reMarkable 2 + AI: Hacks, Tools, and Workflows

Last verified: February 2026

The reMarkable 2 is a full-root-access Linux e-ink tablet. Its zero-distraction philosophy makes it a thinking tool, but its native integrations are minimal. This page covers everything that exists to augment it with AI, from the simplest to the most technical.

See also: AI Ecosystem for the broader mapping of the AI ecosystem (beyond hardware).

Table of Contents

1. remarkable-mcp: direct MCP access via SSH
2. Ghostwriter: Vision-LLM Interface
3. Sync reMarkable → Obsidian
4. OCR + Custom AI Pipeline
5. SSH Access and Community Tools
6. Underused Native Features
7. Official API and Developer Portal
8. Zapier Automation
9. Read-it-later: Web to reMarkable
10. Meeting Notes → AI Summary
11. Zotero to reMarkable (research)
12. Screen sharing as an AI-assisted whiteboard
13. Custom apps and fun hacks
14. AI-augmented workflows to build
15. Where to start

---

1. remarkable-mcp: direct MCP access via SSH

ROI: maximal | Effort: medium | Connection: SSH over USB (no cloud)

Sam Morrow created an MCP server that connects the reMarkable directly to Claude Code, VS Code Copilot, and any MCP-compatible AI assistant.

Attribute	Details
Repo	https://github.com/SamMorrowDrums/remarkable-mcp
Blog	https://sam-morrow.com/blog/building-an-mcp-server-for-remarkable
Connection	SSH over USB (no cloud, no subscription)
Language	Python (FastMCP)

What it does

• Native extraction of typed text (Type Folio / virtual keyboard), instant, no OCR
• Handwriting OCR via Google Cloud Vision (1000 free requests/month)
• Smart search across your entire library
• Text extraction from PDF and EPUB, plus annotations
• Full traversal of documents

Why it’s #1

You can ask Claude “what did I note about X during the January 15 meeting?” It will search through your handwritten notes. The reMarkable becomes a queryable second brain.

Technical stack

FastMCP + rmscene (native .rm parsing) + PyMuPDF (PDF)
+ Google Cloud Vision (OCR) + Paramiko (SSH)

Quick installation

# 1. Enable SSH on the reMarkable
# Settings → Help → Copyrights and licenses → IP + root password

# 2. Clone the repo
git clone https://github.com/SamMorrowDrums/remarkable-mcp
cd remarkable-mcp && pip install -e .

# 3. Add to Claude Code
# In ~/.claude.json or via "claude mcp add"

Configuration in Claude Code

{
  "mcpServers": {
    "remarkable": {
      "command": "python",
      "args": ["-m", "remarkable_mcp"],
      "env": {
        "REMARKABLE_HOST": "10.11.99.1",
        "REMARKABLE_PASSWORD": "<root-password>"
      }
    }
  }
}

---

2. Ghostwriter: Vision-LLM Interface

ROI: experimental | Effort: low (a Rust binary to copy)

Attribute	Details
Repo	https://github.com/awwaiid/ghostwriter
Model	GPT-4o Vision
HN discussion	https://news.ycombinator.com/item?id=42979986

Concept

You write a prompt by hand on the reMarkable. A vision-LLM (GPT-4o) reads your handwriting and drawings and responds directly on the tablet.

Installation

# 1. Download the compiled Rust binary
scp ghostwriter root@10.11.99.1:/home/root/

# 2. SSH in and launch
ssh root@10.11.99.1
chmod +x ghostwriter && ./ghostwriter

Supported interactions

• Handwriting recognition
• Sketch analysis (wireframes, diagrams)
• Small iconographic language
• Gestures

Concrete use case

You draw an architecture diagram, you write “optimize this.” The LLM analyzes it visually and responds. A fascinating prototype for pen-based human-AI interaction.

Honest limitation: The reMarkable’s native drawing app is minimal (no free text placement in the response).

---

3. Sync reMarkable → Obsidian

ROI: high if you use Obsidian | Effort: low to medium

Option A: Scrybble (the most complete)

Attribute	Details
Site	scrybble.ink
Obsidian plugin	Community plugin (vault settings)
Hosting	Self-hosted or Scrybble server
Discussion	https://forum.obsidian.md/t/scrybble-sync-plugin/103194

What it does:

• Sync notebooks, PDFs, ePubs to Obsidian vault
• Extract PDF/ePub highlights as Markdown
• Extract typed text as Markdown
• Full rendering of notebooks as PDF in the vault
• Organization by page with tags

Use case: Academic research, meeting note-taking with Obsidian search behind it.

Option B: Custom Cloud Sync plugin

• Demo: https://www.youtube.com/watch?v=EsRdi8J9Cnc
• “remarkable insert” command, pulls files from the reMarkable cloud
• PDF in the rm/ folder, embedded in Obsidian notes
• Automatic re-fetch when you make changes on the tablet

---

4. OCR + Custom AI Pipeline

ROI: high for a custom workflow | Effort: medium-high

rmirror + Claude API (recommended pattern)

Source: https://news.ycombinator.com/item?id=47110872 (February 2026)

Concept: A macOS background agent that:

1. Syncs notebooks from the reMarkable
2. OCRs via Claude API (better than Tesseract for handwriting with context)
3. Pushes the transcribed notes to Notion as searchable pages

Why Claude beats Tesseract for OCR: Claude understands context, corrects malformed words, and structures lists and tables automatically.

