gather gather gather
Jackdaw is here to collect all information in your domain, store it in a SQL database and show you nice graphs on how your domain objects interact with each-other an how a potential attacker may exploit these interactions. It also comes with a handy feature to help you in a password-cracking project by storing/looking up/reporting hashes/passowrds/users.
enum. This can be done automatically on windows by double-clicking on the executable, or typing
nestto get the web interface. By default it is served under
http://127.0.0.1:5000/nestthere is a SWAGGER documented API under
Generate graph cacheonly once to get the edge information in a cache file. it might take a while but in the command line you will see a progress bar.
graph viewbut don't forget to select the corrrect cache number on the top left.
This section will be regurarly updated based on user feedback.
No advice here, while some improvements can be done in code there is nothing that a generic can do.
Graph data cache file generation must be done on each graph once (and only once) which can take a while using the default sqlalchemy tool.
Performance and speed can be significantly (over 50x more speed and 20x less memory) if you use the sqlite backend AND put the "sqlite3" command line utility somewhere in the
PATH. I'd recommend this to every user.
Now here comes the big tradeoff part. Early implementation of Jackdaw used the
networkx module as the graph backend since it is completely written in Python. But this came at a really significant memory and speed cost. To have Jackdaw pure Python, this option still exists however using the
igraph backend is now the default.
igraph is a C++ library with Python bindings. It has precompiled wheels for Windows and major linux distros but if you use Jackdaw on something else (embedded systems/mobile phones/web browsers) you will either need to switch back to
networkx or suffer with the hours long compilation time.
No need to pre-initialise the database, it will be done automatically.
Using the distributed binary you can just double-click on
Using as a python script
jackdaw --sql sqlite:///<full path here>/test.db dbinit
ON LINUX SYSTEMS
<full path here> includes the firest
/ so you will have
//// four (4) dashes before the file name. Don't get freaked out.
jackdaw --sql sqlite:///test.db enum 'ldap+sspi-ntlm://<domain>\<placeholder>@10.10.10.2' 'smb+sspi-ntlm://<domain>\<placeholder>@10.10.10.2'
The passowrd is
jackdaw --sql sqlite:///test.db enum 'ldap+ntlm-password://TEST\victim:Passw0rdfirstname.lastname@example.org' 'smb+ntlm-password://TEST\victim:Passw0rdemail@example.com'
The passowrd is
jackdaw --sql sqlite:///test.db ldap 'ldap+ntlm-password://TEST\victim:Passw0rdfirstname.lastname@example.org'
jackdaw --sql sqlite:///<FULL PATH TO DB> nest
http://127.0.0.1:5000/ui for the API
Please see the
Building the UI section further down to learn how to build the UI. Once built:
http://127.0.0.1:5000/nest for the graph interface (shows the graph, but far from working)
LDAP enumeration phase acquires data on AD info, User, Machine, OU, Group objects which will be reprezented as a node in the graph, and as a separate table in the DB. Additionally all afforementioned objects' Security Descriptior will be parsed and the ACLs for the DACL added to the DB. This, together with the memebership information will be represented as edges in the garph. Additionally custom SQL queries can be performed on any of the afforementioned data types when needed.
SMB enumeration phase acquires data on shares, localgroups, sessions, NTLM data via connecting to each machine in the domain (which is acquired via LDAP)
Kerberos module does automatic kerberoasting and aspreproasting
The framework allows users to upload LSASS memory dumps to store credentials and extend the session information table. Both will be used as additional edges in the graph (shared password and session respectively). The framework also uses this information to create a password report on weak/shared/cracked credentials.
The framework allows users to upload impacket's DCSYNC files to store credentials. This be used as additional edges in the graph (shared password). The framework also uses this information to create a password report on weak/shared/cracked credentials.
The framework allows manually extending the available DB in every aspect. Example: when user session information on a given computer is discovered (outside of the automatic enumeration) there is a possibility to manually upload these sessions, which will populate the DB and also the result graph
The framework can generate a graph using the available information in the database and plot it via the web UI (nest). Furthermore the graph generation and path canculations can be invoked programmatically, either by using the web API (/ui endpoint) or the grph object's functions.
The framework can identify common AD misconfigurations without graph generation. Currently only via the web API.
User anomalies detection involve detection of insecure UAC permissions and extensive user description values. This feature set is expected to grow in the future as new features will be implemented.
Machine anomalies detection involve detection of insecure UAC permissions, non-mandatory SMB singing, outdated OS version, out-of-domain machines. This feature set is expected to grow in the future as new features will be implemented.
The framework does not performing any cracking, only organizing the hashes and the cracking results
currently main focus is on impacket and aiosmb's dcsync results !NT and LM hashes only!
Sample porcess is the following:
note form author: This feature was implemented for both attackers and defenders. Personally I don't see much added value on either side, since at the point one obtained the NT hash of a user it's just as good as the password... Nonetheless, more and more companies are performing password strength excercises, and this feature would help them. As for attackers: it is just showing off at this point, but be my guest. Maybe scare management for extra points.
This project is in experimental phase! This means multiple things:
Jackdaw uses SQLAlchemy ORM module, which gives you the option to use any SQL DB backend you like. The tests are mainly done on SQLite for ovbious reasons. There will be no backend-specific commands used in this project that would limit you.
THIS IS ONLY NEEDED IF YOU INSTALL VIA GIT AND/OR CHANGE SOMETHING IN THE UI CODE
The UI was written in React. Before first use/installation you have to build it. For this, you will need
npm installed. Then:
npm run build
Once done with the above, the UI is ready to play with.
"If I have seen further it is by standing on the shoulders of Giants."
Zsolt Imre (https://github.com/keymandll)