yangg1224/smart_remarkable — explained in plain English
Analysis updated 2026-05-18
Get an AI-drawn answer to a handwritten question right on your reMarkable tablet.
Lasso a region of your notes and have the AI answer only that selection.
Turn a rough sketch into a more detailed, redrawn illustration.
Test the whole trigger-to-draw pipeline offline on a desktop before using it on hardware.
| yangg1224/smart_remarkable | c4pt0r/pie | runwayml/avtensor | |
|---|---|---|---|
| Stars | 60 | 60 | 60 |
| Language | Rust | Rust | Rust |
| Setup difficulty | hard | easy | moderate |
| Complexity | 4/5 | 2/5 | 4/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires a reMarkable tablet, SSH access, and an LLM API key such as ANTHROPIC_API_KEY.
Smart Remarkable connects a large language model to the reMarkable tablet, a device people use to read and take handwritten notes with a stylus. It is a fork of another project called ghostwriter, with new features added on top. The core idea is simple: you write or draw something on the tablet, trigger the tool with a tap, and it takes a screenshot of the page, sends it to an AI model that can see images, and writes or draws the answer back onto the screen in the same style, either as typed text or as pen strokes that look handwritten. One major feature is Select Mode. You tap a trigger corner, then draw a box around a piece of handwriting you want answered, then a second box showing where the answer should appear. The AI reads only that selected region and draws its response as real ink inside your chosen box, which means you can later move or resize that answer using the tablet's own built in selection tool, just like any other note. The project also adds two small buttons directly into the tablet's native selection menu, next to the usual cut and copy options: an LLM button that starts Select Mode instantly on whatever you have selected, and a Draw button. Tapping Draw on a piece of handwritten text sketches a small illustration under it, while tapping Draw on an existing sketch erases it and redraws an improved version in the same spot. There is also an option to route these drawings through a separate image generating model instead of having the chat AI draw simple line art itself, which produces more detailed pictures that are then traced into pen strokes. Other features include automatically detecting whether the tablet is held upside down and adjusting accordingly, a small web page for viewing and editing settings while the tool runs, and an offline testing mode that lets developers simulate the whole write, screenshot, and draw cycle on a desktop computer without needing the physical tablet. The tool is written in Rust and works with Anthropic, OpenAI, or Google models, though thinking mode and web search are Anthropic only.
A Rust tool that connects an AI vision model to the reMarkable tablet, letting you get typed or hand-drawn answers to what you write or sketch.
Mainly Rust. The stack also includes Rust, Anthropic API, OpenAI API.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.