reMarkable

Turn your reMarkable tablet into an AI-enabled second brain with remarkable-mcp. This MCP server lets AI tools read, search, and traverse your entire library, including handwritten notes via OCR, so you can research, write, and brainstorm across your notes with ease.

How to use

You run a local MCP server on your device and connect to it from your AI client. Choose SSH mode for fast, offline operation or use the cloud option if you cannot enable developer features on the tablet. Once the server is running, you can ask your AI assistant to read documents, search across your library, extract text from notebooks, and fetch page imagery for context. All core tools are read-only and return structured results to drive your AI workflow.

Typical usage patterns include: browsing your library to locate documents, performing batch searches across many documents, enabling OCR to convert handwriting to searchable text, and requesting page images for diagrams or UI sketches. You can also filter the scope to a specific folder to keep AI access focused on work content.

How to install

Prerequisites you need installed before starting: a computer or development environment capable of running the MCP client, and access to your reMarkable device. You will also need an OCR backend key if you plan to use Google Vision for handwriting recognition.

# SSH mode (recommended)
# 1) Ensure developer mode is enabled on your reMarkable and connect via USB
# 2) Install the MCP client and run the server with SSH access
# 3) Provide a Google Vision API key for handwriting OCR (recommended)

{
  "servers": {
    "remarkable_ssh": {
      "command": "uvx",
      "args": ["remarkable-mcp", "--ssh"],
      "env": {
        "GOOGLE_VISION_API_KEY": "YOUR_GOOGLE_VISION_API_KEY"
      }
    }
  }
}

Configuration and options

Control where the MCP server can access on your device, how OCR behaves, and how images are rendered. You can tailor the setup to balance performance, privacy, and accessibility for your AI workflows.

Root path filtering allows you to limit the server to a specific folder. This confines all operations to that path, improving focus and security.

Example to limit access to the /Work folder and set a custom OCR key:

{
  "servers": {
    "remarkable": {
      "command": "uvx",
      "args": ["remarkable-mcp", "--ssh"],
      "env": {
        "REMARKABLE_ROOT_PATH": "/Work",
        "GOOGLE_VISION_API_KEY": "your-api-key"
      }
    }
  }
}

OCR and image options

Choose the OCR backend that best fits your setup. You can use Google Vision for high accuracy, or opt into a privacy-preserving sampling mode that leverages your client’s AI model. Tesseract is available as an offline fallback for printed text.

Example to enable sampling OCR by default:

{
  "env": {
    "REMARKABLE_OCR_BACKEND": "sampling"
  }
}

SSH vs Cloud comparison

SSH mode gives you the fastest access and works offline, with raw PDFs/EPUBs accessible locally. Cloud mode uses a token from a cloud integration and requires a Connect subscription.

No matter the mode, you can access tools like reading, searching, and extracting text, and you can combine these with your AI workflow.

Advanced configuration

You can further tailor how the MCP server operates by setting advanced environment variables or adjusting how results are returned to your client.

Use cases

Research and writing: transfer knowledge from handwritten notes into structured documents. Daily review: summarize notes, identify action items, and detect patterns. Knowledge management: build a personal knowledge system by linking notes to AI workflows.

Development and troubleshooting notes

If you run into issues, ensure you have the correct OCR key, valid token if using cloud mode, and that you are using an available MCP client that supports the server’s features.

Available tools

remarkable_read

Read and extract text from documents with pagination and search capabilities.

remarkable_browse

Navigate folders or search by document name within your library.

remarkable_search

Search content across multiple documents in a single call.

remarkable_recent

Get recently modified documents for quick access.

remarkable_status

Check the connection status of the MCP server.

remarkable_image

Get PNG or SVG images of pages, with OCR support and sampling options.