Skip to content

Latest commit

 

History

History
850 lines (640 loc) · 29.2 KB

security.md

File metadata and controls

850 lines (640 loc) · 29.2 KB
Raw

Security, Native Capabilities, and Your Responsibility

As web developers, we usually enjoy the strong security net of the browser - the risks associated with the code we write are relatively small. Our websites are granted limited powers in a sandbox, and we trust that our users enjoy a browser built by a large team of engineers that is able to quickly respond to newly discovered security threats.

When working with Electron, it is important to understand that Electron is not a web browser. It allows you to build feature-rich desktop applications with familiar web technologies, but your code wields much greater power. JavaScript can access the filesystem, user shell, and more. This allows you to build high quality native applications, but the inherent security risks scale with the additional powers granted to your code.

With that in mind, be aware that displaying arbitrary content from untrusted sources poses a severe security risk that Electron is not intended to handle. In fact, the most popular Electron apps (Atom, Slack, Visual Studio Code, etc) display primarily local content (or trusted, secure remote content without Node integration) – if your application executes code from an online source, it is your responsibility to ensure that the code is not malicious.

Reporting Security Issues

For information on how to properly disclose an Electron vulnerability, see SECURITY.md

Chromium Security Issues and Upgrades

While Electron strives to support new versions of Chromium as soon as possible, developers should be aware that upgrading is a serious undertaking - involving hand-editing dozens or even hundreds of files. Given the resources and contributions available today, Electron will often not be on the very latest version of Chromium, lagging behind by several weeks or a few months.

We feel that our current system of updating the Chromium component strikes an appropriate balance between the resources we have available and the needs of the majority of applications built on top of the framework. We definitely are interested in hearing more about specific use cases from the people that build things on top of Electron. Pull requests and contributions supporting this effort are always very welcome.

Security Is Everyone's Responsibility

It is important to remember that the security of your Electron application is the result of the overall security of the framework foundation (Chromium, Node.js), Electron itself, all NPM dependencies and your code. As such, it is your responsibility to follow a few important best practices:

  • Keep your application up-to-date with the latest Electron framework release. When releasing your product, you’re also shipping a bundle composed of Electron, Chromium shared library and Node.js. Vulnerabilities affecting these components may impact the security of your application. By updating Electron to the latest version, you ensure that critical vulnerabilities (such as nodeIntegration bypasses) are already patched and cannot be exploited in your application.

  • Evaluate your dependencies. While NPM provides half a million reusable packages, it is your responsibility to choose trusted 3rd-party libraries. If you use outdated libraries affected by known vulnerabilities or rely on poorly maintained code, your application security could be in jeopardy.

  • Adopt secure coding practices. The first line of defense for your application is your own code. Common web vulnerabilities, such as Cross-Site Scripting (XSS), have a higher security impact on Electron applications hence it is highly recommended to adopt secure software development best practices and perform security testing.

Isolation For Untrusted Content

A security issue exists whenever you receive code from an untrusted source (e.g. a remote server) and execute it locally. As an example, consider a remote website being displayed inside a default BrowserWindow. If an attacker somehow manages to change said content (either by attacking the source directly, or by sitting between your app and the actual destination), they will be able to execute native code on the user's machine.

⚠️ Under no circumstances should you load and execute remote code with Node.js integration enabled. Instead, use only local files (packaged together with your application) to execute Node.js code. To display remote content, use the <webview> tag or BrowserView, make sure to disable the nodeIntegration and enable contextIsolation.

Electron Security Warnings

From Electron 2.0 on, developers will see warnings and recommendations printed to the developer console. They only show up when the binary's name is Electron, indicating that a developer is currently looking at the console.

You can force-enable or force-disable these warnings by setting ELECTRON_ENABLE_SECURITY_WARNINGS or ELECTRON_DISABLE_SECURITY_WARNINGS on either process.env or the window object.

Checklist: Security Recommendations

You should at least follow these steps to improve the security of your application:

  1. Only load secure content
  2. Disable the Node.js integration in all renderers that display remote content
  3. Enable context isolation in all renderers that display remote content
  4. Use ses.setPermissionRequestHandler() in all sessions that load remote content
  5. Do not disable webSecurity
  6. Define a Content-Security-Policy and use restrictive rules (i.e. script-src 'self')
  7. Do not set allowRunningInsecureContent to true
  8. Do not enable experimental features
  9. Do not use enableBlinkFeatures
  10. <webview>: Do not use allowpopups
  11. <webview>: Verify options and params
  12. Disable or limit navigation
  13. Disable or limit creation of new windows
  14. Do not use openExternal with untrusted content
  15. Disable the remote module
  16. Filter the remote module

To automate the detection of misconfigurations and insecure patterns, it is possible to use electronegativity. For additional details on potential weaknesses and implementation bugs when developing applications using Electron, please refer to this guide for developers and auditors

1) Only Load Secure Content

Any resources not included with your application should be loaded using a secure protocol like HTTPS. In other words, do not use insecure protocols like HTTP. Similarly, we recommend the use of WSS over WS, FTPS over FTP, and so on.

