Missing CORS leads to Complete Account Takeover

Hello Mates,

This vulnerability is one of my best finding till date. It has severity of P2 according to Bugcrowd VRT.

So, the story begins like this, I was testing on this target and found that there was an application wide CSRF in the domain. Now, as per bugcrowd this vulnerability alone has severity of P2. But when I reported it, as usual, it got duplicated.

But yet the vulnerability was present so I thought if I chain this bug with some other bug I might find something interesting.

I immediately fired up my Burpsuite and started looking for some more bugs to chain it with this.

One thing that I do most often when I start with my testing is when I enter the credentials and click login I turn on my burp proxy and when the login functionality is completed and the landing page is completely loaded in the browser. I turn off my proxy for a while and take a look at my Burp History. This methodology of mine gives me a good understanding of what requests the application sends and response is received on login(just login).

So as I told I stared looking in the burp history tab. I was looking every response returned. Something that captured my attention was an HTML response. I saw that in the main domain like “stage.redacted.com/” the HTML contained authentication token and all the session related data.

Now since there was no CSRF , I confirmed that if somehow I get this data(authentication token in particular) I will be able to perform a complete account takeover of any user. But CSRFs are generally exploited to make victim perform some action. But this authentication token was just visible in the victim’s browser when the HTML gets loaded after the user logs in. There was actually nothing that I can change in the victim’s page.

Being a programmer, I have a habit of dividing bigger (seemingly impossible)problems into smaller and smaller problems upto an extent that it becomes solvable.

So currently I have CSRF at my hand as a tool and my problem is to get the data from user’s browser and send it to my server. Using CSRF all I can do was make a request on user’s behalf but that too will only work on the victim’s PC in which he is logged in. This is when Missing CORS came to rescue. I noticed the “Access-Control-Allow-Credential: True” in the response header.

Now I have CSRF and Missing CORS at my hand as tools. I thought what if I craft a malicious page and send it to the victim.

This malicious page will have two callback AJAX requests. First request will issue a get request to https://stage.redacted.com/ the response of which will contain HTML that have authentication token in it and the Second request will add that HTML response to my server URL and send a request to my server. Using the logs I can get the HTML data and thus the authentication token.

Now I may be a newbie in bug hunting but when it comes to programming I am a different person altogether.And creating this script was a piece of cake.

This is the script I wrote, easy right?

It was sending me the auth token I required.

Here for this vulnerability to take place two things were very important CSRF and Missing CORS. Without this we would have to find some different method.


  • Even if your report gets duplicate(which obviously means that the company have not resolved the issue yet), there are still high chances you can escalate or chain the vulnerability.
  • Never ignore any type of bugs even if it of low severity, chaining it may result in high severity bugs.
  • Even if you can’t perform any action such as password/email change, reset password etc. using CSRF, there may be some other GET based request that may lead you to P1 or P2. Be Creative.
  • If you find any interesting exploit of a bug / blog / write-ups share it with community, like this one .

Lastly, encourage new bug hunters like me, by clapping and leaving some tips of yours and sharing it on your twitter feed(okay the sharing part was for some reach. ;-) )

See you until next time.

Do ping me on twitter. Here

I am Jordin