remnote

The RemNote MCP Server acts as a bridge between AI agents and RemNote via the MCP protocol, enabling actions like creating notes, searching your knowledge base, reading and updating notes, and appending journal entries. It connects to an accompanying RemNote Automation Bridge plugin that runs inside RemNote (or RemNote in the browser/desktop) and exposes RemNote API calls over a WebSocket. The MCP server exposes an HTTP endpoint (port 3001 by default) for agent sessions and a WebSocket bridge (port 3002) to the RemNote plugin. Multiple AI agents can connect simultaneously, each with its own MCP session while sharing the same RemNote bridge. To expose the HTTP endpoint to the internet for cloud-based AI clients, you can use tunneling tools like ngrok, while the WebSocket bridge remains local for security.

Available MCP tools include remnote_create_note, remnote_search, remnote_search_by_tag, remnote_read_note, remnote_update_note, remnote_append_journal, remnote_get_playbook, and remnote_status. These tools map directly to common RemNote operations such as creating notes with optional hierarchy and tags, performing full-text searches, reading note content with depth options, updating content and tags, and appending timestamped journal entries. The Tools Reference provides detailed usage examples for each tool. To configure clients like Claude Code CLI, Claude Cowork, Accomplish, or other MCP clients, follow the configuration guides in the Documentation section.

To begin, install the server globally, install the RemNote Automation Bridge plugin in RemNote, start the MCP server, and then configure your AI client to connect via the server’s HTTP endpoint. Examples from the Quick Start show how to install, run, and connect: after installation, run remnote-mcp-server to start the HTTP and WebSocket services, then point your AI client at the server’s MCP endpoint and configure credentials and session settings as needed.

Prerequisites:

• Node.js and npm installed on your machine
• Access to install global npm packages
• RemNote and the RemNote Automation Bridge plugin installed and configured to connect to ws://127.0.0.1:3002

Steps:

1. Install the MCP server globally (or use npx to run it directly):

# Option A: install globally and run
npm install -g remnote-mcp-server
remnote-mcp-server

# Option B: run via npx without global install
npx -y remnote-mcp-server

2. Install and configure the RemNote Automation Bridge plugin in RemNote, ensuring it connects to ws://127.0.0.1:3002

3. Start the server and verify it’s listening (as shown in the README):

RemNote MCP Server v0.2.1 listening { wsPort: 3002, httpPort: 3001 }

4. Configure your AI client (e.g., Claude Code CLI, ChatGPT Apps, Accomplish) to connect to the MCP HTTP endpoint (http://localhost:3001 by default) and manage sessions per agent

Prerequisites recap:

• Node.js and npm in your environment
• RemNote and RemNote Automation Bridge plugin set up and running
• Network access to localhost ports 3001 (HTTP) and 3002 (WebSocket)

Notes and tips:

• Version compatibility is important: use a remnote-mcp-server version compatible with your RemNote Automation Bridge plugin version (0.x semver).
• For remote access (Claude Cowork or other cloud agents), expose the HTTP endpoint securely (e.g., ngrok) while keeping the WebSocket bridge local.
• The server supports multiple agents concurrently; each agent gets its own MCP session.
• If you encounter connectivity issues, verify that the RemNote Bridge plugin is connected to the WebSocket port (3002) and that the MCP HTTP server (3001) is reachable from the client.
• Check the Tools Reference for exact syntax and examples for each MCP tool (e.g., remnote_create_note, remnote_search).
• When upgrading RemNote or the bridge, re-check compatibility guides to adjust versions accordingly.