Released as Proof of Concept for Blue and Purple teams to more effectively use BloodHoundAD in continual security life-cycles by utilizing the BloodHoundAD pathfinding engine to identify Active Directory security vulnerabilities resulting from business operations, procedures, policies and legacy service operations.
PlumHound operates by wrapping BloodHoundAD's powerhouse graphical Neo4J backend cypher queries into operations-consumable reports. Analyzing the output of PlumHound can steer security teams in identifying and hardening common Active Directory configuration vulnerabilities and oversights.
- Release and call to Action
- Background
- Sample Reports
- PlumHound Examples
- Database Connection
- HTML Report Design Output and Variables
- TaskList Files
- Execution Modes
- Additional Modules
- Logging
- Hat-Tips & Acknowledgments
- Presentations
- Installation Requirements
- Environment Setup Instructions
- Known Issues
- Collaboration
- License
- Additional Inclusions:
The initial PlumHound code was released on May 14th, 2020 during a Black Hills Information Security webcast, A Blue Teams Perspective on Red Team Tools. The webcast was recorded and is available on YouTube. A Blue Team's Perspective on Red Team Tools.
The PlumHound Framework yields itself to community involvement in the creation and proliferation of "TaskLists" (work) that can be shared and used across different organizations. TaskLists contain jobs for PlumHound to do (queries to run, reports to write). A second PlumHound community repo has been created to allow for the open sharing of TaskLists (see Plumhound-Tasks)
Looking for more tasks and templates? Checkout PlumHound-Tasks for the community driven marketplace of PlumHound reporting taskslists and report designs
A client of ours working on hardening their Active Directory infrastructure asked us about vulnerabilities that can be found by using BloodHound. They had heard of the effectiveness of BloodHoundAD in Red-Team's hands and was told that BloodHound would identify all types of security mis-alignments and mis-configurations in their Active Directory environment. We helped them through analysis of their BloodHound dataset and it became quickly evident that BloodHoundAD's pathfinding graphical database was not designed for the fast-passed analytical security team accustom to reading reports and action items.
In fact, one of our cypher queries determined that 96% of their 3000 users had a path to Domain Admin with an average of just 4 steps. However, that graphical query rendered over 10,000 paths to Domain Admin. Finding the actual cause of the short-paths to DA wasn't as easy as just loading data into BloodHound or putting Cobalt Strike on Auto-Pilot with BloodHound Navigation.
Hence, PlumHound was created out of a need to retrieve consumable data from BloodHoundAD's pathfinding engine. Data that could yield itself to inferring actionable work for security teams to harden their environments.
We include some sample reports in this respository, however the data represents a fictious small Active Directory deployment atypical of a production environment. The default tasks tasklist will produce reports with ease. Here are a sample of some less-exciting reports generated from a small lab environment. A few samples of produced reports are shown below.
The archive of sample reports is found here: SampleReports
The default tasklist generates the following reports:
- Domains
- Domain Trusts
- Domain Controllers
- Domain Controllers - Read Only
- Enterprise Admins
- Schema Admins
- Domain Admins
- Admin Groups
- Admin Groups Direct Population
- Domain Users
- Domain Computers
- Domain Groups
- OUs By Computer Member Count
- OUs By User Member Count
- OUs By Group Member Count
- Cert Publishers
- DA Sessions
- EA Sessions
- User Sessions Count
- HighValue Group Members (Limited to 1000)
- Protected Users Group
- Admins Without Sensitive Protection Flag
- Kerberoastable Users
- Pre-Windows 2000 Compatibility Access Direct Members
- RDPable Servers
- Domain Controller Owners
- Domain Controller Owned by Users
- Unconstrained Delegation Users with SPN
- Unconstrained Delegation Computers with SPN
