Next generation penetration test tool
GyoiThon at Black Hat ASIA 2018 Arsenal.
GyoiThon is a growing penetration test tool using Machine Learning.
GyoiThon identifies the software installed on web server (OS, Middleware, Framework, CMS, etc...) based on the learning data. After that, it executes valid exploits for the identified software using Metasploit. Finally, it generates reports of scan results. GyoiThon executes the above processing automatically.
- Processing steps
GyoiThon executes the above "Step1" - "Step4" fully automatically.
User's only operation is to input the top URL of the target web server in GyoiThon.
It is very easy!
You can identify vulnerabilities of the web servers without taking time and effort.
GyoiThon gathers several HTTP responses of target website while crawling.
The following are example of HTTP responses gathered by GyoiThon.
- Example.1
HTTP/1.1 200 OK
Date: Tue, 06 Mar 2018 03:01:57 GMT
Connection: close
Content-Type: text/html; charset=UTF-8
Etag: "409ed-183-53c5f732641c0"
Content-Length: 15271
...snip...
- Example.2
HTTP/1.1 200 OK
Date: Tue, 06 Mar 2018 06:56:17 GMT
Connection: close
Content-Type: text/html; charset=UTF-8
Set-Cookie: f00e68432b68050dee9abe33c389831e=0eba9cd0f75ca0912b4849777677f587;
path=/;
Content-Length: 37496
...snip...
- Example.3
HTTP/1.1 200 OK
Date: Tue, 06 Mar 2018 04:19:19 GMT
Connection: close
Content-Type: text/html; charset=UTF-8
Content-Length: 11819
...snip...
<script src="/core/misc/drupal.js?v=8.3.1"></script>
GyoiThon identifies product name installed on web server using following two methods.
By using Machine Learning (Naive Bayes), GyoiThon identifies software based on a combination of slightly different features (Etag value, Cookie value, specific HTML tag etc.) for each software. Naive Bayes is learned using the training data which example below (Training data). Unlike the signature base, Naive Bayes is stochastically identified based on various features included in HTTP response when it cannot be identified software in one feature.
- Example.1
Etag: "409ed-183-53c5f732641c0"
GyoiThon can identify the web server software Apache.
This is because GyoiThon learns features of Apache such as "Etag header value (409ed-183-53c5f732641c0). In our survey, Apache use combination of numeral and lower case letters as the Etag value. And, Etag value is separated 4-5 digits and 3-4 digits and 12 digits, final digit is 0 in many cases.
- Example.2
Set-Cookie: f00e68432b68050dee9abe33c389831e=0eba9cd0f75ca0912b4849777677f587;
GyoiThon can identify the CMS Joomla!.
This is because GyoiThon learns features of Joomla! such as "Cookie name (f00e6 ... 9831e) " and "Cookie value (0eba9 ... 7f587). In our survey, Joomla! uses 32 lower case letters as the Cookie name and Cookie value in many cases.
- Joomla! (CMS)
Set-Cookie: ([a-z0-9]{32})=[a-z0-9]{26,32};
Set-Cookie: [a-z0-9]{32}=([a-z0-9]{26,32});
...snip...
- HeartCore (Japanese famous CMS)
Set-Cookie:.*=([A-Z0-9]{32});.*
<meta name=["'](author)["'] content=["']{2}.*
...snip...
- Apache (Web server software)
Etag:.*".*-[0-9a-z]{3,4}-[0-9a-z]{13}")[\r\n]
...snip...
Of course, GyoiThon can identify software by string matching also used in traditional penetration test tools. Examples are shown below.
- Example.3
<script src="/core/misc/drupal.js?v=8.3.1"></script>
GyoiThon can identify the CMS Drupal.
It is very easy.
GyoiThon executes exploit corresponding to the identified software using Metasploit and it checks whether the software is affected by the vulnerability.
