MCP Remote Auth Demo - an OAuth 2.0 authorization server demonstrating MCP protocol integration with Cloudflare Workers
Single tool 'add' with minimal but present definition. Tool naming is generic (lacks action verb clarity), description is extremely brief (17 chars), and parameter descriptions are adequate but sparse. No output schema documented. Input schema is visible with proper Zod validation but lacks richness. No error handling guidance, no composition patterns, and no security considerations evident.
Add two numbers
Tool name 'add' lacks action verb prefix and is ambiguous. Does it add numbers, add items to a list, add users to a group? LLMs will struggle to disambiguate if other tools are added.
Tool description is only 17 characters ('Add two numbers'). Baseline for A-grade tools is 50-200 chars. Missing WHEN to use, prerequisites, and return value details.
Output schema is not documented. LLM cannot see what structure to expect from the response. Current implementation returns { content: [{ type: 'text', text: String(...) }] } but this is hidden from the schema definition.
Parameter descriptions are minimal. 'First number to add' and 'Second number to add' are adequate but lack constraint information (range, precision, limits). Baseline for A-grade param descriptions is 40-120 chars with format/range details.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 40 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 42 | - | v1 |
No error handling guidance. What happens if inputs are invalid, null, Infinity, or NaN? No recovery hints provided to LLM.