Why?

HTTPS has three main benefits:

  1. It authenticates the remote server, ensuring your app connects to the correct host instead of an impersonator.
  2. It ensures data integrity, asserting that the data was not modified while in transit between your application and the host.
  3. It encrypts the traffic between your user and the destination host, making it more difficult to eavesdrop on the information sent between your app and the host.

How?

// Bad
browserWindow.loadURL('http://example.com')

// Good
browserWindow.loadURL('https://example.com')
<!-- Bad -->
<script crossorigin src="http://example.com/react.js"></script>
<link rel="stylesheet" href="http://example.com/style.css">

<!-- Good -->
<script crossorigin src="https://example.com/react.js"></script>
<link rel="stylesheet" href="https://example.com/style.css">

2) Do not enable Node.js Integration for Remote Content

This recommendation is the default behavior in Electron since 5.0.0.

It is paramount that you do not enable Node.js integration in any renderer (BrowserWindow, BrowserView, or <webview>) that loads remote content. The goal is to limit the powers you grant to remote content, thus making it dramatically more difficult for an attacker to harm your users should they gain the ability to execute JavaScript on your website.

After this, you can grant additional permissions for specific hosts. For example, if you are opening a BrowserWindow pointed at `https://example.com/", you can give that website exactly the abilities it needs, but no more.

Why?

A cross-site-scripting (XSS) attack is more dangerous if an attacker can jump out of the renderer process and execute code on the user's computer. Cross-site-scripting attacks are fairly common - and while an issue, their power is usually limited to messing with the website that they are executed on. Disabling Node.js integration helps prevent an XSS from being escalated into a so-called "Remote Code Execution" (RCE) attack.

How?

// Bad
const mainWindow = new BrowserWindow({
  webPreferences: {
    nodeIntegration: true,
    nodeIntegrationInWorker: true
  }
})

mainWindow.loadURL('https://example.com')
// Good
const mainWindow = new BrowserWindow({
  webPreferences: {
    preload: path.join(app.getAppPath(), 'preload.js')
  }
})

mainWindow.loadURL('https://example.com')
<!-- Bad -->
<webview nodeIntegration src="page.html"></webview>

<!-- Good -->
<webview src="page.html"></webview>

When disabling Node.js integration, you can still expose APIs to your website that do consume Node.js modules or features. Preload scripts continue to have access to require and other Node.js features, allowing developers to expose a custom API to remotely loaded content.

In the following example preload script, the later loaded website will have access to a window.readConfig() method, but no Node.js features.

const { readFileSync } = require('fs')

window.readConfig = function () {
  const data = readFileSync('./config.json')
  return data
}

3) Enable Context Isolation for Remote Content

Context isolation is an Electron feature that allows developers to run code in preload scripts and in Electron APIs in a dedicated JavaScript context. In practice, that means that global objects like Array.prototype.push or JSON.parse cannot be modified by scripts running in the renderer process.

Electron uses the same technology as Chromium's Content Scripts to enable this behavior.

Even when you use nodeIntegration: false to enforce strong isolation and prevent the use of Node primitives, contextIsolation must also be used.

Why?

Context isolation allows each the scripts on running in the renderer to make changes to its JavaScript environment without worrying about conflicting with the scripts in the Electron API or the preload script.

While still an experimental Electron feature, context isolation adds an additional layer of security. It creates a new JavaScript world for Electron APIs and preload scripts, which mitigates so-called "Prototype Pollution" attacks.

At the same time, preload scripts still have access to the document and window objects. In other words, you're getting a decent return on a likely very small investment.

How?