- Running example
[*] exploit/multi/http/struts_code_exec_exception_delegator, target: 1, payload: linux/x86/shell/reverse_nonx_tcp, result: failure
[*] exploit/multi/http/struts_code_exec_exception_delegator, target: 1, payload: linux/x86/shell/reverse_tcp, result: failure
[*] exploit/multi/http/struts_code_exec_exception_delegator, target: 1, payload: linux/x86/shell/reverse_tcp_uuid, result: failure
[*] exploit/multi/http/struts_code_exec_exception_delegator, target: 1, payload: linux/x86/shell_bind_ipv6_tcp, result: failure
[*] exploit/multi/http/struts_code_exec_exception_delegator, target: 1, payload: linux/x86/shell_bind_tcp, result: failure
...snip...
[*] exploit/linux/http/apache_continuum_cmd_exec, target: 0, payload: generic/custom, result: failure
[*] exploit/linux/http/apache_continuum_cmd_exec, target: 0, payload: generic/debug_trap, result: failure
[*] exploit/linux/http/apache_continuum_cmd_exec, target: 0, payload: generic/shell_bind_tcp, result: failure
[*] exploit/linux/http/apache_continuum_cmd_exec, target: 0, payload: generic/shell_reverse_tcp, result: failure
[*] exploit/linux/http/apache_continuum_cmd_exec, target: 0, payload: generic/tight_loop, result: bingo!!
GyoiThon generates a report that summarizes vulnerabilities.
Report's style is html.
- sample
https://www.youtube.com/watch?v=jmi43eZOE9w
Firstly, you initialize metasploit db (postgreSQL) using msfdb command.
root@kali:~# msfdb init
You launch Metasploit on the remote server that installed Metasploit Framework such as Kali Linux.
root@kali:~# msfconsole
______________________________________________________________________________
| |
| METASPLOIT CYBER MISSILE COMMAND V4 |
|______________________________________________________________________________|
\\ / /
\\ . / / x
\\ / /
\\ / + /
\\ + / /
* / /
/ . /
X / / X
/ ###
/ # % #
/ ###
. /
. / . * .
/
*
+ *
^
#### __ __ __ ####### __ __ __ ####
#### / \\ / \\ / \\ ########### / \\ / \\ / \\ ####
################################################################################
################################################################################
# WAVE 4 ######## SCORE 31337 ################################## HIGH FFFFFFFF #
################################################################################
https://metasploit.com
=[ metasploit v4.16.15-dev ]
+ -- --=[ 1699 exploits - 968 auxiliary - 299 post ]
+ -- --=[ 503 payloads - 40 encoders - 10 nops ]
+ -- --=[ Free Metasploit Pro trial: http://r-7.co/trymsp ]
msf >
You launch RPC Server of Metasploit following.
msf> load msgrpc ServerHost=192.168.220.144 ServerPort=55553 User=test Pass=test1234
[*] MSGRPC Service: 192.168.220.144:55553
[*] MSGRPC Username: test
[*] MSGRPC Password: test1234
[*] Successfully loaded plugin: msgrpc
| msgrpc options | description |
|---|---|
| ServerHost | IP address of your server that launched Metasploit. Above example is 192.168.220.144. |
| ServerPort | Any port number of your server that launched Metasploit. Above example is 55553. |
| User | Any user name using authentication (default => msf). Above example is test. |
| Pass | Any password using authentication (default => random string). Above example is test1234. |
You have to change following value in config.ini
...snip...
[GyoiExploit]
server_host : 192.168.220.144
server_port : 55553
msgrpc_user : test
msgrpc_pass : test1234
timeout : 10
LHOST : 192.168.220.144
LPORT : 4444
...snip...
| config | description |
|---|---|
| server_host | IP address of your server that launched Metasploit. Your setting value ServerHost in Step2. |
| server_port | Any port number of your server that launched Metasploit. Your setting value ServerPort in Step2. |
| msgrpc_user | Metasploit's user name using authentication. Your setting value User in Step2. |
| msgrpc_pass | Metasploit's password using authentication. Your setting value Pass in Step2. |
| LHOST | IP address of your server that launched Metasploit. Your setting value ServerHost in Step2. |
GyoiThon accesses target server using host.txt.
So, you have to edit host.txt before executing GyoiThon.
