Execution

MITRE ATT&CK – Execution

1.Kravtext (engelskt original)#

The adversary is trying to run malicious code.

Execution consists of techniques that result in adversary-controlled code running on a local or remote system. Techniques that run malicious code are often paired with techniques from all other tactics to achieve broader goals, like exploring a network or stealing data. For example, an adversary might use a remote access tool to run a PowerShell script that does Remote System Discovery.

T1047 – Windows Management Instrumentation#

Adversaries may abuse Windows Management Instrumentation (WMI) to execute malicious commands and payloads. WMI is designed for programmers and is the infrastructure for management data and operations on Windows systems.(Citation: WMI 1-3) WMI is an administration feature that provides a uniform environment to access Windows system components.

The WMI service enables both local and remote access, though the latter is facilitated by Remote Services such as Distributed Component Object Model and Windows Remote Management.(Citation: WMI 1-3) Remote WMI over DCOM operates using port 135, whereas WMI over WinRM operates over port 5985 when using HTTP and 5986 for HTTPS.(Citation: WMI 1-3) (Citation: Mandiant WMI)

An adversary can use WMI to interact with local and remote systems and use it as a means to execute various behaviors, such as gathering information for Discovery as well as Execution of commands and payloads.(Citation: Mandiant WMI) For example, wmic.exe can be abused by an adversary to delete shadow copies with the command wmic.exe Shadowcopy Delete (i.e., Inhibit System Recovery).(Citation: WMI 6)

Note: wmic.exe is deprecated as of January of 2024, with the WMIC feature being “disabled by default” on Windows 11+. WMIC will be removed from subsequent Windows releases and replaced by PowerShell as the primary WMI interface.(Citation: WMI 7,8) In addition to PowerShell and tools like wbemtool.exe, COM APIs can also be used to programmatically interact with WMI via C++, .NET, VBScript, etc.(Citation: WMI 7,8)

T1053 – Scheduled Task/Job#

Adversaries may abuse task scheduling functionality to facilitate initial or recurring execution of malicious code. Utilities exist within all major operating systems to schedule programs or scripts to be executed at a specified date and time. A task can also be scheduled on a remote system, provided the proper authentication is met (ex: RPC and file and printer sharing in Windows environments). Scheduling a task on a remote system typically may require being a member of an admin or otherwise privileged group on the remote system.(Citation: TechNet Task Scheduler Security)

Adversaries may use task scheduling to execute programs at system startup or on a scheduled basis for persistence. These mechanisms can also be abused to run a process under the context of a specified account (such as one with elevated permissions/privileges). Similar to System Binary Proxy Execution, adversaries have also abused task scheduling to potentially mask one-time execution under a trusted system process.(Citation: ProofPoint Serpent)

T1053.002 – At

Adversaries may abuse the at utility to perform task scheduling for initial or recurring execution of malicious code. The at utility exists as an executable within Windows, Linux, and macOS for scheduling tasks at a specified time and date. Although deprecated in favor of Scheduled Task’s schtasks in Windows environments, using at requires that the Task Scheduler service be running, and the user to be logged on as a member of the local Administrators group. In addition to explicitly running the at command, adversaries may also schedule a task with at by directly leveraging the Windows Management Instrumentation Win32_ScheduledJob WMI class.(Citation: Malicious Life by Cybereason)

On Linux and macOS, at may be invoked by the superuser as well as any users added to the <code>at.allow</code> file. If the <code>at.allow</code> file does not exist, the <code>at.deny</code> file is checked. Every username not listed in <code>at.deny</code> is allowed to invoke at. If the <code>at.deny</code> exists and is empty, global use of at is permitted. If neither file exists (which is often the baseline) only the superuser is allowed to use at.(Citation: Linux at)

Adversaries may use at to execute programs at system startup or on a scheduled basis for Persistence. at can also be abused to conduct remote Execution as part of Lateral Movement and/or to run a process under the context of a specified account (such as SYSTEM).

In Linux environments, adversaries may also abuse at to break out of restricted environments by using a task to spawn an interactive system shell or to run system commands. Similarly, at may also be used for Privilege Escalation if the binary is allowed to run as superuser via <code>sudo</code>.(Citation: GTFObins at)

T1053.003 – Cron

Adversaries may abuse the <code>cron</code> utility to perform task scheduling for initial or recurring execution of malicious code.(Citation: 20 macOS Common Tools and Techniques) The <code>cron</code> utility is a time-based job scheduler for Unix-like operating systems. The <code> crontab</code> file contains the schedule of cron entries to be run and the specified times for execution. Any <code>crontab</code> files are stored in operating system-specific file paths.

An adversary may use <code>cron</code> in Linux or Unix environments to execute programs at system startup or on a scheduled basis for Persistence. In ESXi environments, cron jobs must be created directly via the crontab file (e.g., /var/spool/cron/crontabs/root).(Citation: CloudSEK ESXiArgs 2023)

T1053.005 – Scheduled Task

Adversaries may abuse the Windows Task Scheduler to perform task scheduling for initial or recurring execution of malicious code. There are multiple ways to access the Task Scheduler in Windows. The schtasks utility can be run directly on the command line, or the Task Scheduler can be opened through the GUI within the Administrator Tools section of the Control Panel.(Citation: Stack Overflow) In some cases, adversaries have used a .NET wrapper for the Windows Task Scheduler, and alternatively, adversaries have used the Windows netapi32 library and Windows Management Instrumentation (WMI) to create a scheduled task. Adversaries may also utilize the Powershell Cmdlet Invoke-CimMethod, which leverages WMI class PS_ScheduledTask to create a scheduled task via an XML path.(Citation: Red Canary - Atomic Red Team)

An adversary may use Windows Task Scheduler to execute programs at system startup or on a scheduled basis for persistence. The Windows Task Scheduler can also be abused to conduct remote Execution as part of Lateral Movement and/or to run a process under the context of a specified account (such as SYSTEM). Similar to System Binary Proxy Execution, adversaries have also abused the Windows Task Scheduler to potentially mask one-time execution under signed/trusted system processes.(Citation: ProofPoint Serpent)

Adversaries may also create “hidden” scheduled tasks (i.e. Hide Artifacts) that may not be visible to defender tools and manual queries used to enumerate tasks. Specifically, an adversary may hide a task from schtasks /query and the Task Scheduler by deleting the associated Security Descriptor (SD) registry value (where deletion of this value must be completed using SYSTEM permissions).(Citation: SigmaHQ)(Citation: Tarrask scheduled task) Adversaries may also employ alternate methods to hide tasks, such as altering the metadata (e.g., Index value) within associated registry keys.(Citation: Defending Against Scheduled Task Attacks in Windows Environments)

T1053.006 – Systemd Timers

Adversaries may abuse systemd timers to perform task scheduling for initial or recurring execution of malicious code. Systemd timers are unit files with file extension <code>.timer</code> that control services. Timers can be set to run on a calendar event or after a time span relative to a starting point. They can be used as an alternative to Cron in Linux environments.(Citation: archlinux Systemd Timers Aug 2020) Systemd timers may be activated remotely via the <code>systemctl</code> command line utility, which operates over SSH.(Citation: Systemd Remote Control)

Each <code>.timer</code> file must have a corresponding <code>.service</code> file with the same name, e.g., <code>example.timer</code> and <code>example.service</code>. <code>.service</code> files are Systemd Service unit files that are managed by the systemd system and service manager.(Citation: Linux man-pages: systemd January 2014) Privileged timers are written to <code>/etc/systemd/system/</code> and <code>/usr/lib/systemd/system</code> while user level are written to <code>~/.config/systemd/user/</code>.

An adversary may use systemd timers to execute malicious code at system startup or on a scheduled basis for persistence.(Citation: Arch Linux Package Systemd Compromise BleepingComputer 10JUL2018)(Citation: gist Arch package compromise 10JUL2018)(Citation: acroread package compromised Arch Linux Mail 8JUL2018) Timers installed using privileged paths may be used to maintain root level persistence. Adversaries may also install user level timers to achieve user level persistence.(Citation: Falcon Sandbox smp: 28553b3a9d)

T1053.007 – Container Orchestration Job

Adversaries may abuse task scheduling functionality provided by container orchestration tools such as Kubernetes to schedule deployment of containers configured to execute malicious code. Container orchestration jobs run these automated tasks at a specific date and time, similar to cron jobs on a Linux system. Deployments of this type can also be configured to maintain a quantity of containers over time, automating the process of maintaining persistence within a cluster.

In Kubernetes, a CronJob may be used to schedule a Job that runs one or more containers to perform specific tasks.(Citation: Kubernetes Jobs)(Citation: Kubernetes CronJob) An adversary therefore may utilize a CronJob to schedule deployment of a Job that executes malicious code in various nodes within a cluster.(Citation: Threat Matrix for Kubernetes)

T1059 – Command and Scripting Interpreter#

Adversaries may abuse command and script interpreters to execute commands, scripts, or binaries. These interfaces and languages provide ways of interacting with computer systems and are a common feature across many different platforms. Most systems come with some built-in command-line interface and scripting capabilities, for example, macOS and Linux distributions include some flavor of Unix Shell while Windows installations include the Windows Command Shell and PowerShell.

There are also cross-platform interpreters such as Python, as well as those commonly associated with client applications such as JavaScript and Visual Basic.

