@eggjs/security

Egg’s default security plugin, generally no need to configure.

Install

npm i @eggjs/security

Usage & configuration

• config.default.js

exports.security = {
  xframe: {
    value: 'SAMEORIGIN',
  },
};

Disable security precautions

To disable some security precautions, set enable property to ‘false’ directly.

For example, disable xframe defense:

exports.security = {
  xframe: {
    enable: false,
  },
};

match & ignore

If you want to set security config open for a certain path, you can configure match option.

For example, just open csp when path contains /example, you can configure with the following configuration:

exports.security = {
  csp: {
    match: '/example',
    // match: /^\/api/, // support regexp
    // match: ctx => ctx.path.startsWith('/api'), // support function
    // match: [ ctx => ctx.path.startsWith('/api'), /^\/foo$/, '/bar'], // support Array
    policy: {
      //...
    },
  },
};

If you want to set security config disable for a certain path, you can configure ignore option.

For example, just disable xframe when path contains /example while our pages can be embedded in cooperative businesses , you can configure with the following configuration:

exports.security = {
  xframe: {
    ignore: '/example',
    // ignore: /^\/api/, // support regexp
    // ignore: ctx => ctx.path.startsWith('/api'), // support function
    // ignore: [ ctx => ctx.path.startsWith('/api'), /^\/foo$/, '/bar'], // support Array
    // ...
  },
};

mention：match has higher priority than ignore

Dynamic configuration for security plugins depend on context

There are times when we want to be more flexible to configure security plugins.For example:

1. To decide whether to enable or disable the xframe security header from the context of the request.
2. To decide csp policies from different request urls.

Then we can configure ctx.securityOptions[name] opts in the custom middleware or controller, then the current request configuration will override the default configuration (new configuration will be merged and override the default project configuration, but only take effect in the current request)

async ctx => {
  // if satisfied some condition
  // change configuration
  ctx.securityOptions.xframe = {
    value: 'ALLOW-FROM: https://domain.com',
  };
  // disable configuration
  ctx.securityOptions.xssProtection = {
    enable: false,
  }
}

Not all security plugins support dynamic configuration, only the following plugins list support

• csp
• hsts
• noopen
• nosniff
• xframe
• xssProtection

And in helper：

• shtml

helper is the same way to configure.

ctx.securityOptions.shtml = {
  whiteList: {
  },
};

Mention

• Security is a big thing, please pay attention to the risk of changes in the security configuration (especially dynamic changes)
• ctx.securityOptions the current request configuration will overrides the default configuration, but it does not make a deep copy，so pay attention to configure csp.policy, it will not be merged.
• If you configure ctx.securityOptions，please write unit tests to ensure the code is correct.

API

ctx.isSafeDomain(domain)

Whether or not the domain is in the whitelist of the configuration. See ctx.redirect.

Note: egg-cors module uses this function internally to determine whether or not send back an Access-Control-Allow-Origin response header with the value of safe domain. Otherwise, ignore the request with an error, No 'Access-Control-Allow-Origin' header is present on the requested resource.

exports.security = {
  domainWhiteList: ['http://localhost:4200']
};

Interface restriction

CSRF

usage

• ctx.csrf getter for CSRF token

Generally used when send POST form request. When page rendering, put ctx.csrf into form hidden field or query string.(_csrf is the key). When submitting the form, please submit with the _csrf token parameter.

Using CSRF when upload by formData

browser:

<form method="POST" action="/upload?_csrf={{ ctx.csrf | safe }}" enctype="multipart/form-data">
  title: <input name="title" />
  file: <input name="file" type="file" />
  <button type="submit">上传</button>
</form>

Using CSRF when request by AJAX

CSRF token will also set to cookie by default, and you can send token through header:

In jQuery:

var csrftoken = Cookies.get('csrftoken');

function csrfSafeMethod(method) {
  // these HTTP methods do not require CSRF protection
  return (/^(GET|HEAD|OPTIONS|TRACE)$/.test(method));
}
$.ajaxSetup({
  beforeSend: function(xhr, settings) {
    if (!csrfSafeMethod(settings.type) && !this.crossDomain) {
      xhr.setRequestHeader('x-csrf-token', csrftoken);
    }
  },
});

