aiskillstore/connecting-to-logseq

Connecting to Logseq

When to Use This Skill

This skill auto-invokes when:

• User wants to connect Claude to their Logseq graph
• Setting up Logseq integration or API tokens
• Troubleshooting connection issues
• Configuring graph paths or backends
• User mentions "connect to logseq", "logseq api", "logseq token"
• Questions about HTTP API, CLI, or MCP server setup

Setup Scripts: See {baseDir}/scripts/ for initialization utilities.

Available Backends

| Backend | Requires Logseq Running | Read | Write | Best For | |---------|------------------------|------|-------|----------| | HTTP API | Yes | Full | Full | Real-time, interactive | | CLI | No | Full | Limited | Offline, batch, CI/CD | | MCP Server | Yes (via HTTP) | Full | Full | Claude-native tools |

Quick Start

1. Enable Logseq HTTP API

In Logseq:

1. Settings → Advanced → Developer mode: ON
2. Settings → Advanced → HTTP APIs server: ON
3. Settings → Advanced → Authorization tokens → Create token

2. Set Environment Variable

export LOGSEQ_API_TOKEN="your-token-here"

3. Initialize Plugin

Run the setup wizard:

python {baseDir}/scripts/init-environment.py

Or use the command: /logseq:init

Backend Details

HTTP API (Primary)

URL: http://127.0.0.1:12315/api

Request Format:

POST /api
Content-Type: application/json
Authorization: Bearer YOUR_TOKEN

{
  "method": "logseq.Editor.getPage",
  "args": ["PageName"]
}

Common Methods:

• logseq.App.getCurrentGraph - Get current graph info
• logseq.Editor.getPage - Get page by name
• logseq.Editor.getBlock - Get block by UUID
• logseq.DB.datascriptQuery - Execute Datalog query
• logseq.Editor.insertBlock - Create new block

CLI (@logseq/cli)

Installation:

npm install -g @logseq/cli

Usage:

# Query local graph
logseq query "[:find ?title :where [?p :block/title ?title]]" --graph ~/logseq/my-graph

# With running Logseq (in-app mode)
logseq query "..." --in-app -a YOUR_TOKEN

MCP Server

The plugin includes a custom MCP server that exposes Logseq operations as Claude tools.

Location: servers/logseq-mcp/

Build:

cd servers/logseq-mcp
npm install
npm run build

Configuration File

Location: .claude/logseq-expert/env.json

{
  "backend": "auto",
  "http": {
    "url": "http://127.0.0.1:12315",
    "token": "${LOGSEQ_API_TOKEN}"
  },
  "cli": {
    "graphPath": "/path/to/graph",
    "inApp": false
  },
  "mcp": {
    "enabled": true
  },
  "preferences": {
    "defaultGraph": null,
    "confirmWrites": false,
    "backupBeforeWrite": false
  }
}

Troubleshooting

"Cannot connect to Logseq"

1. Check Logseq is running with HTTP API enabled
2. Verify port: Default is 12315, check Settings → Advanced
3. Check firewall: Ensure localhost:12315 is accessible
4. Test manually:

   curl -X POST http://127.0.0.1:12315/api \
     -H "Content-Type: application/json" \
     -H "Authorization: Bearer YOUR_TOKEN" \
     -d '{"method":"logseq.App.getCurrentGraph"}'
   

"Authentication failed"

1. Verify token: Check it matches what's in Logseq settings
2. Token format: Ensure no extra whitespace
3. Environment variable: Check echo $LOGSEQ_API_TOKEN

"CLI not found"

1. Install globally: npm install -g @logseq/cli
2. Or use npx: npx @logseq/cli --help
3. Check PATH: Ensure npm global bin is in PATH

"MCP server not working"

1. Build server: cd servers/logseq-mcp && npm run build
2. Check Node.js: Requires Node 18+
3. Verify HTTP API: MCP server uses HTTP API internally

Scripts Reference

| Script | Purpose | |--------|---------| | init-environment.py | Interactive setup wizard | | detect-backend.py | Auto-detect available backends | | test-connection.py | Test connectivity | | preflight-checks.sh | Validate environment |

Run scripts from plugin root:

python logseq-expert/scripts/init-environment.py

Security Notes

• Never commit tokens to version control
• Use environment variables for sensitive data
• Token in config supports ${VAR} syntax for env vars
• HTTP API only listens on localhost by default