DIY pipeline

reMarkable → SSH/USB
  → extract .rm files
  → rmscene parse (native text if Type Folio)
  → Claude Vision API (page screenshots for handwriting)
  → structured text + auto tags
  → Notion/Obsidian/GitHub via API

Parsing tools

Tool	Usage	Link
rmscene	Native parsing of .rm files	https://github.com/ricklupton/rmscene
rmc	Converts .rm to SVG/PNG	https://github.com/ricklupton/rmc
rmapi	Cloud API interface in Go	https://github.com/juruen/rmapi

---

5. SSH Access and Community Tools

Essential foundation for everything else

Enable SSH

# Via the tablet interface:
# Settings → Help → Copyrights and licenses
# → Displays: IP + root password

# USB (direct connection)
ssh root@10.11.99.1

# WiFi (after activation)
rm-ssh-over-wlan on
# or via "Simply Customize It"

Essential tools

Tool	Usage	Link
RMHacks/xovi	Mod framework for rM1/2/Paper Pro	https://www.nilorea.net/2025/08/11/latest-rmhacks-with-xovi-for-remarkable-1-2-paper-pro/
Simply Customize It	GUI to toggle features (WLAN SSH, etc.)	Third-party app
ReMy	GUI to browse/preview/export docs via SSH (no cloud)	https://github.com/bordaigorl/remy
rmirro	Bidirectional PDF sync, tablet ↔ local folder	https://github.com/hersle/rmirro
reStream	Stream the reMarkable’s screen to Mac/PC	https://github.com/rien/reStream
KOReader	Alternative reader (more formats, customizable)	Via SSH
reGitable	Automatic backup via git	awesome-reMarkable

Custom templates

# Create an SVG template, copy via SSH
scp mon-template.svg root@10.11.99.1:/usr/share/remarkable/templates/

# Edit templates.json to register it
ssh root@10.11.99.1 'vi /usr/share/remarkable/templates/templates.json'

Generation tools: ReCalendar.me, Remarkable Grid Generator, Remarkably Planner Builder

---

6. Underused Native Features

Effort: zero | Included in Connect (~6 EUR/month)

Feature	Usage
Handwriting conversion	Select, convert, copy/paste into any app
Cloud sync	Google Drive, Dropbox, OneDrive
Send to Slack	Share meeting notes directly into a channel
Handwriting search (beta AI)	Search through your past handwritten notes
Screen sharing	Share the screen on a PC (presentations, meetings)
Send to email	Send as PDF or PNG

Tip: Handwriting-to-text conversion works well for isolated words but less well for dense cursive sentences. Prefer print handwriting for conversion.

---

7. Official API and Developer Portal

Resource	Link
Developer Portal	https://developer.remarkable.com
Cloud API docs	https://github.com/splitbrain/ReMarkableAPI
Community guide	https://remarkable.guide/
rmfakecloud	Self-hosted Cloud (no Connect subscription)

OS: Linux (Codex), full SSH root access, GPL-compliant. The cross-compiler toolchain allows deploying custom native apps.

rmfakecloud: Open-source alternative to the reMarkable cloud, for self-hosting sync and skipping the Connect subscription.

---

8. Zapier Automation

ROI: medium | Effort: low | No code required

Mechanism: reMarkable, then email (my@remarkable.com), then Zapier intercepts, then automatic action

Possible destinations

Google Drive, Asana, ClickUp, Trello, Slack, WordPress, Evernote, Notion

Free plan

• 100 tasks/month
• 2-step Zaps
• Check every 15 min

Concrete workflows

Meeting notes → PDF auto-uploaded to Google Drive
Sketches → File sent to a Slack channel
Action items → Tasks created in Asana/ClickUp

Source: https://myremarkable.substack.com/p/integrating-remarkable

---

9. Read-it-later: Web to reMarkable

ROI: high for reading | Effort: near zero

Tool	Description
Chrome extension “Read on reMarkable”	Save any web page as EPUB/PDF on the tablet (ads removed)
Goosepaper	Daily RSS feed + news + Wikipedia, formatted for e-ink
remarkable_news	Daily news/comics/images as a screensaver
Instapaper workaround	Download articles as EPUB, import via desktop app

PDF option: Right-click the extension, “Read on reMarkable as PDF” (adjustable margins for annotating).

---

10. Meeting Notes → AI Summary

ROI: high | Effort: very low

Manual workflow (without MCP)

1. Handwritten notes during the meeting
2. Screenshot via the reMarkable mobile app (or cloud sync)
3. Upload the image to Claude/ChatGPT
4. Prompt: "Summarize these notes, extract action items with deadlines and owners"

MCP workflow (with remarkable-mcp installed)

Claude, summarize my notes from today's meeting
→ Claude fetches the files via SSH
→ OCR if needed
→ Structured summary, directly

MCP advantage: Skips the screenshot and upload steps. Works even without a Connect subscription.

