Written by
RSnake with input from
id,
Vacuum and
Robert E Lee. A special thanks to IceShaman to porting it to use multi-threading.
Fierce domain scan was born out of personal frustration after
performing a web application security audit. It is traditionally very
difficult to discover large swaths of a corporate network that is
non-contiguous. It's terribly easy to run a scanner against an IP
range, but if the IP ranges are nowhere near one another you can miss
huge chunks of networks.
First what Fierce is not. Fierce is not an IP scanner, it is not a
DDoS tool, it is not designed to scan the whole internet or perform any
un-targeted attacks. It is meant specifically to locate likely targets
both inside and outside a corporate network. Only those targets are
listed (unless the -nopattern switch is used). No exploitation is
performed (unless you do something intentionally malicious with the
-connect switch). Fierce is a reconnaissance tool. Fierce is a PERL
script that quickly scans domains (usually in just a few minutes,
assuming no network lag) using several tactics.
First it queries your DNS for the DNS servers of the target. It then
switches to using the target's DNS server (you can use a different one
if you want using the -dnsserver switch but this can cause problems if
the server you use won't tell you information about other people's sites
and of course you won't find much relevant internal address space).
Fierce then attempts to dump the SOA records for the domain in the very
slim hope that the DNS server that your target uses may be
misconfigured. Once that fails (because it almost always will) it
attempts to "guess" names that are common amongst a lot of different
companies. Don't ask me where I got the list, it's just a list of names
that id and I have seen all over the place. I thought about adding a
dictionary to this, but I think that would take a lot longer, and given
that very few of the words are dictionary words I don't think this would
add a lot of value.
Next, if it finds anything on any IP address it will scan up and down
a set amount (default 5 but you can expand it with -traverse or
increase it to the entire subnet with -wide) looking for anything else
with the same domain name in it using reverse lookups. If it finds
anything on any of those it will recursively scan until it doesn't find
any more. In this way it ends up looping a lot, and the bigger the
domain is the more you get back. The reason Fierce automatically
switches to using the target's DNS server is so that it can probe the
Intranet (RFC1918) of the target, assuming the target uses a single DNS
server for both their Intranet and external sites.
I also added a random call to something that should fail to test for
wildcard DNS. If it's found, the wildcard is discarded to reduce
erroneous results. That doesn't speed up the scan because it still
needs to check to see if the test resolves back to IP address that the
wildcard is pointing to. However it does reduce false positives.
Also, I've added a "search" option that allows you to find other
non-related domain names. For example, let's say my target's domain is
widget.com but I know they have email addresses like
soandso@widgetcompany.com and own another company called
nutsandbolts.com I can add search queries. This won't scan for those
domains, but if those names pop up, it won't ignore them. Fierce will
report on anything inside the search pattern as long as it matches. If
you want everything I guess you could put a,b,c,...,x,y,z but I'll
probably make something in the future to allow for scanning/reporting
the entire C block once anything is found in it that matches the DNS
string. Here's the syntax:
perl fierce.pl -dns widget.com -search widgetcompany,nutsandbolts
I also realized it can be a little bad about finding everything in a
class C if the target used non-contiguous blocks within the class C. To
deal with that I built in a function to allow a scan (of only C
blocks). This is also really useful for scanning intranets if the DNS
is poorly configured. I might expand on this later.
perl fierce.pl -range 10.10.10.0-255 -dnsserver ns1.example.com
As an alternative, you can use the -wide switch which does a wide
path of reverse lookups after finding any C names that match your query
in the C block. This provides a lot more information but is a lot more
noisy.
perl fierce.pl -dns example.com -wide -file output.txt
Finally, for the web application security folks I added a command to
connect to any http servers on port 80 and perform whatever action you
put into a configuration file. This is really noisy and really slow
(especially on large networks), so I wouldn't recommend trying it unless
you have a few hours with nothing better to do, unless you know there
are only a handful of machines or have already ran this without the
connect scan turned on.
perl fierce.pl -dns example.com -connect headers.txt -fulloutput -file output.txt
Here's what a sample header file might look like. The sample file below is attempting to exploit the
Expect cross site scripting vulnerability:
Fierce also has wordlist support so that you can supply your own
dictionary using the -wordlist keyword. Since the brute force does rely
on matching at least a few internal targets, this could be helpful if
you know that the naming convention has to do with a certain non-obvious
naming convention or uses another language, etc.
perl fierce.pl -dns example.com -wordlist dictionary.txt -file output.txt
Not convinced? Prior to running the scan I had never been to either
mail.ru or rambler.ru (a few of the top Alexa sites in Russia). Since I
don't read Russian, performing an audit against them is far more
difficult. Here's some sample output from the two. In the first
example you can see that mail.ru has a non-contiguous address for it's
mobile.mail.ru than it does for the rest of the site. That would have
been very difficult to locate with any other scanner. In the rambler.ru
example you can see the RFC1918 space 10.* pop up:
- mail.ru - 418 entries and 303 hostnames found.
- rambler.ru - 472 entries and 458 hostnames found.
Trust me, we've found far more interesting sites than these two in
our tests, but I don't want to disparage any companies for their
mistakes. I'm sure you can think of a few companies to test this
against. The results can be pretty amazing. If you don't get many
results, that could be one of three things, 1) you aren't scanning their
corporate domain, you are only scanning their external domain which
they only have one or two machines on 2) it's a very small company or 3)
you typo'd the domain name (I haven't built any checks to make sure the
domain you entered is valid).
Requirements: This is a PERL program requiring the PERL interpreter with the modules
Net::DNS and
Net::hostent. You can install modules using CPAN:
perl -MCPAN -e 'install Net::DNS'
perl -MCPAN -e 'install Net::hostent'
Windows users: You can use Fierce under Windows if you use
Cygwin
with PERL and the above two modules installed. I have not tested this
using ActivePerl in Windows, so I would recommend Cygwin until
ActivePerl can be thoroughly tested. I am/was working on a win32
version of Fierce, but have put the project on hold. If anyone is
interested in picking up where I left off, drop
me a line.
Version: Fierce is currently at version 0.9.9 - Beta 03/24/2007
Download:
fierce.pl
Download:
hosts.txt
(Thanks to Robert E Lee for the help with this and to Michael Thumann's
DNSDigger wordlist).
Getting started:
perl fierce.pl -help
This may some bugs in it. Also this can be a noisy scanner, but in
the tests I've performed it's exceptionally effective at finding
non-contiguous IP blocks and new attack points. This should be
considered a pre-cursor to
nmap,
unicornscan or
nessus
as it gives you enough information to begin a much more thorough scan
with one of those other tools. Also, it can point out DNS entries for
hosts that are no longer up or have not yet been put into production.
Please use Fierce with care and at your own risk.