Wishket-mcp has 5 tools with basic naming and descriptions, but significant gaps in schema completeness and parameter documentation. Tool names follow verb_noun convention (search_projects, scan_new, get_project, list_filters, reset_cache), which is good. However, descriptions are inconsistent in quality and length. Most critically, input schemas visible in the code show parameter types but lack comprehensive descriptions for all parameters. The 'raw' parameter in search_projects is particularly problematic, it accepts unvalidated object input, violating constrained-input patterns. Error handling is minimal; no recovery guidance or actionable error messages are evident. Output schemas are not documented. The server lacks tool annotations (readOnlyHint, destructiveHint) despite having clear risk classifications (READ_ONLY, WRITE, REVERSIBLE).
프로젝트 상세 조회 (JSON-LD 전체 설명 포함). id는 숫자 문자열 (예: "158063").
위시켓 검색 필터 목록 (카테고리, 형태, 기술 등). 캐시 1시간.
seen 캐시 초기화 (다음 scan_new는 전체가 신규). 신중히 사용.
마지막 스캔 이후 신규 프로젝트만 반환 (신규는 seen 캐시에 기록). 기본 최대 3페이지(30건) 조회, 요청 간 5초 지연 (robots Crawl-delay). 첫 실행은 전체가 신규(베이스라인).
위시켓 프로젝트 검색 (캐시 기록 없음, 순수 조회). 페이지당 10건. 기본 카테고리 development.
search_projects accepts 'raw' parameter as unvalidated object, violating constrained-input pattern. LLMs cannot determine valid structure or values.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., project IDs for chaining).
Tool annotations missing. Despite clear risk classifications (READ_ONLY, WRITE, REVERSIBLE), no readOnlyHint, destructiveHint, or idempotentHint are present in tool definitions.
Parameter descriptions incomplete. 'category' and 'form_factors' described as 'semantic validated key' without explaining valid values or how to discover them. Users must call list_filters first, but this dependency is undocumented.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2026-07-28+ | v2 |
No error handling guidance. No recovery hints, retryability classification, or actionable error messages visible in code. Agents cannot self-correct on failures.