Sigma detection rules, expert-reviewed

Masquerading Firewall Allow Rule Masquerading as Windows Defender Update Service via Netsh (via process_creation)

This rule detects netsh adding an advfirewall rule named to impersonate a Windows Defender update service, a defense-evasion step in the Salat Stealer chain that whitelists its own network traffic. Adversaries leverage a benign-looking firewall rule name to allow C2 and exfiltration traffic while blending into legitimate configuration, so this behavior warrants review alongside surrounding activity.

Web Server Worker Process Spawning Command Interpreter

This rule detects the IIS worker process w3wp.exe spawning command shells or administrative utilities, behavior consistent with the China Chopper web shell used in Operation Soft Cell against telecommunications providers. A web server worker launching cmd, wmic or PowerShell strongly indicates web shell command execution.

Malicious Akira Ransomware Encryption Execution (via process_creation)

This rule detects the Akira ransomware encryptor invoked with its characteristic path and share targeting switches. These combined command-line flags drive selective encryption across local and network locations. Detection here indicates active encryption and imminent impact on the environment.

Malicious PowerShell AMSI Bypass via AmsiInitFailed Command Line

This rule detects the amsiInitFailed AMSI-bypass string in a PowerShell command line, a defense-evasion technique used after Citrix Bleed exploitation to disable script scanning before running the LockBit toolkit. Setting amsiInitFailed forces AMSI to skip inspection. Detecting the string exposes an active attempt to blind endpoint script scanning.

Malicious DLL Side-Loading via AVGApplicationFrameHost

This rule detects the AVGApplicationFrameHost.exe binary loading a wsc.dll module, a DLL side-loading chain used by TheWizards APT group to execute the WizardNet backdoor alongside an encrypted log.dat shellcode payload. Abusing a signed AV-related host process to load an attacker DLL evades trust controls and blends malicious execution into legitimate software.

Suspicious PowerShell Self-Delete Of Executable via Process Creation

This rule detects PowerShell used to delete an executable in a wait-and-retry loop. Rhysida ransomware removes its own binary after execution using a PowerShell routine that loops on Test-Path and Start-Sleep until the file is gone, eliminating forensic artifacts.

Suspicious Script Host Execution of JavaScript From a User-Writable Directory (via process_creation)

This rule detects wscript.exe or cscript.exe running a .js or .jse file from a user-writable location such as AppData or Temp, the delivery-and-execution pattern behind JavaScript loaders like Gootloader that arrive through SEO-poisoned downloads. Malicious script-host execution from user directories is a recurring initial-access and execution behavior in the Red Canary Threat Detection Report. Detecting it surfaces the loader stage before it fetches and runs a second-stage payload.

Malicious Logon Script Persistence via UserInitMprLogonScript

This rule detects the creation or modification of the UserInitMprLogonScript value under the user Environment key, a logon persistence technique observed in a martial-law-themed APT campaign that pointed this value at updater.exe to execute at each logon. It captures the abuse of a rarely used logon-script registry entry to survive reboots. Detecting this is important because UserInitMprLogonScript is almost never set by legitimate software and is a well-known persistence primitive.

Malicious Boot Recovery Disable via Bcdedit (via process_creation)

This rule detects bcdedit being used to disable Windows recovery or force boot into ignore-all-failures or safe mode. Ransomware alters the boot configuration so that automatic repair and recovery options cannot restore the system after encryption. Tampering with boot recovery settings from the command line is a strong impact indicator.

Malicious Fileless JavaScript Execution via Deno Data URI (via process_creation)

This rule detects the Deno runtime executing an inline base64 encoded JavaScript payload passed as a data application javascript URI, a fileless execution technique used by attackers abusing alternative runtime environments to evade disk based detection. The allow all flag combined with an inline data URI is a strong indicator of malicious in memory code execution.

Suspicious DLL Sideloading via rundll32 Loading MSADOCG (via process_creation)

This rule detects rundll32.exe loading MSADOCG.DLL, the sideloaded module used to launch the SugarGh0st RAT payload. Sideloading a maliciously named DLL through rundll32 hides code execution behind a signed Windows utility.

Suspicious Wininit Look-Alike Binary Execution from Windows Directory

This rule detects execution of a binary named to closely resemble the legitimate Windows initialization process which Lorenz ransomware uses for persistence from the Windows directory. Observed in NCC Group research unmasking Lorenz ransomware deploying a WinIniw.exe persistence binary. Typosquatting core system process names is a common masquerading tactic that warrants investigation.

Suspicious Active Directory Enumeration via Sysinternals AD Explorer Snapshot (via process_creation)

This rule detects Sysinternals AD Explorer being run with the accepteula and snapshot switches to dump an offline copy of Active Directory, a discovery step used by the RedCurl EarthKapre APT. A snapshot captures the full directory for later credential and target analysis. Automated AD Explorer snapshots outside a sanctioned administrative context indicate reconnaissance by an intruder.

Suspicious Vulnerable Driver EneIo Load for BYOVD (via image_load)

This rule detects the loading of the EneIo.sys driver, a known vulnerable driver abused for bring-your-own-vulnerable-driver privilege escalation. This behavior was observed in the updated MATA campaign against industrial companies to gain kernel-level access and disable defenses. Because a legitimate ENE Technology driver of this name also exists, the match should be triaged against expected motherboard and peripheral software.

Suspicious Environment File Credential Search via findstr (via process_creation)

This rule detects use of findstr or find to search for .env files, the credential-harvesting behavior of the InvisibleFerret backdoor which sweeps drives C through G for secrets. Programmatically hunting environment files across drives targets API keys and database credentials for theft.