A simple Model Context Protocol (MCP) server that provides weather information tools, resources, and prompts.
Single tool with basic structure but significant gaps. get_weather has a verb-noun name (good), but description is only 56 chars, below the 194-char baseline for production tools. Input schema present with type and description, but output is unstructured free-text string. No error handling, no pagination, no output schema documentation. Tool is READ_ONLY (correct risk classification) but lacks idempotent/readOnly annotations. Resources and prompts are present but not evaluated as tools per the rubric.
Return a mock weather report for a given location.
Description too short (56 chars vs 194-char baseline). Lacks WHEN to use, prerequisites, and return value documentation.
Output schema not documented. Returns unstructured string; LLM cannot plan downstream calls or extract structured fields.
No error handling or recovery guidance. If location is invalid or API fails, LLM receives no actionable error message.
Tool lacks annotations (readOnlyHint, idempotentHint). Clients cannot infer safety properties without explicit metadata.
Location parameter accepts free-form string with no validation, format, or constraint hints. LLM may pass invalid values.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | <=2025-11-25 | v2 |