Options

there are some options that you can customize:

exports.security = {
  csrf: {
    type: 'ctoken',             // can be ctoken, referer, all or any, default to ctoken
    useSession: false,          // if useSession set to true, the secret will keep in session instead of cookie
    ignoreJSON: false,          // skip check JSON requests if ignoreJSON set to true
    cookieName: 'csrfToken',    // csrf token's cookie name
    sessionName: 'csrfToken',   // csrf token's session name
    headerName: 'x-csrf-token', // request csrf token's name in header
    bodyName: '_csrf',          // request csrf token's name in body
    queryName: '_csrf',         // request csrf token's name in query
    rotateWhenInvalid: false,   // rotate csrf secret when csrf token invalid. For multi applications which be deployed on the same domain, as tokens from one application may impact others.
    refererWhiteList: [],       // referer white list
    supportedRequests: [        // supported URL path and method, the package will match URL path regex patterns one by one until path matched. We recommend you set {path: /^\//, methods:['POST','PATCH','DELETE','PUT','CONNECT']} as the last rule in the list, which is also the default config.
      {path: /^\//, methods:['POST','PATCH','DELETE','PUT','CONNECT']}
    ],
    cookieOptions: {},          // csrf token's cookie options
  },
}

methods in supportedRequests can be empty, which means if you set supportedRequests: [{path: /.*/, methods:[]}], the whole csrf protection will be disabled.

Rotate CSRF secret

Must call ctx.rotateCsrfSecret() when user login to ensure each user has independent secret.

safe redirect

• ctx.redirect(url) If url is not in the configuration of the white list, the redirect will be prohibited

• ctx.unsafeRedirect(url) Not Recommended;

Security plugin override ctx.redirect method，all redirects will be judged by the domain name.

If you need to use ctx.redirect, you need to do the following configuration in the application configuration file：

exports.security = {
  domainWhiteList:['.domain.com'],  // security whitelist, starts with '.'
};

If user do not configure domainWhiteList or domainWhiteList is empty, it will pass all redirects, equal to ctx.unsafeRedirect(url). domainWhiteList and url are case insensitive.

jsonp

Based on jsonp-body.

Defense:

• The longest callback function name limit of 50 characters.
• Callback function only allows “[“,”]”,”a-zA-Z0123456789_”, “$” “.” to prevent xss or utf-7 attack.

Config：

• callback function default name _callback.
• limit - function name limit, default by 50.

helper

.escape()

String xss filter, the most secure filtering mechanism.

const str = '><script>alert("abc") </script><';
console.log(ctx.helper.escape(str));
// => &gt;&lt;script&gt;alert(&quot;abc&quot;) &lt;/script&gt;&lt;

In nunjucks template, escape by default.

.surl()

url filter.

Used for url in html tags (like <a href=""/><img src=""/>),please do not call under other places.

helper.surl($value)。

Mention: Particular attention, if you need to resolve URL use surl，surl need warpped in quotes, Otherwise will lead to XSS vulnerability.

Example: do not use surl

<a href="$value" />

output:

<a href="http://ww.domain.com<script>" />

Use surl

<a href="helper.surl($value)" />

output:

<a href="http://ww.domain.com&lt;script&gt;" />

protocolWhitelist

If url’s protocol is not in the protocol whitelist, it will return empty string.

Protocol whitelist is http, https, file, data.

So if you want surl support custom protocol, please extend the security protocolWhitelist config :

exports.security = {
  protocolWhitelist: ['test']
};

.sjs()

Used to output variables in javascript(include onload/event),it will do JAVASCRIPT ENCODE for the variable string.It will escape all characters to \x which are not in the whitelist to avoid XSS attack.

const foo = '"hello"';

// not use sjs
console.log(`var foo = "${foo}";`);
// => var foo = ""hello"";

// use sjs
console.log(`var foo = "${ctx.helper.sjs(foo)}";`);
// => var foo = "\x22hello\x22";

.shtml()

If you want to output richtexts in views, you need to use shtml helper. It will do XSS filter, then output html tags to avoid illegal scripts.

shtml is a very complex process, it will effect server performance, so if you do not need to output HTML, please do not use shtml.

Examples:

// js
const value = `<a href="http://www.domain.com">google</a><script>evilcode…</script>`;

// in your view
<html>
<body>
  ${helper.shtml($value)}
</body>
</html>
// => <a href="http://www.domain.com">google</a>&lt;script&gt;evilcode…&lt;/script&gt;

shtml based on xss, and add filter by domain feature.

• default rule
• custom rule http://jsxss.com/zh/options.html

For example, only support a tag, and filter all attributes except for title:

whiteList: {a: ['title']}

options:

config.helper.shtml.domainWhiteList has been deprecated, please use config.security.domainWhiteList instead.

Mention that shtml uses a strict white list mechanism, in addition to filtering out the XSS risk of the string,tags and attrs which are not in the default rule will be filtered.

For example html tag is not in the whitelist.

const html = '<html></html>';

// html
${helper.shtml($html)}

// output none

Commonly used data-xx property is not in the whitelist, so it will be filtered. So please check the applicable scenarios for shtml, it is usually used for rich-text submitted by user.

