The Travelling Pentester: Diaries of the Shortest Path to Compromise
•Download as PPTX, PDF•
8 likes•9,143 views
The document summarizes the work of three pentesters who use tools like BloodHound and PowerView to analyze attack paths within Active Directory environments. It describes how they use these tools to hunt for logged-in users, enumerate administrative privileges, discover group memberships, and analyze object ACLs to escalate privileges and compromise systems. It also discusses defenses organizations can implement and how increased endpoint telemetry could help detect and prevent their techniques.
Report
Share
The Travelling Pentester: Diaries of the Shortest Path to Compromise
- 1. The Travelling Pentester Diaries of the Shortest Path to Compromise
- 2. About Me I am Will Schroeder Job: “Offensive Engineer” at Veris Group’s ATD Co-Founder: Veil-Framework, PowerView, PowerUp, Empire/Empyre, BloodHound Trainer: Black Hat USA 2014-2017 Other: Microsoft PowerShell/CDM MVP Twitter: @harmj0y
- 3. The Bloodhound Gang Rohan Vazarkar Job: Pentester at Veris Group’s ATD Tool creator/dev: BloodHound, Python EmPyre Presenter: BSides, Black Hat Arsenal, DEF CON Trainer: Black Hat USA 2016-2017 Twitter: @CptJesus Andy Robbins Job: Pentest lead at Veris Group’s ATD Tool creator/dev: BloodHound Speaker: BSides, ISSA International, Black Hat Arsenal, DEF CON Trainer: Black Hat USA 2016-2017 Twitter: @_wald0
- 4. tl;dr
- 6. Our (Current) Ops ◇“Assume breach” approach ◇Lots of Active Directory and offensive PowerShell ◇Defenses are getting better- we’ve had to evolve!
- 7. “Fundamentally, if someone wants to get in, they’re getting in…accept that. What we tell clients is: Number one, you’re in the fight, whether you thought you were or not. Number two, you almost certainly are penetrated.” Michael Hayden Former Director of NSA & CIA Microsoft Enterprise Cloud Red Teaming Whitepaper
- 8. “Defenders think in lists. Attackers think in graphs. As long as this is true, attackers win.” John Lambert GM, Microsoft Threat Intelligence Center
- 10. BloodHound ◇Automates the attack path analysis process ◇Components: ￭ PowerShell ingestor ￭ neo4j backend ￭ Cross-platform electron app front end ◇Open source and BSD 3-clause licensed!
- 11. BloodHound Attack Graph Design Vertices represent users, groups, computers, and domains Edges identify group memberships, admin rights, user sessions, and now ACL relationships Paths always lead toward escalating rights. Always.
- 12. Who’s Logged in Where? aka “user-hunting” NetSessionEnum/NetWkstaUserEnum Attacker DC
- 13. Who’s Logged in Where? “Stealth” user-hunting Attacker DC File Server NetSessionEnum sessions sessions
- 14. Who’s Logged in Where? Defenses
- 15. Who’s Logged in Where? Defenses
- 16. Who Can Admin What?
- 17. Who Can Admin What? PowerView
- 18. Who Can Admin What? Defenses “Windows 10 had introduced an option to control the remote access to the SAM, through a specific registry value. On Windows Anniversary update (Windows 10 Version 16074) the default permissions were changed to allow remote access only to administrators.”
- 19. Who Can Admin What? GPO Edition Restricted Groups Group Policy Preferences Group Policy Object OU/site/ domain Contains Server WorkstationLocal Admins
- 20. Who’s in What Groups? ◇Enumerate all groups and extract the members of each ◇PowerView: ￭ Get-DomainGroup | Get-DomainGroupMember ◇BloodHound: ￭ Just pulls the member for all group objects
- 23. ◇Offline (ntds.dit) and some online DACL collection capabilities ◇Backend neo4j database allows for control flow discovery ◇Code released at https://github.com/ANSSI-FR/AD-control- paths Previous DACL Work
- 24. Who Has Rights Over What Objects? ◇By default, any user can enumerate all DACLs for all objects in the domain ￭ Through .NET methods or by specifying ntsecuritydescriptor in the LDAP query props ◇PowerView: Get-DomainObjectACL ◇BloodHound enumerates just the control relationships we care about
- 27. ◇Default Rights ￭ GenericAll - ALL THE RIGHTS ￭ GenericWrite - write all object properties ￭ WriteDacl - modify the DACL for the object ￭ WriteOwner - modify an object owner ￭ WriteProperty Self-Membership/Script-Path – modify group membership/user script path ◇Extended Rights ￭ User-Force-Change-Password BloodHound Currently Collected ACLs
- 28. ◇DS-Replication-Get-Changes-All ◇Modification rights to GPC-File-Sys-Path for GPOs ◇“Kerberoastable” accounts ◇Read rights to ms-MCS-AdmPwd BloodHound Future Collected ACLs
- 30. Case Studies (in Failure) Details have been changed to protect the innocent ;)
- 31. Case #1 1. Service binary rotated the local admin passwords monthly 2. .NET coded, predictable algorithm based on the date and hostname, no salt 3. Pulled apart app, build weaponized code, had admin access to every gold image system 4. Performed the ‘credential shuffle’ by hand with PowerView, took about 2 weeks Local Passwords Are Hard
- 32. Case #2 1. Kerberoasted 2 services accounts, allowing for access to a handful of systems 2. BloodHound analysis determined one user logged into one system we controlled had direct access to 5 systems, but derivative access to hundreds 3. Bonus: all user accounts had reversible encryption set 4. Elevated, hopped down the chain, DCSynced to recover ultimate target’s plaintext, grabbed the objective Kerberos is Hard As Well
- 33. Case #3 1. VULNERABLE SERVICE on terminal-type machines, allowed elevation 2. All terminal servers had the same (and enabled) local admin account 3. No formal trust, but correlated similar accounts between the two accessible domains 4. Developed GPO correlation technique on the engagement to hop to 2 cross-network targets 5. Group Policy Preferences in cross-network target, allowed compromise to a handful of machines GPP and GPOs and extra SIDs, Oh My
- 34. Case #3 6. Quick escalation to elevated domain rights 7. DCSynced to recover krbtgt of child domain 8. Hopped to child domain controller to build a Golden Ticket with extra SIDs 9. Injected and was able to hop up the trust and DCSync the corporate root domain GPP and GPOs and extra SIDs, Oh My
- 35. Sniffing Out ACLs with BloodHound Case #4
- 36. Sniffing Out ACLs with BloodHound Case #4
- 37. How it Could Have Been Prevented LAPS https://technet.microsoft.com/en-us/mt227395.aspx
- 38. How it Could Have Been Prevented https://technet.microsoft.com/en-us/windows-server-docs/security/securing-privileged-access/securing-privileged-access
- 39. How it Could Have Been Prevented ◇Managed service accounts ◇ATA ◇SAMRi10 / NetCease ◇Credential Guard ◇Red Forest Architecture ◇PowerUp ◇GET RID OF GPP ◇Separate forests to enforce trust ◇Centralized logging/analysis ◇Increased endpoint telemetry
- 40. How it Could Have Been Prevented DACLs ¯_(ツ)_/¯
- 41. How We Get Caught Our Biggest Pain Points
- 43. Endpoint Telemetry ◇Command line logging is a huge pain ￭ Many many attacker toolsets end up calling shell commands ◇Mining things like process tree traces at scale can give enormous insight ◇Windows Defender ATP, Sysmon, etc.
- 44. Closing Thoughts
- 45. Thank You! @harmj0y will [at] harmj0y.net blog.harmj0y.net