Adversaries may abuse these technologies in various ways as a means of executing arbitrary commands. Commands and scripts can be embedded in Initial Access payloads delivered to victims as lure documents or as secondary payloads downloaded from an existing C2. Adversaries may also execute commands through interactive terminals/shells, as well as utilize various Remote Services in order to achieve remote Execution.(Citation: Powershell Remote Commands)(Citation: Cisco IOS Software Integrity Assurance - Command History)(Citation: Remote Shell Execution in Python)

T1059.001 – PowerShell

Adversaries may abuse PowerShell commands and scripts for execution. PowerShell is a powerful interactive command-line interface and scripting environment included in the Windows operating system.(Citation: TechNet PowerShell) Adversaries can use PowerShell to perform a number of actions, including discovery of information and execution of code. Examples include the <code>Start-Process</code> cmdlet which can be used to run an executable and the <code>Invoke-Command</code> cmdlet which runs a command locally or on a remote computer (though administrator permissions are required to use PowerShell to connect to remote systems).

PowerShell may also be used to download and run executables from the Internet, which can be executed from disk or in memory without touching disk.

A number of PowerShell-based offensive testing tools are available, including Empire, PowerSploit, PoshC2, and PSAttack.(Citation: Github PSAttack)

PowerShell commands/scripts can also be executed without directly invoking the <code>powershell.exe</code> binary through interfaces to PowerShell’s underlying <code>System.Management.Automation</code> assembly DLL exposed through the .NET framework and Windows Common Language Interface (CLI).(Citation: Sixdub PowerPick Jan 2016)(Citation: SilentBreak Offensive PS Dec 2015)(Citation: Microsoft PSfromCsharp APR 2014)

T1059.002 – AppleScript

Adversaries may abuse AppleScript for execution. AppleScript is a macOS scripting language designed to control applications and parts of the OS via inter-application messages called AppleEvents.(Citation: Apple AppleScript) These AppleEvent messages can be sent independently or easily scripted with AppleScript. These events can locate open windows, send keystrokes, and interact with almost any open application locally or remotely.

Scripts can be run from the command-line via <code>osascript /path/to/script</code> or <code>osascript -e “script here”</code>. Aside from the command line, scripts can be executed in numerous ways including Mail rules, Calendar.app alarms, and Automator workflows. AppleScripts can also be executed as plain text shell scripts by adding <code>#!/usr/bin/osascript</code> to the start of the script file.(Citation: SentinelOne AppleScript)

AppleScripts do not need to call <code>osascript</code> to execute. However, they may be executed from within mach-O binaries by using the macOS Native APIs <code>NSAppleScript</code> or <code>OSAScript</code>, both of which execute code independent of the <code>/usr/bin/osascript</code> command line utility.

Adversaries may abuse AppleScript to execute various behaviors, such as interacting with an open SSH connection, moving to remote machines, and even presenting users with fake dialog boxes. These events cannot start applications remotely (they can start them locally), but they can interact with applications if they’re already running remotely. On macOS 10.10 Yosemite and higher, AppleScript has the ability to execute Native APIs, which otherwise would require compilation and execution in a mach-O binary file format.(Citation: SentinelOne macOS Red Team) Since this is a scripting language, it can be used to launch more common techniques as well such as a reverse shell via Python.(Citation: Macro Malware Targets Macs)

T1059.003 – Windows Command Shell

Adversaries may abuse the Windows command shell for execution. The Windows command shell (cmd) is the primary command prompt on Windows systems. The Windows command prompt can be used to control almost any aspect of a system, with various permission levels required for different subsets of commands. The command prompt can be invoked remotely via Remote Services such as SSH.(Citation: SSH in Windows)

Batch files (ex: .bat or .cmd) also provide the shell with a list of sequential commands to run, as well as normal scripting operations such as conditionals and loops. Common uses of batch files include long or repetitive tasks, or the need to run the same set of commands on multiple systems.

Adversaries may leverage cmd to execute various commands and payloads. Common uses include cmd to execute a single command, or abusing cmd interactively with input and output forwarded over a command and control channel.

T1059.004 – Unix Shell

Adversaries may abuse Unix shell commands and scripts for execution. Unix shells are the primary command prompt on Linux, macOS, and ESXi systems, though many variations of the Unix shell exist (e.g. sh, ash, bash, zsh, etc.) depending on the specific OS or distribution.(Citation: DieNet Bash)(Citation: Apple ZShell) Unix shells can control every aspect of a system, with certain commands requiring elevated privileges.

Unix shells also support scripts that enable sequential execution of commands as well as other typical programming operations such as conditionals and loops. Common uses of shell scripts include long or repetitive tasks, or the need to run the same set of commands on multiple systems.

Adversaries may abuse Unix shells to execute various commands or payloads. Interactive shells may be accessed through command and control channels or during lateral movement such as with SSH. Adversaries may also leverage shell scripts to deliver and execute multiple commands on victims or as part of payloads used for persistence.

Some systems, such as embedded devices, lightweight Linux distributions, and ESXi servers, may leverage stripped-down Unix shells via Busybox, a small executable that contains a variety of tools, including a simple shell.

T1059.005 – Visual Basic

Adversaries may abuse Visual Basic (VB) for execution. VB is a programming language created by Microsoft with interoperability with many Windows technologies such as Component Object Model and the Native API through the Windows API. Although tagged as legacy with no planned future evolutions, VB is integrated and supported in the .NET Framework and cross-platform .NET Core.(Citation: VB .NET Mar 2020)(Citation: VB Microsoft)

Derivative languages based on VB have also been created, such as Visual Basic for Applications (VBA) and VBScript. VBA is an event-driven programming language built into Microsoft Office, as well as several third-party applications.(Citation: Microsoft VBA)(Citation: Wikipedia VBA) VBA enables documents to contain macros used to automate the execution of tasks and other functionality on the host. VBScript is a default scripting language on Windows hosts and can also be used in place of JavaScript on HTML Application (HTA) webpages served to Internet Explorer (though most modern browsers do not come with VBScript support).(Citation: Microsoft VBScript)

Adversaries may use VB payloads to execute malicious commands. Common malicious usage includes automating execution of behaviors with VBScript or embedding VBA content into Spearphishing Attachment payloads (which may also involve Mark-of-the-Web Bypass to enable execution).(Citation: Default VBS macros Blocking )

T1059.006 – Python

Adversaries may abuse Python commands and scripts for execution. Python is a very popular scripting/programming language, with capabilities to perform many functions. Python can be executed interactively from the command-line (via the <code>python.exe</code> interpreter) or via scripts (.py) that can be written and distributed to different systems. Python code can also be compiled into binary executables.(Citation: Zscaler APT31 Covid-19 October 2020)

Python comes with many built-in packages to interact with the underlying system, such as file operations and device I/O. Adversaries can use these libraries to download and execute commands or other scripts as well as perform various malicious behaviors.

T1059.007 – JavaScript

Adversaries may abuse various implementations of JavaScript for execution. JavaScript (JS) is a platform-independent scripting language (compiled just-in-time at runtime) commonly associated with scripts in webpages, though JS can be executed in runtime environments outside the browser.(Citation: NodeJS)

JScript is the Microsoft implementation of the same scripting standard. JScript is interpreted via the Windows Script engine and thus integrated with many components of Windows such as the Component Object Model and Internet Explorer HTML Application (HTA) pages.(Citation: JScrip May 2018)(Citation: Microsoft JScript 2007)(Citation: Microsoft Windows Scripts)

JavaScript for Automation (JXA) is a macOS scripting language based on JavaScript, included as part of Apple’s Open Scripting Architecture (OSA), that was introduced in OSX 10.10. Apple’s OSA provides scripting capabilities to control applications, interface with the operating system, and bridge access into the rest of Apple’s internal APIs. As of OSX 10.10, OSA only supports two languages, JXA and AppleScript. Scripts can be executed via the command line utility <code>osascript</code>, they can be compiled into applications or script files via <code>osacompile</code>, and they can be compiled and executed in memory of other programs by leveraging the OSAKit Framework.(Citation: Apple About Mac Scripting 2016)(Citation: SpecterOps JXA 2020)(Citation: SentinelOne macOS Red Team)(Citation: Red Canary Silver Sparrow Feb2021)(Citation: MDSec macOS JXA and VSCode)

Adversaries may abuse various implementations of JavaScript to execute various behaviors. Common uses include hosting malicious scripts on websites as part of a Drive-by Compromise or downloading and executing these script files as secondary payloads. Since these payloads are text-based, it is also very common for adversaries to obfuscate their content as part of Obfuscated Files or Information.

T1059.008 – Network Device CLI

Adversaries may abuse scripting or built-in command line interpreters (CLI) on network devices to execute malicious command and payloads. The CLI is the primary means through which users and administrators interact with the device in order to view system information, modify device operations, or perform diagnostic and administrative functions. CLIs typically contain various permission levels required for different commands.

Scripting interpreters automate tasks and extend functionality beyond the command set included in the network OS. The CLI and scripting interpreter are accessible through a direct console connection, or through remote means, such as telnet or SSH.

Adversaries can use the network CLI to change how network devices behave and operate. The CLI may be used to manipulate traffic flows to intercept or manipulate data, modify startup configuration parameters to load malicious system software, or to disable security features or logging to avoid detection.(Citation: Cisco Synful Knock Evolution)

T1059.009 – Cloud API

Adversaries may abuse cloud APIs to execute malicious commands. APIs available in cloud environments provide various functionalities and are a feature-rich method for programmatic access to nearly all aspects of a tenant. These APIs may be utilized through various methods such as command line interpreters (CLIs), in-browser Cloud Shells, PowerShell modules like Azure for PowerShell(Citation: Microsoft - Azure PowerShell), or software developer kits (SDKs) available for languages such as Python.

