Evaluating Binary Atlas Against Real-World Malware (XWorm RAT)

Why This Test Was Performed

The main goals were to optimize Binary Atlas using real-world malware instead of synthetic or simplified samples. Rather than building detections only from documentation or known signatures, the aims were to:

Sample Information

Field Value
SHA256 44285aa905260b91338398c20807f4fbc0d0cd191d23553fa7f060f15ff4718c
File Type Windows PE
Runtime .NET Framework / VB.NET
Suspected Family XWorm RAT
Analysis Type Static-only

External Validation & Threat Intelligence

Before analyzing the sample with Binary Atlas, the file was checked using industry malware intelligence platforms and public threat reports.

The sample was flagged as malicious and associated with:

  • Anti-debugging behavior
  • Obfuscation
  • WMI usage
  • RAT behavior

Community reports and threat-analysis platforms identified the malware family and characteristics as:

  • XWorm
  • RAT / Backdoor behavior
  • Windows PE executable
  • VB.NET / .NET Framework based malware

These external signals provided a baseline to compare Binary Atlas findings against real-world detection systems.

Binary Atlas Analysis Results

Privilege & Execution Context

Binary Atlas identified:

Property Value
Privilege Context USER
Execution Level AS_INVOKER
Subsystem WINDOWS_GUI

This indicates the executable runs with the same privileges as the current user and does not request elevated administrator privileges by default.

PE Structure & Import Analysis

The engine successfully extracted PE headers, section information, and import data.

One of the most interesting findings: the binary imports only mscoree.dll -> _CorExeMain.

Why this matters:

  • Confirms the executable is a managed .NET application
  • Demonstrates how .NET malware can hide functionality behind the CLR runtime

Limitations of traditional import analysis in this case:

  • Malicious functionality may not appear in native imports
  • APIs can be resolved dynamically
  • The malware can rely heavily on managed code, reflection, or runtime loading

At the moment, the engine still relies heavily on PE-level inspection and needs deeper .NET-specific analysis capabilities.

Security Mitigations

Binary Atlas detected:

Feature Status
ASLR Enabled
DEP / NX Enabled
CFG Disabled
TLS Callbacks None

The binary enables common mitigations (ASLR, DEP) while Control Flow Guard (CFG) is not enabled.

Static Indicators & C2 Detection

Static analysis makes detecting real C2 infrastructure difficult compared to dynamic systems, but Binary Atlas identified suspicious hardcoded indicators including:

Potential C2 / Indicators:
  • https://api.telegram.org/bot
  • http://ip-api.com/line/?fields=hosting

These suggest remote communication capability, possible Telegram-based C2 usage, and system/network profiling behavior.

Other detected artifacts:

  • Registry keys
  • User-agent strings
  • PowerShell references
  • Socket-related keywords
  • Command execution indicators

Domain Extraction Problems

Binary Atlas currently produces many false-positive domains. For example, internal .NET namespaces such as System.Core and System.Net were incorrectly treated as hardcoded domains.

Needed improvements:

  • Better filtering
  • Improved context awareness
  • More intelligent parsing

Suspicious Keyword Detection

Detected suspicious keywords (examples):
  • download
  • cmd.exe
  • connect
  • powershell
  • remote
  • c2
  • socket
  • telegram
  • exec
  • /c

Weakness: the engine mostly detects words individually rather than reconstructing full commands or execution chains. For example, detecting powershell alone is weaker than detecting a full PowerShell execution chain.

Future improvements should focus on:

  • Full command reconstruction
  • Context-aware string extraction
  • Better behavioral correlation

Signature & Runtime Information

Binary Atlas identified the runtime as .NET Framework and reported an unsigned or invalid digital signature.

Currently the engine still performs full analysis even if a binary contains a valid signature because signature-validation logic is under development.

Import Anomaly Weakness

Import anomaly scoring performed poorly for this managed .NET sample. The engine reported:

Import anomaly score: 0/100

This is misleading because managed .NET malware often minimizes native imports and resolves APIs dynamically.

This sample highlighted the need for:

  • .NET metadata analysis
  • Reflection detection
  • P/Invoke analysis
  • Managed-code inspection

DLL Hijacking Detection

Binary Atlas detected several DLL-related references (examples):

  • kernel32.dll
  • SbieDll.dll
  • NTDll.dll
  • SHCore.dll
  • user32.dll
  • avicap32.dll
  • mscoree.dll

It also identified registry paths related to autoruns, Explorer settings, and class registrations.

The engine classified DLL hijacking severity as HIGH, which may be an overestimate since some DLL references are normal for Windows applications.

Anti-Analysis Detection

The engine detected anti-analysis indicators related to virtualization:

  • VMware indicators
  • VirtualBox indicators

This suggests the malware attempts to detect virtualized or sandboxed environments, matching external threat intelligence.

Persistence Mechanism Detection

Detected persistence-related artifacts and techniques:

  • Scheduled Task detection (schtasks)
  • Startup folder references (get_StartupPath)
  • WMI persistence indicators (Win32_Process, WMI)

The engine classified the sample as persistent malware due to multiple persistence techniques being present.

Overall Threat Classification

Category Result
Threat Level CRITICAL
Detection Confidence 75%
Total Findings 8
Malware Complexity ADVANCED

Binary Atlas Findings

Detection Severity
Ransomware_Generic CRITICAL
Rootkit_Indicators HIGH
Anti-Analysis Evasion HIGH
Command Execution MEDIUM
Persistence Mechanisms MEDIUM
Polymorphic Signature MEDIUM
Hooks and Detours MEDIUM
String Encryption MEDIUM

Current Problems & Limitations

This test revealed several important weaknesses in Binary Atlas including:

  • False positives in ransomware detection
  • Overly broad rootkit indicators
  • Weak .NET import analysis
  • Noisy domain extraction
  • Incomplete command reconstruction
  • Overly aggressive DLL hijacking detection

What the engine needs:

  • More rules
  • Better heuristics
  • Improved context analysis
  • Deeper .NET inspection
  • Smarter correlation logic

Final Thoughts

Binary Atlas is making progress in static malware triage and behavioral inference. The engine successfully identified anti-analysis behavior, persistence indicators, suspicious communication patterns, command execution references, and several malware-related heuristics from a real-world sample using static analysis alone.

At the same time, this report documents important gaps and shows the project is not production-ready. The goal is to:

  • Document real-world testing
  • Openly show weaknesses and detection gaps
  • Encourage contributions from malware researchers, reverse engineers, and detection engineers