Back to Intelligence

Kimsuky's KimJongRAT & Langflow Exploitation: OTX Pulse Analysis — Enterprise Detection Pack

SA
Security Arsenal Team
July 26, 2026
5 min read

Recent OTX pulses indicate a resurgence of credential-focused espionage activity, primarily driven by the North Korean APT group Kimsuky, alongside opportunistic exploitation of AI infrastructure.

Intelligence gathered reveals Kimsuky has evolved its infection chain, leveraging legitimate services like GitHub to distribute KimJongRAT and MeshAgent via shortened URLs in phishing emails targeting Japan and South Korea. Concurrently, threat actors are actively exploiting Langflow (CVE-2026-55255 and CVE-2026-33017) to achieve Remote Code Execution (RCE) and cross-tenant IDOR, facilitating botnet deployment and credential theft within technology sectors. The convergence of these campaigns highlights a shift toward abusing trusted developer platforms and AI pipelines to bypass traditional security controls and exfiltrate sensitive credentials.

Threat Actor / Malware Profile

Kimsuky (APT37)

Objective: Intelligence gathering, credential theft, and persistent access within target networks.

  • Distribution Methods: Spear-phishing with LNK files disguised as work documents; abuse of GitHub Releases to host malicious ZIP files; shortened URLs redirecting to malicious infrastructure.
  • Payloads & Behavior:
    • KimJongRAT: Combines infostealing capabilities (browser data, system info) with remote access. Communicates with C2 via HTTP/HTTPS, often using specific paths like /agent.ashx.
    • XenoRAT & AutoIt Backdoors: Used for sustained access, often utilizing MeshAgent for stealthy remote management.
  • Persistence: Established via Windows Task Scheduler or service registration.
  • C2 Communication: Traffic observed to domains utilizing dynamic DNS (e.g., .servequake.com, .corpsecs.com).

Langflow Exploit Chain (CVE-2026-55255 / CVE-2026-33017)

Objective: Unauthenticated access to AI pipelines and cloud environments for data theft.

  • Technique: Chaining a cross-tenant IDOR (CVE-2026-55255) with an unauthenticated RCE (CVE-2026-33017).
  • Impact: Allows attackers to manipulate AI workflows, deploy web shells (indicated by /slt path), and steal API keys or credentials stored within the environment.

IOC Analysis

The provided IOCs reflect a mix of C2 infrastructure and delivery mechanisms:

  • Domains/Hostnames: lutkdd.corpsecs.com, pxqtkc.corpsecs.com, googleoba.servequake.com, bohyeonsanvil.com, kumhosports.com. These should be blocklisted immediately as they serve as C2 nodes for KimJongRAT and related payloads.
  • File Hashes (SHA256/MD5): Multiple hashes for KimJongRAT droppers, XenoRAT, and Autoit backdoors. Operationalize these via EDR threat feeds and YARA scans for retrospective analysis.
  • URLs: http://googleoba.servequake.com:8443/agent.ashx and http://45.207.216.55:8084/slt. These are critical for proxy/SG detection rules to identify beaconing or exploit attempts.
  • CVEs: CVE-2026-33017, CVE-2026-55255. Use these to prioritize vulnerability scanning for Langflow instances.

Detection Engineering

Sigma Rules

YAML
title: Kimsuky KimJongRAT C2 Beacon Activity
id: b4a3e6c1-8a2f-4b2c-9e5d-1f6a7c8d9e0f
description: Detects potential KimJongRAT C2 communication to known suspicious infrastructure or specific URI patterns observed in recent campaigns.
status: experimental
date: 2026/07/26
author: Security Arsenal
references:
    - https://otx.alienvault.com/pulse/67927d723b8b2b859d3e
logsource:
    category: proxy
definition:
    condition: selection
fields:
    - dest_ip
    - dest_port
    - url
falsepositives:
    - Unknown
level: high
tags:
    - attack.command_and_control
    - attack.t1071.001
selection:
    cs-method|contains: 'GET'
    cs-uri-query|contains:
        - '/agent.ashx'
    dest-host|endswith:
        - '.servequake.com'
        - '.corpsecs.com'
---
title: Langflow RCE Exploitation Attempt
id: a1b2c3d4-e5f6-7890-1234-567890abcdef
description: Detects exploitation attempts against Langflow CVE-2026-33017 involving specific paths or non-standard ports associated with the botnet deployment.
status: experimental
date: 2026/07/26
author: Security Arsenal
references:
    - https://otx.alienvault.com/pulse/66928d723b8b2b859d3e
logsource:
    category: web
product: apache
definition:
    condition: selection
fields:
    - client_ip
    - url
falsepositives:
    - Legitimate administrative access
level: critical
tags:
    - attack.initial_access
    - attack.t1190
selection:
    c-uri|contains: '/slt'
    c-port: 8084
---
title: Kimsuky Persistence via Scheduled Task
id: e5f6g7h8-i9j0-1234-5678-90abcdef1234
description: Identifies suspicious scheduled task creation patterns associated with XenoRAT and AutoIt backdoors used by Kimsuky.
status: experimental
date: 2026/07/26
author: Security Arsenal
references:
    - https://otx.alienvault.com/pulse/65919d723b8b2b859d3e
logsource:
    category: process_creation
product: windows
definition:
    condition: selection
fields:
    - CommandLine
    - ParentImage
falsepositives:
    - Legitimate software updates
level: high
tags:
    - attack.persistence
    - attack.t1053.005
selection:
    Image|endswith:
        - '\schtasks.exe'
    CommandLine|contains:
        - '/create'
        - '/sc'
        - '/tn'
    CommandLine|contains:
        - 'AutoIt3'
        - 'PowerShell -EncodedCommand'