Cloud API functionality may allow for administrative access across all major services in a tenant such as compute, storage, identity and access management (IAM), networking, and security policies.

With proper permissions (often via use of credentials such as Application Access Token and Web Session Cookie), adversaries may abuse cloud APIs to invoke various functions that execute malicious actions. For example, CLI and PowerShell functionality may be accessed through binaries installed on cloud-hosted or on-premises hosts or accessed through a browser-based cloud shell offered by many cloud platforms (such as AWS, Azure, and GCP). These cloud shells are often a packaged unified environment to use CLI and/or scripting modules hosted as a container in the cloud environment.

T1059.010 – AutoHotKey & AutoIT

Adversaries may execute commands and perform malicious tasks using AutoIT and AutoHotKey automation scripts. AutoIT and AutoHotkey (AHK) are scripting languages that enable users to automate Windows tasks. These automation scripts can be used to perform a wide variety of actions, such as clicking on buttons, entering text, and opening and closing programs.(Citation: AutoIT)(Citation: AutoHotKey)

Adversaries may use AHK (.ahk) and AutoIT (.au3) scripts to execute malicious code on a victim’s system. For example, adversaries have used for AHK to execute payloads and other modular malware such as keyloggers. Adversaries have also used custom AHK files containing embedded malware as Phishing payloads.(Citation: Splunk DarkGate)

These scripts may also be compiled into self-contained executable payloads (.exe).(Citation: AutoIT)(Citation: AutoHotKey)

T1059.011 – Lua

Adversaries may abuse Lua commands and scripts for execution. Lua is a cross-platform scripting and programming language primarily designed for embedded use in applications. Lua can be executed on the command-line (through the stand-alone lua interpreter), via scripts (<code>.lua</code>), or from Lua-embedded programs (through the <code>struct lua_State</code>).(Citation: Lua main page)(Citation: Lua state)

Lua scripts may be executed by adversaries for malicious purposes. Adversaries may incorporate, abuse, or replace existing Lua interpreters to allow for malicious Lua command execution at runtime.(Citation: PoetRat Lua)(Citation: Lua Proofpoint Sunseed)(Citation: Cyphort EvilBunny)(Citation: Kaspersky Lua)

T1059.012 – Hypervisor CLI

Adversaries may abuse hypervisor command line interpreters (CLIs) to execute malicious commands. Hypervisor CLIs typically enable a wide variety of functionality for managing both the hypervisor itself and the guest virtual machines it hosts.

For example, on ESXi systems, tools such as esxcli and vim-cmd allow administrators to configure firewall rules and log forwarding on the hypervisor, list virtual machines, start and stop virtual machines, and more.(Citation: Broadcom ESXCLI Reference)(Citation: Crowdstrike Hypervisor Jackpotting Pt 2 2021)(Citation: LOLESXi) Adversaries may be able to leverage these tools in order to support further actions, such as File and Directory Discovery or Data Encrypted for Impact.

T1059.013 – Container CLI/API

Adversaries may abuse built-in CLI tools or API calls to execute malicious commands in containerized environments.

The Docker CLI is used for managing containers via an exposed API point from the dockerd daemon. Some common examples of Docker CLI include Docker Desktop CLI and Docker Compose, but users are also able to use SDKs to interact with the API. For example, Docker SDK for Python can be used to run commands within a Python application.(Citation: Docker Desktop CLI)

Adversaries may leverage the Docker CLI, API, or SDK to pull or build Docker images (i.e., Ingress Tool Transfer, Build Image on Host), run containers (i.e., Deploy Container), or execute commands inside running containers (i.e., Container Administration Command). In some cases, threat actors may pull legitimate images that include scripts or tools that they can leverage - for example, using an image that includes the curl command to download payloads.(Citation: Intezer) Adversaries may also utilize docker inspect and docker ps to scan for cloud environment variables and other running containers (i.e., Container and Resource Discovery).(Citation: Cisco Talos Blog)(Citation: aquasec)

Kubernetes is responsible for the management and orchestration of containers across clusters. The Kubernetes control plane, which manages the state of the cluster and is responsible for scheduling, communication, and resource monitoring, can be invoked directly via the API or indirectly via CLI tools such as kubectl. It may also be accessed within client libraries such as Go or Python. By utilizing the API, administrators can interact with resources within the cluster such as listing or creating pods, which is a group of one or more containers. Adversaries call the API server via curl or other tools, allowing them to obtain further information about the environment such as pods, deployments, daemonsets, namespaces, or sysvars.(Citation: aquasec) They may also run various commands regarding resource management.

T1072 – Software Deployment Tools#

Adversaries may gain access to and use centralized software suites installed within an enterprise to execute commands and move laterally through the network. Configuration management and software deployment applications may be used in an enterprise network or cloud environment for routine administration purposes. These systems may also be integrated into CI/CD pipelines. Examples of such solutions include: SCCM, HBSS, Altiris, AWS Systems Manager, Microsoft Intune, Azure Arc, and GCP Deployment Manager.

Access to network-wide or enterprise-wide endpoint management software may enable an adversary to achieve remote code execution on all connected systems. The access may be used to laterally move to other systems, gather information, or cause a specific effect, such as wiping the hard drives on all endpoints.

SaaS-based configuration management services may allow for broad Cloud Administration Command on cloud-hosted instances, as well as the execution of arbitrary commands on on-premises endpoints. For example, Microsoft Configuration Manager allows Global or Intune Administrators to run scripts as SYSTEM on on-premises devices joined to Entra ID.(Citation: SpecterOps Lateral Movement from Azure to On-Prem AD 2020) Such services may also utilize Web Protocols to communicate back to adversary owned infrastructure.(Citation: Mitiga Security Advisory: SSM Agent as Remote Access Trojan)

Network infrastructure devices may also have configuration management tools that can be similarly abused by adversaries.(Citation: Fortinet Zero-Day and Custom Malware Used by Suspected Chinese Actor in Espionage Operation)

The permissions required for this action vary by system configuration; local credentials may be sufficient with direct access to the third-party system, or specific domain credentials may be required. However, the system may require an administrative account to log in or to access specific functionality.

T1106 – Native API#

Adversaries may interact with the native OS application programming interface (API) to execute behaviors. Native APIs provide a controlled means of calling low-level OS services within the kernel, such as those involving hardware/devices, memory, and processes.(Citation: NT API Windows)(Citation: Linux Kernel API) These native APIs are leveraged by the OS during system boot (when other system components are not yet initialized) as well as carrying out tasks and requests during routine operations.

Adversaries may abuse these OS API functions as a means of executing behaviors. Similar to Command and Scripting Interpreter, the native API and its hierarchy of interfaces provide mechanisms to interact with and utilize various components of a victimized system.

Native API functions (such as <code>NtCreateProcess</code>) may be directed invoked via system calls / syscalls, but these features are also often exposed to user-mode applications via interfaces and libraries.(Citation: OutFlank System Calls)(Citation: CyberBit System Calls)(Citation: MDSec System Calls) For example, functions such as the Windows API <code>CreateProcess()</code> or GNU <code>fork()</code> will allow programs and scripts to start other processes.(Citation: Microsoft CreateProcess)(Citation: GNU Fork) This may allow API callers to execute a binary, run a CLI command, load modules, etc. as thousands of similar API functions exist for various system operations.(Citation: Microsoft Win32)(Citation: LIBC)(Citation: GLIBC)

Higher level software frameworks, such as Microsoft .NET and macOS Cocoa, are also available to interact with native APIs. These frameworks typically provide language wrappers/abstractions to API functionalities and are designed for ease-of-use/portability of code.(Citation: Microsoft NET)(Citation: Apple Core Services)(Citation: MACOS Cocoa)(Citation: macOS Foundation)

Adversaries may use assembly to directly or in-directly invoke syscalls in an attempt to subvert defensive sensors and detection signatures such as user mode API-hooks.(Citation: Redops Syscalls) Adversaries may also attempt to tamper with sensors and defensive tools associated with API monitoring, such as unhooking monitored functions via Disable or Modify Tools.

T1127 – Trusted Developer Utilities Proxy Execution#

Adversaries may take advantage of trusted developer utilities to proxy execution of malicious payloads. There are many utilities used for software development related tasks that can be used to execute code in various forms to assist in development, debugging, and reverse engineering.(Citation: engima0x3 DNX Bypass)(Citation: engima0x3 RCSI Bypass)(Citation: Exploit Monday WinDbg)(Citation: LOLBAS Tracker) These utilities may often be signed with legitimate certificates that allow them to execute on a system and proxy execution of malicious code through a trusted process that effectively bypasses application control solutions.

Smart App Control is a feature of Windows that blocks applications it considers potentially malicious from running by verifying unsigned applications against a known safe list from a Microsoft cloud service before executing them.(Citation: Microsoft Smart App Control) However, adversaries may leverage “reputation hijacking” to abuse an operating system’s trust of safe, signed applications that support the execution of arbitrary code. By leveraging Trusted Developer Utilities Proxy Execution to run their malicious code, adversaries may bypass Smart App Control protections.(Citation: Elastic Security Labs)

T1127.001 – MSBuild

Adversaries may use MSBuild to proxy execution of code through a trusted Windows utility. MSBuild.exe (Microsoft Build Engine) is a software build platform used by Visual Studio. It handles XML formatted project files that define requirements for loading and building various platforms and configurations.(Citation: MSDN MSBuild)

Adversaries can abuse MSBuild to proxy execution of malicious code. The inline task capability of MSBuild that was introduced in .NET version 4 allows for C# or Visual Basic code to be inserted into an XML project file.(Citation: MSDN MSBuild)(Citation: Microsoft MSBuild Inline Tasks 2017) MSBuild will compile and execute the inline task. MSBuild.exe is a signed Microsoft binary, so when it is used this way it can execute arbitrary code and bypass application control defenses that are configured to allow MSBuild.exe execution.(Citation: LOLBAS Msbuild)

T1127.002 – ClickOnce

Adversaries may use ClickOnce applications (.appref-ms and .application files) to proxy execution of code through a trusted Windows utility.(Citation: Burke/CISA ClickOnce BlackHat) ClickOnce is a deployment that enables a user to create self-updating Windows-based .NET applications (i.e, .XBAP, .EXE, or .DLL) that install and run from a file share or web page with minimal user interaction. The application launches as a child process of DFSVC.EXE, which is responsible for installing, launching, and updating the application.(Citation: SpectorOps Medium ClickOnce)

Because ClickOnce applications receive only limited permissions, they do not require administrative permissions to install.(Citation: Microsoft Learn ClickOnce) As such, adversaries may abuse ClickOnce to proxy execution of malicious code without needing to escalate privileges.