- Unconstrained Delegation Computers with SPN Non-DC
- Constrained Delegation - All
- Constrained Delegation - Users
- Constrained Delegation - Users NonDA
- Constrained Delegation - Computers NonDC
- DCSync Direct Relationships
- DCSync Privileged Non-DCs
- DCSync Privileged Non-DAs
- Permissions for Everyone and Authenticated Users
- Users (Non-Admin) With AddMember Group Permissions
- Direct Relationships - Everyone
- Direct Relationships - Authenticated Users
- Direct Relationships - Pre-Windows 2000 Compatibility Access
- Direct Relationships - Domain Users
- Direct Relationships - Domain Computers
- Direct Relationships - Users
- Direct Relationships - Guests
- PasswordResetter Groups Count
- GPOs
- GPO OU Links
- GPO Owners Summary Count
- GPO Owners Detail
- GPO Owners - Non-DAs
- GPOS - Non-DA Users With Interesting GPO Permissions
- GPO Creator Owners
- RDPable Groups
- RDPable Groups Count
- Computers Vulnerable to Local Admin Enumeration
- Local Admins - Groups
- Local Admins - Groups Count
- Local Admins - Users
- Local Admins - Users Count
- Local Admins - Computers
- Local Admins - Computers Count
- Computers with MSSQL SPN
- Computers with Descriptions
- Operating Systems Unsupported
- Operating Systems Unsupported Count
- Operating Systems Count
- LAPS Deployment Count
- LAPS Deployment OS Count
- LAPS Not Enabled
- Users Sessions
- Users Sessions Count
- Users With < 1 Day Old Passwords
- Users With < 7 Day Old Passwords
- Users With < 30 Day Old Passwords
- Users With > 6 Month Old Passwords
- Users With > 1 Year Old Passwords
- Users With > 5 Year Old Passwords
- Users With > 10 Year Old Passwords
- Users With > 15 Year Old Passwords
- Users With > 20 Year Old Passwords
- Users Password Never Expires Exception
- Users Password Not Required
- Users Password Not Required and Never Set
- Users With Userpassword Attribute
- User No Kerb Needed (AS-REP Roastable Users)
- Users Not Active 6 Months
- Users Not Active 12 Months
- Users Not Active 60 Months
- Users Not Active 120 Months
- Users Never Active
- Hunt - Users with Pass or PW in Description
- Hunt - Users with Change in Description
- Hunt - Users with VPN Group
- Hunt - Computers with Pass or PW in Description
- Owned Groups
- Owned Users
- Owned Computers
- Owned Users Groups
- Owned Computers Groups
- Owned Objects
- Owned Objects Admin To
- Owned Objects Member Of
- Certificate Templates
- Certificate Authorities
- Certificate Enrollment Rights
- Certificate Template ESC1 Vuln
- Certificate Template ESC2 Vuln
- Certificate Template ESC3 Vuln
- Certificate Template ESC6 Vuln
- Certificate Template ESC8 Vuln
- Report Index
Use the default username, password, server, and execute the "Easy" task, to test connectivity. This will output all Active Directory user objects from the Neo4J database.
python3 PlumHound.py --easy
Execute PlumHound with the Default TaskList using Default Credentials (neo4j:neo4jneo4j) and Database.
python3 PlumHound.py -x tasks/default.tasks
The same, but quiet the output (-v 0), specify the Neo4J server, username, and password instead of using defaults.
python3 PlumHound.py -x tasks/default.tasks -s "bolt://127.0.0.1:7687" -u "neo4j" -p "neo4jneo4j" -v 0
Using label. The supported labels are User
, Group
, Computer
, OU
and GPO
. This function will assume the target group is "DOMAIN ADMINS".
python3 PlumHound.py -ap user
NOTE: The above syntax implies you are using the default values for sever
, user
and password
or that you have hard-coded them in the script.
specify start node
and end node
python3 PlumHound.py -ap "domain [email protected]" "domain [email protected]"
NOTE: To use BlueHound Path Analyzer logic you need to get a copy of the Python script from https://github.com/scoubi/BlueHound
The Busiest Path(s) function takes two parameters
Find the shortest path that give users Domain Admin (top 5)
PlumHound.py -bp short 5
Find the busiest path irregardless of length (top 5) to give users Domain Admin
PlumHound.py -bp all 5
usage: PlumHound.py [-h] [-s SERVER] [-u USERNAME] [-p PASSWORD] [--UseEnc]
(--easy | -x TASKFILE | -q,--QuerySingle QUERYSINGLE | -bp,--BusiestPath BUSIESTPATH [BUSIESTPATH ...] | -ap,--AnalyzePath ANALYZEPATH [ANALYZEPATH ...])
