Loading ToolBench
MCP server providing Korean financial data tools: DART electronic disclosures, KRX stock prices, and semantic document search via RAG
Server has 4 tools with complete input schemas and descriptions. Naming follows verb_noun pattern (get_*, search_*). Descriptions are present but vary in quality (72-280 chars). All parameters have type definitions and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Error handling is minimal (generic error strings, no recovery guidance). No tool annotations (readOnlyHint, idempotentHint). Parameter descriptions lack format constraints and validation rules. Missing pagination guidance for search results.
기업의 재무제표(매출액, 영업이익, 당기순이익 등)를 조회합니다.
KRX 주가 데이터를 조회합니다.
DART 전자공시 시스템에서 공시를 검색합니다.
인덱싱된 공시 전문(全文)에서 시맨틱(의미 기반) 검색을 수행합니다. 키워드 매칭이 아닌 벡터 유사도 검색이므로 개념적 질문에 적합합니다. 공시 목록 검색(키워드 기반)은 search_disclosures를 사용하세요.
Output schemas not documented. LLMs cannot predict response structure, field names, or types. Agents cannot reliably extract data for downstream tool calls.
Error handling lacks recovery guidance. Errors return generic strings (e.g. 'DART API 네트워크 오류') without actionable next steps. LLMs cannot self-correct or retry intelligently.
Parameter descriptions lack format constraints. 'year' accepts string but no validation that it is YYYY format. 'start_date' and 'end_date' lack YYYY-MM-DD format specification. LLMs may pass invalid dates.
search_disclosures and search_documents lack pagination parameters (limit, offset, page). Large result sets will blow context window. No guidance on result limits.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, idempotentHint). All tools are read-only but this is not declared. Agents cannot distinguish safe tools from destructive ones.