// Main process
const mainWindow = new BrowserWindow({
  webPreferences: {
    contextIsolation: true,
    preload: path.join(app.getAppPath(), 'preload.js')
  }
})
// Preload script

// Set a variable in the page before it loads
webFrame.executeJavaScript('window.foo = "foo";')

// The loaded page will not be able to access this, it is only available
// in this context
window.bar = 'bar'

document.addEventListener('DOMContentLoaded', () => {
  // Will log out 'undefined' since window.foo is only available in the main
  // context
  console.log(window.foo)

  // Will log out 'bar' since window.bar is available in this context
  console.log(window.bar)
})

4) Handle Session Permission Requests From Remote Content

You may have seen permission requests while using Chrome: They pop up whenever the website attempts to use a feature that the user has to manually approve ( like notifications).

The API is based on the Chromium permissions API and implements the same types of permissions.

Why?

By default, Electron will automatically approve all permission requests unless the developer has manually configured a custom handler. While a solid default, security-conscious developers might want to assume the very opposite.

How?

const { session } = require('electron')

session
  .fromPartition('some-partition')
  .setPermissionRequestHandler((webContents, permission, callback) => {
    const url = webContents.getURL()

    if (permission === 'notifications') {
      // Approves the permissions request
      callback(true)
    }

    // Verify URL
    if (!url.startsWith('https://example.com/')) {
      // Denies the permissions request
      return callback(false)
    }
  })

5) Do Not Disable WebSecurity

Recommendation is Electron's default

You may have already guessed that disabling the webSecurity property on a renderer process (BrowserWindow, BrowserView, or <webview>) disables crucial security features.

Do not disable webSecurity in production applications.

Why?

Disabling webSecurity will disable the same-origin policy and set allowRunningInsecureContent property to true. In other words, it allows the execution of insecure code from different domains.

How?

// Bad
const mainWindow = new BrowserWindow({
  webPreferences: {
    webSecurity: false
  }
})
// Good
const mainWindow = new BrowserWindow()
<!-- Bad -->
<webview disablewebsecurity src="page.html"></webview>

<!-- Good -->
<webview src="page.html"></webview>

6) Define a Content Security Policy

A Content Security Policy (CSP) is an additional layer of protection against cross-site-scripting attacks and data injection attacks. We recommend that they be enabled by any website you load inside Electron.

Why?

CSP allows the server serving content to restrict and control the resources Electron can load for that given web page. https://example.com should be allowed to load scripts from the origins you defined while scripts from https://evil.attacker.com should not be allowed to run. Defining a CSP is an easy way to improve your application's security.

The following CSP will allow Electron to execute scripts from the current website and from apis.example.com.

// Bad
Content-Security-Policy: '*'

// Good
Content-Security-Policy: script-src 'self' https://apis.example.com

CSP HTTP Header

Electron respects the Content-Security-Policy HTTP header which can be set using Electron's webRequest.onHeadersReceived handler:

const { session } = require('electron')

session.defaultSession.webRequest.onHeadersReceived((details, callback) => {
  callback({
    responseHeaders: {
      ...details.responseHeaders,
      'Content-Security-Policy': ['default-src \'none\'']
    }
  })
})

CSP Meta Tag

CSP's preferred delivery mechanism is an HTTP header, however it is not possible to use this method when loading a resource using the file:// protocol. It can be useful in some cases, such as using the file:// protocol, to set a policy on a page directly in the markup using a <meta> tag:

<meta http-equiv="Content-Security-Policy" content="default-src 'none'">

webRequest.onHeadersReceived([filter, ]listener)

7) Do Not Set allowRunningInsecureContent to true

Recommendation is Electron's default

By default, Electron will not allow websites loaded over HTTPS to load and execute scripts, CSS, or plugins from insecure sources (HTTP). Setting the property allowRunningInsecureContent to true disables that protection.

Loading the initial HTML of a website over HTTPS and attempting to load subsequent resources via HTTP is also known as "mixed content".

Why?

Loading content over HTTPS assures the authenticity and integrity of the loaded resources while encrypting the traffic itself. See the section on only displaying secure content for more details.

How?

// Bad
const mainWindow = new BrowserWindow({
  webPreferences: {
    allowRunningInsecureContent: true
  }
})
// Good
const mainWindow = new BrowserWindow({})

8) Do Not Enable Experimental Features

Recommendation is Electron's default

Advanced users of Electron can enable experimental Chromium features using the experimentalFeatures property.

Why?

Experimental features are, as the name suggests, experimental and have not been enabled for all Chromium users. Furthermore, their impact on Electron as a whole has likely not been tested.

Legitimate use cases exist, but unless you know what you are doing, you should not enable this property.

How?