[-t TITLE] [--of OUTFILE] [--op PATH] [--ox {stdout,HTML,CSV}] [--HTMLHeader HTMLHEADER] [--HTMLFooter HTMLFOOTER] [--HTMLCSS HTMLCSS]
[-v VERBOSE]
BloodHound Wrapper for Blue/Purple Teams; v01.070a
optional arguments:
-h, --help show this help message and exit
--easy Test Database Connection, Returns Domain Users to stdout
-x TASKFILE, --TaskFile TASKFILE
Specify a PlumHound TaskList File
-q,--QuerySingle QUERYSINGLE
Specify a Single Cypher Query
-bp,--BusiestPath BUSIESTPATH [BUSIESTPATH ...]
Find the X Shortest Paths that give the most users a path to Domain Admins. Need to specified [short|all] for shortestpath and the
number of results. Ex: PlumHound -bp all 3
-ap,--AnalyzePath ANALYZEPATH [ANALYZEPATH ...]
Analyze 'Attack Paths' between two nodes and find which path needs to be remediated to brake the path.
DATABASE:
-s SERVER, --server SERVER
Neo4J Server
-u USERNAME, --username USERNAME
Neo4J Database Useranme
-p PASSWORD, --password PASSWORD
Neo4J Database Password
--UseEnc Use encryption when connecting.
OUTPUT:
Output Options (For single cypher queries only. --These options are ignored when -x or --easy is specified.
-t TITLE, --title TITLE
Report Title for Single Query [HTML,CSV,Latex]
--of OUTFILE, --OutFile OUTFILE
Specify a Single Cypher Query
--op PATH, --OutPath PATH
Specify an Output Path for Reports
--ox {stdout,HTML,CSV}, --OutFormat {stdout,HTML,CSV}
Specify the type of output
HTML:
Options for HTML Output (For single queries or TaskLists
--HTMLHeader HTMLHEADER
HTML Header (file) of Report
--HTMLFooter HTMLFOOTER
HTML Footer (file) of Report
--HTMLCSS HTMLCSS Specify a CSS template for HTML Output
VERBOSESet verbosity:
-v VERBOSE, --verbose VERBOSE
Verbosity 0-1000, 0 = quiet
For more information see https://plumhound.DefensiveOrigins.com
PlumHound needs to connect to the Neo4J graphing database where BloodHoundAD data was loaded.
DATABASE:
-s SERVER, --server SERVER
Neo4J Server
-u USERNAME, --username USERNAME
Neo4J Database Useranme
-p PASSWORD, --password PASSWORD
Neo4J Database Password
--UseEnc Use encryption when connect
PlumHound paramters are set by default. You can override the default by including the argument.
Argument/Parameter | Default |
---|---|
SERVER | bolt://localhost:7687 |
USERNAME | neo4j |
PASSWORD | neo4jneo4j |
HTML output includes the ability to use HTML Headers, Footers, and CSS to modify the design of the report. Additionaly, variables can be added to the HTML Header and Footer files that are replaced at runtime.
HTML:
Options for HTML Output (For single queries or TaskLists
--HTMLHeader HTMLHEADER
HTML Header (file) of Report
--HTMLFooter HTMLFOOTER
HTML Footer (file) of Report
--HTMLCSS HTMLCSS Specify a CSS template for HTML Output
Argument/Parameter | Default |
---|---|
HTMLHeader | templates/head.html |
HTMLFooter | templates/tail.html |
HTMLCSS | templates/html.css |
Variable | Output |
---|---|
--------PH_TITLE------- | Report Tile from --Title or TaskList/Job |
--------PH_DATE------- | Python date.today() |
This allows the HTML output to be dynamic and tailored to your specification.
The PlumHound Repo includes a sample TaskList that exports some basic BloodHoundAD Cypher queries to an HTML Report. The included tasks\Default.tasks sample shows the basic syntax of the TaskList files. The TaskList Files allow PlumHound to be fully scripted with batch jobs after the SharpHound dataset has been imported not BloodHoundAD on Neo4j. Looking for more tasks and templates? Checkout PlumHound-Tasks for the community driven marketplace of PlumHound reporting taskslists and report designs.
The TaskList file syntax is as follows. Note that any cypher query containing a double quote must be modified to use a single quote instead of double.