KQL (Microsoft Sentinel)

KQL — Microsoft Sentinel / Defender
// Hunt for KimJongRAT and Langflow related network activity
let IOCs = dynamic([
    "googleoba.servequake.com", "lutkdd.corpsecs.com", "pxqtkc.corpsecs.com",
    "bohyeonsanvil.com", "kumhosports.com", "45.207.216.55"
]);
let MaliciousHashes = dynamic([
    "9758e76b601798a30d903bf05052a53df80451e5c156548ce9da828f608b6470",
    "221a39856b37e3c682f62427f1e6b965b36a2405764689c914672770a01a1fa9"
]);
// Network Connections
DeviceNetworkEvents
| where RemoteUrl in (IOCs) or RemoteIP in (IOCs)
| project Timestamp, DeviceName, InitiatingProcessAccountName, RemoteUrl, RemoteIP, RemotePort
| order by Timestamp desc
| extend Alert = "C2 Communication Detected"
union (
    DeviceProcessEvents
    | where SHA256 in (MaliciousHashes) or MD5 in ("03e4bef86f3e3e6ea23eb6f017af0c98", "05c07339603994b36dcfefcce720d03d")
    | project Timestamp, DeviceName, FileName, ProcessCommandLine, SHA256, MD5
    | order by Timestamp desc
    | extend Alert = "Malware Execution Detected"
)

PowerShell Hunt Script

PowerShell
<#
.SYNOPSIS
    Hunt script for Kimsuky artifacts and scheduled tasks.
.DESCRIPTION
    Checks for scheduled tasks registered with AutoIt/PowerShell patterns and compares file hashes against the latest OTX pulse.
#>

$IndicatorHashes = @(
    "9758e76b601798a30d903bf05052a53df80451e5c156548ce9da828f608b6470",
    "221a39856b37e3c682f62427f1e6b965b36a2405764689c914672770a01a1fa9",
    "107b5aa3c4ef30b9b832e0a10b1efb1dcf433158bc6af8d890d66c0c9ed50d21"
)

Write-Host "[+] Checking for suspicious Scheduled Tasks..." -ForegroundColor Cyan

$SuspiciousTasks = Get-ScheduledTask | Where-Object { 
    $_.Actions.Execute -like "*powershell*" -or 
    $_.Actions.Execute -like "*autoit*" -or
    $_.Actions.Arguments -like "*/sc*" -and $_.Actions.Arguments -like "*/tn*"
}

foreach ($Task in $SuspiciousTasks) {
    Write-Host "[!] Suspicious Task Found: $($Task.TaskName)" -ForegroundColor Yellow
    Write-Host "    Action: $($Task.Actions.Execute) $($Task.Actions.Arguments)"
}

Write-Host "[+] Scanning common drop directories for malware hashes..." -ForegroundColor Cyan

$Paths = @("$env:TEMP", "$env:USERPROFILE\Downloads", "C:\ProgramData", "C:\Windows\Temp")

foreach ($Path in $Paths) {
    if (Test-Path $Path) {
        Get-ChildItem -Path $Path -Recurse -ErrorAction SilentlyContinue | Get-FileHash -ErrorAction SilentlyContinue | Where-Object { $IndicatorHashes -contains $_.Hash } | ForEach-Object {
            Write-Host "[!] MALICIOUS FILE DETECTED: $($_.Path)" -ForegroundColor Red
            Write-Host "    Hash: $($_.Hash)"
        }
    }
}

Write-Host "[+] Hunt complete." -ForegroundColor Green


# Response Priorities

*   **Immediate:** Block all domains listed in the IOC Analysis section at the perimeter firewall and proxy. Isolate endpoints exhibiting connections to `45.207.216.55` or `servequake.com`.
*   **24h:** Conduct a credential audit for users who may have interacted with the phishing emails (shortened URLs/GitHub links). Reset credentials for accounts accessed from endpoints identified as infected.
*   **1 week:** Patch all instances of Langflow to address CVE-2026-55255 and CVE-2026-33017. Review GitHub access logs and implement stricter policies for the use of GitHub Releases in internal software deployment pipelines.

Related Resources

Security Arsenal Incident Response Managed SOC & MDR Services AlertMonitor Threat Detection From The Dark Side Intel Hub

darkwebotx-pulsedarkweb-credentialskimsukykimjongratxenoratcredential-theftlangflow

Is your security operations ready?

Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.