API を介して LogSeq と対話するための MCP サーバー。
サーバーは、LogSeq と対話するための複数のツールを実装しています。
list_graphs: 利用可能なすべてのグラフを一覧表示します
list_pages: 現在のグラフ内のすべてのページを一覧表示します
get_page_content: 1ページのコンテンツを返す
検索: すべてのページのコンテンツを検索
create_page: 新しいページを作成する
update_page: 既存のページのコンテンツを更新する
delete_page: ページを削除する
まずClaudeにLogSeqを使うように指示するのが良いでしょう。そうすれば、常にツールが呼び出されるようになります。
プロンプトの例:
最新の会議メモの内容を取得して要約する
プロジェクトXが言及されているすべてのページを検索し、その背景を説明します。
今日の会議メモを新しいページとして作成する
プロジェクトのステータスページを最新のアップデートで更新します
Related MCP server: Logseq MCP Server
LogSeq API 設定を使用して環境を構成するには、次の 2 つの方法があります。
サーバー設定に追加（推奨）
{
"mcp-logseq": {
"command": "uvx",
"args": [
"mcp-logseq"
],
"env": {
"LOGSEQ_API_TOKEN": "<your_api_token_here>",
"LOGSEQ_API_URL": "http://localhost:12315"
}
}
}必要な変数を含む.envファイルを作業ディレクトリに作成します。
LOGSEQ_API_TOKEN=your_token_here
LOGSEQ_API_URL=http://localhost:12315配布用のパッケージを準備するには:
依存関係を同期し、ロックファイルを更新します。
uv syncMCPサーバーはstdio経由で実行されるため、デバッグが困難になる場合があります。最適なデバッグ環境を実現するには、 MCP Inspectorの使用をお勧めします。
MCP インスペクターは、次のようにnpm経由で起動できます。
npx @modelcontextprotocol/inspector uv --directory /path/to/mcp-logseq run mcp-logseqCreate a new page in Logseq with properly structured blocks.
Fails if a page with the same title already exists (use update_page to modify existing pages). This makes retries safe: re-sending a create_page that timed out will not create numbered duplicates like "Page(1)".
Markdown content is automatically parsed into Logseq's block hierarchy:
Headings (# ## ###) create nested sections
Lists (- or 1.) become proper block trees
Code blocks are preserved as single blocks
YAML frontmatter (---) becomes page properties
Example content:
---
tags: [project, active]
priority: high
---
# Project Title
Introduction paragraph.
## Tasks
- Task 1
- Subtask A
- Task 2Inputs: title (required), content, properties.
Delete a block from LogSeq by its UUID.
Inputs: block_uuid (required).
Delete a page from LogSeq.
Inputs: page_name (required).
Find all pages that have a specific property, optionally filtered by value. Simpler alternative to the full query DSL.
Inputs: property_name (required), limit, property_value.
Get a single block by its UUID. Returns the block content, properties, and child blocks (recursively). Useful for inspecting a specific block after finding its UUID via search or query.
Inputs: block_uuid (required), format, include_children.
Get all pages and blocks that link to a specific page (backlinks/linked references).
Inputs: page_name (required), include_content.
Get the content of a specific page from LogSeq.
Inputs: page_name (required), format, max_depth, resolve_refs.
Get all pages within a namespace hierarchy (flat list). Use this to discover subpages of a parent page.
Inputs: namespace (required).
Get pages within a namespace as a hierarchical tree structure. Useful for understanding the full page hierarchy.
Inputs: namespace (required).
Insert a new block as a child or sibling of an existing block, enabling nested hierarchical structures
Inputs: content (required), parent_block_uuid (required), sibling, properties.
Lists all pages in a LogSeq graph, each with its last-updated time (UTC). To find recently modified pages, use sort='updated' (most recent first) and/or updated_since.
Inputs: sort, limit, updated_since, include_journals.
Execute a Logseq DSL query to search pages and blocks. Supports property queries, tag queries, task queries, and logical combinations. Raw datalog ({:query ...} / [:find ...]) is not supported. See https://docs.logseq.com/#/page/queries for query syntax.
Inputs: query (required), limit, format, result_type.
Rename an existing page. All references throughout the graph will be automatically updated.
Inputs: new_name (required), old_name (required).
Search for content across LogSeq pages, blocks, and files
Inputs: query (required), limit, format, include_files, +2 more.
Set properties on a block in Logseq DB-mode. Properties must be defined on the block's tag/class. Use property display names (e.g. 'Content status', not the internal ident).
Inputs: block_uuid (required), properties (required).
Update the content of an existing LogSeq block by UUID.
Inputs: content (required), block_uuid (required).
Update a page in Logseq with new content and/or properties.
Supports two modes:
append: Add new blocks after existing content (default)
replace: Clear all existing blocks and add new content
Markdown is parsed into proper block hierarchy just like create_page. YAML frontmatter in content will be merged with explicit properties.
Inputs: page_name (required), mode, content, properties.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 17 tools
Each tool targets a distinct resource and action: pages, blocks, search/query, namespaces, and properties. Overlaps like query vs find_pages_by_property are explicitly framed as simple alternatives, so an agent can choose correctly.
Mostly consistent verb_noun snake_case (list_pages, get_block, create_page), but 'query' and 'search' are bare verbs, a minor deviation from the pattern.
17 tools is slightly above the ideal 3-15 range for a cohesive set, though each covers a distinct operation. The count feels a bit heavy but not excessive for LogSeq's complexity.
Core CRUD for pages and blocks is covered, plus searching, querying, backlinks, and namespace navigation. Minor gaps remain (e.g., explicit block creation on a page, move/reorder operations), but agents can work around them.
Query your app's logs, metrics and incidents from your AI assistant.
Search, read, and safely update Markdown notes in your connected Phasoric knowledge vaults.
Notes, kanban tasks, and a capture inbox: read and create Flow data with a personal token.
MCP-native open-source Notion alternative: read & write pages, databases and kanban boards.