["Report Title","[Output-Format]","[Output-File]","[CypherQuery]"]
Output-Format options for TaskLists:
- HTML: Generates a HTML Report
- CSV: Generates a CSV of data
- HTMLCSV: Generates both an HTML report and CSV data.
- NOTE: DO not include a file-extension in the Output-File paramter.
Additional Modules in TaskLists:
- Report Index: to use the Report-Index module in a TaskList, use CypherQuery "REPORT-INDEX"
The default.tasks file includes multiple tasks that instruct PlumHound to create reports using the specified "HTML" output format, output filename, and specific BloodHoundAD Neo4JS cypher Query. See the Tasks mode below in the Execution modes section for additional information.
PlumHound.py -x tasks/default.tasks
There are multiple execution modes to get PlumHound to do work.
- Tasks Mode
- Single Query Mode
- BusiestPath Mode (BlueHound Module)
- AnalyzePath Mode (BlueHound Module)
By specifying a task list, PlumHound can be programmed to run multiple cypher queries and export to multiple formats. For more information on the tasks file, see the section above.
This mode allows you to run a single query and export to standard output, HTML, or CSV.
PlumHound.py -q "MATCH (n:User) RETURN n.name, n.displayname LIMIT 10"
PlumHound 1.4
For more information: https://github.com/plumhound
--------------------------------------
Server: bolt://localhost:7687
User: neo4j
Password: *****
Encryption: False
Timeout: 300
--------------------------------------
Task: Single Query
Query Title: Adhoc Query
Query Format: STDOUT
Query Path: reports//
on 1:
on 1: n.name n.displayname
------------------------ ----------------
[email protected]
[email protected]
[email protected]
[email protected] HelpdeskUser
[email protected] HelpdeskUser
[email protected] FileServiceAdmin
[email protected] ITAdmin
[email protected] ITAdmin
[email protected] Silver Sandy
on 1:
Executing Tasks |██████████████████████████████████████████████████| Tasks 1 / 1 in 0.2s (4.40/s)
Completed 1 of 1 tasks.
By turning the verbosity to zero, this can act as a direct Neo4J handler for queries.
PlumHound.py -q "MATCH (n:User) RETURN n.name, n.displayname LIMIT 10" -v 0
on 1:
on 1: n.name n.displayname
------------------------ ----------------
[email protected]
[email protected]
[email protected]
[email protected] HelpdeskUser
[email protected] HelpdeskUser
[email protected] FileServiceAdmin
[email protected] ITAdmin
[email protected] ITAdmin
[email protected] Silver Sandy
on 1:
Executing Tasks |██████████████████████████████████████████████████| Tasks 1 / 1 in 0.2s (4.33/s)
The single query can also be used to generate HTML reports instead of outputting to STDOUT.
PlumHound.py -q "MATCH (n:User) RETURN n.name, n.displayname LIMIT 10" --title "Domain Users (Limit 10)" --of "DomainUserstest.html" --op "reports//" --ox HTML
The Busiest Path(s) function takes two parameters
1- all
or short
either you want to use shortestpath
or allshorteshpaths
algorithym.
2- The number of results you want to return. ex: Top 5
PlumHound.py -bp short 5
[*]Building Task List
[51, '[email protected]']
[51, '[email protected]']
[50, '[email protected]']
[49, '[email protected]']
[49, '[email protected]']
[*]Tasks Generation Completed
Tasks: []
The Analyze Path takes either a label
or a start node
and end node
and loop through all the paths finding which relationship(s) need to be broken in order to break the whole path. This is useful when you want to provide your AD Admins with concrete actions they can take in order to improuve your overall AD Security Posture.
PlumHound.py -ap group
[...]
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
Removing the relationship CanRDP between [email protected] and COMP00886.BTV.ORG breaks the path!
Removing the relationship HasSession between COMP00886.BTV.ORG and [email protected] breaks the path!
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
Removing the relationship ExecuteDCOM between [email protected] and COMP00629.BTV.ORG breaks the path!
Removing the relationship HasSession between COMP00629.BTV.ORG and [email protected] breaks the path!
Removing the relationship MemberOf between [email protected] and [email protected] breaks the path!
Removing the relationship MemberOf between [email protected] and [email protected] breaks the path!