// Bad
const mainWindow = new BrowserWindow({
  webPreferences: {
    experimentalFeatures: true
  }
})
// Good
const mainWindow = new BrowserWindow({})

9) Do Not Use enableBlinkFeatures

Recommendation is Electron's default

Blink is the name of the rendering engine behind Chromium. As with experimentalFeatures, the enableBlinkFeatures property allows developers to enable features that have been disabled by default.

Why?

Generally speaking, there are likely good reasons if a feature was not enabled by default. Legitimate use cases for enabling specific features exist. As a developer, you should know exactly why you need to enable a feature, what the ramifications are, and how it impacts the security of your application. Under no circumstances should you enable features speculatively.

How?

// Bad
const mainWindow = new BrowserWindow({
  webPreferences: {
    enableBlinkFeatures: ['ExecCommandInJavaScript']
  }
})
// Good
const mainWindow = new BrowserWindow()

10) Do Not Use allowpopups

Recommendation is Electron's default

If you are using <webview>, you might need the pages and scripts loaded in your <webview> tag to open new windows. The allowpopups attribute enables them to create new BrowserWindows using the window.open() method. <webview> tags are otherwise not allowed to create new windows.

Why?

If you do not need popups, you are better off not allowing the creation of new BrowserWindows by default. This follows the principle of minimally required access: Don't let a website create new popups unless you know it needs that feature.

How?

<!-- Bad -->
<webview allowpopups src="page.html"></webview>

<!-- Good -->
<webview src="page.html"></webview>

11) Verify WebView Options Before Creation

A WebView created in a renderer process that does not have Node.js integration enabled will not be able to enable integration itself. However, a WebView will always create an independent renderer process with its own webPreferences.

It is a good idea to control the creation of new <webview> tags from the main process and to verify that their webPreferences do not disable security features.

Why?

Since <webview> live in the DOM, they can be created by a script running on your website even if Node.js integration is otherwise disabled.

Electron enables developers to disable various security features that control a renderer process. In most cases, developers do not need to disable any of those features - and you should therefore not allow different configurations for newly created <webview> tags.

How?

Before a <webview> tag is attached, Electron will fire the will-attach-webview event on the hosting webContents. Use the event to prevent the creation of webViews with possibly insecure options.

app.on('web-contents-created', (event, contents) => {
  contents.on('will-attach-webview', (event, webPreferences, params) => {
    // Strip away preload scripts if unused or verify their location is legitimate
    delete webPreferences.preload
    delete webPreferences.preloadURL

    // Disable Node.js integration
    webPreferences.nodeIntegration = false

    // Verify URL being loaded
    if (!params.src.startsWith('https://example.com/')) {
      event.preventDefault()
    }
  })
})

Again, this list merely minimizes the risk, it does not remove it. If your goal is to display a website, a browser will be a more secure option.

12) Disable or limit navigation

If your app has no need to navigate or only needs to navigate to known pages, it is a good idea to limit navigation outright to that known scope, disallowing any other kinds of navigation.

Why?

Navigation is a common attack vector. If an attacker can convince your app to navigate away from its current page, they can possibly force your app to open web sites on the Internet. Even if your webContents are configured to be more secure (like having nodeIntegration disabled or contextIsolation enabled), getting your app to open a random web site will make the work of exploiting your app a lot easier.

A common attack pattern is that the attacker convinces your app's users to interact with the app in such a way that it navigates to one of the attacker's pages. This is usually done via links, plugins, or other user-generated content.

How?

If your app has no need for navigation, you can call event.preventDefault() in a will-navigate handler. If you know which pages your app might navigate to, check the URL in the event handler and only let navigation occur if it matches the URLs you're expecting.

We recommend that you use Node's parser for URLs. Simple string comparisons can sometimes be fooled - a startsWith('https://example.com') test would let https://example.com.attacker.com through.

const URL = require('url').URL

app.on('web-contents-created', (event, contents) => {
  contents.on('will-navigate', (event, navigationUrl) => {
    const parsedUrl = new URL(navigationUrl)

    if (parsedUrl.origin !== 'https://example.com') {
      event.preventDefault()
    }
  })
})

13) Disable or limit creation of new windows

If you have a known set of windows, it's a good idea to limit the creation of additional windows in your app.

Why?

Much like navigation, the creation of new webContents is a common attack vector. Attackers attempt to convince your app to create new windows, frames, or other renderer processes with more privileges than they had before; or with pages opened that they couldn't open before.

If you have no need to create windows in addition to the ones you know you'll need to create, disabling the creation buys you a little bit of extra security at no cost. This is commonly the case for apps that open one BrowserWindow and do not need to open an arbitrary number of additional windows at runtime.

