MCP server and client implementation for Spring AI chatbot CLI, featuring tool calling, RAG, multi-agent orchestration, and MCP protocol integration
This is a Spring AI educational codebase, not an MCP server. No MCP protocol implementation detected. Tools are defined as Spring AI function callbacks with basic schemas and Korean-language descriptions. Naming follows verb_noun patterns (current_datetime, calculate, discount_calculator) but descriptions lack LLM-optimization guidance. Input schemas are present but minimal, most parameters lack detailed constraints, ranges, or format specifications. No output schemas documented. Error handling is absent from visible code. Security considerations (secret injection, permission gates) not evident. Overall structure suggests educational examples rather than production-grade agent tools.
두 숫자에 대해 add, subtract, multiply, divide 중 하나의 연산을 수행합니다.
오늘 날짜를 yyyy-MM-dd 형식으로 반환합니다.
현재 날짜와 시간을 IANA 시간대 기준으로 반환합니다. 시간대가 없으면 Asia/Seoul을 사용합니다.
고객 ID로 고객 연락처와 지원 등급을 조회합니다. 모델에는 이메일이 마스킹된 결과만 전달합니다.
정가와 할인율을 받아 할인 후 금액을 계산합니다.
전체 값에서 일부 값이 차지하는 비율을 퍼센트로 계산합니다.
툴 컨텍스트에 담긴 사용자/대화 정보를 사용해 현재 학습 세션을 요약합니다.
Not an MCP server, this is a Spring AI educational codebase. No MCP protocol transport, tool registration, or remote accessibility. Cannot be used as an MCP server without complete reimplementation.
Descriptions are in Korean and lack LLM-optimization. Descriptions should be 50 - 200 characters, action-oriented, and in English for broad LLM compatibility. Current descriptions (e.g., '현재 날짜와 시간을 IANA 시간대 기준으로 반환합니다...') are 40 - 80 chars but lack clarity on WHEN to use the tool vs alternatives.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 32 | 2026-07-28+ | v2 |
Input schemas lack constraints, ranges, and format specifications. 'operation' parameter in calculate accepts free-form strings instead of an enum (add|subtract|multiply|divide). 'zoneId' in current_datetime has no pattern or validation. Parameters need explicit constraints to prevent LLM hallucination.
No output schemas documented. Tools return unstructured responses with no field definitions. LLMs cannot plan downstream calls or extract structured data without knowing the response shape.
No error handling or recovery guidance visible. Tools do not document what happens on invalid input (e.g., invalid timezone, division by zero, missing customer ID). LLMs need actionable error messages to self-correct.