A usage error will limit functions, and also affect the performance of the server. Such scenes are generally forums, comments, etc.

Even if the forum does not support the HTML content input, do not use this helper, you can directly use escape instead.

.spath()

If you want to use users input for a file path, please use spath for security check. If path is illegal, it will return null.

Illegal path:

• relative path starts with ..
• absolute path starts with /
• above path try to use url encode to bypass the check

const foo = '/usr/local/bin';
console.log(ctx.helper.spath(foo2));
// => null

.sjson()

json encode.

If you want to output json in javascript without encoding, it will be a risk for XSS. sjson supports json encode，it will iterate all keys in json, then escape all characters in the value to \x to avoid XSS attack, and keep the json structure unchanged. If you want to output json string in your views, please use ${ctx.helper.sjson(var)}to escape.

it has a very complex process and will lost performance, so avoid the use as far as possible

example:

  <script>
    window.locals = ${ctx.helper.sjson(locals)};
  </script>

.cliFilter()

It will cause remote command execution vulnerability, when user submit the implementation of the command by browser.because the server does not filter for the implementation of the function, resulting in the execution of the command can usually lead to the invasion of the server.

If you want to get user submit for command’s parameter, please use cliFilter。

before fix:

  cp.exec("bash /home/admin/ali-knowledge-graph-backend/initrun.sh " + port);

after fix:

  cp.exec("bash /home/admin/ali-knowledge-graph-backend/initrun.sh " + ctx.helper.cliFilter(port));

.escapeShellArg()

Escape command line arguments. Add single quotes around a string and quotes/escapes any existing single quotes allowing you to pass a string directly to a shell function and having it be treated as a single safe argument.

const ip = '127.0.0.1 && cat /etc/passwd'
const cmd = 'ping -c 1 ' + this.helper.escapeShellArg(ip);

console.log(cmd);
//ping -c 1 '127.0.0.1 && cat /etc/passwd'

.escapeShellCmd()

Command line escape to remove the following characters from the entered command line: #&;`|*?~<>^()[]{}$;'", 0x0A and 0xFF

const ip = '127.0.0.1 && cat /etc/passwd'
const cmd = 'ping -c 1 ' + this.helper.escapeShellCmd(ip);

console.log(cmd);
//ping -c 1 127.0.0.1  cat /etc/passwd

Security Headers

Refer to lusca, appreciate their work.

hsts Strict-Transport-Security

Disabled by default. If your website based on https, we recommend you should enable it.

• maxAge one year by default 365 * 24 * 3600
• includeSubdomains false by default

csp

Default disabled. If you need to enable it, please contact your security engineers and determine the opening strategy

• policy policies used by csp

X-Download-Options:noopen

Default enabled, disable IE download dialog automatically opens download file and will cause XSS

X-Content-Type-Options:nosniff

Disable IE8’s auto MIME sniffing. E.g.: take text/plain as text/html by mistake and render it, especially when there’s something untrusted in the local service.

X-Frame-Options

Defaulting to “SAMEORIGIN”, only allows iframe to embed by the same origin.

• value Defaulting to SAMEORIGIN

X-XSS-Protection

• disable Defaulting to false, the same as 1; mode=block.

SSRF Protection

In a Server-Side Request Forgery (SSRF) attack, the attacker can abuse functionality on the server to read or update internal resources.

egg-security provide ctx.safeCurl, app.safeCurl and agent.safeCurl to provide http request(like ctx.curl, app.curl and agent.curl) with SSRF protection.

Configuration

• ipBlackList(Array) - specific which IP addresses are illegal when requested with safeCurl.
• ipExceptionList(Array) - specific which IP addresses are legal within ipBlackList. hostnameExceptionList(Array) - specifies which hostnames are legal within ipBlackList.
• checkAddress(Function) - determine the ip by the function’s return value, false means illegal ip.

// config/config.default.js
exports.security = {
  ssrf: {
    // support both cidr subnet or specific IP
    ipBlackList: [
      '10.0.0.0/8',
      '127.0.0.1',
      '0.0.0.0/32',
    ],
    // support both cidr subnet or specific IP
    ipExceptionList: [
      '10.1.1.1',
      '10.10.0.1/24',
    ],
    // legal hostname
    hostnameExceptionList: [
      'example.com',
    ],
    // checkAddress has higher priority than ipBlackList
    checkAddress(ip) {
      return ip !== '127.0.0.1';
    }
  },
};

Other

• Forbid trace track http methods.

License

MIT

Contributors

Made with contributors-img.