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
Removing the relationship AdminTo between [email protected] and COMP01055.BTV.ORG breaks the path!
Removing the relationship HasSession between COMP01055.BTV.ORG and [email protected] breaks the path!
Removing the relationship MemberOf between [email protected] and [email protected] breaks the path!
Removing the relationship MemberOf between [email protected] and [email protected] breaks the path!
Removing the relationship AdminTo between [email protected] and COMP00658.BTV.ORG breaks the path!
Removing the relationship AllowedToDelegate between COMP00658.BTV.ORG and COMP01387.BTV.ORG breaks the path!
Removing the relationship AllowedToDelegate between COMP01387.BTV.ORG and COMP00275.BTV.ORG breaks the path!
Removing the relationship HasSession between COMP00275.BTV.ORG and [email protected] breaks the path!
Removing the relationship MemberOf between [email protected] and DOMAIN [email protected] breaks the path!
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
---------------------------------------------------------------------
Analyzing paths between [email protected] and DOMAIN [email protected]
---------------------------------------------------------------------
[...]
Valid pre-sets are ALL
, AZUser
, AZGroup
, AZApp
, AZServicePrincipal
, AZDevice
.
All presets search for relationships with Global Administrator
.
Plumhound.py -az ALL
[...]
Additional modules will be documented here.
The report indexer builds an HTML report of all the completed jobs. Add the following job to a task list. The parameter = "REPORT-INDEX" instructs PlumHound to generate an HTML index of all the successfully exported tasks in that run of tasks.
["Report Index","HTML","Reports.html","REPORT-INDEX"]
The Taskzipper module will create a zipfile of all of the reports generated from a tasklist. Be sure that the Zip Module is included as the last task in a last list. The paramter = "ZIP-TASKS" instructs PlumHound to generate a ZIP file of all the sucessfuly exported tatsks in that run of tasks.
["Zip Tasks","ZIP","Reports.zip","ZIP-TASKS"]
By default, PlumHound generates a log in file log\PlumHound.log
- Hausec's Cypher Query CheatSheet gave us a headstart on some decent pathfinding cypher queries. | Git
- SadProcessor's Blue Hands on BloodHound gave us a detailed primer on BloodHoundAD's ability to lead a BlueTeam to water. | Git.
- Additional work by SadProcessor with Cypher Dog 3.0 shows similar POC via utilizing BloodHoundAD's Cypher Queries with a RestAPI endpoint via PowerShell. PlumHound operates similarly however written in python and designed for stringing multiple queries into consumable reports designed to infer actionable items.
- BloodHoundAD: We wouldn't be talking about this at all if it weren't for the original BloodHoundAD work. BloodHound is developed by @_wald0, @CptJesus, and @harmj0y.
- "Band-aids don't fix dank domains." BadBlood saved us a ton of time building realistic-enough AD domains for testing. @davidprowe
- BloodHound from Red to Blue - Scoubi- Mathieu Saulnier Mathieu merged his BlueHound project with PlumHound in 2020 as a extension of features. Mathieu is an active collaborator of PlumHound and the BlueHound path-finding extension.
- BHIS Webcast: A Blue Team's Perspective on Red Team Hack Tools
- Sector Session: BlueHound Path Destroyer
- BHIS Blog: PlumHound Reporting Engine for BloodHoundAD
apt-get install python3
pip3 install -r requirements.txt
- Tested with Python 3.10.6 and 3.11.0rc1*
- Install Neo4J
- Install BloodhoundAD
- Import AD dataset into BloodhoundAD to be parsed
- Use PlumHound to Report
- Please report as issue on GitHub
Help PlumHound grow and be a great tool for Blue and Purple Teams. We've created the initial proof of concept and are committed to continuing the maturity of PlumHound to leverage the power of BloodHoundAD into continual security improvement processes. Community involvement is what makes this industry great!
- Black Hills Information Security - @BHInfoSecurity - Discord
- Defensive Origins - @DefensiveOGs - Git
- Kent Ickler - @Krelkci - Git
- Jordan Drysdale - @Rev10D - Git
- Mathieu Saulnier - @Scoubi - Git
- table-sort-js https://github.com/LeeWannacott/table-sort-js provides sorting for HTML reports (in templates/head.html).