Logseq MCP Tools

A Model Context Protocol (MCP) server that provides AI assistants with structured access to your Logseq knowledge graph.

Overview

This project creates an MCP server that allows AI assistants like Claude to interact with your Logseq knowledge base. It provides tools for:

• Retrieving a list of all pages
• Getting content from specific pages
• Generating journal summaries for flexible date ranges
• Extracting linked pages and exploring connections

Installation

Installing via Smithery

To install Logseq Tools for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @joelhooks/logseq-mcp-tools --client claude

1. Clone this repository
2. Install dependencies using npm, yarn, or pnpm:

# Using npm
npm install

# Using yarn
yarn install

# Using pnpm
pnpm install

3. Copy the environment template and configure your Logseq token:

cp .env.template .env
# Edit .env with your Logseq authentication token

Configuration

This project includes a .env.template file that you can copy and rename to .env.

You can find your Logseq auth token by:

1. Opening Logseq
2. Enabling the HTTP API in Settings > Features > Enable HTTP API
3. Setting your authentication token in Settings > Features > HTTP API Authentication Token

Usage

Running the MCP Server

The server can be started using:

# Using the npm script
npm start

# Or directly with tsx
npx tsx index.ts

Connecting with Claude

Claude Desktop

Follow the Claude MCP Quickstart guide:

1. Important: Install Node.js globally via Homebrew (or whatever):

brew install node

2. Install the Claude desktop app
3. Open the Claude menu and select "Settings..."
4. Click on "Developer" in the left sidebar, then click "Edit Config"
5. This will open your claude_desktop_config.json file. Replace its contents with:

{
  "mcpServers": {
    "logseq": {
      "command": "npx",
      "args": [
        "tsx",
        "/path/to/your/index.ts"
      ]
    }
  }
}

IMPORTANT: Replace /path/to/your/index.ts with the exact absolute path to your index.ts file (e.g., /Users/username/Code/logseq-mcp-tools/index.ts)

6. Save the file and restart Claude Desktop

Now you can chat with Claude and ask it to use your Logseq data:

• "Show me my recent journal entries"
• "Summarize my notes from last week"
• "Find all pages related to [topic]"

Claude in Cursor

Follow the Cursor MCP documentation:

1. Open Cursor
2. Add a new MCP service from settings
3. Enter the following command:

npx tsx "/path/to/index.ts"

4. Give your service a name like "Logseq Tools"

Now you can use Claude in Cursor with your Logseq data.

Claude in Anthropic API (generic)

When using the Claude API or CLI tools, you can add the MCP service with:

claude mcp add "logseq" npx tsx "/path/to/index.ts"

Jetbrains Junie Setup

To run this project with JetBrains Junie, follow these steps:

Build the Docker image:

docker build -t logseq-mcp .
# run in project root

Then add the following config to your Junie MCP configuration:

    "logseq": {
      "command": "c:\\Program Files\\Docker\\Docker\\resources\\bin\\docker.exe",
      "args": [
        "run",
        "-i",
        "--rm",
        "--network=host",
        "-e",
        "LOGSEQ_TOKEN=<Your Token>",
        "-e",
        "LOGSEQ_HOST=host.docker.internal",
        "logseq-mcp"
      ]
    }

Available Tools

getAllPages

Retrieves a list of all pages in your Logseq graph.

getPage

Gets the content of a specific page.

Parameters:

• pageName: The name of the page to retrieve

getJournalSummary

Generates a summary of journal entries for a specified date range.

Parameters:

• dateRange: Natural language date range like "today", "this week", "last month", "this year", etc.

This tool will:

• Collect journal entries in the specified range
• Format them in a readable way
• Extract and analyze referenced pages/concepts
• Show the most frequently referenced concepts

createPage

Creates a new page in your Logseq graph.

Parameters:

• pageName: Name for the new page
• content: (Optional) Initial content for the page

searchPages

Searches for pages by name.

Parameters:

• query: Search query to filter pages by name

getBacklinks

Finds all pages that reference a specific page.

Parameters:

• pageName: The page name for which to find backlinks

analyzeGraph

Performs a comprehensive analysis of your knowledge graph.

Parameters:

• daysThreshold: (Optional) Number of days to look back for "recent" content (default: 30)

Features:

• Identifies frequently referenced pages
• Tracks recent updates
• Discovers page clusters and connections
• Lists outstanding tasks
• Suggests potential updates needed

findKnowledgeGaps

Analyzes your knowledge graph to identify potential gaps and areas for improvement.

Parameters:

• minReferenceCount: (Optional) Minimum references to consider (default: 3)
• includeOrphans: (Optional) Include orphaned pages in analysis (default: true)

Features:

• Identifies missing pages that are frequently referenced
• Finds underdeveloped pages that need expansion
• Lists orphaned pages with no incoming links
• Provides summary statistics

analyzeJournalPatterns

Analyzes patterns in your journal entries over time.

Parameters:

• timeframe: (Optional) Time period to analyze (e.g., "last 30 days", "this year")
• includeMood: (Optional) Analyze mood patterns if present (default: true)
• includeTopics: (Optional) Analyze topic patterns (default: true)

Features:

• Topic trends and evolution
• Mood pattern analysis
• Habit tracking statistics
• Project progress tracking

smartQuery

Executes natural language queries using Logseq's DataScript capabilities.

Parameters:

• request: Natural language description of what you want to find
• includeQuery: (Optional) Include the generated Datalog query in results
• advanced: (Optional) Use advanced analysis features

Features:

• Page connections and relationships
• Content clustering
• Task progress analysis
• Concept evolution tracking

suggestConnections

Uses AI to analyze your graph and suggest interesting connections.

Parameters:

• minConfidence: (Optional) Minimum confidence score for suggestions (0-1, default: 0.6)
• maxSuggestions: (Optional) Maximum number of suggestions to return (default: 10)
• focusArea: (Optional) Topic or area to focus suggestions around

Features:

• Discovers potential connections between pages
• Identifies knowledge synthesis opportunities
• Suggests exploration paths based on recent interests
• Provides confidence scores for suggestions

Development

The server is built using:

• Model Context Protocol TypeScript SDK
• Zod for parameter validation
• Logseq HTTP API for data access

To extend with new tools, add additional server.tool() definitions in index.ts.

Troubleshooting

Common Issues

Node.js Version Managers (fnm, nvm, etc.)

If you're using a Node.js version manager like fnm or nvm, Claude Desktop won't be able to access the Node.js binaries properly, as it runs outside of your shell environment where the PATH is modified.

Solution: Install a system-wide Node.js with Homebrew:

brew install node

This ensures Node.js is available to all applications, including Claude Desktop.

Basic Troubleshooting Steps

• Ensure Logseq is running with the HTTP API enabled
• Verify your auth token in .env matches the one set in Logseq
• Check that the path to your index.ts file is correct in the Claude configuration
• Try running npx tsx index.ts directly in your terminal to verify it works

Viewing Logs in Claude Desktop

Monitor logs in real-time:

# macOS
tail -n 20 -F ~/Library/Logs/Claude/mcp*.log

For more detailed debugging information, refer to the official MCP debugging documentation.