ClickOnce may be abused in a number of ways. For example, an adversary may rely on User Execution. When a user visits a malicious website, the .NET malware is disguised as legitimate software and a ClickOnce popup is displayed for installation.(Citation: NetSPI ClickOnce)

Adversaries may also abuse ClickOnce to execute malware via a Rundll32 script using the command rundll32.exe dfshim.dll,ShOpenVerbApplication1.(Citation: LOLBAS /Dfsvc.exe)

Additionally, an adversary can move the ClickOnce application file to a remote user’s startup folder for continued malicious code deployment (i.e., Registry Run Keys / Startup Folder).(Citation: Burke/CISA ClickOnce BlackHat)(Citation: Burke/CISA ClickOnce Paper)

T1127.003 – JamPlus

Adversaries may use JamPlus to proxy the execution of a malicious script. JamPlus is a build utility tool for code and data build systems. It works with several popular compilers and can be used for generating workspaces in code editors such as Visual Studio.(Citation: JamPlus manual)

Adversaries may abuse the JamPlus build utility to execute malicious scripts via a .jam file, which describes the build process and required dependencies. Because the malicious script is executed from a reputable developer tool, it may subvert application control security systems such as Smart App Control.(Citation: Cyble)(Citation: Elastic Security Labs)

T1129 – Shared Modules#

Adversaries may execute malicious payloads via loading shared modules. Shared modules are executable files that are loaded into processes to provide access to reusable code, such as specific custom functions or invoking OS API functions (i.e., Native API).

Adversaries may use this functionality as a way to execute arbitrary payloads on a victim system. For example, adversaries can modularize functionality of their malware into shared objects that perform various functions such as managing C2 network communications or execution of specific actions on objective.

The Linux & macOS module loader can load and execute shared objects from arbitrary local paths. This functionality resides in dlfcn.h in functions such as dlopen and dlsym. Although macOS can execute .so files, common practice uses .dylib files.(Citation: Apple Dev Dynamic Libraries)(Citation: Linux Shared Libraries)(Citation: RotaJakiro 2021 netlab360 analysis)(Citation: Unit42 OceanLotus 2017)

The Windows module loader can be instructed to load DLLs from arbitrary local paths and arbitrary Universal Naming Convention (UNC) network paths. This functionality resides in NTDLL.dll and is part of the Windows Native API which is called from functions like LoadLibrary at run time.(Citation: Microsoft DLL)

T1197 – BITS Jobs#

Adversaries may abuse BITS jobs to persistently execute code and perform various background tasks. Windows Background Intelligent Transfer Service (BITS) is a low-bandwidth, asynchronous file transfer mechanism exposed through Component Object Model (COM).(Citation: Microsoft COM)(Citation: Microsoft BITS) BITS is commonly used by updaters, messengers, and other applications preferred to operate in the background (using available idle bandwidth) without interrupting other networked applications. File transfer tasks are implemented as BITS jobs, which contain a queue of one or more file operations.

The interface to create and manage BITS jobs is accessible through PowerShell and the BITSAdmin tool.(Citation: Microsoft BITS)(Citation: Microsoft BITSAdmin)

Adversaries may abuse BITS to download (e.g. Ingress Tool Transfer), execute, and even clean up after running malicious code (e.g. Indicator Removal). BITS tasks are self-contained in the BITS job database, without new files or registry modifications, and often permitted by host firewalls.(Citation: CTU BITS Malware June 2016)(Citation: Mondok Windows PiggyBack BITS May 2007)(Citation: Symantec BITS May 2007) BITS enabled execution may also enable persistence by creating long-standing jobs (the default maximum lifetime is 90 days and extendable) or invoking an arbitrary program when a job completes or errors (including after system reboots).(Citation: PaloAlto UBoatRAT Nov 2017)(Citation: CTU BITS Malware June 2016)

BITS upload functionalities can also be used to perform Exfiltration Over Alternative Protocol.(Citation: CTU BITS Malware June 2016)

T1203 – Exploitation for Client Execution#

Adversaries may exploit software vulnerabilities in client applications to execute code. Vulnerabilities can exist in software due to unsecure coding practices that can lead to unanticipated behavior. Adversaries can take advantage of certain vulnerabilities through targeted exploitation for the purpose of arbitrary code execution. Oftentimes the most valuable exploits to an offensive toolkit are those that can be used to obtain code execution on a remote system because they can be used to gain access to that system. Users will expect to see files related to the applications they commonly used to do work, so they are a useful target for exploit research and development because of their high utility.

Several types exist:

Browser-based Exploitation#

Web browsers are a common target through Drive-by Compromise and Spearphishing Link. Endpoint systems may be compromised through normal web browsing or from certain users being targeted by links in spearphishing emails to adversary controlled sites used to exploit the web browser. These often do not require an action by the user for the exploit to be executed.

Office Applications#

Common office and productivity applications such as Microsoft Office are also targeted through Phishing. Malicious files will be transmitted directly as attachments or through links to download them. These require the user to open the document or file for the exploit to run.

Common Third-party Applications#

Other applications that are commonly seen or are part of the software deployed in a target network may also be used for exploitation. Applications such as Adobe Reader and Flash, which are common in enterprise environments, have been routinely targeted by adversaries attempting to gain access to systems. Depending on the software and nature of the vulnerability, some may be exploited in the browser or require the user to open a file. For instance, some Flash exploits have been delivered as objects within Microsoft Office documents.

T1204 – User Execution#

An adversary may rely upon specific actions by a user in order to gain execution. Users may be subjected to social engineering to get them to execute malicious code by, for example, opening a malicious document file or link. These user actions will typically be observed as follow-on behavior from forms of Phishing.

While User Execution frequently occurs shortly after Initial Access it may occur at other phases of an intrusion, such as when an adversary places a file in a shared directory or on a user’s desktop hoping that a user will click on it. This activity may also be seen shortly after Internal Spearphishing.

Adversaries may also deceive users into performing actions such as:

• Enabling Remote Access Tools, allowing direct control of the system to the adversary
• Running malicious JavaScript in their browser, allowing adversaries to Steal Web Session Cookies(Citation: Talos Roblox Scam 2023)(Citation: Krebs Discord Bookmarks 2023)
• Downloading and executing malware for User Execution
• Coerceing users to copy, paste, and execute malicious code manually(Citation: Reliaquest-execution)(Citation: proofpoint-selfpwn)

For example, tech support scams can be facilitated through Phishing, vishing, or various forms of user interaction. Adversaries can use a combination of these methods, such as spoofing and promoting toll-free numbers or call centers that are used to direct victims to malicious websites, to deliver and execute payloads containing malware or Remote Access Tools.(Citation: Telephone Attack Delivery)

T1204.001 – Malicious Link

An adversary may rely upon a user clicking a malicious link in order to gain execution. Users may be subjected to social engineering to get them to click on a link that will lead to code execution. This user action will typically be observed as follow-on behavior from Spearphishing Link. Clicking on a link may also lead to other execution techniques such as exploitation of a browser or application vulnerability via Exploitation for Client Execution. Links may also lead users to download files that require execution via Malicious File.

T1204.002 – Malicious File

An adversary may rely upon a user opening a malicious file in order to gain execution. Users may be subjected to social engineering to get them to open a file that will lead to code execution. This user action will typically be observed as follow-on behavior from Spearphishing Attachment. Adversaries may use several types of files that require a user to execute them, including .doc, .pdf, .xls, .rtf, .scr, .exe, .lnk, .pif, .cpl, .reg, and .iso.(Citation: Mandiant Trojanized Windows 10)

Adversaries may employ various forms of Masquerading and Obfuscated Files or Information to increase the likelihood that a user will open and successfully execute a malicious file. These methods may include using a familiar naming convention and/or password protecting the file and supplying instructions to a user on how to open it.(Citation: Password Protected Word Docs)

While Malicious File frequently occurs shortly after Initial Access it may occur at other phases of an intrusion, such as when an adversary places a file in a shared directory or on a user’s desktop hoping that a user will click on it. This activity may also be seen shortly after Internal Spearphishing.

