Edit 2022
I've published a larger post on the history of Electron and it's security that provides additional context on the changes that affect how security was approached in different framework versions (and what's the best approach to take).
Original answer
I hope this answer gets some attention, because a large majority of answers here leave large security holes in your electron app. In fact this answer is essentially what you should be doing to use require() in your electron apps. (There is just a new electron API that makes it a little bit cleaner in v7).
I wrote a detailed explanation/solution in github using the most current electron apis of how you can require() something, but I'll explain briefly here why you should follow an approach using a preload script, contextBridge and ipc.
The problem
Electron apps are great because we get to use node, but this power is a double-edged sword. If we are not careful, we give someone access to node through our app, and with node a bad actor can corrupt your machine or delete your operating system files (among other things, I imagine).
As brought up by @raddevus in a comment, this is necessary when loading remote content. If your electron app is entirely offline/local, then you are probably okay simply turning on nodeIntegration:true. I still would, however, opt to keep nodeIntegration:false to act as a safeguard for accidental/malicious users using your app, and prevent any possible malware that might ever get installed on your machine from interacting with your electron app and using the nodeIntegration:true attack vector (incredibly rare, but could happen)!
What does the problem look like
This problem manifests when you (any one of the below):
- Have
nodeIntegration:true enabled
- Use the
remote module
All of these problems give uninterrupted access to node from your renderer process. If your renderer process is ever hijacked, you can consider all is lost.
What our solution is
The solution is to not give the renderer direct access to node (ie. require()), but to give our electron main process access to require, and anytime our renderer process needs to use require, marshal a request to the main process.
The way this works in the latest versions (7+) of Electron is on the renderer side we set up ipcRenderer bindings, and on the main side we set up ipcMain bindings. In the ipcMain bindings we set up listener methods that use modules we require(). This is fine and well because our main process can require all it wants.
We use the contextBridge to pass the ipcRenderer bindings to our app code (to use), and so when our app needs to use the required modules in main, it sends a message via IPC (inter-process-communication) and the main process runs some code, and we then send a message back with our result.
Roughly, here's what you want to do.
main.js
const {
app,
BrowserWindow,
ipcMain
} = require("electron");
const path = require("path");
const fs = require("fs");
// Keep a global reference of the window object, if you don't, the window will
// be closed automatically when the JavaScript object is garbage collected.
let win;
async function createWindow() {
// Create the browser window.
win = new BrowserWindow({
width: 800,
height: 600,
webPreferences: {
nodeIntegration: false, // is default value after Electron v5
contextIsolation: true, // protect against prototype pollution
enableRemoteModule: false, // turn off remote
preload: path.join(__dirname, "preload.js") // use a preload script
}
});
// Load app
win.loadFile(path.join(__dirname, "dist/index.html"));
// rest of code..
}
app.on("ready", createWindow);
ipcMain.on("toMain", (event, args) => {
fs.readFile("path/to/file", (error, data) => {
// Do something with file contents
// Send result back to renderer process
win.webContents.send("fromMain", responseObj);
});
});
preload.js
const {
contextBridge,
ipcRenderer
} = require("electron");
// Expose protected methods that allow the renderer process to use
// the ipcRenderer without exposing the entire object
contextBridge.exposeInMainWorld(
"api", {
send: (channel, data) => {
// whitelist channels
let validChannels = ["toMain"];
if (validChannels.includes(channel)) {
ipcRenderer.send(channel, data);
}
},
receive: (channel, func) => {
let validChannels = ["fromMain"];
if (validChannels.includes(channel)) {
// Deliberately strip event as it includes `sender`
ipcRenderer.on(channel, (event, ...args) => func(...args));
}
}
}
);
index.html
<!doctype html>
<html lang="en-US">
<head>
<meta charset="utf-8"/>
<title>Title</title>
</head>
<body>
<script>
window.api.receive("fromMain", (data) => {
console.log(`Received ${data} from main process`);
});
window.api.send("toMain", "some data");
</script>
</body>
</html>
Disclaimer
I'm the author of secure-electron-template, a secure template to build electron apps. I care about this topic, and have been working on this for a few weeks (at this point in time).
