⚠️ This package is not for production use yet.

An experimental grammar for parsing XSS attacks.

This package provides a single function for detecting (not mitigating) XSS threats in your JavaScript strings:

const scan = require('@customcommander/xss-lang');

scan("javascript:alert('XSS!')");
/*
{ threat: "js_url"
, raw: "javascript:alert('XSS!')"
, found: "javascript:alert('XSS!')"
}
*/

Your web application will be attacked. Period.

With this parser I wanted to understand the different threats a web application can be exposed to. Knowing what's coming in and out of your application can help you detect and react to potential attacks.

Say we need to create a function that adds links to a page:

function add_link(link) {
  const p = document.createElement('p');
  p.innerHTML = `<a href="${link}">CLICK ME</a>`;
  document.body.appendChild(p);
}

To add a link to a page we simply need to call the function as such:

add_link("https://example.com");

However the attentive reader will notice that this function is vulnerable to XSS attacks:

add_link("javascript:alert('XSS!')");

When the user clicks on that link, an alert box will be displayed. This example is harmless but it demonstrates that code can be executed with full user privileges. If that code comes from an untrusted source there is a high probability that it will cause some damage.

Here's a naive approach to mitigate this threat:

function add_link(link) {
  if (link.startsWith('javascript:')){
    return;
  }
  const p = document.createElement('p');
  p.innerHTML = `<a href="${link}">CLICK ME</a>`;
  document.body.appendChild(p);
}

However this will defeat the check yet still allow code to be executed:

add_link("jaVAscRIpt:alert('XSS!')");

And so will this:

add_link("jaVA	scRIpt:alert('XSS!')");

There's plenty more variations of this than you can handle...