- sample of host.txt
target server => 192.168.220.148
target port => 80
target path => /oscommerce/catalog/
192.168.220.148 80 /oscommerce/catalog/
You have to separate IP address, port number and target path using single space.
| Note |
|---|
Current gyoithon.py is provisional version that without crawling function. We'll upgrade gyoithon.py by April 9. Then, target path will be unnecessary. |
You execute GyoiThon following command.
local@client:~$ python gyoithon.py
Please check scan report using any web browser.
local@client:~$ firefox "gyoithon root path"/classifier4gyoithon/report/gyoithon_report.html
signatures path includes four files corresponding to each product categories.
local@client:~$ ls "gyoithon root path"/signatures/
signature_cms.txt
signature_framework.txt
signature_os.txt
signature_web.txt
signature_cms.txt
It includes string matching patterns of CMS.signature_framework.txt
It includes string matching patterns of FrameWork.signature_os.txt
It includes string matching patterns of Operating System.signature_web.txt
It includes string matching patterns of Web server software.
If you want to add new string matching patterns, you add new string matching patterns at last line in each file.
ex) How to add new string matching pattern of CMS at signature_cms.txt.
tikiwiki@(Powered by TikiWiki)
wordpress@<.*=(.*/wp-).*/.*>
wordpress@(<meta name="generator" content="WordPress).*>
...snip...
typo@.*(href="fileadmin/templates/).*>
typo@(<meta name="generator" content="TYPO3 CMS).*>
"new product name"@"regex pattern"
[EOF]
| Note |
|---|
Above new product name must be a name that Metasploit can identify. And you have to separate new product name and regex pattern using @. |
signatures path includes four files corresponding to each product categories.
local@client:~$ ls "gyoithon root path"/classifier4gyoithon/train_data/
train_cms_in.txt
train_framework_in.txt
train_os_in.txt
train_web_in.txt
train_cms_in.txt
It includes learning data of CMS.train_framework_in.txt
It includes learning data of FrameWork.train_os_in.txt
It includes learning data of Operating System.train_web_in.txt
It includes learning data of Web server software.
If you want to add new learning data, you add learning data at last line in each file.
ex) How to add new learning data of CMS at train_cms_in.txt.
joomla@(Set-Cookie: [a-z0-9]{32}=.*);
joomla@(Set-Cookie: .*=[a-z0-9]{26,32});
...snip...
xoops@(xoops\.js)
xoops@(xoops\.css)
"new product name"@"regex pattern"
[EOF]
| Note |
|---|
Above new product name must be a name that Metasploit can identify. And you have to separate new product name and regex pattern using @. |
And you have to delete trained data (*.pkl).
local@client:~$ ls "gyoithon root path"/classifier4gyoithon/trained_data/
train_cms_out.pkl
train_framework_out.pkl
train_web_out.pkl
local@client:~$ rm "gyoithon root path"/classifier4gyoithon/trained_data/*.pkl
When GyoiThon exploits, it uses default value of Exploit module options.
If you want to change option values, please input any value to "user_specify" in exploit_tree.json as following.
"unix/webapp/joomla_media_upload_exec": {
"targets": {
"0": [
"generic/custom",
"generic/shell_bind_tcp",
"generic/shell_reverse_tcp",
...snip...
"TARGETURI": {
"type": "string",
"required": true,
"advanced": false,
"evasion": false,
"desc": "The base path to Joomla",
"default": "/joomla",
"user_specify": "/my_original_dir/"
},
Above example is to change value of TARGETURI option in exploit module "exploit/unix/webapp/joomla_media_upload_exec" to "/my_original_dir/" from "/joomla".
- Kali Linux 2017.3 (for Metasploit)
- Memory: 8.0GB
- Metasploit Framework 4.16.15-dev
- ubuntu 16.04 LTS (Host OS)
- CPU: Intel(R) Core(TM) i5-5200U 2.20GHz
- Memory: 8.0GB
- Python 3.6.1（Anaconda3）
- docopt 0.6.2
- jinja2 2.10
- msgpack-python 0.4.8
- pandas 0.20.3