Provides weather information for US locations using the National Weather Service API
mcp-quickstart is a basic quickstart/demo server with 7 tools across calculator and weather domains. All tools have descriptions and input schemas present, but quality is inconsistent. Tool names are mostly action-verb based (add, subtract, multiply, divide, get_alerts, get_forecast) but descriptions are extremely brief (8-50 chars, well below the 34-char p10 baseline for even minimal quality). Parameter descriptions are present and typed, but sparse. The server lacks any output schema documentation, error handling guidance, or composition patterns. All tools are READ_ONLY with no destructive operations. This is appropriate for a quickstart but falls short of production quality. The show_calculator tool (no parameters, generic UI description) is particularly weak. No tool annotations (readOnlyHint, etc.) are present despite being available in the spec. STDIO transport caps protocol readiness to 50 max.
Add two numbers.
Divide a by b.
Get weather alerts for a US state.
Get weather forecast for a location.
Multiply two numbers.
Display an interactive calculator with buttons in the chat.
Subtract b from a.
Tool descriptions critically short (8-50 chars). Rubric baseline p10=34 chars, avg=194 chars. Tools like 'add' (19 chars), 'subtract' (20 chars), 'multiply' (19 chars) provide minimal context for LLM selection. Descriptions lack WHEN to use, dependencies, or expected outputs.
No output schemas documented. Tools return results but LLMs cannot plan downstream calls without knowing the response structure. Critical for composition and error handling guidance.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all tools being READ_ONLY. Modern MCP spec (2026-07-28) expects annotations for protocol clarity.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 58 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 38 | - | v1 |
show_calculator is poorly named and described. Name does not follow verb_noun pattern (suggest 'display_calculator'). Description 'Display an interactive calculator with buttons in the chat' is vague about when to call it vs other calculator tools or expected user interaction model.
No error handling guidance. Tools provide no documentation on what errors they return, when they are retryable, or what the LLM should do if a call fails (e.g., if coordinates are invalid for get_forecast or state code is malformed for get_alerts).
get_alerts parameter 'state' has minimal description: 'Two-letter US state code (e.g. CA, NY)'. No enum constraint provided despite being a fixed, enumerable set. This invites hallucinated state codes like 'California' instead of 'CA'.
Arithmetic tool descriptions do not mention their purpose or when to use them. A description of 'Add two numbers' does not tell the LLM whether this is for calculations, financial operations, or UI display. Lacks context to compete with other tools in a crowded agent tool palette.