Recommended template

Paper Pro Move Meeting Notebook: 60 meetings, 5 interlinked pages per meeting (overview + notes + action items + follow-up).

---

11. Zotero to reMarkable (research)

ROI: high if you read papers | Effort: medium

Tool	Usage
Zotero2reMarkable Bridge	Sync PDFs from Zotero with highlight support
KOReader + Toltec + Zotero plugin	Better 2-column PDF reading, bidirectional sync
sync_zotero_remarkable	Lighter alternative

Honest limitation: reMarkable is a closed system. Zotero integration requires workarounds. Not as smooth as an Android e-reader with native Zotero. Functional but with friction.

---

12. Screen sharing as an AI-assisted whiteboard

ROI: presentation/facilitation | Effort: zero (native Connect feature)

• Screen Share: Your live handwriting appears on the external screen/virtual meeting
• Laser pointer: Pen near the top of the screen activates a laser pointer
• Combo workflow: Screen Share + a colleague feeding your notes into ChatGPT in real time = augmented whiteboard

Price: Included in Connect (~$30/year US, ~6 EUR/month EU).

---

13. Custom apps and fun hacks

App/Hack	Description
Ephemeris	Daily agenda generated from your calendars (Python)
Remarcal	Sync Google/Outlook/Apple calendars to reMarkable
reMarkable keywriter	Distraction-free keyboard notes app
remarkable-wikipedia	Offline Wikipedia reader
whiteboard-hypercard	Live collaboration, shared drawing
NetSurf	Minimal web browser (via SSH)
pdf2remarkable	Upload PDFs to the cloud from the command line
send-to-remarkable	Upload docs by email (send-to-Kindle style)
libreMarkable	Framework for developing native apps
oxide/remux/draft	Launchers for multitasking
latex-yearly-planner	Annual planner generated in LaTeX

Full catalog: https://github.com/reHackable/awesome-reMarkable

---

14. AI-augmented workflows to build

Workflows not yet packaged but achievable with the building blocks available.

A. AI-analyzed logbook

Every evening → write 1 page of reflection on the reMarkable
→ remarkable-mcp + Claude → weekly analysis of patterns, emotions, decisions
→ Output: insights in Obsidian with a connection graph

B. Assisted inbox processing

Papers/articles read and annotated on reMarkable
→ OCR via Claude Vision → structured summaries
→ Automatic tags + filing in Obsidian/Notion

C. Sketch-to-code

Draw a UI wireframe on reMarkable
→ Screenshot → Claude Vision → HTML/React code

D. Automatic flashcards

Course/reading notes on reMarkable
→ remarkable-mcp → Claude extracts the key concepts
→ Automatically generates Anki flashcards

E. Automated daily standup

TODOs written every morning (custom template)
→ OCR → Slack/email automatically formatted
→ End of day: check off items, diff sent

F. Brainstorm capture → Mind map

Freely scribbled ideas
→ Claude Vision analyzes the spatial layout + text
→ Generates a structured mind map (Mermaid/Markmap)

---

15. Where to start

Phase 1: This weekend (2h)

1. Enable SSH via Settings → Help → Copyrights and licenses
2. Install remarkable-mcp and connect it to Claude Code
3. Test: ask Claude to search through your notes

Phase 2: The following week

4. If you use Obsidian, install Scrybble
5. Try Ghostwriter (10 min to install, fun guaranteed)

Phase 3: When you want to go further

6. Build a custom OCR pipeline with Claude Vision API
7. Explore rmfakecloud to skip the Connect subscription

---

Sources

GitHub projects

• https://github.com/SamMorrowDrums/remarkable-mcp (MCP server, Nov 2025)
• https://github.com/awwaiid/ghostwriter (Vision-LLM interface)
• https://github.com/reHackable/awesome-reMarkable (community catalog)
• https://github.com/hersle/rmirro (sync without cloud)
• https://github.com/bordaigorl/remy (GUI SSH)
• https://github.com/rien/reStream (screen streaming)
• https://github.com/splitbrain/ReMarkableAPI (Cloud API docs)
• https://github.com/ricklupton/rmscene (native .rm parsing)

Articles and discussions

• https://sam-morrow.com/blog/building-an-mcp-server-for-remarkable
• https://news.ycombinator.com/item?id=47110872 (rmirror + Claude OCR, Feb 2026)
• https://news.ycombinator.com/item?id=42979986 (Ghostwriter HN)
• https://news.ycombinator.com/item?id=46099997 (Hacking reMarkable 2, HN 2025)
• https://sgt.hootr.club/blog/hacking-on-the-remarkable-2/ (SSH hacking guide)
• https://myremarkable.substack.com/p/integrating-remarkable (Zapier integration)

Obsidian

• https://forum.obsidian.md/t/scrybble-sync-plugin/103194
• https://www.youtube.com/watch?v=EsRdi8J9Cnc (Cloud sync demo)

Official

• https://developer.remarkable.com (Developer Portal, SDK, API)
• https://remarkable.guide/ (Community guide)