T1204.003 – Malicious Image

Adversaries may rely on a user running a malicious image to facilitate execution. Amazon Web Services (AWS) Amazon Machine Images (AMIs), Google Cloud Platform (GCP) Images, and Azure Images as well as popular container runtimes such as Docker can be backdoored. Backdoored images may be uploaded to a public repository via Upload Malware, and users may then download and deploy an instance or container from the image without realizing the image is malicious, thus bypassing techniques that specifically achieve Initial Access. This can lead to the execution of malicious code, such as code that executes cryptocurrency mining, in the instance or container.(Citation: Summit Route Malicious AMIs)

Adversaries may also name images a certain way to increase the chance of users mistakenly deploying an instance or container from the image (ex: Match Legitimate Resource Name or Location).(Citation: Aqua Security Cloud Native Threat Report June 2021)

T1204.004 – Malicious Copy and Paste

An adversary may rely upon a user copying and pasting code in order to gain execution. Users may be subjected to social engineering to get them to copy and paste code directly into a Command and Scripting Interpreter. One such strategy is “ClickFix,” in which adversaries present users with seemingly helpful solutions—such as prompts to fix errors or complete CAPTCHAs—that instead instruct the user to copy and paste malicious code.

Malicious websites, such as those used in Drive-by Compromise, may present fake error messages or CAPTCHA prompts that instruct users to open a terminal or the Windows Run Dialog box and execute an arbitrary command. These commands may be obfuscated using encoding or other techniques to conceal malicious intent. Once executed, the adversary will typically be able to establish a foothold on the victim’s machine.(Citation: CloudSEK Lumma Stealer 2024)(Citation: Sekoia ClickFake 2025)(Citation: Reliaquest CAPTCHA 2024)(Citation: AhnLab LummaC2 2025)

Adversaries may also leverage phishing emails for this purpose. When a user attempts to open an attachment, they may be presented with a fake error and offered a malicious command to paste as a solution, consistent with the “ClickFix” strategy.(Citation: Proofpoint ClickFix 2024)(Citation: AhnLab Malicioys Copy Paste 2024)

Tricking a user into executing a command themselves may help to bypass email filtering, browser sandboxing, or other mitigations designed to protect users against malicious downloaded files.

T1204.005 – Malicious Library

Adversaries may rely on a user installing a malicious library to facilitate execution. Threat actors may Upload Malware to package managers such as NPM and PyPi, as well as to public code repositories such as GitHub. User may install libraries without realizing they are malicious, thus bypassing techniques that specifically achieve Initial Access. This can lead to the execution of malicious code, such as code that establishes persistence, steals data, or mines cryptocurrency.(Citation: Datadog Security Labs Malicious PyPi Packages 2024)(Citation: Fortinet Malicious NPM Packages 2023)

In some cases, threat actors may compromise and backdoor existing popular libraries (i.e., Compromise Software Dependencies and Development Tools). Alternatively, they may create entirely new packages and leverage behaviors such as typosquatting to encourage users to install them.

T1559 – Inter-Process Communication#

Adversaries may abuse inter-process communication (IPC) mechanisms for local code or command execution. IPC is typically used by processes to share data, communicate with each other, or synchronize execution. IPC is also commonly used to avoid situations such as deadlocks, which occurs when processes are stuck in a cyclic waiting pattern.

Adversaries may abuse IPC to execute arbitrary code or commands. IPC mechanisms may differ depending on OS, but typically exists in a form accessible through programming languages/libraries or native interfaces such as Windows Dynamic Data Exchange or Component Object Model. Linux environments support several different IPC mechanisms, two of which being sockets and pipes.(Citation: Linux IPC) Higher level execution mediums, such as those of Command and Scripting Interpreters, may also leverage underlying IPC mechanisms. Adversaries may also use Remote Services such as Distributed Component Object Model to facilitate remote IPC execution.(Citation: Fireeye Hunting COM June 2019)

T1559.001 – Component Object Model

Adversaries may use the Windows Component Object Model (COM) for local code execution. COM is an inter-process communication (IPC) component of the native Windows application programming interface (API) that enables interaction between software objects, or executable code that implements one or more interfaces.(Citation: Fireeye Hunting COM June 2019) Through COM, a client object can call methods of server objects, which are typically binary Dynamic Link Libraries (DLL) or executables (EXE).(Citation: Microsoft COM) Remote COM execution is facilitated by Remote Services such as Distributed Component Object Model (DCOM).(Citation: Fireeye Hunting COM June 2019)

Various COM interfaces are exposed that can be abused to invoke arbitrary execution via a variety of programming languages such as C, C++, Java, and Visual Basic.(Citation: Microsoft COM) Specific COM objects also exist to directly perform functions beyond code execution, such as creating a Scheduled Task/Job, fileless download/execution, and other adversary behaviors related to privilege escalation and persistence.(Citation: Fireeye Hunting COM June 2019)(Citation: ProjectZero File Write EoP Apr 2018)

T1559.002 – Dynamic Data Exchange

Adversaries may use Windows Dynamic Data Exchange (DDE) to execute arbitrary commands. DDE is a client-server protocol for one-time and/or continuous inter-process communication (IPC) between applications. Once a link is established, applications can autonomously exchange transactions consisting of strings, warm data links (notifications when a data item changes), hot data links (duplications of changes to a data item), and requests for command execution.

Object Linking and Embedding (OLE), or the ability to link data between documents, was originally implemented through DDE. Despite being superseded by Component Object Model, DDE may be enabled in Windows 10 and most of Microsoft Office 2016 via Registry keys.(Citation: BleepingComputer DDE Disabled in Word Dec 2017)(Citation: Microsoft ADV170021 Dec 2017)(Citation: Microsoft DDE Advisory Nov 2017)

Microsoft Office documents can be poisoned with DDE commands, directly or through embedded files, and used to deliver execution via Phishing campaigns or hosted Web content, avoiding the use of Visual Basic for Applications (VBA) macros.(Citation: SensePost PS DDE May 2016)(Citation: Kettle CSV DDE Aug 2014)(Citation: Enigma Reviving DDE Jan 2018)(Citation: SensePost MacroLess DDE Oct 2017) Similarly, adversaries may infect payloads to execute applications and/or commands on a victim device by way of embedding DDE formulas within a CSV file intended to be opened through a Windows spreadsheet program.(Citation: OWASP CSV Injection)(Citation: CSV Excel Macro Injection )

DDE could also be leveraged by an adversary operating on a compromised machine who does not have direct access to a Command and Scripting Interpreter. DDE execution can be invoked remotely via Remote Services such as Distributed Component Object Model (DCOM).(Citation: Fireeye Hunting COM June 2019)

T1559.003 – XPC Services

Adversaries can provide malicious content to an XPC service daemon for local code execution. macOS uses XPC services for basic inter-process communication between various processes, such as between the XPC Service daemon and third-party application privileged helper tools. Applications can send messages to the XPC Service daemon, which runs as root, using the low-level XPC Service <code>C API</code> or the high level <code>NSXPCConnection API</code> in order to handle tasks that require elevated privileges (such as network connections). Applications are responsible for providing the protocol definition which serves as a blueprint of the XPC services. Developers typically use XPC Services to provide applications stability and privilege separation between the application client and the daemon.(Citation: creatingXPCservices)(Citation: Designing Daemons Apple Dev)

Adversaries can abuse XPC services to execute malicious content. Requests for malicious execution can be passed through the application’s XPC Services handler.(Citation: CVMServer Vuln)(Citation: Learn XPC Exploitation) This may also include identifying and abusing improper XPC client validation and/or poor sanitization of input parameters to conduct Exploitation for Privilege Escalation.

T1569 – System Services#

Adversaries may abuse system services or daemons to execute commands or programs. Adversaries can execute malicious content by interacting with or creating services either locally or remotely. Many services are set to run at boot, which can aid in achieving persistence (Create or Modify System Process), but adversaries can also abuse services for one-time or temporary execution.

T1569.001 – Launchctl

Adversaries may abuse launchctl to execute commands or programs. Launchctl interfaces with launchd, the service management framework for macOS. Launchctl supports taking subcommands on the command-line, interactively, or even redirected from standard input.(Citation: Launchctl Man)

Adversaries use launchctl to execute commands and programs as Launch Agents or Launch Daemons. Common subcommands include: <code>launchctl load</code>,<code>launchctl unload</code>, and <code>launchctl start</code>. Adversaries can use scripts or manually run the commands <code>launchctl load -w “%s/Library/LaunchAgents/%s”</code> or <code>/bin/launchctl load</code> to execute Launch Agents or Launch Daemons.(Citation: Sofacy Komplex Trojan)(Citation: 20 macOS Common Tools and Techniques)

T1569.002 – Service Execution

Adversaries may abuse the Windows service control manager to execute malicious commands or payloads. The Windows service control manager (<code>services.exe</code>) is an interface to manage and manipulate services.(Citation: Microsoft Service Control Manager) The service control manager is accessible to users via GUI components as well as system utilities such as <code>sc.exe</code> and Net.

PsExec can also be used to execute commands or payloads via a temporary Windows service created through the service control manager API.(Citation: Russinovich Sysinternals) Tools such as PsExec and <code>sc.exe</code> can accept remote servers as arguments and may be used to conduct remote execution.

Adversaries may leverage these mechanisms to execute malicious content. This can be done by either executing a new or modified service. This technique is the execution used in conjunction with Windows Service during service persistence or privilege escalation.

T1569.003 – Systemctl

Adversaries may abuse systemctl to execute commands or programs. Systemctl is the primary interface for systemd, the Linux init system and service manager. Typically invoked from a shell, Systemctl can also be integrated into scripts or applications.

Adversaries may use systemctl to execute commands or programs as Systemd Services. Common subcommands include: systemctl start, systemctl stop, systemctl enable, systemctl disable, and systemctl status.(Citation: Red Hat Systemctl 2022)

T1574 – Hijack Execution Flow#

Adversaries may execute their own malicious payloads by hijacking the way operating systems run programs. Hijacking execution flow can be for the purposes of persistence, since this hijacked execution may reoccur over time. Adversaries may also use these mechanisms to elevate privileges or evade defenses, such as application control or other restrictions on execution.

