Daniel Coughlin wrote:

> The only difference between CSS attacks with POST vs GET methods is ease.
> Its easier with the GET method because, as you have noticed, the attacker
> can see the GET request in the location field of the browser.
> POST requests are not visible this way so may appear more secure, BUT THEY
> ARE NOT. Say you have a guestbook that accepts POST request. (First of all -
> if your client code doesnt filter script tags, we could just type a mean
> javascript into the text entry fields). Now lets imagine you *do* filter
> script tags out of the input with your client code -The attacker can still
> 1)turn off javascript to break through you client side filtering or 2)use a
> perl or python script to bypass the client all together. Now everyone
> viewing your guestbook will get hit with the mean javascript. POSTing to a
> form via a perl script is as simple as looking at the form source code.
>

Well said.


> The moral is is to do client as well as server side checking for malicious
> code.

I would say that doing client-side checking should be done purely for
user convienience and not security.  "never trust client-side data".


Reply via email to