FastMCP + Mangum + Cognito OAuth sample MCP server for AWS Lambda with passkey authentication integration
Single tool 'add' has a well-formed JSON Schema with proper type definitions and parameter descriptions. Tool name is a clear verb. Description is present but extremely brief (Japanese; translated: '2つの数値を加算します' = 'Adds two numbers'). The tool is READ_ONLY, simple, and correctly implements a basic arithmetic operation. The server uses FastMCP with HTTP+Mangum/Lambda, supports OAuth via OIDCProxy, and integrates Cognito. Tool registration is explicit and visible in lambda/main.py. The single tool prevents comprehensive pattern evaluation, but the implementation is clean and follows MCP conventions.
2つの数値を加算します。
Tool description is extremely brief (15 chars; baseline minimum ~34 chars for adequate context). LLMs may struggle to assess when to invoke this tool relative to other arithmetic operations.
No output schema documented. LLMs cannot predict the return type or structure. While 'add(int, int) -> int' is obvious, explicit schema documentation is a baseline for 80+ scoring.
Tool description is in Japanese only ('2つの数値を加算します'). English descriptions are required for LLM clarity in non-Japanese systems.
Only one tool present. The server is a minimal sample demonstrating OAuth + Lambda integration, not a feature-complete MCP server. Cannot evaluate composition patterns (tool chains, complementary tools).
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 57 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 54 | - | v1 |