There are many ways an adversary may hijack the flow of execution, including by manipulating how the operating system locates programs to be executed. How the operating system locates libraries to be used by a program can also be intercepted. Locations where the operating system looks for programs/resources, such as file directories and in the case of Windows the Registry, could also be poisoned to include malicious payloads.

T1574.001 – DLL

Adversaries may abuse dynamic-link library files (DLLs) in order to achieve persistence, escalate privileges, and evade defenses. DLLs are libraries that contain code and data that can be simultaneously utilized by multiple programs. While DLLs are not malicious by nature, they can be abused through mechanisms such as side-loading, hijacking search order, and phantom DLL hijacking.(Citation: unit 42)

Specific ways DLLs are abused by adversaries include:

DLL Sideloading#

Adversaries may execute their own malicious payloads by side-loading DLLs. Side-loading involves hijacking which DLL a program loads by planting and then invoking a legitimate application that executes their payload(s).

Side-loading positions both the victim application and malicious payload(s) alongside each other. Adversaries likely use side-loading as a means of masking actions they perform under a legitimate, trusted, and potentially elevated system or software process. Benign executables used to side-load payloads may not be flagged during delivery and/or execution. Adversary payloads may also be encrypted/packed or otherwise obfuscated until loaded into the memory of the trusted process.

Adversaries may also side-load other packages, such as BPLs (Borland Package Library).(Citation: kroll bpl)

Adversaries may chain DLL sideloading multiple times to fragment functionality hindering analysis. Adversaries using multiple DLL files can split the loader functions across different DLLs, with a main DLL loading the separated export functions. (Citation: Virus Bulletin) Spreading loader functions across multiple DLLs makes analysis harder, since all files must be collected to fully understand the malware’s behavior. Another method implements a “loader-for-a-loader”, where a malicious DLL’s sole role is to load a second DLL (or a chain of DLLs) that contain the real payload. (Citation: Sophos)

DLL Search Order Hijacking#

Adversaries may execute their own malicious payloads by hijacking the search order that Windows uses to load DLLs. This search order is a sequence of special and standard search locations that a program checks when loading a DLL. An adversary can plant a trojan DLL in a directory that will be prioritized by the DLL search order over the location of a legitimate library. This will cause Windows to load the malicious DLL when it is called for by the victim program.(Citation: unit 42)

DLL Redirection#

Adversaries may directly modify the search order via DLL redirection, which after being enabled (in the Registry or via the creation of a redirection file) may cause a program to load a DLL from a different location.(Citation: Microsoft redirection)(Citation: Microsoft - manifests/assembly)

Phantom DLL Hijacking#

Adversaries may leverage phantom DLL hijacking by targeting references to non-existent DLL files. They may be able to load their own malicious DLL by planting it with the correct name in the location of the missing module.(Citation: Hexacorn DLL Hijacking)(Citation: Hijack DLLs CrowdStrike)

DLL Substitution#

Adversaries may target existing, valid DLL files and substitute them with their own malicious DLLs, planting them with the same name and in the same location as the valid DLL file.(Citation: Wietze Beukema DLL Hijacking)

Programs that fall victim to DLL hijacking may appear to behave normally because malicious DLLs may be configured to also load the legitimate DLLs they were meant to replace, evading defenses.

Remote DLL hijacking can occur when a program sets its current directory to a remote location, such as a Web share, before loading a DLL.(Citation: dll pre load owasp)(Citation: microsoft remote preloading)

If a valid DLL is configured to run at a higher privilege level, then the adversary-controlled DLL that is loaded will also be executed at the higher level. In this case, the technique could be used for privilege escalation.

T1574.004 – Dylib Hijacking

Adversaries may execute their own payloads by placing a malicious dynamic library (dylib) with an expected name in a path a victim application searches at runtime. The dynamic loader will try to find the dylibs based on the sequential order of the search paths. Paths to dylibs may be prefixed with <code>@rpath</code>, which allows developers to use relative paths to specify an array of search paths used at runtime based on the location of the executable. Additionally, if weak linking is used, such as the <code>LC_LOAD_WEAK_DYLIB</code> function, an application will still execute even if an expected dylib is not present. Weak linking enables developers to run an application on multiple macOS versions as new APIs are added.

Adversaries may gain execution by inserting malicious dylibs with the name of the missing dylib in the identified path.(Citation: Wardle Dylib Hijack Vulnerable Apps)(Citation: Wardle Dylib Hijacking OSX 2015)(Citation: Github EmpireProject HijackScanner)(Citation: Github EmpireProject CreateHijacker Dylib) Dylibs are loaded into an application’s address space allowing the malicious dylib to inherit the application’s privilege level and resources. Based on the application, this could result in privilege escalation and uninhibited network access. This method may also evade detection from security products since the execution is masked under a legitimate process.(Citation: Writing Bad Malware for OSX)(Citation: wardle artofmalware volume1)(Citation: MalwareUnicorn macOS Dylib Injection MachO)

T1574.005 – Executable Installer File Permissions Weakness

Adversaries may execute their own malicious payloads by hijacking the binaries used by an installer. These processes may automatically execute specific binaries as part of their functionality or to perform other actions. If the permissions on the file system directory containing a target binary, or permissions on the binary itself, are improperly set, then the target binary may be overwritten with another binary using user-level permissions and executed by the original process. If the original process and thread are running under a higher permissions level, then the replaced binary will also execute under higher-level permissions, which could include SYSTEM.

Another variation of this technique can be performed by taking advantage of a weakness that is common in executable, self-extracting installers. During the installation process, it is common for installers to use a subdirectory within the <code>%TEMP%</code> directory to unpack binaries such as DLLs, EXEs, or other payloads. When installers create subdirectories and files they often do not set appropriate permissions to restrict write access, which allows for execution of untrusted code placed in the subdirectories or overwriting of binaries used in the installation process. This behavior is related to and may take advantage of DLL search order hijacking.

Adversaries may use this technique to replace legitimate binaries with malicious ones as a means of executing code at a higher permissions level. Some installers may also require elevated privileges that will result in privilege escalation when executing adversary controlled code. This behavior is related to Bypass User Account Control. Several examples of this weakness in existing common installers have been reported to software vendors.(Citation: mozilla_sec_adv_2012) (Citation: Executable Installers are Vulnerable) If the executing process is set to run at a specific time or during a certain event (e.g., system bootup) then this technique can also be used for persistence.

T1574.006 – Dynamic Linker Hijacking

Adversaries may execute their own malicious payloads by hijacking environment variables the dynamic linker uses to load shared libraries. During the execution preparation phase of a program, the dynamic linker loads specified absolute paths of shared libraries from various environment variables and files, such as <code>LD_PRELOAD</code> on Linux or <code>DYLD_INSERT_LIBRARIES</code> on macOS.(Citation: TheEvilBit DYLD_INSERT_LIBRARIES)(Citation: Timac DYLD_INSERT_LIBRARIES)(Citation: Gabilondo DYLD_INSERT_LIBRARIES Catalina Bypass) Libraries specified in environment variables are loaded first, taking precedence over system libraries with the same function name.(Citation: Man LD.SO)(Citation: TLDP Shared Libraries)(Citation: Apple Doco Archive Dynamic Libraries) Each platform’s linker uses an extensive list of environment variables at different points in execution. These variables are often used by developers to debug binaries without needing to recompile, deconflict mapped symbols, and implement custom functions in the original library.(Citation: Baeldung LD_PRELOAD)

Hijacking dynamic linker variables may grant access to the victim process’s memory, system/network resources, and possibly elevated privileges. On Linux, adversaries may set <code>LD_PRELOAD</code> to point to malicious libraries that match the name of legitimate libraries which are requested by a victim program, causing the operating system to load the adversary’s malicious code upon execution of the victim program. For example, adversaries have used LD_PRELOAD to inject a malicious library into every descendant process of the sshd daemon, resulting in execution under a legitimate process. When the executing sub-process calls the execve function, for example, the malicious library’s execve function is executed rather than the system function execve contained in the system library on disk. This allows adversaries to Hide Artifacts from detection, as hooking system functions such as execve and readdir enables malware to scrub its own artifacts from the results of commands such as ls, ldd, iptables, and dmesg.(Citation: ESET Ebury Oct 2017)(Citation: Intezer Symbiote 2022)(Citation: Elastic Security Labs Pumakit 2024)

Hijacking dynamic linker variables may grant access to the victim process’s memory, system/network resources, and possibly elevated privileges.

T1574.007 – Path Interception by PATH Environment Variable

Adversaries may execute their own malicious payloads by hijacking environment variables used to load libraries. The PATH environment variable contains a list of directories (User and System) that the OS searches sequentially through in search of the binary that was called from a script or the command line.

Adversaries can place a malicious program in an earlier entry in the list of directories stored in the PATH environment variable, resulting in the operating system executing the malicious binary rather than the legitimate binary when it searches sequentially through that PATH listing.