How?

webContents will emit the new-window event before creating new windows. That event will be passed, amongst other parameters, the url the window was requested to open and the options used to create it. We recommend that you use the event to scrutinize the creation of windows, limiting it to only what you need.

const { shell } = require('electron')

app.on('web-contents-created', (event, contents) => {
  contents.on('new-window', async (event, navigationUrl) => {
    // In this example, we'll ask the operating system
    // to open this event's url in the default browser.
    event.preventDefault()

    await shell.openExternal(navigationUrl)
  })
})

14) Do not use openExternal with untrusted content

Shell's openExternal allows opening a given protocol URI with the desktop's native utilities. On macOS, for instance, this function is similar to the open terminal command utility and will open the specific application based on the URI and filetype association.

Why?

Improper use of openExternal can be leveraged to compromise the user's host. When openExternal is used with untrusted content, it can be leveraged to execute arbitrary commands.

How?

//  Bad
const { shell } = require('electron')
shell.openExternal(USER_CONTROLLED_DATA_HERE)
//  Good
const { shell } = require('electron')
shell.openExternal('https://example.com/index.html')

15) Disable the remote module

The remote module provides a way for the renderer processes to access APIs normally only available in the main process. Using it, a renderer can invoke methods of a main process object without explicitly sending inter-process messages. If your desktop application does not run untrusted content, this can be a useful way to have your renderer processes access and work with modules that are only available to the main process, such as GUI-related modules (dialogs, menus, etc.).

However, if your app can run untrusted content and even if you sandbox your renderer processes accordingly, the remote module makes it easy for malicious code to escape the sandbox and have access to system resources via the higher privileges of the main process. Therefore, it should be disabled in such circumstances.

Why?

remote uses an internal IPC channel to communicate with the main process. "Prototype pollution" attacks can grant malicious code access to the internal IPC channel, which can then be used to escape the sandbox by mimicking remote IPC messages and getting access to main process modules running with higher privileges.

Additionally, it's possible for preload scripts to accidentally leak modules to a sandboxed renderer. Leaking remote arms malicious code with a multitude of main process modules with which to perform an attack.

Disabling the remote module eliminates these attack vectors. Enabling context isolation also prevents the "prototype pollution" attacks from succeeding.

How?

// Bad if the renderer can run untrusted content
const mainWindow = new BrowserWindow({})
// Good
const mainWindow = new BrowserWindow({
  webPreferences: {
    enableRemoteModule: false
  }
})
<!-- Bad if the renderer can run untrusted content  -->
<webview src="page.html"></webview>

<!-- Good -->
<webview enableremotemodule="false" src="page.html"></webview>

16) Filter the remote module

If you cannot disable the remote module, you should filter the globals, Node, and Electron modules (so-called built-ins) accessible via remote that your application does not require. This can be done by blocking certain modules entirely and by replacing others with proxies that expose only the functionality that your app needs.

Why?

Due to the system access privileges of the main process, functionality provided by the main process modules may be dangerous in the hands of malicious code running in a compromised renderer process. By limiting the set of accessible modules to the minimum that your app needs and filtering out the others, you reduce the toolset that malicious code can use to attack the system.

Note that the safest option is to fully disable the remote module. If you choose to filter access rather than completely disable the module, you must be very careful to ensure that no escalation of privilege is possible through the modules you allow past the filter.

How?

const readOnlyFsProxy = require(/* ... */) // exposes only file read functionality

const allowedModules = new Set(['crypto'])
const proxiedModules = new Map(['fs', readOnlyFsProxy])
const allowedElectronModules = new Set(['shell'])
const allowedGlobals = new Set()

app.on('remote-require', (event, webContents, moduleName) => {
  if (proxiedModules.has(moduleName)) {
    event.returnValue = proxiedModules.get(moduleName)
  }
  if (!allowedModules.has(moduleName)) {
    event.preventDefault()
  }
})

app.on('remote-get-builtin', (event, webContents, moduleName) => {
  if (!allowedElectronModules.has(moduleName)) {
    event.preventDefault()
  }
})

app.on('remote-get-global', (event, webContents, globalName) => {
  if (!allowedGlobals.has(globalName)) {
    event.preventDefault()
  }
})

app.on('remote-get-current-window', (event, webContents) => {
  event.preventDefault()
})

app.on('remote-get-current-web-contents', (event, webContents) => {
  event.preventDefault()
})

app.on('remote-get-guest-web-contents', (event, webContents, guestWebContents) => {
  event.preventDefault()
})