BOOK THIS SPACE FOR AD
ARTICLE ADA few months ago, I received an invitation to a private bug bounty program on HackerOne. Initially, I did my usual testing and I discovered and reported various bugs, such as BAC, leaking “Authorization Tokens” of other users, open redirects, and HTML injection in clients’ emails.
After reporting these bugs to the program, I noticed that XSS is considered Out Of Scope according to their policy. The program business is “an online content management system and website builder”. Upon creating an account, users receive a unique subdomain like <YOUR-SUB>.target.comthat can be customized. Visitors to these subdomains do not need to be logged in.
Considering the app’s structure, XSS is limited to being self, and the program excludes XSS on <YOUR-SUB>.target.com from their scope. This encouraged me to explore finding a self-XSS and chaining it with another bug to show an Impact.
I discovered several chains for XSS that escalated its impact. Since there is only one resolved XSS chain report, I will only write a write-up about it. When the other reports are resolved, I plan to write separate write-ups for each of them. Now, let’s dive into the story.
Finding Self-XSS didn’t take much time.
Now, it’s time to find another bug to chain with XSS. I began testing the website with the goal of identifying bugs to chain with the XSS. One of the bugs I tested for was CORS misconfiguration. I prefer testing it using the CORS* Burp Extension.
To use this extension, simply click on the “Activate CORS* checkbox” and browse your target to let the extension do its magic. It will attempt different CORS bypass techniques for all requests passing through your Burp proxy. All you need to do is to check the results.
After activating the extension I found a possible CORS misconfiguration in www.target.com/api/me. This API request is responsible for fetching user PII such as first name, last name, email, and account ID.
When sending a request with
Origin: 7odamoo-target.comThe response was
Access-Control-Allow-Credentials: trueAccess-Control-Allow-Origin: 7odamoo-target.com
Awesome, seems I got a bug now, in normal cases I should register the domain 7odamoo-target.comand upload a CORS PoC on it to show the exploit to the team, but when I checked the cookie flags I found that the SameSite cookie flag was set to Strict. This means the browser would not include the cookie in the request unless the origin is allowed.
I think you get what I will do next, I will use the Out-Of-Scope XSS to send the request so the cookie will be included.
So I Injected this script on my own subdomain <MY-SUB>.target.com so when anyone visits my subdomain a request will be sent to www.target.com/api/meand his PII will be displayed on an alert Popup as a PoC
function cors() {var xhttp = new XMLHttpRequest();
xhttp.onreadystatechange = function() {
if (this.readyState == 4 && this.status == 200) {
document.getElementById("demo").innerHTML = alert(this.responseText);
}
};
xhttp.onload = cors;
xhttp.open("GET", "https://www.target.com/api/me", true);
xhttp.withCredentials = true;
xhttp.send();
}
cors();
Cool, I successfully chained the CORS Misconfiguration with the Out-Of-Scope XSS to bypass the origin check for the SameSite cookie.
The team promptly addressed the vulnerability by fixing the CORS misconfiguration, as they no longer allowing *.target.comto read the response from www.target.com
As highlighted earlier in the write-up, once the remaining XSS chains are resolved, I’ll write separate write-ups for each of them.