Loading ToolBench

Reading the current corpus…

logseq-mcp MCP server · C · ToolBench

← Browse

logseq-mcp

github.com/dearcloud09/logseq-mcp

MCP server for Logseq graph integration

2026-07-28+STDIO15 tools

60/100

Definition Quality79

Protocol Readiness50

Supportability36

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

Server has 15 tools with complete input schemas and descriptions in Korean. Naming follows verb_noun convention (list_, read_, create_, update_, delete_, search_, get_, append_, add_). Schemas use Zod validation with security limits (MAX_PATH_LENGTH=500, MAX_CONTENT_LENGTH=10MB). However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions are present but generic (e.g., '페이지 경로 또는 이름' lacks format/constraint details). Error handling is minimal, no recovery guidance or categorization. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE). Composition is sound: tools are single-responsibility and chainable (e.g., list_pages → read_page). Security: no secrets in params, but no audit logging or permission gates visible.

Tools (15)

add_articlewritesource verified80/100

오늘 저널에 아티클 정보 추가. 대화 정리, 웹 아티클, 읽은 글 등을 기록

add_bookwritesource verified80/100

오늘 저널에 책 정보 추가. 읽은 책, 읽고 싶은 책 기록

add_exhibitionwritesource verified80/100

오늘 저널에 전시회 정보 추가. 본 전시회, 보고 싶은 전시회 기록

add_moviewritesource verified80/100

오늘 저널에 영화 정보 추가. 본 영화, 보고 싶은 영화 기록

append_to_pagewritesource verified77/100

기존 페이지 끝에 내용 추가

create_journalwritesource verified78/100

오늘 또는 특정 날짜의 저널 페이지 생성

create_pagewritesource verified78/100

새 페이지 생성. Logseq 프로퍼티 포함 가능

delete_pagedestructivesource verified77/100

페이지 삭제

Issues (5)

HIGHtool ↗

Output schemas not documented. LLMs cannot predict response structure (fields, types, pagination). Forces agents to guess what data is returned and how to chain tools.

list_pagesread_pagesearch_pagesget_backlinksget_graph

HIGHtool-annotations ↗

No tool annotations despite clear risk levels. delete_page is DESTRUCTIVE but lacks destructiveHint. read_page is READ_ONLY but lacks readOnlyHint. Agents cannot distinguish safe from dangerous operations.

delete_pageread_pagelist_pagescreate_pageupdate_page

MEDIUMtool-description ↗

Parameter descriptions lack format/constraint details. 'date' params say 'YYYY-MM-DD' in description but no validation shown. 'path' accepts 'pages/note' OR 'note' but ambiguity not resolved in description. LLMs cannot reliably format inputs.

read_pageget_journalcreate_journaladd_articleadd_book

Recommendations

Document output schemas for all tools. Example for list_pages: 'Returns {pages: [{name, path, tags, links, backlinks}], total_count}'. LLMs need to know what fields to extract and how to chain results.
Add tool annotations to MCP tool definitions. Mark delete_page with destructiveHint=true, read_page with readOnlyHint=true, create_page with idempotentHint=false. Use MCP SDK's tool annotation API.
Enhance parameter descriptions with explicit format/constraint rules. Change 'date (YYYY-MM-DD, 기본값: 오늘)' to 'ISO 8601 date string (YYYY-MM-DD). Defaults to today if omitted. Example: 2024-01-15'. Include valid ranges for numeric params (depth: 0-10).
Implement error categorization and recovery guidance. Return structured errors: {code, message, retryable, nextSteps}. Example: 'Page not found. Try search_pages() to find the correct name, or create_page() to add it.'
Add dry-run support to delete_page. Accept optional 'dry_run=true' parameter. When true, return what would be deleted without actually deleting. Prevents accidental data loss.

Score history

Overall score trend

First recorded score · v2 rubric

60/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	C	60	2026-07-28+	v2