Like Node-Webkit, atom-shell provides a platform to write desktop applications with JavaScript and HTML, and has Node integration to grant access to low level system in web pages.
But there are also fundamental differences between the two projects that make atom-shell a completely separate product from Node-Webkit:
1. Entry of application
In Node-Webkit, the main entry of an application is a web page, you specify a
main page in the package.json and it would be opened in a browser window as
the application's main window.
While in atom-shell, the entry point is a JavaScript script, instead of providing a URL directly, you need to manually create a browser window and load html file in it with corresponding API. You also need to listen to window events to decide when to quit the application.
So atom-shell works more like the Node.js runtime, and APIs are more low level, you can also use atom-shell for web testing purpose like phantomjs.
2. Build system
In order to avoid the complexity of building the whole Chromium, atom-shell uses libchromiumcontent to access Chromium's Content API, libchromiumcontent is a single, shared library that includes the Chromium Content module and all its dependencies. So users don't need a powerful machine to build atom-shell.
3. Node integration
In Node-Webkit, the Node integration in web pages requires patching Chromium to
work, while in atom-shell we chose a different way to integrate libuv loop to
each platform's message loop to avoid hacking Chromium, see the
node_bindings code for how that was done.
4. Multi-context
If you are an experienced Node-Webkit user, you should be familiar with the concept of Node context and web context, these concepts were invented because of how the Node-Webkit was implemented.
By using the multi-context feature of Node, atom-shell doesn't introduce a new JavaScript context in web pages.