Visual Studio Code Extension (Beta)

Our VS Code Extension can utilize AI to perform more accurate OCR on maths equations with context perception for users using vscode writing LaTeX.
This is a usage guide, for developers please visit our repository.

WARNING: THE EXTENSION IS UNDER HEAVY DEVELOPMENT AND OFFERS ZERO WARRANTY. USE AT YOUR OWN RISK. SEE LICENSE.

Features

The VSCode extension of WriteTex offers a more accurate OCR result by feeding the AI model with the context of the current document. The user can simply write on there iPad/tablet, and the results will be placed at your cursor in VSCode. The context is built from lines within a radius of the cursor.

Installation

WARNING: ONLY INSTALL THE EXTENSION IF YOUR DEVICE IS BEHIND A FIREWALL. ONLY ENABLE THE EXTENSION IN A TRUSTED NETWORK. DO NOT INSTALL ON SERVER.

• For Microsoft Visual Studio Code users: Search in the extensions page for "WriteTex" or install from the VS Code Marketplace.
• For VSCode forks (Cursor, Trae, Antigravity, Kiro, Qoder, Void, VSCodium, Windsurf, Codex etc): Try searching for "WriteTex" in the extensions page of your fork. If it is not found, download our .vsix file from the releases page and install it manually via extensions -> 3 dots on top right -> Install from VSIX.

Configure Settings

Once installed, our icon will appear in your sidebar. Click on it to configure the settings.

AI API Configuration

This part is similar to other platform's setup. A simple example is shown below:

Setting	Value
API Endpoint	https://api.openai.com/v1
Model	gpt-5
API Key	sk-...

Custom Instructions
Some additional prompt to guide the AI model. For example, you can add "Use \mathbb{i} for the imaginary unit."

Context

• Context Line Radius Here you can control the radius of the context. A number of 10 means the context is built from 10 lines before and 10 lines after the cursor.
• Context Char limit The maximum number of characters in the context. A number of 1000 means the context is limited to 1000 characters.

Service Management

The server status is indicated at the top. You can start/stop the server by clicking the corresponding buttons. Start the server for the mobile app of WriteTex to connect to this VSCode instance. ONLY ONE SERVER CAN BE RUNNING AT A TIME. If you start the server in one instance of VSCode, you must stop it before starting it in another instance.

Connect to extension

Make sure your extension is enabled and the server is running. Make sure only one project has the extension server running.

From iPhone/iPad

Obtain the IP address of the computer running the VSCode extension. Go to Task Manager and click on your network, look for IPv4 Address In settings, set the AI API to:

Setting	Value
API Endpoint	http://your-computer-ip:50905/v1
Model	your model
API Key	writetex

For iOS v2.1 (pending release): Make sure your extension is enabled and the server is running. On your iOS device, go to AI API Settings -> VSCode Extension. Toggle on Use VSCode, look for your device's name in the discover list and select it. If your device is not listed, try manual connection by entering the your computer's IP address in Other....

From Android

Obtain the IP address of the computer running the VSCode extension. Go to Task Manager and click on your network, look for IPv4 Address In settings, set the AI API to:

Setting	Value
API Endpoint	http://your-computer-ip:50905
Model	your model
API Key	writetex

From Windows/MacOS

In settings, set the AI API to:

Setting	Value
API Endpoint	http://127.0.0.1:50905
Model	your model
API Key	writetex