MCP сервер для надання інформації про Monad гаманці в тестнет мережі
This server exposes blockchain wallet queries via HTTP REST endpoints, not as MCP tools. The source code shows Express routes (wallet.js) that handle HTTP requests, not MCP tool definitions. Tool names, descriptions, and schemas are inferred from route handlers and parameter validation logic, not from explicit MCP tool registration. No MCP protocol implementation is visible, no tool definitions with JSON Schema, no MCP request/response handling, no capabilities negotiation. Descriptions are in Ukrainian and extremely brief (10 - 20 chars), below the 20-char floor. Input schemas exist in code but are not formally declared in MCP format. This is a REST API masquerading as an MCP server.
Отримання тільки баланса нативного токену
Отримання баланса конкретного токена
Отримання всієї інформації про гаманець
No MCP tool registration detected. Tools are inferred from Express routes, not explicitly registered via MCP protocol. Tool definitions must be declared in MCP format with JSON Schema input/output.
Tool descriptions are under 20 characters and in Ukrainian ('Отримання всієї інформації про гаманець'). Descriptions must be in English, 50 - 200 chars, and explain WHAT the tool does, WHEN to use it, and what it returns.
Input schemas are embedded in route validation logic, not formally declared as JSON Schema in MCP tool definitions. Parameter descriptions exist in code comments but are not exposed to the MCP client.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 23 | 2026-07-28+ | v2 |
No output schema documentation. Callers cannot predict response structure (e.g., getWalletInfo returns {address, nativeBalance, tokens, tokenErrors}). LLMs need documented output schemas to plan downstream calls.
Error handling returns generic HTTP status codes (400, 500) with minimal context. Errors do not guide recovery (e.g., 'Invalid address' does not suggest what a valid address looks like). Per pattern:recovery-guide, errors must tell the LLM what to do next.