An MCP server that exposes business domain tools for payment transactions and random number generation, integrated with a .NET AI agent framework
This MCP server exposes 2 tools with basic schemas and descriptions. Both tools have input schemas with typed parameters and defaults, meeting minimum definition standards. However, descriptions are generic and lack depth required for reliable LLM tool selection. GetRecentPayments lacks actionable context about when to use it vs alternatives, and parameter descriptions are minimal. GetRandomNumber is a toy tool with adequate but minimal documentation. Neither tool has output schema documentation, and there is no error handling guidance. The server does not implement tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite both tools being READ_ONLY. Per-tool scores average to 52, solidly in the 'poor' range with room for improvement in description quality and output schema documentation.
Generates a random number between the specified minimum and maximum values.
Returns recent payment transactions from the payment system
Output schemas are not documented. LLMs do not know what fields GetRecentPayments returns, forcing them to guess or invoke with confidence issues.
Tool descriptions lack WHEN/WHY context. 'Returns recent payment transactions' does not explain when to call it instead of searching for a specific transaction. 'Generates a random number' is vague about use cases, unit tests? game mechanics? mock data?
GetRandomNumber is a toy tool with no practical business value in a payment system context. Its presence suggests testing infrastructure not yet removed; unclear if this is intentional.
No tool annotations present despite both tools being READ_ONLY. readOnlyHint should be set to true for safe caching and retry semantics.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | F | 45 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 40 | - | v1 |
Parameter 'count' in GetRecentPayments has default=7 with no documented range or max. An LLM could request 10,000 payments, straining the backend. No limits enforced in schema.
No error handling guidance. If GetRecentPayments fails (auth timeout, backend down), the error response does not tell the LLM what to do next.