For example, on Windows if an adversary places a malicious program named “net.exe” in C:\example path, which by default precedes C:\Windows\system32\net.exe in the PATH environment variable, when “net” is executed from the command-line the C:\example path will be called instead of the system’s legitimate executable at C:\Windows\system32\net.exe. Some methods of executing a program rely on the PATH environment variable to determine the locations that are searched when the path for the program is not given, such as executing programs from a Command and Scripting Interpreter.(Citation: ExpressVPN PATH env Windows 2021)

Adversaries may also directly modify the $PATH variable specifying the directories to be searched. An adversary can modify the $PATH variable to point to a directory they have write access. When a program using the $PATH variable is called, the OS searches the specified directory and executes the malicious binary. On macOS, this can also be performed through modifying the $HOME variable. These variables can be modified using the command-line, launchctl, Unix Shell Configuration Modification, or modifying the /etc/paths.d folder contents.(Citation: uptycs Fake POC linux malware 2023)(Citation: nixCraft macOS PATH variables)(Citation: Elastic Rules macOS launchctl 2022)

T1574.008 – Path Interception by Search Order Hijacking

Adversaries may execute their own malicious payloads by hijacking the search order used to load other programs. Because some programs do not call other programs using the full path, adversaries may place their own file in the directory where the calling program is located, causing the operating system to launch their malicious software at the request of the calling program.

Search order hijacking occurs when an adversary abuses the order in which Windows searches for programs that are not given a path. Unlike DLL search order hijacking, the search order differs depending on the method that is used to execute the program. (Citation: Microsoft CreateProcess) (Citation: Windows NT Command Shell) (Citation: Microsoft WinExec) However, it is common for Windows to search in the directory of the initiating program before searching through the Windows system directory. An adversary who finds a program vulnerable to search order hijacking (i.e., a program that does not specify the path to an executable) may take advantage of this vulnerability by creating a program named after the improperly specified program and placing it within the initiating program’s directory.

For example, “example.exe” runs “cmd.exe” with the command-line argument <code>net user</code>. An adversary may place a program called “net.exe” within the same directory as example.exe, “net.exe” will be run instead of the Windows system utility net. In addition, if an adversary places a program called “net.com” in the same directory as “net.exe”, then <code>cmd.exe /C net user</code> will execute “net.com” instead of “net.exe” due to the order of executable extensions defined under PATHEXT. (Citation: Microsoft Environment Property)

Search order hijacking is also a common practice for hijacking DLL loads and is covered in DLL.

T1574.009 – Path Interception by Unquoted Path

Adversaries may execute their own malicious payloads by hijacking vulnerable file path references. Adversaries can take advantage of paths that lack surrounding quotations by placing an executable in a higher level directory within the path, so that Windows will choose the adversary’s executable to launch.

Service paths (Citation: Microsoft CurrentControlSet Services) and shortcut paths may also be vulnerable to path interception if the path has one or more spaces and is not surrounded by quotation marks (e.g., <code>C:\unsafe path with space\program.exe</code> vs. <code>“C:\safe path with space\program.exe”</code>). (Citation: Help eliminate unquoted path) (stored in Windows Registry keys) An adversary can place an executable in a higher level directory of the path, and Windows will resolve that executable instead of the intended executable. For example, if the path in a shortcut is <code>C:\program files\myapp.exe</code>, an adversary may create a program at <code>C:\program.exe</code> that will be run instead of the intended program. (Citation: Windows Unquoted Services) (Citation: Windows Privilege Escalation Guide)

This technique can be used for persistence if executables are called on a regular basis, as well as privilege escalation if intercepted executables are started by a higher privileged process.

T1574.010 – Services File Permissions Weakness

Adversaries may execute their own malicious payloads by hijacking the binaries used by services. Adversaries may use flaws in the permissions of Windows services to replace the binary that is executed upon service start. These service processes may automatically execute specific binaries as part of their functionality or to perform other actions. If the permissions on the file system directory containing a target binary, or permissions on the binary itself are improperly set, then the target binary may be overwritten with another binary using user-level permissions and executed by the original process. If the original process and thread are running under a higher permissions level, then the replaced binary will also execute under higher-level permissions, which could include SYSTEM.

Adversaries may use this technique to replace legitimate binaries with malicious ones as a means of executing code at a higher permissions level. If the executing process is set to run at a specific time or during a certain event (e.g., system bootup) then this technique can also be used for persistence.

T1574.011 – Services Registry Permissions Weakness

Adversaries may execute their own malicious payloads by hijacking the Registry entries used by services. Flaws in the permissions for Registry keys related to services can allow adversaries to redirect the originally specified executable to one they control, launching their own code when a service starts. Windows stores local service configuration information in the Registry under <code>HKLM\SYSTEM\CurrentControlSet\Services</code>. The information stored under a service’s Registry keys can be manipulated to modify a service’s execution parameters through tools such as the service controller, sc.exe, PowerShell, or Reg. Access to Registry keys is controlled through access control lists and user permissions. (Citation: Registry Key Security)(Citation: malware_hides_service)

If the permissions for users and groups are not properly set and allow access to the Registry keys for a service, adversaries may change the service’s binPath/ImagePath to point to a different executable under their control. When the service starts or is restarted, the adversary-controlled program will execute, allowing the adversary to establish persistence and/or privilege escalation to the account context the service is set to execute under (local/domain account, SYSTEM, LocalService, or NetworkService).

Adversaries may also alter other Registry keys in the service’s Registry tree. For example, the <code>FailureCommand</code> key may be changed so that the service is executed in an elevated context anytime the service fails or is intentionally corrupted.(Citation: Kansa Service related collectors)(Citation: Tweet Registry Perms Weakness)

The <code>Performance</code> key contains the name of a driver service’s performance DLL and the names of several exported functions in the DLL.(Citation: microsoft_services_registry_tree) If the <code>Performance</code> key is not already present and if an adversary-controlled user has the <code>Create Subkey</code> permission, adversaries may create the <code>Performance</code> key in the service’s Registry tree to point to a malicious DLL.(Citation: insecure_reg_perms)

Adversaries may also add the <code>Parameters</code> key, which can reference malicious drivers file paths. This technique has been identified to be a method of abuse by configuring DLL file paths within the <code>Parameters</code> key of a given services registry configuration. By placing and configuring the <code>Parameters</code> key to reference a malicious DLL, adversaries can ensure that their code is loaded persistently whenever the associated service or library is invoked.

For example, the registry path(Citation: MDSec) <code>HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\WinSock2\Parameters</code>(Citation: hexacorn)(Citation: gendigital) contains the <code>AutodiaDLL</code> value, which specifies the DLL to be loaded for autodial funcitionality. An adversary could set the <code>AutodiaDLL</code> to point to a hijacked or malicious DLL:

<code>“AutodialDLL”=“c:\temp\foo.dll”</code>

This ensures persistence, as it causes the DLL (in this case, foo.dll) to be loaded each time the Winsock 2 library is invoked.

T1574.012 – COR_PROFILER

Adversaries may leverage the COR_PROFILER environment variable to hijack the execution flow of programs that load the .NET CLR. The COR_PROFILER is a .NET Framework feature which allows developers to specify an unmanaged (or external of .NET) profiling DLL to be loaded into each .NET process that loads the Common Language Runtime (CLR). These profilers are designed to monitor, troubleshoot, and debug managed code executed by the .NET CLR.(Citation: Microsoft Profiling Mar 2017)(Citation: Microsoft COR_PROFILER Feb 2013)

The COR_PROFILER environment variable can be set at various scopes (system, user, or process) resulting in different levels of influence. System and user-wide environment variable scopes are specified in the Registry, where a Component Object Model (COM) object can be registered as a profiler DLL. A process scope COR_PROFILER can also be created in-memory without modifying the Registry. Starting with .NET Framework 4, the profiling DLL does not need to be registered as long as the location of the DLL is specified in the COR_PROFILER_PATH environment variable.(Citation: Microsoft COR_PROFILER Feb 2013)

Adversaries may abuse COR_PROFILER to establish persistence that executes a malicious DLL in the context of all .NET processes every time the CLR is invoked. The COR_PROFILER can also be used to elevate privileges (ex: Bypass User Account Control) if the victim .NET process executes at a higher permission level, as well as to hook and impair defenses provided by .NET processes.(Citation: RedCanary Mockingbird May 2020)(Citation: Red Canary COR_PROFILER May 2020)(Citation: Almond COR_PROFILER Apr 2019)(Citation: GitHub OmerYa Invisi-Shell)(Citation: subTee .NET Profilers May 2017)

T1574.013 – KernelCallbackTable

Adversaries may abuse the <code>KernelCallbackTable</code> of a process to hijack its execution flow in order to run their own payloads.(Citation: Lazarus APT January 2022)(Citation: FinFisher exposed ) The <code>KernelCallbackTable</code> can be found in the Process Environment Block (PEB) and is initialized to an array of graphic functions available to a GUI process once <code>user32.dll</code> is loaded.(Citation: Windows Process Injection KernelCallbackTable)

An adversary may hijack the execution flow of a process using the <code>KernelCallbackTable</code> by replacing an original callback function with a malicious payload. Modifying callback functions can be achieved in various ways involving related behaviors such as Reflective Code Loading or Process Injection into another process.

A pointer to the memory address of the <code>KernelCallbackTable</code> can be obtained by locating the PEB (ex: via a call to the <code>NtQueryInformationProcess()</code> Native API function).(Citation: NtQueryInformationProcess) Once the pointer is located, the <code>KernelCallbackTable</code> can be duplicated, and a function in the table (e.g., <code>fnCOPYDATA</code>) set to the address of a malicious payload (ex: via <code>WriteProcessMemory()</code>). The PEB is then updated with the new address of the table. Once the tampered function is invoked, the malicious payload will be triggered.(Citation: Lazarus APT January 2022)

The tampered function is typically invoked using a Windows message. After the process is hijacked and malicious code is executed, the <code>KernelCallbackTable</code> may also be restored to its original state by the rest of the malicious payload.(Citation: Lazarus APT January 2022) Use of the <code>KernelCallbackTable</code> to hijack execution flow may evade detection from security products since the execution can be masked under a legitimate process.

