There is a new repo: https://github.com/njpipeorgan/wolfram-language-notebook
The author's introduction:
Using Wolfram Notebooks in VS Code
The release of Wolfram Engine in 2019 marks the separation of the Wolfram language's runtime library and its notebook interface. The subtlety of this separation is that the Wolfram Engine is free, yet essentially has the functionality of both the kernel and the front end (just not the full notebook interface). That is, we can simply create an interface that can be adapted to the kernel to get a similar experience to using Mathematica .
Recently, I wrote a VS Code extension, Wolfram Language Notebook (Github), which makes it easy to write and run code in VS Code using the notebook interface. You can install this extension in VS Code's Marketplace at
Wolfram Language Notebook - Visual Studio Marketplace
To use this extension, you just need to install Wolfram Engine or Wolfram Mathematica (v12.0+). If you can run wolframscript from the command line or shell, you are ready to go. If you want to use the remote kernel, then you need to install Engine on the remote machine, and just have ssh on the local machine.
How to use it
First create a new notebook: Search for and run Create New Wolfram Language Notebook in the Command Palette, or just create a new file with the .wlnb suffix.
Next, run the Manage Kernels command and select Use wolframscript. this will add wolframscript to the kernel configuration and then connect to this kernel; the status of the connection will be shown in the status bar.
You can then add code units, write some Wolfram Language code, and run it.
Syntax highlighting: The notebook will highlight some simple Wolfram Language syntax, common built-in functions, and the full names of special characters (such as [Alpha]).
Auto-completion and function descriptions: The notebook prompts for auto-completion of built-in functions. Inputting or hovering over built-in functions will show their usage descriptions.
Output rendering: The notebook renders common expressions to HTML and images to bitmaps.
Export notebooks: You can export .wlnb files to Wolfram notebooks, which include Markdown, input, and output.
Remote kernel: Notebooks can be configured with a remote kernel, and then booted and connected to the kernel via ssh. All calculations will be done remotely, but the code and output will be saved locally.
Principle and Development
Structure of Wolfram Language Notebook
The Wolfram Language Notebook interface is based on the VS Code Notebook API, which is not yet fully finalized and includes an interface for passing messages between the notebook interface, renderer, and controller. As a result, the extension can only send results from the kernel in one direction and cannot receive feedback from the notebook, so some important features (such as Manipulate) are not yet available. This issue will be addressed in a later release.
The rendering of Graphics and Graphics3D is another area that needs improvement. Because of the many options, rendering expressions directly is not a realistic approach, so the current approach is to export them as bitmaps and display them . In a future version, there may be a better way to display these images. (If you have a good method, feel free to contact me.)
You can download the Wolfram Engine here, register your Wolfram ID, and get a license. The use of the Wolfram Language is described here, and the examples in it can basically be run in the Wolfram Language Notebook, except for the interactive interface and natural language input.
Although you can use the Wolfram Language Notebook itself in almost any way, computations performed through the Wolfram Engine are still limited by the Wolfram Engine license.