Test whether your malware sandbox would be detected by adversarial samples before deploying it.
Verify that your debugging environment is invisible to anti-debugging techniques used by real malware.
Check if a virtual machine setup leaks hardware or registry identifiers that malware would notice.
Benchmark anti-malware tooling against a known set of real-world evasion techniques.
Pre-built binaries available from the releases page, no compilation needed to get started.
Al-khaser is a test program for security researchers and malware analysts. It simulates techniques that real malware uses to detect and avoid analysis environments, allowing people who build security tools to check whether those tools would be spotted by an adversary. The program runs a battery of checks against the system it is running on. These checks look for signs that a debugger is attached, that the machine is a virtual machine rather than real hardware, or that the program is running inside a sandboxed analysis environment. When real malware detects any of these conditions it typically stops doing malicious things to avoid being studied. Al-khaser lets security tool builders verify that their setups would not trigger those checks. The checks cover several categories. Anti-debugging checks look for software and hardware breakpoints, monitoring processes, and other signals that a reverse engineer is watching the program run. Virtual machine detection looks for registry keys, driver files, and hardware identifiers left behind by VirtualBox, VMware, QEMU, Hyper-V, and several other platforms. Sandbox detection includes timing tests (checking whether the system is accelerating sleep calls, which some sandboxes do to speed up analysis), human interaction checks such as whether the mouse has moved or whether disk sizes look realistic, and detection of known analysis tools by name. The project is written in C++ and targets Windows. Pre-built binaries are available from the releases page. Running the program with no arguments exercises all checks, specific categories can be selected individually via command-line flags. The intended audience is people building or testing anti-malware software, sandbox environments, or malware analysis systems. It is described as a proof-of-concept with good intentions rather than an actual malicious tool, and contributions of newly observed real-world techniques are welcome.
← ayoubfaouzi on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.