T1574.014 – AppDomainManager

Adversaries may execute their own malicious payloads by hijacking how the .NET AppDomainManager loads assemblies. The .NET framework uses the AppDomainManager class to create and manage one or more isolated runtime environments (called application domains) inside a process to host the execution of .NET applications. Assemblies (.exe or .dll binaries compiled to run as .NET code) may be loaded into an application domain as executable code.(Citation: Microsoft App Domains)

Known as “AppDomainManager injection,” adversaries may execute arbitrary code by hijacking how .NET applications load assemblies. For example, malware may create a custom application domain inside a target process to load and execute an arbitrary assembly. Alternatively, configuration files (.config) or process environment variables that define .NET runtime settings may be tampered with to instruct otherwise benign .NET applications to load a malicious assembly (identified by name) into the target process.(Citation: PenTestLabs AppDomainManagerInject)(Citation: PwC Yellow Liderc)(Citation: Rapid7 AppDomain Manager Injection)

T1609 – Container Administration Command#

Adversaries may abuse a container administration service to execute commands within a container. A container administration service such as the Docker daemon, the Kubernetes API server, or the kubelet may allow remote management of containers within an environment.(Citation: Docker Daemon CLI)(Citation: Kubernetes API)(Citation: Kubernetes Kubelet)

In Docker, adversaries may specify an entrypoint during container deployment that executes a script or command, or they may use a command such as <code>docker exec</code> to execute a command within a running container.(Citation: Docker Entrypoint)(Citation: Docker Exec) In Kubernetes, if an adversary has sufficient permissions, they may gain remote execution in a container in the cluster via interaction with the Kubernetes API server, the kubelet, or by running a command such as <code>kubectl exec</code>.(Citation: Kubectl Exec Get Shell)

T1610 – Deploy Container#

Adversaries may deploy a container into an environment to facilitate execution or evade defenses. In some cases, adversaries may deploy a new container to execute processes associated with a particular image or deployment, such as processes that execute or download malware. In others, an adversary may deploy a new container configured without network rules, user limitations, etc. to bypass existing defenses within the environment. In Kubernetes environments, an adversary may attempt to deploy a privileged or vulnerable container into a specific node in order to Escape to Host and access other containers running on the node. (Citation: AppSecco Kubernetes Namespace Breakout 2020)

Containers can be deployed by various means, such as via Docker’s <code>create</code> and <code>start</code> APIs or via a web application such as the Kubernetes dashboard or Kubeflow. (Citation: Docker Container)(Citation: Kubernetes Dashboard)(Citation: Kubeflow Pipelines) In Kubernetes environments, containers may be deployed through workloads such as ReplicaSets or DaemonSets, which can allow containers to be deployed across multiple nodes.(Citation: Kubernetes Workload Management) Adversaries may deploy containers based on retrieved or built malicious images or from benign images that download and execute malicious payloads at runtime.(Citation: Aqua Build Images on Hosts)

T1648 – Serverless Execution#

Adversaries may abuse serverless computing, integration, and automation services to execute arbitrary code in cloud environments. Many cloud providers offer a variety of serverless resources, including compute engines, application integration services, and web servers.

Adversaries may abuse these resources in various ways as a means of executing arbitrary commands. For example, adversaries may use serverless functions to execute malicious code, such as crypto-mining malware (i.e. Resource Hijacking).(Citation: Cado Security Denonia) Adversaries may also create functions that enable further compromise of the cloud environment. For example, an adversary may use the IAM:PassRole permission in AWS or the iam.serviceAccounts.actAs permission in Google Cloud to add Additional Cloud Roles to a serverless cloud function, which may then be able to perform actions the original user cannot.(Citation: Rhino Security Labs AWS Privilege Escalation)(Citation: Rhingo Security Labs GCP Privilege Escalation)

Serverless functions can also be invoked in response to cloud events (i.e. Event Triggered Execution), potentially enabling persistent execution over time. For example, in AWS environments, an adversary may create a Lambda function that automatically adds Additional Cloud Credentials to a user and a corresponding CloudWatch events rule that invokes that function whenever a new user is created.(Citation: Backdooring an AWS account) This is also possible in many cloud-based office application suites. For example, in Microsoft 365 environments, an adversary may create a Power Automate workflow that forwards all emails a user receives or creates anonymous sharing links whenever a user is granted access to a document in SharePoint.(Citation: Varonis Power Automate Data Exfiltration)(Citation: Microsoft DART Case Report 001) In Google Workspace environments, they may instead create an Apps Script that exfiltrates a user’s data when they open a file.(Citation: Cloud Hack Tricks GWS Apps Script)(Citation: OWN-CERT Google App Script 2024)

T1651 – Cloud Administration Command#

Adversaries may abuse cloud management services to execute commands within virtual machines. Resources such as AWS Systems Manager, Azure RunCommand, and Runbooks allow users to remotely run scripts in virtual machines by leveraging installed virtual machine agents. (Citation: AWS Systems Manager Run Command)(Citation: Microsoft Run Command)

If an adversary gains administrative access to a cloud environment, they may be able to abuse cloud management services to execute commands in the environment’s virtual machines. Additionally, an adversary that compromises a service provider or delegated administrator account may similarly be able to leverage a Trusted Relationship to execute commands in connected virtual machines.(Citation: MSTIC Nobelium Oct 2021)

T1674 – Input Injection#

Adversaries may simulate keystrokes on a victim’s computer by various means to perform any type of action on behalf of the user, such as launching the command interpreter using keyboard shortcuts, typing an inline script to be executed, or interacting directly with a GUI-based application. These actions can be preprogrammed into adversary tooling or executed through physical devices such as Human Interface Devices (HIDs).

For example, adversaries have used tooling that monitors the Windows message loop to detect when a user visits bank-specific URLs. If detected, the tool then simulates keystrokes to open the developer console or select the address bar, pastes malicious JavaScript from the clipboard, and executes it - enabling manipulation of content within the browser, such as replacing bank account numbers during transactions.(Citation: BleepingComputer BackSwap)(Citation: welivesecurity BackSwap)

Adversaries have also used malicious USB devices to emulate keystrokes that launch PowerShell, leading to the download and execution of malware from adversary-controlled servers.(Citation: BleepingComputer USB)

T1675 – ESXi Administration Command#

Adversaries may abuse ESXi administration services to execute commands on guest machines hosted within an ESXi virtual environment. Persistent background services on ESXi-hosted VMs, such as the VMware Tools Daemon Service, allow for remote management from the ESXi server. The tools daemon service runs as vmtoolsd.exe on Windows guest operating systems, vmware-tools-daemon on macOS, and vmtoolsd on Linux.(Citation: Broadcom VMware Tools Services)

Adversaries may leverage a variety of tools to execute commands on ESXi-hosted VMs – for example, by using the vSphere Web Services SDK to programmatically execute commands and scripts via APIs such as StartProgramInGuest, ListProcessesInGuest, ListFileInGuest, and InitiateFileTransferFromGuest.(Citation: Google Cloud Threat Intelligence VMWare ESXi Zero-Day 2023)(Citation: Broadcom Running Guest OS Operations) This may enable follow-on behaviors on the guest VMs, such as File and Directory Discovery, Data from Local System, or OS Credential Dumping.

T1677 – Poisoned Pipeline Execution#

Adversaries may manipulate continuous integration / continuous development (CI/CD) processes by injecting malicious code into the build process. There are several mechanisms for poisoning pipelines:

• In a <b>Direct Pipeline Execution</b> scenario, the threat actor directly modifies the CI configuration file (e.g., gitlab-ci.yml in GitLab). They may include a command to exfiltrate credentials leveraged in the build process to a remote server, or to export them as a workflow artifact.(Citation: Unit 42 Palo Alto GitHub Actions Supply Chain Attack 2025)(Citation: OWASP CICD-SEC-4)
• In an <b>Indirect Pipeline Execution</b> scenario, the threat actor injects malicious code into files referenced by the CI configuration file. These may include makefiles, scripts, unit tests, and linters.(Citation: OWASP CICD-SEC-4)
• In a <b>Public Pipeline Execution</b> scenario, the threat actor does not have direct access to the repository but instead creates a malicious pull request from a fork that triggers a part of the CI/CD pipeline. For example, in GitHub Actions, the pull_request_target trigger allows workflows running from forked repositories to access secrets. If this trigger is combined with an explicit pull request checkout and a location for a threat actor to insert malicious code (e.g., an npm build command), a threat actor may be able to leak pipeline credentials.(Citation: Unit 42 Palo Alto GitHub Actions Supply Chain Attack 2025)(Citation: GitHub Security Lab GitHub Actions Security 2021) Similarly, threat actors may craft pull requests with malicious inputs (such as branch names) if the build pipeline treats those inputs as trusted.(Citation: Wiz Ultralytics AI Library Hijack 2024)(Citation: Synactiv Hijacking GitHub Runners)(Citation: GitHub Security Labs GitHub Actions Security Part 2 2021) Finally, if a pipeline leverages a self-hosted runner, a threat actor may be able to execute arbitrary code on a host inside the organization’s network.(Citation: John Stawinski PyTorch Supply Chain Attack 2024)

By poisoning CI/CD pipelines, threat actors may be able to gain access to credentials, laterally move to additional hosts, or input malicious components to be shipped further down the pipeline (i.e., Supply Chain Compromise).

© The MITRE Corporation, MITRE ATT&CK® 19.2. This content is reproduced under the MITRE ATT&CK Terms of Use with attribution. attack.mitre.org. Source.