BloodHound is a tool to graphically map Active Directory and discover attack paths.
One of the glorious design features of AD is that everyone in the domain needs to know where everything is. So when you get user credentials and/or a shell, you can basically map the entire domain without breaking any rules. Any user can query Active Directory for computers, domain controllers, groups and sessions.
Now we can use this brilliant feature to collect a ton of information and create a cool GUI map of the entire AD which can be queried using BloodHound. There are two software requirements, you need
BloodHoundand a database to store the data in. The recommended choice is
neo4j, see below for further instructions.
- Run neo4j with
- Navigate to
localhost:7474in your browser
- Log in with username and password
- Set a new password for the neo4j account
- Open the file
neo4j.conffrom the neo4j installation directory and set the following parameters to make any host be able to access the database.dbms.connector.http.enabled=truedbms.connector.http.listen_address=0.0.0.0:7474
- Restart neo4j with
- Access neo4j in the browser at
Neo4j can be started with powershell on windows.
- Spawn an administrative powershell with -bypass exec
- Navigate to the neo4j/bin directory
- Likewise to Linux, log in to
localhost:7474from your browser and change the password.
Similar procedure as linux. Neo4j does not support Java 9, so Java SDK must be version 8 and not 9. Install java 8 with cask in Homebrew:
brew cask install java8
- Launch BloodHound and log in to the neo4j database with credentials previously set
To collect data in a format Bloodhound can read is called ingestion. There are several ways of doing this and different types of collection methods. The most useable is the C# ingestor called SharpHound and a Powershell ingestor called Invoke-BloodHound. Both are bundled with the latest release.
What I recommend doing if you have internal network access is to run Bloodhound using
runas /netonlyfrom your own machine and not from a host you are not in the control of. This way you're not cluttering a domain joined machine with files, you will not trigger antivirus and you don't have to exfiltrate the data either, so its generally less noisy.
runas /netonly /FQDN\user\<username> powershell
Example with the domain
testlab.localand a username
runas /netonly /testlab.local\user\testuser powershell
Type in the password of testuser when prompted. This should spawn a new Powershell window. This window will use the local DNS settings to find the nearest domain controller and perform the various LDAP queries Bloodhound performs. First, from a powershell shell with execution policy set to bypass, import the powershell module
Then, start collecting data. This command specifies to collect all kinds of information, compress it into a ZIP and remove stray CSV files generated during ingestion.
Invoke-BloodHound -CollectionMethod All -CompressData -RemoveCSV
You will now have get a ZIP file containing CSV files in the directory where you ran the ingestor from. This ZIP file can from v2.0 be dragged and dropped straight into the BloodHound interface.
You should immediately see data being populated into the database and the interface.
You can now play with BloodHound to create really some really cool maps. You can also perform queries to show the shortest path to DA, etc. See the default queries and SpectreOps blog posts for inspiration.
Explanations of things found under Node info https://github.com/BloodHoundAD/BloodHound/wiki/Users https://posts.specterops.io/sharphound-target-selection-and-api-usage-bba517b9e69b https://github.com/porterhau5/BloodHound https://porterhau5.com/blog/representing-password-reuse-in-bloodhound/ https://porterhau5.com/blog/extending-bloodhound-track-and-visualize-your-compromise/ http://threat.tevora.com/lay-of-the-land-with-bloodhound/